[
{
    "_id": "699824906c61c5bef9ef7d7d",
    "projectId": "243532",
    "projectName": "Unlocking Indias AI Potential: Key Strategies for Growth and Innovation",
    "projectDescription": "Unlocking India's AI Potential: Key Strategies for Growth and Innovation",
    "projectCreatedDate": "2026-02-20T09:08:30.866Z",
    "projectUpdatedDate": "2026-02-20T09:08:30.866Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-20T12:58:42.761Z",
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "Content",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "Driving_India_s_AI_Future__Growth__Innovation__and_Impact(MP3_128K)[1].mp3",
            "col3": " trusted governance. So this conversation here and the couple of conversations we are going to have over the next one hour or so are aligned with the summit's goal of bridging the global AI divide. So AI drives economic growth, social empowerment and of course global leadership for India. This is not just a presentation, it is a call to action. I'm your host, Mridhu Bhandari.  senior anchor and consulting editor at Network 18 with brands like CNBC and Forbes India, and I'll be guiding you through this next 55-minute journey that we're on. To set the tone of this morning, we're going to begin with framing the execution pathway of AI adoption and scaling it up from an industry vantage point. Our leadership keynote theme today is architecting India's AI leadership, a blueprint for transformation.  To deliver this note, please join me in welcoming on stage Dr. Vivek Mohindra, Special Advisor to the Vice Chairman and COO of Dell Technologies Global. Dr. Mohindra, please join us here. Thank you, Mridhu, and thank you everyone for joining us for the unveiling of this important blueprint. As we have heard over the most of this week, India is at the cusp of very significant changes.  and progress on the back of AI with a very bold aspirations which are not only bold for India, but they're very bold when you put it in the context of global aspirations that lots of other countries have. Dell has been, had a presence in India for over 30 years. We have partnered very closely with several government.  agencies as well as companies and the broader ecosystem to bring the broader set of capabilities that we have, which are across the board covering server storage networking PCs. And we are the number one AI infrastructure provider to enterprises globally. So leveraging our global presence and leveraging our deep knowledge of India, we have put all of that thought into putting forth as Mizu described,  an AI blueprint, which is a practical guide for what we think not only the country, but also companies need to do to be able to take advantage of this particular opportunity. The growth in terms of compute expected on the back of AI in India is expected to be well greater than 10 exoplobs. And that is a significant amount of growth. And the AI workloads in India are growing at over 30%  compound annual growth rate over the next few years, which is extremely significant. So as we step back and look at what are the key elements of what a country and companies need to do, there really are three key elements. The first element is investments. And the investment really goes at the heart of the compute infrastructure that a country needs to put in place to ensure that everybody has access to that infrastructure, including  MSMAs who sometimes do not have the capacity to be able to put their infrastructure in place. Investment also includes energy infrastructure because without energy there is really no compute infrastructure you can put in place which can run on that. So those are some of the key areas of the invest pillar of that and there are several other areas that I will encourage you to read through our blueprint.  that you will see both from a policy perspective and practical perspective that we think needs to get done. The innovate side really comes down to areas like skilling, which I know when Minister Chaudhary joins us, we will get into that in quite some detail. But innovating around how the skilling occurs all the way from schools to colleges to workforce entering employment and employers themselves,  what role they play across the whole spectrum of mediums to deliver that skilling is a key part of the innovate pillar. And then the last one revolves all around governance aspects of that. And governance covers multiple areas. One of the key areas within governance is fundamentally the regulatory framework that needs to exist and that countries need to put in place.  The pace of change is so significant with AI and how rapidly the technologies are moving that one of the fundamental balances that countries need to strike vis-a-vis regulations is striking the balance between innovation and responsibility while anchoring it to responsibility. That is one of the key regulatory principles that needs to be in place.  And the regulations have to be agile because the technology is moving at such a fast pace that you cannot anchor the regulatory framework to yesterday's technologies. And at the heart of it, I hope what you will take away from our blueprint is realizing sovereign AI potential for any country, including India, is really about the public-private partnership. And it's really about marrying  the public resources with private innovation and that really is what the key to unlocking the full potential of AI and sovereign AI is in this country. So again I would encourage you to read through the blueprint and we look forward to your feedback and we look forward to partnering closely with Indian ecosystem to help India realize its aspirations with AI. Thank you very much. Thank you so much Dr. Mohindra. I'm going to request you to please  stay back on stage. I'd also like to invite Manish Gupta, President and Managing Director of Dell Technologies India to join us here. This is the big moment, ladies and gentlemen. We are ready for the unveiling of the Dell Technologies Blueprint to accelerate India's AI growth. Yes, that's a photo moment for everyone. Thank you. Thank you so much, gentlemen.  This blueprint advances India's vision of Viksit Bharat 2047, positioning AI as a foundational engine for productivity, modernized public services, opportunity expansion, and strategic autonomy. It centers on three pillars that we've been discussing. Invest, invest in sovereign scalable compute and data foundations. Innovate,  with collaboration and with a future-ready workforce and evolve. Evolve into a responsible, agile, security-first governance structure. So our next panel today will go inside this blueprint and India's AI future to unpack how to convert this ambition into nation-scale execution. And that's quite a mean feat for a country as diverse  and as huge as India. So let's welcome the panelists for all the tough questions this morning. Raj Gopal, AS, Managing Director and Chief Executive Officer of NextGen Cloud Technologies. Bhaskar Chakravarthy, Dean of Global Business, the Fletcher School of Law and Diplomacy at Tufts University. Please have a seat, sir. And once again on stage, Manish Gupta, President and Managing Director of Dell Technologies India. And I will be moderating this session for you. Welcome, gentlemen.  We are here this morning to really translate the invest, innovate, and evolve pillars into very actionable steps that we all can take together to grow India's AI ecosystem. So I'll begin with targeted questions to each of you. And of course, you are free to jump in to add thoughts to each other. It's a candid free flowing conversation. Mr. Raj Gopal, if I can start with you. Startups and MSMEs.  They are the engine of our economy. They are also the engine of our innovation, especially as far as AI is concerned, but access to very reliable, affordable AI compute and cloud at scale continues to be a barrier for many of the small and medium enterprises. Now, in your opinion, what are some of the policies, some of the infrastructure, some of the market interventions that we need today to really unlock this access at scale?  I think across many other countries that we have seen, India has got a much comprehensive approach to this. I mean, in terms of actually they started this India emission, which is across seven pillars. I actually don't see many startups actually using the facilities that are there in the sense that you could apply for GPU infrastructure and you would get it at a subsidized rate. And some of them even got 100% of the GPUs that they need.  So I think from India's side, we've got a little less number of GPUs compared to what we really need. Maybe we need about 200,000 GPUs now, and we have about 450,000 now. So we all need to really invest more and then deploy more. But most important thing that you should see is that there is a good ecosystem. There is a system policy available.  for MSMEs and startups to leverage this. So there are a lot of innovative AI solutions being built. Most importantly, I also see that government actually setting pace in terms of actually leveraging some of these. We ourselves, I mean, we did one job for government which is very, very unique.  Like we serve Election Commission of India, they came to us and said, can you deduplicate and look at all the photographs that we have? This is like 90 crore pictures, right? So humanly it was not possible to deduplicate. You can't check one photograph with 90 crore others. We did that in a matter of 51 hours, and then we responded to them as to whether they had applications and all that. That deserves some applause. That's a humongous task.  What I see in this country is that I don't think we will be pure play. This chatbot, I mean, the generative AI, the way it has been envisaged, I mean, will be the primary use case. I think we are going far beyond what AI can be applied in terms of actually improving the productivity of citizen services and also give use cases that these small and medium enterprises can actually use.  In terms of actually enabling more GPUs, I think we need a lot of money.  100th of what US is investing or even less. So for us to do more, I think we need to remove certain bottlenecks that are there. One of the things I believe that can be done is, I'm sure everybody's familiar, we all pay GST. But basically when we import servers, when we pay GST on it, and then when we deliver servers, I get that as an input. And we do the, I mean, we only pay the value added piece back to the government.  The government gets the GST either way and I get an input. So the way to, one thing that we could look at is whether we can waive off the GST upfront and just take the GST when the services are being delivered. What that would do is it will deliver, it will reduce my upfront infrastructure cost to buy about.  you know, you know, 18%. You know, I don't have to fund that upfront and then pay interest on it, you know, or raise equity and, you know, pay more expectation on that. I mean, these things could really help. I mean, there are some of these things that government should look at and actually facilitate this piece. The last point is that they've given tax holiday for delivering services to the world, but I think India has got a lot more to do within India.  And then just look at world market and I believe that we should get, Indian service providers should get the same benefits as the global providers would get when, you know, they host services in India. So maybe a GST waiver and some income tax benefits could be good.  Okay, GST waiver, income tax benefits, demands from the industry coming in. Professor Bhaskar, if I can come to you now. You've often argued that nations need to compete not just on technology, but on trust, on institutional strength, and of course,  very, very inclusive digital participation. And that's very critical for a country like India because we are a country of many countries. The rich poor divide is quite huge. There are a lot of bridges to gap here. There are a lot of gaps to bridge here. My apologies. Now, as India accelerates AI, what do you think are some of the biggest non-technical bottlenecks that we should be looking at addressing, which you believe could really limit this  momentum that we are on currently and if we need larger societal good, what are some of the non-technical barriers that we need to immediately resolve? Thank you for the question and thank you for the invitation to join this terrific panel.  The issue of what are the non-technical elements, it's great that you have included that question in this discussion because in all the excitement around the technical infrastructure, which is of course enormous that's happening right here in India, it's no secret that this is one of the biggest talent pools in the world, growing very, very fast.  years, one in three developers are going to be in India, the largest mobile data pool anywhere in the world, the third largest data pool, you know, anywhere in the world once you take mobile and everything else together, growth and compute, growth and energy, growth and workloads, all that is happening, you know, which is fantastic. Now, when you think about what are the other elements that drive demand?  But we have found, we study 125 countries and try to understand the role of technology in shaping lives and livelihoods across 135 countries. The single most important determinant of what keeps a country on trajectory in terms of both the momentum of growth and also the state of their digital evolution is the demand side. So when I think about the demand side, obviously the core infrastructure which has been talked about is  enormously important and is going to continue to be a major contributor. A second part which has been talked about a lot in the Indian context is the distribution infrastructure. With DPI and all the different platforms associated with it, we know that there's a very powerful distribution system. Now there's a third infrastructure which is the non-technical part.  And that is what I would call the trust infrastructure. Now when you think about trust, it's a bit of a slippery concept. It's really hard to define. Each one of us in our heads has an idea of what trust is. But if we force somebody to define it, we'll struggle. The best thing about trust is I know what trust is when it is not there, when it is missing. And then you have to ask the question from a human perspective, what really is trust? And how do I?  bake that into the policy systems, into the technical systems, into the marketing systems, into the narrative around the India ecosystem, which will then keep moving the demand going. And trust ultimately has to do with, do people have confidence? The people who are the grantors of trust, do they have confidence that this invisible  transaction that I'm engaging in, whether I'm putting my data into a system or whether I have entered my financial information and I'm expecting something on the other side, that this whole thing is going to be completed, and it's going to be completed in a way that is reliable, that is repeatable, and will not take advantage of me. So when you think about that whole trust ecosystem, India starts in a great place. Relative to where I come from,  The United States, India is a far more trusting country in terms of trust in digital systems overall and certainly in terms of AI. There's a tremendous level of enthusiasm in terms of embracing all things AI. And we've seen this right here, just the sheer numbers of people who've attended the conference. It shows a level of trust that is probably unmatched anywhere in the world. Now, this is a tremendous asset to start with. The challenge is that the institutional side of trust  is still in the development process in India. So if you think about data governance, privacy, security, we're making progress, but we need to be much further along. Other aspects of trust has to do with the fact that, say, the India AI mission, that is developed at the union level, at the center level. But the actual exercising of trust, the granting of trust happens at the district level.  The district varies depending on whether I'm in Telangana or I'm in Jharkhand. And at the union level, the principles that I've got in place need to be sensitive to how it's being experienced from the ground up. So there are many different facets of trust that we need to work on and put in place, including transparency, which AI is, you know, we are facing a challenge regarding transparency across the world. So this is an issue. And then, you know, an approach to having redress and grievance systems and then literacy.  You know, people need to be able to understand how to use this exciting technology and also protect themselves. I'll pause and have a conversation. Absolutely. Thank you for that wonderful perspective, Professor. Coming to you, Manish, now the Dell Technologies Blueprint really calls for tighter alignment between policymakers, between industry, academia, institutional capacity. How can frameworks like this one really ensure growth that is both  globally competitive for India, but also locally inclusive because there is a lot of regional growth. There is a lot of geography that needs to be taken into the fold of AI when we are talking about being globally competitive as well. Thank you for that question. Just before I go in there, I would just add or maybe speak on a couple of topics that Professor just spoke about.  While he spoke more from a non-technical standpoint, he also brought in a technical aspect to it around the entire governance. And it's not just non-technical in today's world. It's driven by data privacy and all of the things around. But equally on literacy, there's also explainability. That trust comes really inherently once you have got explainability and people are aware on what outcomes are coming. Is that explainable and do they understand that? So it's a very, very interesting world that we are in.  Now, back to the blueprint that we were talking about and Vivek articulated that beautifully well across the three pillars of invest in data center capacity, invest in energy infrastructure, invest in people which goes back into the innovate side of it because like we just discussed, we've possibly got the largest pool of engineers around AI and that's  that skilling on innovate part is what's going to differentiate us what's going to make it really real and practically doable within the industry and would differentiate us versus other nations equally make it make it make it more democratized and ubiquitous across the nation and lastly it's really about how do you continue to build in the guardrails how do you build the trust like we just discussed to ensure that there is the the ecosystem knows  that this entire process can be trusted and can be built upon. We've also got to remember the sustainability aspect of it, which is where, as you look at the blueprint, you will see us talk about the fact that energy efficiency, sustainability on data centers, and new architectural models are becoming super important, and that's something that NextGen under Raj Gopal's leadership has demonstrated in building highly sustainable  more energy efficient data centers that will allow us to use our energy resources in the best possible manner while democratizing the access of compute capacity, while democratizing the access of data center capacity to organizations and verticals of various sizes.  So that's going to be the critical pillars around which we really believe that there's practicality in adopting and differentiating ourselves on the AI arena as we go forward. Right. I'm going to pick up on that data center piece and come to you Raj Gopal. Now, given the scale that  India will really need for competing globally. What would it take to truly build sovereign cost-efficient AI infrastructure that's not just available to large enterprises, but is also very affordable for the long tail of innovators that we have in this country.  Yeah, if you see the data in the data center industry It's been pretty concentrated in Mumbai and little bit in Chennai and you know, the other markets didn't really take off as much as they should have So what we are trying to do is to have I mean our current plan is to put about 100 megawatts of data centers in about six states  And what I see is that going forward, I mean, this could be the model that can be built one where each state has got a capacity because these states itself have got so much of consumption that can happen primarily because if you start looking at applying AI into elevating the quality of education in India, which will be one of the first things that will get rolled out at scale in India.  and also the healthcare aspects of it and citizen services. These things can require lots of computing capacity. So we are working with few state governments to actually see if we can bring a total transformation, actually consolidate their applications, bring about a data lake and then apply intelligence to it and take it to masses.  The way I think it will evolve is that there will be many more regions where data centers would spring up. And then when they are distributed, they need to be interconnected. We have very good interconnect system.  and not just the telcos but you have things like you know we have railway networks we have power networks which can actually assist the good connectivity between them when these things come into play actually we can have a pretty distributed a good amount of compute in india that can actually serve this this aspect but you must be aware that you know this game is actually i mean if you see my context i mean i have four diamonds to work on  One, of course, is geopolitics. I mean, there is quite a lot that is happening. We need to ensure that we have access to the technologies that we want to bring to India so that India actually works on the best available infrastructure. It is slightly better now, but there were restrictions before.  The other aspect is the amount of money. I mean, so I think it requires multiple billion dollars of investments to do, and that should be facilitated. That should really come into the country. When we have the money piece sorted out and then we build the infrastructure,  A good thing that we can leverage is open source. I mean, when we leverage open source, we can actually combine the infrastructure and open source and bring down the cost of compute so much that it is actually palatable for the Indian citizens to use. Because it's not about serving this 2%, 3% of the population which pays the income tax. It's about serving the 90% others. The moment we succeed in doing this, I think the talent also  The good talent that we have, we have access to good talent in India, but we don't have good talent. The quantity is missing. I mean, so we have good people, but you know, you need many more good people. They're going all over the world. We want to bring them back.  These things, these people can come back when these money and infrastructure fall in place. Right. That would ensure that India is playing a role which is actually pretty balanced, leveraging the global technologies and leveraging local talent and actually sets a blueprint for all the other countries which don't have, which other countries which are, which may miss out this revolution, right, and become digital colonies of the top two countries that are investing heavily in AI.  So I personally believe we have a good blueprint. And the blueprint can be applied to multiple countries. And we are well on the path. And I would prefer a distributed development of data centers across the country so that we are closer to the users.  We serve the other 90% or 95% of the population that is not being served now. Absolutely. Well, Professor, coming to you next, studies on digital competitiveness have consistently shown that institutional capacity often determines whether technology adoption really translates into economic value or not.  in the Indian context, what do we really need to do to really strengthen institutional muscle here that ensures that AI drives very, very inclusive growth rather than, you know, deepening the tech divide because there are already many divides that we are battling with. Yeah, so when you talk about institutions, the institutions need to be fit for purpose. So I think it's helpful to  break down this thing called AI that we all know that it's not just one thing. It is a general purpose technology that's finding its way into many different uses. And when you think about the uses, we talked about small and medium businesses, of course, the backbone of the adoption of AI in the enterprise level. But if you think about some of the biggest use cases in India, agriculture, skilling,  healthcare. Those are the three biggest ones, just in terms of sheer numbers of people affected and the impact that you're going to have. It may not necessarily have the same revenue impact, but it has a much larger societal impact. We just think about agriculture. If 40% of the crops grown in India, primarily among smallholder farmers, if 40% of those crops are destroyed by pests, and if I can  fix that problem, get that 40% number down to 30% or 20%, suddenly I have improved farmer well-being and income by an enormous amount. So if we just think about that one transformation.  using essentially very simple AI. It doesn't have to involve complex databases. The compute is done at the edge. You are dealing with people who have intermittent access to the internet, and you can make it work. Now, in that context, the question is, what is the institutional capacity, the institutional infrastructure, the trust infrastructure I need? Basically, the farmer needs to start  from the point of view that I can use my phone as a tool for farming  which is something completely new. People have learned farming from earlier generations. How do I think about the phone as a tool for farming? Just crossing that chasm requires a couple of seasons of using the phone and then seeing the outcomes. The impact of pests has gone down and it has made a practical difference to my life. Now then you have to ask the question from a farmer's perspective, what's preventing them?  from doing that. And that's where the institution becomes relevant. That's where the explainability aspect also becomes relevant. For a farmer to understand how the system works, essentially, the picture I take of my crop needs to match up with an image database that is out there. And essentially, with a very simple interface, I can match the pest, and I can then end up with a solution to the problem. So the same thing for skill building.  literacy. If I can see my ability to speak, my ability to read in multiple languages, improve, you know, suddenly my trust goes up. So what is the minimum amount of institutional safeguards I need to provide that. Then I come to something like healthcare, when people have a much bigger chasm to cross.  That's where people have a lot of concerns about, should I be putting my information into the system? How is this going to be used? Can I trust the phone, where I've relied on a doctor, or I've relied on a wise person in my community? I've relied on my mother for maternal health care advice. So how do I cross that chasm, being able to provide the foundational trust elements?  is going to be important so the answer to your question very long answer is it depends it depends on the usage as is the case to a lot of questions about india  It depends, right? Well, Manish, you know, globally we are seeing nations tie AI strategy to strategic autonomy. Now, whether it comes to the semiconductor ecosystem or the supply chain, you know, the strategic autonomy is becoming extremely important for countries. For India, what are the two or three foundational capabilities that you think we need to build domestically in this decade to ensure that we are  true creators of AI value and not just consumers? Awesome. Great question, Ritu. And we as a nation have proven ourselves to be phenomenal adopters of technology. And the best example in my mind comes on UPI or digital payment. 10 years back, 11 years back, we were just not there. And today, we are by far the largest consumers or the number of digital payments and the value of digital payments within India is multiple times of the second  economy that does this right so that's a great example of how we have been able to localize democratize and proliferate the use of technology so within that I would really put on three hats here the first and just you know inverting the pyramids not starting with technology but starting people we've really got to think away from the users to the to the developers you know it's got to move from 1 billion users  to 1 billion or 1 million or 10 million developers. And that's the skill set, that's the IP that we are going to bring in because we've got that largest talent pool residing within the country. The second, and again, I heard you talk about semiconductor and supply chain. I think we've got to adopt the best that's available there globally. But equally, we've got to move, we've got to not just think about made in India, but talk about trusted in India.  and which is where work with organizations such as AISI Artificial Intelligence Safety Institute to ensure that we are putting the right guardrails, putting the right governance policy and the entire institutional framework to ensure that the AI that we are building here is trusted. And lastly, it goes back to the same thing. And maybe I'll use the same example. You know, we had the UPI of money. We need to have UPI of AI.  where we are building that at scale using the data sources that we have. We have the largest ones. And some of the initiatives that the government has taken, India AI Mission, but equally think about AI Cosh. There are more than 7,000 data sets that are now available to organizations of all sizes. Use that to ensure that we are developing for the country at the population scale through academia, through...  private sector, through startups, through MSMEs all coming together. And that really requires a consistent API layer that's bringing, theoretically, maybe even all of the data center and the compute capacity that we are creating as a part of AI mission to be one single layer that can be consumed by anybody and everybody across the nation to start to innovate on that, to start to develop on that. So going back, I know a long answer, but if I were to summarize three things there, UPI of money to UPI of AI. Made in India.  transitioning to trusted in India and you know from a developer users to from billion users to maybe a million or 10 million developers made in India but made for the world absolutely all right so I'm going to ask each one of you for  a few concise takeaways today. Now, the blueprint that we are talking about that Dell Technologies has just unveiled talks about agile trusted AI governance with sectoral baselines, test beds, strong institutional coordination. Yet, globally, what we've seen is that speed often beats caution.  And we are seeing some of the scary stuff coming out with AI as well. A lot of the experiments, a lot of agentic AI experiments that people are doing across the world, some of them call for caution. Now, in the Indian context, how do we stay globally competitive while also operationalizing very, very stringent safeguards? And where should we really draw the line between  the speed of innovation or innovation velocity and regulatory discipline. So, Braj Gopal, if we can start with you, please. If you take the birth of GNI, I mean, it actually, I don't think none of those rules were actually followed and, you know, it was built on every data that was available in whatever form.  Personally, you know, in most places, I'm actually trying to tell people just it's not about ignoring the risk factors of it, but I think the regulation should not.  curtail the innovation in the thinking that we should be restrictive about whatever we are working with. And so one of the most important things is I think we should have less regulation in this place because I think overall the benefit, you should look at AI like a utility. I mean, you will have more good with AI than with bad. Yes, there are things that can be handled as we go along.  If you see what we do in cloud today, we haven't been able to sort out our security and you know data protection postures even today. It's an evolving journey and I think we will continuously catch up with the bad factors around the AI adoption and that's a journey it cannot start or stop or it can be implemented at a point in time. So we should keep.  Looking at those aspects and keep putting with the regulations or technology interventions to ensure that we handle those pieces, but we should go fast forward with implementation and adoption of AI. And I see a lot of Indian enterprises really being reluctant in terms of adopting it, especially the larger ones. But if you see in India, I think government will set the pace and the startups and the MSMEs will actually catch on from there.  And the large enterprises will actually struggle to catch up with the amount of innovation that's happening in these two spaces. Where does that reluctance come from? What are the top three reasons large organizations are reluctant? One is, of course, the fact that it's not easy to adopt and transform a large organization. And perhaps startups and SMEs have the benefit of the agility and the small scale that they're at. What else?  So I think the first issue is not about security and all that. I mean, about these aspects. But most importantly, I think a lot of people are struggling to imagine where to apply AI. And I think the moment we can understand that, you will start seeing that the benefits are far outweigh the negative aspects of it. So I think that's the first thing that people should look at is to not just look at leveraging the  I mean, but actually really look at where you can deploy. I talked about that deduplication piece there are like we are working on more than 150 projects and not all of them are, you know, bot based. So that imagination is what is important and one that once that imagination comes you will out. I mean the benefits will outweigh the negative aspects of whatever we are seeing. All right, Professor final takeaway from you.  on, you know, speed versus caution. Yeah, so if you think about speed, I like to always like to use the analogy of a car and a road. You can think about the speed that you can build into the car, the velocity of the Ferrari and a lot of the conversations that have happened.  not necessarily in this room, but in other rooms, is about the Ferraris, whether I'm talking about agentic AI or AI optimized for certain applications and the technical aspects, you know, really, really important. Now, if you take the Ferrari and you bring it into the Indian context, maybe it's a Maruti or something else that I need to be talking about. But then the question is...  What's the road on which this Ferrari is going? If it's a dirt road full of potholes, even a Ferrari is not going to go very fast. So much of our conversation here is about that dirt road. And what are the things, what are the potholes that we need to fix? There's one elephant on the table that we did not address, and I'm just going to leave it at that, which is when we talk about trust, there's a whole bunch of things you can do from an institutional standpoint to build trust, transparency, explainability, and so on.  There's a huge issue that we need to think about, which is what is going to be the impact on jobs? What's going to be the impact on jobs? This is the youngest major country in the world. It's also one of the least employed country in the world. And now with AI coming in, is that going to help boost jobs or is it going to take jobs away? If we don't fix that problem, get ahead of that. All the trust we are talking about, all the institutions you build.  could come down. So part of the policy infrastructure here is to figure out what is the post-AI jobs picture. Absolutely. Manish, final word to you. So I honestly don't think that these are opposing forces. Agility versus security, and particularly in this side of technology, you cannot have them act as opposing. It's really about building the frameworks that are going to take both of them together.  This is fast evolving as a technology, but equally as institutions, we'll have to be faster than that in evolving. I think the government has done a phenomenal job in building some of the frameworks around that, and the institutions, AISI as one example, but on the privacy side, DPDP or DEPA, all of those acts being there, are good frameworks to start with, and I'll just index back on the question that you had asked Raj Gopal earlier.  on what is the hesitation from enterprises in adopting. I don't think it's necessarily about security. It's really about saying how many of those have real use cases. While the real use cases exist, how many of them are able to monetize or are able to see them scale from experimentation or pilots into production. I think that's a job that we as industry folks who understand the technology, who are innovating in this space, really need to bring to the table so that we can...  bring this to the floor across the nation and enterprises and organizations of all sizes and academia and public. I think that's where this will get practical. But equally, these are not opposing forces, they go hand in hand.  Well, thank you gentlemen for that absolutely incredible conversation. The takeaway is clear that investing, innovating and of course, innovating by expanding skills pipelines and accelerating AI deployment is going to be key to India's sovereign AI infrastructure and appreciate you joining us here and taking the time today. We are also very delighted to now be joined by Honorable Sri Jayan Chaudhary Ji, Minister of State.  for Education and Minister of Skill Development. Huge round of applause. We are going to have him up here shortly. Thank you, gentlemen. Thank you very much for joining us. So if we can have you up here for a quick photo op and we will then continue the conversation. Thank you. Thank you, everyone. Thank you. Thank you, gentlemen. Manish, if I can please request you to felicitate our speakers. Mr. Raj Gopal.  Let's have a huge round of applause for our panelists here today. Professor Bhaskar Chakraborty, thank you so much for joining us here today. If you all can just get off the stage for two minutes, we are getting it ready for our next conversation. Thank you so much. Well, ladies and gentlemen, time to move on. Now, if India's AI ambition is to translate into real economic growth,  It's obviously not going to be any one entity's job. It is not going to be driven only by the government or by the industry alone. It will be driven by partnership. India has the talent, the digital backbone and the momentum. The real question though is how do we scale AI responsibly, securely and inclusively?  So our next fireside chat conversation will explore what a powerful public-private model regarding AI could really look like. And for this, I'm delighted to welcome two very eminent leaders who are instrumental in shaping the journey, both from policy and industry perspectives. We have, of course, Honorable Sri Jayan Chaudharyji, Minister of State for Education and Minister of Skill Development and Entrepreneurship, Independent Charge, Government of India.  And we have Dr. Vivek Mohindra, Special Advisor to the Vice Chairman and COO Dell Technologies Global. If I can please have both of you up here for a quick conversation. Thank you so much. Thank you very much. Thank you, gentlemen. Well, it's quite clear that public-private partnership is going to be critical to AI scaling and adoption in India. You know, Chaudhary, if I can start with you, how can PPP models really accelerate  large-scale AI infrastructure. What have been some of the on-ground experiences you've seen so far? Of course, the government has been moving at breakneck speed when it comes to deploying more technology, giving a more Philip to innovation in India. How are you really ensuring trust, resilience, and long-term national competitiveness as AI becomes very mainstream with this event in India?  In the Indian context, as the audience is aware, we had a lot of catching up to do. And it's fair to say that a lot of what we are seeing around us in AI has been facilitated by creating an ecosystem in a short span of time. Perhaps we may enjoy a second mover advantage with regards to this technology. And that has come about only because of a strong top-down  emphasis and push. The only reason why this event is happening here in India is because the leadership at the top understood very quickly the value and the potential of this new technology that we should not view it as a disruption but view it as an opportunity to leapfrog legacy problems, deficit problems and provide access and equity to our citizens and dignity.  to our workers. And that's why the Prime Minister, you know, the last event in France, shared that leadership space. Every opportunity he gets, he talks about skilling, about young people, about the potential for AI and the enunciation in Manav that this technology needs to be human centric. I think that has given us a real emphasis and a push.  for the academia for our industry for our vibrant startups systems to really think about what they are doing in this space. I think that is the background to the event that we are all witnessing thousands and thousands of people casual visitors apart from those who are already entrenched in technology and this message that goes out is that 1 billion strong young people in a developing country.  are already thinking about what AI means to them and what they can do in this space not just be consumers, but also be producers and innovators and thinkers and creators now PPP in this domain for me and when you think about man of being human centric citizen centric the P that really matters is the people and in that context, it's important that you have the broad architecture, which is open.  This is something that India has stood for from the beginning when there was a lot of debate about what should be the policy that enables AI, but there was also a lot of fear around AI about trust factors, about privacy, data, sovereignty, multiple issues about the human interface, the augmented human worker, what this means for education, for the future of jobs.  A lot of those issues were being discussed and debated and India said that yes, it's good to have a strategy and out of that strategy and experiences will evolve a robust policy. It is essential to have guardrails, but it is a starting point currently. We don't want to infringe upon the possibility of innovation and India took that approach and we had open access to whatever compute, you know India AI mission.  was set up with a target of 18,000 GPUs and in a short span they were surpassed it. It's about 38,000 and a roadmap is by the end of this year. It's going to cross one lakh threefold now think about it. All of this compute for cell facility that has been created is a model of PPP. It has to be housed in educational institutions. So that real research can happen in our premier educational institutions. This is the great time when academia.  is more important than it perhaps ever was in the Indian context. In the Indian context, academia was partly separated from industry and the real economy in our minds. But now every Indian citizen is realizing the value of research and innovation. Every family is saying that no, this is important. We must value it and every educational institution is saying that we are not divorced from the market and needs of our community and society and nationhood building.  So that engagement with nationhood building and concept of technology is deeply immersed thanks to the efforts of India AI mission and here I'll just leave one data point to you. What is the cost of this compute facility? It's being provided to for startups for researchers at 65 rupees for an hour. You pay 300 rupees for a couple of hours PVR cinema ticket. So it's probably the world's cheapest compute facility, which is open.  We are celebrating Sarvam. Let's not focus on for-profit, not for-profit, because everything has to be for people. If you look at Sarvam, that's also in my mind a PPP. It's been incubated by IIT Madras. It has been supported by AI mission. So that's another example, because you're right. Government cannot invest in from data, to energy, to compute, to innovation. It really has to come from  our citizens, our researchers, our technologies. It's an effective mission. It is. It is. Absolutely. Well, Dr. Mohindra, getting your point of view from a Dell perspective. Now, if we look at PPP as far as job enablement is concerned, because that's the big concern that citizens have, that what's going to happen to our jobs. And of course, skilling is part of that journey. How can Dell partner with  ministry of skill development and entrepreneurship. What are you all doing from a future skill labs perspective? How are you accelerating AI apprenticeships? So that basically the jobs also move beyond the metros because tier two, tier three towns is where we are looking at a lot of talent sitting there, but we are looking at a lack of access of sorts when it comes to skilling.  Yeah, I think that's a great question to do and I think, honorable minister, good to see you again. It reminds me of our discussion we had back in last time we met in October of 2024. I know we missed each other and often and we covered very similar ground. I think at the heart of it, it does come down to and I commend India on the progress it's made in making access to all these GPUs available. It is an industry academia partnership working closely with the  the minister here. And our view is when you think about skilling, there are three different levels. You have to think about schooling, you have to think about college level, and you have to think about employment. People entering the workforce are employed today. And you think about the delivery of this through online, in-person, incubation. So those are the two big dimensions.  And from our perspective, we are very excited to partner on extending it to tier two, tier three, working closely with the minister and other institutions in India. And the core of it, having accessibility to this GPU at such an amazing price point, really unlocks the potential. And I think as Minister and I talked about last time we met, and I predicted the rise of agentic technologies that will lower the bar even further in terms of being able to provide the skilling.  So those are some of the ways that we have been working and we look forward to partnering. Right. Well, finally, to both of you now, as we embed AI into all our critical sectors, whether it's BFSI or telecom or agriculture, health care, education, now governance has to move from intent to very, very strong operational safeguards. What does zero trust AI architecture then practically look like at a national level? Minister, if we can start with you.  Well zero trust it's it's an interesting terminology What that for me the way I look at it? It means that you have to be able to verify each and every protocol in your design and In India, we generate a lot of data everything and Indian citizens are quite open about access globally privacy  has been a major concern and sometimes it becomes also an impediment for governance because those data points aren't being collected, analyzed, researched. In the Indian context, citizens are okay about sharing their data and I'm saying this with the knowledge that you know, crores of upper IDs have been formed, created using consent, but we have not received any blowback.  in a way from from students families saying that why are you because they understand that we are able to with technology customize and tailor experience in the classroom for every student where no student can get left behind. What that means for the employability for the knowledge acquisition for that student for the quality of the educational experience is immense. So I think but once you're collecting the data, there is a lot of  effort that needs to be put in so that that trust is maintained from zero trust to a hundred percent trust in in the public mind. So I think our data sets need to be segmented. There are protocols in within government of India in education. We're thinking of creating a complete AI stack, which means the anonymized data sets will be made available for researchers for the creating the value for the layers of innovation for enabling.  startups to engage with that data that government of India and the citizens have shared similarly in skill. We have skill India digital hub, which is also looking at creating those data sets which can then really help us unleash the next wave of innovation and requirement that we have in skilling. So I think once you have that system design in place, it can be achieved the Prime Minister spoke about a label for content that you know that  It should be verifiable and legal. This technology and consumer awareness is a big aspect of it on how we engage with these tools and how we understand what the outcome of our engagement. How true is it? How is it verifiable? Where are these AI models trained? Is there any bias in that data set? All that knowledge needs to be out there for the consumers. I feel also that there needs to be audit trail for our new AI models.  Maybe in the future, you could have CAG come out with the audit report of all the AI models. So it's a brave new future, but it's a balance. For partnership at scale, you need an architecture with trust. Absolutely. Final 30 seconds to you, Dr. Mohindra. I think the minister said it very eloquently. I would extend the notion to zero trust should extend to start with data.  go into AI models, the usability, the cybersecurity elements, and the identity and access management. Those would be the ways I would extend it. And practically, it means, beyond the governance framework, having things like a national risk registry, observability, being able to report whenever there is an infraction and auditability. But, Minister, you said it very eloquently, and I think our AI blueprint has more details that would be worth looking at.  Absolutely. Thank you very much. Thank you, Minister Chaudhary. Thank you, Dr. Mohindra. Well, clear takeaways from that conversation, public-private partnerships, future-ready workforce pipelines that reach much beyond the metros, and, of course, zero-trust guardrails that we all need for a very, very defined, AI-led future for India. Thank you very much for joining us here today. Appreciate you taking the time. Dr. Mohindra, if you can please...  . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  And you can, of course, download the Dell Technologies Blueprint by scanning the QR code right here. It's got a lot of interesting insights. So go ahead, scan that QR code, and take a look at the Blueprint. Thank you, gentlemen. Thank you, everyone, for being an immensely participative audience here today. I'm sorry we could not have any time for audience questions, but appreciate you joining us here today. Thank you very much.",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771578476314.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:27",
                    "text": " trusted governance. So this conversation here and the couple of conversations we are going to have over the next one hour or so are aligned with the summit's goal of bridging the global AI divide. So AI drives economic growth, social empowerment and of course global leadership for India. This is not just a presentation, it is a call to action. I'm your host, Mridhu Bhandari."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:27",
                    "end": "00:00:56",
                    "text": " senior anchor and consulting editor at Network 18 with brands like CNBC and Forbes India, and I'll be guiding you through this next 55-minute journey that we're on. To set the tone of this morning, we're going to begin with framing the execution pathway of AI adoption and scaling it up from an industry vantage point. Our leadership keynote theme today is architecting India's AI leadership, a blueprint for transformation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:56",
                    "end": "00:01:27",
                    "text": " To deliver this note, please join me in welcoming on stage Dr. Vivek Mohindra, Special Advisor to the Vice Chairman and COO of Dell Technologies Global. Dr. Mohindra, please join us here. Thank you, Mridhu, and thank you everyone for joining us for the unveiling of this important blueprint. As we have heard over the most of this week, India is at the cusp of very significant changes."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:27",
                    "end": "00:01:52",
                    "text": " and progress on the back of AI with a very bold aspirations which are not only bold for India, but they're very bold when you put it in the context of global aspirations that lots of other countries have. Dell has been, had a presence in India for over 30 years. We have partnered very closely with several government."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:52",
                    "end": "00:02:21",
                    "text": " agencies as well as companies and the broader ecosystem to bring the broader set of capabilities that we have, which are across the board covering server storage networking PCs. And we are the number one AI infrastructure provider to enterprises globally. So leveraging our global presence and leveraging our deep knowledge of India, we have put all of that thought into putting forth as Mizu described,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:21",
                    "end": "00:02:51",
                    "text": " an AI blueprint, which is a practical guide for what we think not only the country, but also companies need to do to be able to take advantage of this particular opportunity. The growth in terms of compute expected on the back of AI in India is expected to be well greater than 10 exoplobs. And that is a significant amount of growth. And the AI workloads in India are growing at over 30%"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:51",
                    "end": "00:03:20",
                    "text": " compound annual growth rate over the next few years, which is extremely significant. So as we step back and look at what are the key elements of what a country and companies need to do, there really are three key elements. The first element is investments. And the investment really goes at the heart of the compute infrastructure that a country needs to put in place to ensure that everybody has access to that infrastructure, including"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:20",
                    "end": "00:03:48",
                    "text": " MSMAs who sometimes do not have the capacity to be able to put their infrastructure in place. Investment also includes energy infrastructure because without energy there is really no compute infrastructure you can put in place which can run on that. So those are some of the key areas of the invest pillar of that and there are several other areas that I will encourage you to read through our blueprint."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:48",
                    "end": "00:04:17",
                    "text": " that you will see both from a policy perspective and practical perspective that we think needs to get done. The innovate side really comes down to areas like skilling, which I know when Minister Chaudhary joins us, we will get into that in quite some detail. But innovating around how the skilling occurs all the way from schools to colleges to workforce entering employment and employers themselves,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:17",
                    "end": "00:04:44",
                    "text": " what role they play across the whole spectrum of mediums to deliver that skilling is a key part of the innovate pillar. And then the last one revolves all around governance aspects of that. And governance covers multiple areas. One of the key areas within governance is fundamentally the regulatory framework that needs to exist and that countries need to put in place."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:44",
                    "end": "00:05:10",
                    "text": " The pace of change is so significant with AI and how rapidly the technologies are moving that one of the fundamental balances that countries need to strike vis-a-vis regulations is striking the balance between innovation and responsibility while anchoring it to responsibility. That is one of the key regulatory principles that needs to be in place."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:10",
                    "end": "00:05:38",
                    "text": " And the regulations have to be agile because the technology is moving at such a fast pace that you cannot anchor the regulatory framework to yesterday's technologies. And at the heart of it, I hope what you will take away from our blueprint is realizing sovereign AI potential for any country, including India, is really about the public-private partnership. And it's really about marrying"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:38",
                    "end": "00:06:12",
                    "text": " the public resources with private innovation and that really is what the key to unlocking the full potential of AI and sovereign AI is in this country. So again I would encourage you to read through the blueprint and we look forward to your feedback and we look forward to partnering closely with Indian ecosystem to help India realize its aspirations with AI. Thank you very much. Thank you so much Dr. Mohindra. I'm going to request you to please"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:12",
                    "end": "00:06:50",
                    "text": " stay back on stage. I'd also like to invite Manish Gupta, President and Managing Director of Dell Technologies India to join us here. This is the big moment, ladies and gentlemen. We are ready for the unveiling of the Dell Technologies Blueprint to accelerate India's AI growth. Yes, that's a photo moment for everyone. Thank you. Thank you so much, gentlemen."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:50",
                    "end": "00:07:20",
                    "text": " This blueprint advances India's vision of Viksit Bharat 2047, positioning AI as a foundational engine for productivity, modernized public services, opportunity expansion, and strategic autonomy. It centers on three pillars that we've been discussing. Invest, invest in sovereign scalable compute and data foundations. Innovate,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:20",
                    "end": "00:07:46",
                    "text": " with collaboration and with a future-ready workforce and evolve. Evolve into a responsible, agile, security-first governance structure. So our next panel today will go inside this blueprint and India's AI future to unpack how to convert this ambition into nation-scale execution. And that's quite a mean feat for a country as diverse"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:46",
                    "end": "00:08:32",
                    "text": " and as huge as India. So let's welcome the panelists for all the tough questions this morning. Raj Gopal, AS, Managing Director and Chief Executive Officer of NextGen Cloud Technologies. Bhaskar Chakravarthy, Dean of Global Business, the Fletcher School of Law and Diplomacy at Tufts University. Please have a seat, sir. And once again on stage, Manish Gupta, President and Managing Director of Dell Technologies India. And I will be moderating this session for you. Welcome, gentlemen."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:32",
                    "end": "00:08:59",
                    "text": " We are here this morning to really translate the invest, innovate, and evolve pillars into very actionable steps that we all can take together to grow India's AI ecosystem. So I'll begin with targeted questions to each of you. And of course, you are free to jump in to add thoughts to each other. It's a candid free flowing conversation. Mr. Raj Gopal, if I can start with you. Startups and MSMEs."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:59",
                    "end": "00:09:30",
                    "text": " They are the engine of our economy. They are also the engine of our innovation, especially as far as AI is concerned, but access to very reliable, affordable AI compute and cloud at scale continues to be a barrier for many of the small and medium enterprises. Now, in your opinion, what are some of the policies, some of the infrastructure, some of the market interventions that we need today to really unlock this access at scale?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:30",
                    "end": "00:09:59",
                    "text": " I think across many other countries that we have seen, India has got a much comprehensive approach to this. I mean, in terms of actually they started this India emission, which is across seven pillars. I actually don't see many startups actually using the facilities that are there in the sense that you could apply for GPU infrastructure and you would get it at a subsidized rate. And some of them even got 100% of the GPUs that they need."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:59",
                    "end": "00:10:25",
                    "text": " So I think from India's side, we've got a little less number of GPUs compared to what we really need. Maybe we need about 200,000 GPUs now, and we have about 450,000 now. So we all need to really invest more and then deploy more. But most important thing that you should see is that there is a good ecosystem. There is a system policy available."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:25",
                    "end": "00:10:45",
                    "text": " for MSMEs and startups to leverage this. So there are a lot of innovative AI solutions being built. Most importantly, I also see that government actually setting pace in terms of actually leveraging some of these. We ourselves, I mean, we did one job for government which is very, very unique."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:45",
                    "end": "00:11:14",
                    "text": " Like we serve Election Commission of India, they came to us and said, can you deduplicate and look at all the photographs that we have? This is like 90 crore pictures, right? So humanly it was not possible to deduplicate. You can't check one photograph with 90 crore others. We did that in a matter of 51 hours, and then we responded to them as to whether they had applications and all that. That deserves some applause. That's a humongous task."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:14",
                    "end": "00:11:43",
                    "text": " What I see in this country is that I don't think we will be pure play. This chatbot, I mean, the generative AI, the way it has been envisaged, I mean, will be the primary use case. I think we are going far beyond what AI can be applied in terms of actually improving the productivity of citizen services and also give use cases that these small and medium enterprises can actually use."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:43",
                    "end": "00:11:51",
                    "text": " In terms of actually enabling more GPUs, I think we need a lot of money."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:51",
                    "end": "00:12:18",
                    "text": " 100th of what US is investing or even less. So for us to do more, I think we need to remove certain bottlenecks that are there. One of the things I believe that can be done is, I'm sure everybody's familiar, we all pay GST. But basically when we import servers, when we pay GST on it, and then when we deliver servers, I get that as an input. And we do the, I mean, we only pay the value added piece back to the government."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:18",
                    "end": "00:12:38",
                    "text": " The government gets the GST either way and I get an input. So the way to, one thing that we could look at is whether we can waive off the GST upfront and just take the GST when the services are being delivered. What that would do is it will deliver, it will reduce my upfront infrastructure cost to buy about."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:38",
                    "end": "00:13:07",
                    "text": " you know, you know, 18%. You know, I don't have to fund that upfront and then pay interest on it, you know, or raise equity and, you know, pay more expectation on that. I mean, these things could really help. I mean, there are some of these things that government should look at and actually facilitate this piece. The last point is that they've given tax holiday for delivering services to the world, but I think India has got a lot more to do within India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:07",
                    "end": "00:13:26",
                    "text": " And then just look at world market and I believe that we should get, Indian service providers should get the same benefits as the global providers would get when, you know, they host services in India. So maybe a GST waiver and some income tax benefits could be good."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:26",
                    "end": "00:13:43",
                    "text": " Okay, GST waiver, income tax benefits, demands from the industry coming in. Professor Bhaskar, if I can come to you now. You've often argued that nations need to compete not just on technology, but on trust, on institutional strength, and of course,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:43",
                    "end": "00:14:12",
                    "text": " very, very inclusive digital participation. And that's very critical for a country like India because we are a country of many countries. The rich poor divide is quite huge. There are a lot of bridges to gap here. There are a lot of gaps to bridge here. My apologies. Now, as India accelerates AI, what do you think are some of the biggest non-technical bottlenecks that we should be looking at addressing, which you believe could really limit this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:12",
                    "end": "00:14:29",
                    "text": " momentum that we are on currently and if we need larger societal good, what are some of the non-technical barriers that we need to immediately resolve? Thank you for the question and thank you for the invitation to join this terrific panel."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:29",
                    "end": "00:14:52",
                    "text": " The issue of what are the non-technical elements, it's great that you have included that question in this discussion because in all the excitement around the technical infrastructure, which is of course enormous that's happening right here in India, it's no secret that this is one of the biggest talent pools in the world, growing very, very fast."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:53",
                    "end": "00:15:14",
                    "text": " years, one in three developers are going to be in India, the largest mobile data pool anywhere in the world, the third largest data pool, you know, anywhere in the world once you take mobile and everything else together, growth and compute, growth and energy, growth and workloads, all that is happening, you know, which is fantastic. Now, when you think about what are the other elements that drive demand?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:14",
                    "end": "00:15:43",
                    "text": " But we have found, we study 125 countries and try to understand the role of technology in shaping lives and livelihoods across 135 countries. The single most important determinant of what keeps a country on trajectory in terms of both the momentum of growth and also the state of their digital evolution is the demand side. So when I think about the demand side, obviously the core infrastructure which has been talked about is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:43",
                    "end": "00:16:04",
                    "text": " enormously important and is going to continue to be a major contributor. A second part which has been talked about a lot in the Indian context is the distribution infrastructure. With DPI and all the different platforms associated with it, we know that there's a very powerful distribution system. Now there's a third infrastructure which is the non-technical part."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:04",
                    "end": "00:16:32",
                    "text": " And that is what I would call the trust infrastructure. Now when you think about trust, it's a bit of a slippery concept. It's really hard to define. Each one of us in our heads has an idea of what trust is. But if we force somebody to define it, we'll struggle. The best thing about trust is I know what trust is when it is not there, when it is missing. And then you have to ask the question from a human perspective, what really is trust? And how do I?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:32",
                    "end": "00:16:60",
                    "text": " bake that into the policy systems, into the technical systems, into the marketing systems, into the narrative around the India ecosystem, which will then keep moving the demand going. And trust ultimately has to do with, do people have confidence? The people who are the grantors of trust, do they have confidence that this invisible"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:60",
                    "end": "00:17:28",
                    "text": " transaction that I'm engaging in, whether I'm putting my data into a system or whether I have entered my financial information and I'm expecting something on the other side, that this whole thing is going to be completed, and it's going to be completed in a way that is reliable, that is repeatable, and will not take advantage of me. So when you think about that whole trust ecosystem, India starts in a great place. Relative to where I come from,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:28",
                    "end": "00:17:56",
                    "text": " The United States, India is a far more trusting country in terms of trust in digital systems overall and certainly in terms of AI. There's a tremendous level of enthusiasm in terms of embracing all things AI. And we've seen this right here, just the sheer numbers of people who've attended the conference. It shows a level of trust that is probably unmatched anywhere in the world. Now, this is a tremendous asset to start with. The challenge is that the institutional side of trust"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:56",
                    "end": "00:18:22",
                    "text": " is still in the development process in India. So if you think about data governance, privacy, security, we're making progress, but we need to be much further along. Other aspects of trust has to do with the fact that, say, the India AI mission, that is developed at the union level, at the center level. But the actual exercising of trust, the granting of trust happens at the district level."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:22",
                    "end": "00:18:52",
                    "text": " The district varies depending on whether I'm in Telangana or I'm in Jharkhand. And at the union level, the principles that I've got in place need to be sensitive to how it's being experienced from the ground up. So there are many different facets of trust that we need to work on and put in place, including transparency, which AI is, you know, we are facing a challenge regarding transparency across the world. So this is an issue. And then, you know, an approach to having redress and grievance systems and then literacy."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:52",
                    "end": "00:19:21",
                    "text": " You know, people need to be able to understand how to use this exciting technology and also protect themselves. I'll pause and have a conversation. Absolutely. Thank you for that wonderful perspective, Professor. Coming to you, Manish, now the Dell Technologies Blueprint really calls for tighter alignment between policymakers, between industry, academia, institutional capacity. How can frameworks like this one really ensure growth that is both"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:21",
                    "end": "00:19:47",
                    "text": " globally competitive for India, but also locally inclusive because there is a lot of regional growth. There is a lot of geography that needs to be taken into the fold of AI when we are talking about being globally competitive as well. Thank you for that question. Just before I go in there, I would just add or maybe speak on a couple of topics that Professor just spoke about."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:47",
                    "end": "00:20:14",
                    "text": " While he spoke more from a non-technical standpoint, he also brought in a technical aspect to it around the entire governance. And it's not just non-technical in today's world. It's driven by data privacy and all of the things around. But equally on literacy, there's also explainability. That trust comes really inherently once you have got explainability and people are aware on what outcomes are coming. Is that explainable and do they understand that? So it's a very, very interesting world that we are in."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:14",
                    "end": "00:20:37",
                    "text": " Now, back to the blueprint that we were talking about and Vivek articulated that beautifully well across the three pillars of invest in data center capacity, invest in energy infrastructure, invest in people which goes back into the innovate side of it because like we just discussed, we've possibly got the largest pool of engineers around AI and that's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:37",
                    "end": "00:21:06",
                    "text": " that skilling on innovate part is what's going to differentiate us what's going to make it really real and practically doable within the industry and would differentiate us versus other nations equally make it make it make it more democratized and ubiquitous across the nation and lastly it's really about how do you continue to build in the guardrails how do you build the trust like we just discussed to ensure that there is the the ecosystem knows"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:06",
                    "end": "00:21:34",
                    "text": " that this entire process can be trusted and can be built upon. We've also got to remember the sustainability aspect of it, which is where, as you look at the blueprint, you will see us talk about the fact that energy efficiency, sustainability on data centers, and new architectural models are becoming super important, and that's something that NextGen under Raj Gopal's leadership has demonstrated in building highly sustainable"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:34",
                    "end": "00:21:53",
                    "text": " more energy efficient data centers that will allow us to use our energy resources in the best possible manner while democratizing the access of compute capacity, while democratizing the access of data center capacity to organizations and verticals of various sizes."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:53",
                    "end": "00:22:11",
                    "text": " So that's going to be the critical pillars around which we really believe that there's practicality in adopting and differentiating ourselves on the AI arena as we go forward. Right. I'm going to pick up on that data center piece and come to you Raj Gopal. Now, given the scale that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:11",
                    "end": "00:22:29",
                    "text": " India will really need for competing globally. What would it take to truly build sovereign cost-efficient AI infrastructure that's not just available to large enterprises, but is also very affordable for the long tail of innovators that we have in this country."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:29",
                    "end": "00:22:51",
                    "text": " Yeah, if you see the data in the data center industry It's been pretty concentrated in Mumbai and little bit in Chennai and you know, the other markets didn't really take off as much as they should have So what we are trying to do is to have I mean our current plan is to put about 100 megawatts of data centers in about six states"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:51",
                    "end": "00:23:16",
                    "text": " And what I see is that going forward, I mean, this could be the model that can be built one where each state has got a capacity because these states itself have got so much of consumption that can happen primarily because if you start looking at applying AI into elevating the quality of education in India, which will be one of the first things that will get rolled out at scale in India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:16",
                    "end": "00:23:43",
                    "text": " and also the healthcare aspects of it and citizen services. These things can require lots of computing capacity. So we are working with few state governments to actually see if we can bring a total transformation, actually consolidate their applications, bring about a data lake and then apply intelligence to it and take it to masses."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:43",
                    "end": "00:23:58",
                    "text": " The way I think it will evolve is that there will be many more regions where data centers would spring up. And then when they are distributed, they need to be interconnected. We have very good interconnect system."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:58",
                    "end": "00:24:26",
                    "text": " and not just the telcos but you have things like you know we have railway networks we have power networks which can actually assist the good connectivity between them when these things come into play actually we can have a pretty distributed a good amount of compute in india that can actually serve this this aspect but you must be aware that you know this game is actually i mean if you see my context i mean i have four diamonds to work on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:26",
                    "end": "00:24:46",
                    "text": " One, of course, is geopolitics. I mean, there is quite a lot that is happening. We need to ensure that we have access to the technologies that we want to bring to India so that India actually works on the best available infrastructure. It is slightly better now, but there were restrictions before."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:46",
                    "end": "00:25:08",
                    "text": " The other aspect is the amount of money. I mean, so I think it requires multiple billion dollars of investments to do, and that should be facilitated. That should really come into the country. When we have the money piece sorted out and then we build the infrastructure,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:08",
                    "end": "00:25:33",
                    "text": " A good thing that we can leverage is open source. I mean, when we leverage open source, we can actually combine the infrastructure and open source and bring down the cost of compute so much that it is actually palatable for the Indian citizens to use. Because it's not about serving this 2%, 3% of the population which pays the income tax. It's about serving the 90% others. The moment we succeed in doing this, I think the talent also"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:33",
                    "end": "00:25:49",
                    "text": " The good talent that we have, we have access to good talent in India, but we don't have good talent. The quantity is missing. I mean, so we have good people, but you know, you need many more good people. They're going all over the world. We want to bring them back."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:49",
                    "end": "00:26:18",
                    "text": " These things, these people can come back when these money and infrastructure fall in place. Right. That would ensure that India is playing a role which is actually pretty balanced, leveraging the global technologies and leveraging local talent and actually sets a blueprint for all the other countries which don't have, which other countries which are, which may miss out this revolution, right, and become digital colonies of the top two countries that are investing heavily in AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:18",
                    "end": "00:26:35",
                    "text": " So I personally believe we have a good blueprint. And the blueprint can be applied to multiple countries. And we are well on the path. And I would prefer a distributed development of data centers across the country so that we are closer to the users."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:35",
                    "end": "00:26:59",
                    "text": " We serve the other 90% or 95% of the population that is not being served now. Absolutely. Well, Professor, coming to you next, studies on digital competitiveness have consistently shown that institutional capacity often determines whether technology adoption really translates into economic value or not."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:59",
                    "end": "00:27:26",
                    "text": " in the Indian context, what do we really need to do to really strengthen institutional muscle here that ensures that AI drives very, very inclusive growth rather than, you know, deepening the tech divide because there are already many divides that we are battling with. Yeah, so when you talk about institutions, the institutions need to be fit for purpose. So I think it's helpful to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:26",
                    "end": "00:27:55",
                    "text": " break down this thing called AI that we all know that it's not just one thing. It is a general purpose technology that's finding its way into many different uses. And when you think about the uses, we talked about small and medium businesses, of course, the backbone of the adoption of AI in the enterprise level. But if you think about some of the biggest use cases in India, agriculture, skilling,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:55",
                    "end": "00:28:24",
                    "text": " healthcare. Those are the three biggest ones, just in terms of sheer numbers of people affected and the impact that you're going to have. It may not necessarily have the same revenue impact, but it has a much larger societal impact. We just think about agriculture. If 40% of the crops grown in India, primarily among smallholder farmers, if 40% of those crops are destroyed by pests, and if I can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:24",
                    "end": "00:28:36",
                    "text": " fix that problem, get that 40% number down to 30% or 20%, suddenly I have improved farmer well-being and income by an enormous amount. So if we just think about that one transformation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:36",
                    "end": "00:29:06",
                    "text": " using essentially very simple AI. It doesn't have to involve complex databases. The compute is done at the edge. You are dealing with people who have intermittent access to the internet, and you can make it work. Now, in that context, the question is, what is the institutional capacity, the institutional infrastructure, the trust infrastructure I need? Basically, the farmer needs to start"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:06",
                    "end": "00:29:11",
                    "text": " from the point of view that I can use my phone as a tool for farming"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:11",
                    "end": "00:29:37",
                    "text": " which is something completely new. People have learned farming from earlier generations. How do I think about the phone as a tool for farming? Just crossing that chasm requires a couple of seasons of using the phone and then seeing the outcomes. The impact of pests has gone down and it has made a practical difference to my life. Now then you have to ask the question from a farmer's perspective, what's preventing them?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:37",
                    "end": "00:30:07",
                    "text": " from doing that. And that's where the institution becomes relevant. That's where the explainability aspect also becomes relevant. For a farmer to understand how the system works, essentially, the picture I take of my crop needs to match up with an image database that is out there. And essentially, with a very simple interface, I can match the pest, and I can then end up with a solution to the problem. So the same thing for skill building."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:07",
                    "end": "00:30:26",
                    "text": " literacy. If I can see my ability to speak, my ability to read in multiple languages, improve, you know, suddenly my trust goes up. So what is the minimum amount of institutional safeguards I need to provide that. Then I come to something like healthcare, when people have a much bigger chasm to cross."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:26",
                    "end": "00:30:51",
                    "text": " That's where people have a lot of concerns about, should I be putting my information into the system? How is this going to be used? Can I trust the phone, where I've relied on a doctor, or I've relied on a wise person in my community? I've relied on my mother for maternal health care advice. So how do I cross that chasm, being able to provide the foundational trust elements?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:51",
                    "end": "00:31:01",
                    "text": " is going to be important so the answer to your question very long answer is it depends it depends on the usage as is the case to a lot of questions about india"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:01",
                    "end": "00:31:29",
                    "text": " It depends, right? Well, Manish, you know, globally we are seeing nations tie AI strategy to strategic autonomy. Now, whether it comes to the semiconductor ecosystem or the supply chain, you know, the strategic autonomy is becoming extremely important for countries. For India, what are the two or three foundational capabilities that you think we need to build domestically in this decade to ensure that we are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:29",
                    "end": "00:31:59",
                    "text": " true creators of AI value and not just consumers? Awesome. Great question, Ritu. And we as a nation have proven ourselves to be phenomenal adopters of technology. And the best example in my mind comes on UPI or digital payment. 10 years back, 11 years back, we were just not there. And today, we are by far the largest consumers or the number of digital payments and the value of digital payments within India is multiple times of the second"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:59",
                    "end": "00:32:26",
                    "text": " economy that does this right so that's a great example of how we have been able to localize democratize and proliferate the use of technology so within that I would really put on three hats here the first and just you know inverting the pyramids not starting with technology but starting people we've really got to think away from the users to the to the developers you know it's got to move from 1 billion users"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:26",
                    "end": "00:32:55",
                    "text": " to 1 billion or 1 million or 10 million developers. And that's the skill set, that's the IP that we are going to bring in because we've got that largest talent pool residing within the country. The second, and again, I heard you talk about semiconductor and supply chain. I think we've got to adopt the best that's available there globally. But equally, we've got to move, we've got to not just think about made in India, but talk about trusted in India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:55",
                    "end": "00:33:22",
                    "text": " and which is where work with organizations such as AISI Artificial Intelligence Safety Institute to ensure that we are putting the right guardrails, putting the right governance policy and the entire institutional framework to ensure that the AI that we are building here is trusted. And lastly, it goes back to the same thing. And maybe I'll use the same example. You know, we had the UPI of money. We need to have UPI of AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:22",
                    "end": "00:33:46",
                    "text": " where we are building that at scale using the data sources that we have. We have the largest ones. And some of the initiatives that the government has taken, India AI Mission, but equally think about AI Cosh. There are more than 7,000 data sets that are now available to organizations of all sizes. Use that to ensure that we are developing for the country at the population scale through academia, through..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:46",
                    "end": "00:34:16",
                    "text": " private sector, through startups, through MSMEs all coming together. And that really requires a consistent API layer that's bringing, theoretically, maybe even all of the data center and the compute capacity that we are creating as a part of AI mission to be one single layer that can be consumed by anybody and everybody across the nation to start to innovate on that, to start to develop on that. So going back, I know a long answer, but if I were to summarize three things there, UPI of money to UPI of AI. Made in India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:16",
                    "end": "00:34:34",
                    "text": " transitioning to trusted in India and you know from a developer users to from billion users to maybe a million or 10 million developers made in India but made for the world absolutely all right so I'm going to ask each one of you for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:34",
                    "end": "00:34:54",
                    "text": " a few concise takeaways today. Now, the blueprint that we are talking about that Dell Technologies has just unveiled talks about agile trusted AI governance with sectoral baselines, test beds, strong institutional coordination. Yet, globally, what we've seen is that speed often beats caution."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:54",
                    "end": "00:35:20",
                    "text": " And we are seeing some of the scary stuff coming out with AI as well. A lot of the experiments, a lot of agentic AI experiments that people are doing across the world, some of them call for caution. Now, in the Indian context, how do we stay globally competitive while also operationalizing very, very stringent safeguards? And where should we really draw the line between"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:20",
                    "end": "00:35:42",
                    "text": " the speed of innovation or innovation velocity and regulatory discipline. So, Braj Gopal, if we can start with you, please. If you take the birth of GNI, I mean, it actually, I don't think none of those rules were actually followed and, you know, it was built on every data that was available in whatever form."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:42",
                    "end": "00:35:54",
                    "text": " Personally, you know, in most places, I'm actually trying to tell people just it's not about ignoring the risk factors of it, but I think the regulation should not."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:54",
                    "end": "00:36:20",
                    "text": " curtail the innovation in the thinking that we should be restrictive about whatever we are working with. And so one of the most important things is I think we should have less regulation in this place because I think overall the benefit, you should look at AI like a utility. I mean, you will have more good with AI than with bad. Yes, there are things that can be handled as we go along."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:20",
                    "end": "00:36:43",
                    "text": " If you see what we do in cloud today, we haven't been able to sort out our security and you know data protection postures even today. It's an evolving journey and I think we will continuously catch up with the bad factors around the AI adoption and that's a journey it cannot start or stop or it can be implemented at a point in time. So we should keep."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:43",
                    "end": "00:37:13",
                    "text": " Looking at those aspects and keep putting with the regulations or technology interventions to ensure that we handle those pieces, but we should go fast forward with implementation and adoption of AI. And I see a lot of Indian enterprises really being reluctant in terms of adopting it, especially the larger ones. But if you see in India, I think government will set the pace and the startups and the MSMEs will actually catch on from there."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:13",
                    "end": "00:37:39",
                    "text": " And the large enterprises will actually struggle to catch up with the amount of innovation that's happening in these two spaces. Where does that reluctance come from? What are the top three reasons large organizations are reluctant? One is, of course, the fact that it's not easy to adopt and transform a large organization. And perhaps startups and SMEs have the benefit of the agility and the small scale that they're at. What else?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:39",
                    "end": "00:38:07",
                    "text": " So I think the first issue is not about security and all that. I mean, about these aspects. But most importantly, I think a lot of people are struggling to imagine where to apply AI. And I think the moment we can understand that, you will start seeing that the benefits are far outweigh the negative aspects of it. So I think that's the first thing that people should look at is to not just look at leveraging the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:07",
                    "end": "00:38:35",
                    "text": " I mean, but actually really look at where you can deploy. I talked about that deduplication piece there are like we are working on more than 150 projects and not all of them are, you know, bot based. So that imagination is what is important and one that once that imagination comes you will out. I mean the benefits will outweigh the negative aspects of whatever we are seeing. All right, Professor final takeaway from you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:35",
                    "end": "00:38:53",
                    "text": " on, you know, speed versus caution. Yeah, so if you think about speed, I like to always like to use the analogy of a car and a road. You can think about the speed that you can build into the car, the velocity of the Ferrari and a lot of the conversations that have happened."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:53",
                    "end": "00:39:11",
                    "text": " not necessarily in this room, but in other rooms, is about the Ferraris, whether I'm talking about agentic AI or AI optimized for certain applications and the technical aspects, you know, really, really important. Now, if you take the Ferrari and you bring it into the Indian context, maybe it's a Maruti or something else that I need to be talking about. But then the question is..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:11",
                    "end": "00:39:36",
                    "text": " What's the road on which this Ferrari is going? If it's a dirt road full of potholes, even a Ferrari is not going to go very fast. So much of our conversation here is about that dirt road. And what are the things, what are the potholes that we need to fix? There's one elephant on the table that we did not address, and I'm just going to leave it at that, which is when we talk about trust, there's a whole bunch of things you can do from an institutional standpoint to build trust, transparency, explainability, and so on."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:36",
                    "end": "00:40:01",
                    "text": " There's a huge issue that we need to think about, which is what is going to be the impact on jobs? What's going to be the impact on jobs? This is the youngest major country in the world. It's also one of the least employed country in the world. And now with AI coming in, is that going to help boost jobs or is it going to take jobs away? If we don't fix that problem, get ahead of that. All the trust we are talking about, all the institutions you build."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:01",
                    "end": "00:40:27",
                    "text": " could come down. So part of the policy infrastructure here is to figure out what is the post-AI jobs picture. Absolutely. Manish, final word to you. So I honestly don't think that these are opposing forces. Agility versus security, and particularly in this side of technology, you cannot have them act as opposing. It's really about building the frameworks that are going to take both of them together."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:27",
                    "end": "00:40:55",
                    "text": " This is fast evolving as a technology, but equally as institutions, we'll have to be faster than that in evolving. I think the government has done a phenomenal job in building some of the frameworks around that, and the institutions, AISI as one example, but on the privacy side, DPDP or DEPA, all of those acts being there, are good frameworks to start with, and I'll just index back on the question that you had asked Raj Gopal earlier."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:55",
                    "end": "00:41:20",
                    "text": " on what is the hesitation from enterprises in adopting. I don't think it's necessarily about security. It's really about saying how many of those have real use cases. While the real use cases exist, how many of them are able to monetize or are able to see them scale from experimentation or pilots into production. I think that's a job that we as industry folks who understand the technology, who are innovating in this space, really need to bring to the table so that we can..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:20",
                    "end": "00:41:37",
                    "text": " bring this to the floor across the nation and enterprises and organizations of all sizes and academia and public. I think that's where this will get practical. But equally, these are not opposing forces, they go hand in hand."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:37",
                    "end": "00:42:06",
                    "text": " Well, thank you gentlemen for that absolutely incredible conversation. The takeaway is clear that investing, innovating and of course, innovating by expanding skills pipelines and accelerating AI deployment is going to be key to India's sovereign AI infrastructure and appreciate you joining us here and taking the time today. We are also very delighted to now be joined by Honorable Sri Jayan Chaudhary Ji, Minister of State."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:06",
                    "end": "00:43:20",
                    "text": " for Education and Minister of Skill Development. Huge round of applause. We are going to have him up here shortly. Thank you, gentlemen. Thank you very much for joining us. So if we can have you up here for a quick photo op and we will then continue the conversation. Thank you. Thank you, everyone. Thank you. Thank you, gentlemen. Manish, if I can please request you to felicitate our speakers. Mr. Raj Gopal."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:23",
                    "end": "00:44:03",
                    "text": " Let's have a huge round of applause for our panelists here today. Professor Bhaskar Chakraborty, thank you so much for joining us here today. If you all can just get off the stage for two minutes, we are getting it ready for our next conversation. Thank you so much. Well, ladies and gentlemen, time to move on. Now, if India's AI ambition is to translate into real economic growth,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:03",
                    "end": "00:44:27",
                    "text": " It's obviously not going to be any one entity's job. It is not going to be driven only by the government or by the industry alone. It will be driven by partnership. India has the talent, the digital backbone and the momentum. The real question though is how do we scale AI responsibly, securely and inclusively?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:27",
                    "end": "00:44:56",
                    "text": " So our next fireside chat conversation will explore what a powerful public-private model regarding AI could really look like. And for this, I'm delighted to welcome two very eminent leaders who are instrumental in shaping the journey, both from policy and industry perspectives. We have, of course, Honorable Sri Jayan Chaudharyji, Minister of State for Education and Minister of Skill Development and Entrepreneurship, Independent Charge, Government of India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:56",
                    "end": "00:45:33",
                    "text": " And we have Dr. Vivek Mohindra, Special Advisor to the Vice Chairman and COO Dell Technologies Global. If I can please have both of you up here for a quick conversation. Thank you so much. Thank you very much. Thank you, gentlemen. Well, it's quite clear that public-private partnership is going to be critical to AI scaling and adoption in India. You know, Chaudhary, if I can start with you, how can PPP models really accelerate"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:33",
                    "end": "00:46:00",
                    "text": " large-scale AI infrastructure. What have been some of the on-ground experiences you've seen so far? Of course, the government has been moving at breakneck speed when it comes to deploying more technology, giving a more Philip to innovation in India. How are you really ensuring trust, resilience, and long-term national competitiveness as AI becomes very mainstream with this event in India?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:01",
                    "end": "00:46:31",
                    "text": " In the Indian context, as the audience is aware, we had a lot of catching up to do. And it's fair to say that a lot of what we are seeing around us in AI has been facilitated by creating an ecosystem in a short span of time. Perhaps we may enjoy a second mover advantage with regards to this technology. And that has come about only because of a strong top-down"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:31",
                    "end": "00:46:60",
                    "text": " emphasis and push. The only reason why this event is happening here in India is because the leadership at the top understood very quickly the value and the potential of this new technology that we should not view it as a disruption but view it as an opportunity to leapfrog legacy problems, deficit problems and provide access and equity to our citizens and dignity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:60",
                    "end": "00:47:26",
                    "text": " to our workers. And that's why the Prime Minister, you know, the last event in France, shared that leadership space. Every opportunity he gets, he talks about skilling, about young people, about the potential for AI and the enunciation in Manav that this technology needs to be human centric. I think that has given us a real emphasis and a push."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:26",
                    "end": "00:47:54",
                    "text": " for the academia for our industry for our vibrant startups systems to really think about what they are doing in this space. I think that is the background to the event that we are all witnessing thousands and thousands of people casual visitors apart from those who are already entrenched in technology and this message that goes out is that 1 billion strong young people in a developing country."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:54",
                    "end": "00:48:25",
                    "text": " are already thinking about what AI means to them and what they can do in this space not just be consumers, but also be producers and innovators and thinkers and creators now PPP in this domain for me and when you think about man of being human centric citizen centric the P that really matters is the people and in that context, it's important that you have the broad architecture, which is open."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:25",
                    "end": "00:48:51",
                    "text": " This is something that India has stood for from the beginning when there was a lot of debate about what should be the policy that enables AI, but there was also a lot of fear around AI about trust factors, about privacy, data, sovereignty, multiple issues about the human interface, the augmented human worker, what this means for education, for the future of jobs."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:51",
                    "end": "00:49:20",
                    "text": " A lot of those issues were being discussed and debated and India said that yes, it's good to have a strategy and out of that strategy and experiences will evolve a robust policy. It is essential to have guardrails, but it is a starting point currently. We don't want to infringe upon the possibility of innovation and India took that approach and we had open access to whatever compute, you know India AI mission."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:20",
                    "end": "00:49:50",
                    "text": " was set up with a target of 18,000 GPUs and in a short span they were surpassed it. It's about 38,000 and a roadmap is by the end of this year. It's going to cross one lakh threefold now think about it. All of this compute for cell facility that has been created is a model of PPP. It has to be housed in educational institutions. So that real research can happen in our premier educational institutions. This is the great time when academia."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:50",
                    "end": "00:50:19",
                    "text": " is more important than it perhaps ever was in the Indian context. In the Indian context, academia was partly separated from industry and the real economy in our minds. But now every Indian citizen is realizing the value of research and innovation. Every family is saying that no, this is important. We must value it and every educational institution is saying that we are not divorced from the market and needs of our community and society and nationhood building."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:19",
                    "end": "00:50:50",
                    "text": " So that engagement with nationhood building and concept of technology is deeply immersed thanks to the efforts of India AI mission and here I'll just leave one data point to you. What is the cost of this compute facility? It's being provided to for startups for researchers at 65 rupees for an hour. You pay 300 rupees for a couple of hours PVR cinema ticket. So it's probably the world's cheapest compute facility, which is open."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:52",
                    "end": "00:51:21",
                    "text": " We are celebrating Sarvam. Let's not focus on for-profit, not for-profit, because everything has to be for people. If you look at Sarvam, that's also in my mind a PPP. It's been incubated by IIT Madras. It has been supported by AI mission. So that's another example, because you're right. Government cannot invest in from data, to energy, to compute, to innovation. It really has to come from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:21",
                    "end": "00:51:48",
                    "text": " our citizens, our researchers, our technologies. It's an effective mission. It is. It is. Absolutely. Well, Dr. Mohindra, getting your point of view from a Dell perspective. Now, if we look at PPP as far as job enablement is concerned, because that's the big concern that citizens have, that what's going to happen to our jobs. And of course, skilling is part of that journey. How can Dell partner with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:48",
                    "end": "00:52:13",
                    "text": " ministry of skill development and entrepreneurship. What are you all doing from a future skill labs perspective? How are you accelerating AI apprenticeships? So that basically the jobs also move beyond the metros because tier two, tier three towns is where we are looking at a lot of talent sitting there, but we are looking at a lack of access of sorts when it comes to skilling."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:13",
                    "end": "00:52:43",
                    "text": " Yeah, I think that's a great question to do and I think, honorable minister, good to see you again. It reminds me of our discussion we had back in last time we met in October of 2024. I know we missed each other and often and we covered very similar ground. I think at the heart of it, it does come down to and I commend India on the progress it's made in making access to all these GPUs available. It is an industry academia partnership working closely with the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:43",
                    "end": "00:53:05",
                    "text": " the minister here. And our view is when you think about skilling, there are three different levels. You have to think about schooling, you have to think about college level, and you have to think about employment. People entering the workforce are employed today. And you think about the delivery of this through online, in-person, incubation. So those are the two big dimensions."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:05",
                    "end": "00:53:34",
                    "text": " And from our perspective, we are very excited to partner on extending it to tier two, tier three, working closely with the minister and other institutions in India. And the core of it, having accessibility to this GPU at such an amazing price point, really unlocks the potential. And I think as Minister and I talked about last time we met, and I predicted the rise of agentic technologies that will lower the bar even further in terms of being able to provide the skilling."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:34",
                    "end": "00:54:03",
                    "text": " So those are some of the ways that we have been working and we look forward to partnering. Right. Well, finally, to both of you now, as we embed AI into all our critical sectors, whether it's BFSI or telecom or agriculture, health care, education, now governance has to move from intent to very, very strong operational safeguards. What does zero trust AI architecture then practically look like at a national level? Minister, if we can start with you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:04",
                    "end": "00:54:31",
                    "text": " Well zero trust it's it's an interesting terminology What that for me the way I look at it? It means that you have to be able to verify each and every protocol in your design and In India, we generate a lot of data everything and Indian citizens are quite open about access globally privacy"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:31",
                    "end": "00:54:58",
                    "text": " has been a major concern and sometimes it becomes also an impediment for governance because those data points aren't being collected, analyzed, researched. In the Indian context, citizens are okay about sharing their data and I'm saying this with the knowledge that you know, crores of upper IDs have been formed, created using consent, but we have not received any blowback."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:58",
                    "end": "00:55:25",
                    "text": " in a way from from students families saying that why are you because they understand that we are able to with technology customize and tailor experience in the classroom for every student where no student can get left behind. What that means for the employability for the knowledge acquisition for that student for the quality of the educational experience is immense. So I think but once you're collecting the data, there is a lot of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:25",
                    "end": "00:55:54",
                    "text": " effort that needs to be put in so that that trust is maintained from zero trust to a hundred percent trust in in the public mind. So I think our data sets need to be segmented. There are protocols in within government of India in education. We're thinking of creating a complete AI stack, which means the anonymized data sets will be made available for researchers for the creating the value for the layers of innovation for enabling."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:54",
                    "end": "00:56:24",
                    "text": " startups to engage with that data that government of India and the citizens have shared similarly in skill. We have skill India digital hub, which is also looking at creating those data sets which can then really help us unleash the next wave of innovation and requirement that we have in skilling. So I think once you have that system design in place, it can be achieved the Prime Minister spoke about a label for content that you know that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:24",
                    "end": "00:56:53",
                    "text": " It should be verifiable and legal. This technology and consumer awareness is a big aspect of it on how we engage with these tools and how we understand what the outcome of our engagement. How true is it? How is it verifiable? Where are these AI models trained? Is there any bias in that data set? All that knowledge needs to be out there for the consumers. I feel also that there needs to be audit trail for our new AI models."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:53",
                    "end": "00:57:22",
                    "text": " Maybe in the future, you could have CAG come out with the audit report of all the AI models. So it's a brave new future, but it's a balance. For partnership at scale, you need an architecture with trust. Absolutely. Final 30 seconds to you, Dr. Mohindra. I think the minister said it very eloquently. I would extend the notion to zero trust should extend to start with data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:22",
                    "end": "00:57:49",
                    "text": " go into AI models, the usability, the cybersecurity elements, and the identity and access management. Those would be the ways I would extend it. And practically, it means, beyond the governance framework, having things like a national risk registry, observability, being able to report whenever there is an infraction and auditability. But, Minister, you said it very eloquently, and I think our AI blueprint has more details that would be worth looking at."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:49",
                    "end": "00:58:18",
                    "text": " Absolutely. Thank you very much. Thank you, Minister Chaudhary. Thank you, Dr. Mohindra. Well, clear takeaways from that conversation, public-private partnerships, future-ready workforce pipelines that reach much beyond the metros, and, of course, zero-trust guardrails that we all need for a very, very defined, AI-led future for India. Thank you very much for joining us here today. Appreciate you taking the time. Dr. Mohindra, if you can please..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:18",
                    "end": "00:59:16",
                    "text": " . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:16",
                    "end": "00:59:49",
                    "text": " And you can, of course, download the Dell Technologies Blueprint by scanning the QR code right here. It's got a lot of interesting insights. So go ahead, scan that QR code, and take a look at the Blueprint. Thank you, gentlemen. Thank you, everyone, for being an immensely participative audience here today. I'm sorry we could not have any time for audience questions, but appreciate you joining us here today. Thank you very much."
                }
            ]
        }
    ],
    "projectDeleted": false,
    "codeFrameColumn": [
        {
            "col1": "col1",
            "col2": "col3",
            "researchObjective": "",
            "filterColumn": [],
            "netSetting": true,
            "subNetSetting": true,
            "categorySetting": true,
            "sentimentSetting": true,
            "thinkingBudgetSetting": 2000,
            "questionType": "OPENEND",
            "otherBrand": false,
            "existingCodeFrame": [],
            "isList": false,
            "codeFrameWithDataList": [],
            "codeFrameList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 11672,
                        "CandidatesTokenCount": 2781,
                        "TotalTokenCount": 15960,
                        "thoughtsTokenCount": 1507,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Net": "India's AI Vision & Strategy",
                                "Subnets": [
                                    {
                                        "Subnet": "National Blueprint & Goals",
                                        "Categories": [
                                            {
                                                "Category": "Architecting India's AI Leadership Blueprint",
                                                "Summary": "Dell Technologies' blueprint for accelerating India's AI growth and leadership."
                                            },
                                            {
                                                "Category": "Viksit Bharat 2047 alignment",
                                                "Summary": "Positioning AI as a foundational engine to achieve India's vision of Viksit Bharat by 2047."
                                            },
                                            {
                                                "Category": "Human-centric AI (Manav)",
                                                "Summary": "Emphasizing AI development and deployment that prioritizes human well-being and citizen-centric approaches."
                                            }
                                        ],
                                        "Summary": "The overarching strategic documents and aspirations guiding India's AI journey."
                                    },
                                    {
                                        "Subnet": "Role of AI",
                                        "Categories": [
                                            {
                                                "Category": "Economic growth & productivity",
                                                "Summary": "AI drives economic growth, improves productivity, and modernizes public services."
                                            },
                                            {
                                                "Category": "Social empowerment & public services",
                                                "Summary": "Leveraging AI for social empowerment, improving education, healthcare, and citizen services."
                                            },
                                            {
                                                "Category": "Global leadership & strategic autonomy",
                                                "Summary": "AI as a driver for India's global leadership and strategic autonomy in technology."
                                            }
                                        ],
                                        "Summary": "The multifaceted impact and objectives of AI adoption in India."
                                    },
                                    {
                                        "Subnet": "Global Competitiveness & Autonomy",
                                        "Categories": [
                                            {
                                                "Category": "True AI creators vs. consumers",
                                                "Summary": "India's ambition to be a creator of AI value, not just a consumer of technology."
                                            },
                                            {
                                                "Category": "Trusted in India (vs. Made in India)",
                                                "Summary": "Emphasis on building AI systems that are not just locally developed but also trusted through robust governance."
                                            },
                                            {
                                                "Category": "Blueprint for other countries",
                                                "Summary": "India's AI model serving as a blueprint for other nations to avoid becoming 'digital colonies'."
                                            }
                                        ],
                                        "Summary": "How India aims to position itself on the global AI stage, focusing on creation, trust, and leadership."
                                    }
                                ],
                                "Summary": "The high-level strategic direction and overarching goals for AI development and adoption in India."
                            },
                            {
                                "Net": "AI Infrastructure & Investment",
                                "Subnets": [
                                    {
                                        "Subnet": "Compute & Data Foundations",
                                        "Categories": [
                                            {
                                                "Category": "Sovereign scalable compute",
                                                "Summary": "The need for national control over large-scale computing resources for AI."
                                            },
                                            {
                                                "Category": "Data centers (distributed, energy-efficient)",
                                                "Summary": "Developing a network of geographically distributed, sustainable, and energy-efficient data centers across states."
                                            },
                                            {
                                                "Category": "GPU availability (India AI Mission)",
                                                "Summary": "Government initiatives like India AI Mission to increase access to GPUs for research and development."
                                            }
                                        ],
                                        "Summary": "The essential hardware and data resources required for AI development and deployment."
                                    },
                                    {
                                        "Subnet": "Energy Infrastructure",
                                        "Categories": [
                                            {
                                                "Category": "Energy requirements for compute",
                                                "Summary": "Acknowledging that compute infrastructure is reliant on robust and sufficient energy supply."
                                            },
                                            {
                                                "Category": "Sustainability & efficiency",
                                                "Summary": "Focus on energy-efficient data centers to manage high energy consumption and environmental impact."
                                            }
                                        ],
                                        "Summary": "The critical role of energy in powering AI infrastructure and the emphasis on sustainable practices."
                                    },
                                    {
                                        "Subnet": "Cost & Accessibility",
                                        "Categories": [
                                            {
                                                "Category": "Affordable AI compute for MSMEs/startups",
                                                "Summary": "Ensuring that small and medium enterprises have access to reliable and affordable AI computing power."
                                            },
                                            {
                                                "Category": "GST waiver & tax benefits for infrastructure",
                                                "Summary": "Industry demands for policy interventions like GST waivers and income tax benefits to reduce upfront costs for AI infrastructure."
                                            },
                                            {
                                                "Category": "Cheapest compute facility (65 rupees/hour)",
                                                "Summary": "India's provision of highly affordable GPU access (e.g., 65 rupees/hour) to foster innovation."
                                            }
                                        ],
                                        "Summary": "Strategies and policies aimed at making AI infrastructure accessible and financially viable for a broad range of users."
                                    }
                                ],
                                "Summary": "The foundational requirements and financial considerations for building a robust AI ecosystem in India."
                            },
                            {
                                "Net": "Talent & Workforce Development",
                                "Subnets": [
                                    {
                                        "Subnet": "Skilling & Education Pipelines",
                                        "Categories": [
                                            {
                                                "Category": "Skilling from schools to workforce",
                                                "Summary": "Developing comprehensive skilling programs that span from foundational education to professional workforce training."
                                            },
                                            {
                                                "Category": "Academia-industry integration",
                                                "Summary": "Fostering closer collaboration between educational institutions and the industry to align curricula with market needs."
                                            },
                                            {
                                                "Category": "Future Skill Labs & Apprenticeships",
                                                "Summary": "Initiatives like Future Skill Labs and AI apprenticeships to accelerate practical AI expertise."
                                            }
                                        ],
                                        "Summary": "Developing the necessary skills across the population to support India's AI ambitions."
                                    },
                                    {
                                        "Subnet": "Developer Ecosystem",
                                        "Categories": [
                                            {
                                                "Category": "From users to developers (1 billion to 10 million)",
                                                "Summary": "Shifting focus from being a large consumer base to cultivating a massive pool of AI developers."
                                            },
                                            {
                                                "Category": "Leveraging open source",
                                                "Summary": "Utilizing open-source technologies to reduce development costs and foster collaboration in AI innovation."
                                            },
                                            {
                                                "Category": "Bringing back talent to India",
                                                "Summary": "Creating an environment that attracts and retains skilled Indian talent currently working abroad."
                                            }
                                        ],
                                        "Summary": "Cultivating a robust community of AI innovators and creators."
                                    },
                                    {
                                        "Subnet": "Job Impact & Creation",
                                        "Categories": [
                                            {
                                                "Category": "Post-AI jobs picture",
                                                "Summary": "Analyzing and strategizing for the evolving landscape of employment in an AI-driven economy."
                                            },
                                            {
                                                "Category": "Boosting jobs vs. taking jobs away",
                                                "Summary": "Addressing concerns about AI's potential to displace jobs and focusing on its capacity to create new employment opportunities."
                                            }
                                        ],
                                        "Summary": "Addressing the socio-economic implications of AI on employment and workforce transformation."
                                    }
                                ],
                                "Summary": "Strategies for nurturing human capital and ensuring a skilled workforce for the AI era."
                            },
                            {
                                "Net": "AI Governance & Trust",
                                "Subnets": [
                                    {
                                        "Subnet": "Regulatory Frameworks",
                                        "Categories": [
                                            {
                                                "Category": "Balance innovation & responsibility",
                                                "Summary": "Striking a critical balance between encouraging technological advancement and ensuring ethical, responsible AI development."
                                            },
                                            {
                                                "Category": "Agile regulations",
                                                "Summary": "Developing regulatory policies that can adapt quickly to the fast pace of AI technological change."
                                            },
                                            {
                                                "Category": "Zero Trust AI architecture",
                                                "Summary": "Implementing a security model where no entity is inherently trusted, requiring verification for every interaction within AI systems."
                                            }
                                        ],
                                        "Summary": "The policy and legal structures required to guide responsible AI development."
                                    },
                                    {
                                        "Subnet": "Data Privacy & Security",
                                        "Categories": [
                                            {
                                                "Category": "Data governance & privacy progress",
                                                "Summary": "Efforts to establish robust frameworks for managing data, including privacy regulations like DPDP."
                                            },
                                            {
                                                "Category": "Cybersecurity elements",
                                                "Summary": "Integrating strong cybersecurity measures to protect AI systems and the data they process."
                                            },
                                            {
                                                "Category": "India's openness to data sharing",
                                                "Summary": "Acknowledging Indian citizens' willingness to share data for public good, balanced with trust and safeguards."
                                            }
                                        ],
                                        "Summary": "Ensuring the protection of sensitive information and the integrity of AI systems."
                                    },
                                    {
                                        "Subnet": "Societal Trust & Inclusivity",
                                        "Categories": [
                                            {
                                                "Category": "Trust infrastructure (confidence, reliability)",
                                                "Summary": "Building a foundational sense of confidence and reliability in digital and AI systems among the populace."
                                            },
                                            {
                                                "Category": "Redress & grievance systems",
                                                "Summary": "Establishing mechanisms for citizens to report issues, seek recourse, and resolve problems related to AI applications."
                                            },
                                            {
                                                "Category": "Transparency & explainability",
                                                "Summary": "Ensuring that AI models and their outcomes are understandable and transparent to users and stakeholders."
                                            },
                                            {
                                                "Category": "Literacy & consumer awareness",
                                                "Summary": "Educating the public on how to use AI technologies effectively and safely, including understanding potential biases."
                                            }
                                        ],
                                        "Summary": "Fostering public acceptance and ensuring equitable benefits from AI adoption."
                                    }
                                ],
                                "Summary": "The ethical, legal, and social considerations for developing and deploying AI in India."
                            },
                            {
                                "Net": "Public-Private Partnership (PPP)",
                                "Subnets": [
                                    {
                                        "Subnet": "Collaboration Models",
                                        "Categories": [
                                            {
                                                "Category": "Public-private marriage of resources",
                                                "Summary": "Combining government resources with private sector innovation to maximize AI potential."
                                            },
                                            {
                                                "Category": "Industry-academia-government alignment",
                                                "Summary": "Fostering tighter collaboration between these three pillars for holistic AI ecosystem development."
                                            },
                                            {
                                                "Category": "Public resources with private innovation",
                                                "Summary": "Leveraging government-provided infrastructure and data with private sector creativity for scalable solutions."
                                            }
                                        ],
                                        "Summary": "Different forms of collaborative efforts between government, industry, and academia."
                                    },
                                    {
                                        "Subnet": "Scaling & Adoption Mechanisms",
                                        "Categories": [
                                            {
                                                "Category": "India AI Mission (GPUs)",
                                                "Summary": "Government-led initiative to provide access to GPUs for research and startups, accelerating AI adoption."
                                            },
                                            {
                                                "Category": "AI Cosh (datasets)",
                                                "Summary": "Creating platforms like AI Cosh to make thousands of datasets available for AI development and innovation."
                                            },
                                            {
                                                "Category": "UPI of AI (consistent API layer)",
                                                "Summary": "Vision for a standardized, accessible API layer for AI, akin to the UPI for digital payments, to democratize access."
                                            }
                                        ],
                                        "Summary": "Methods and initiatives to expand the reach and utilization of AI technologies across the nation."
                                    },
                                    {
                                        "Subnet": "Funding & Resource Sharing",
                                        "Categories": [
                                            {
                                                "Category": "Multi-billion dollar investments",
                                                "Summary": "Attracting substantial financial investments to fuel AI infrastructure development and innovation."
                                            },
                                            {
                                                "Category": "Subsidized GPU access",
                                                "Summary": "Providing affordable access to high-performance computing resources to reduce barriers for innovators."
                                            },
                                            {
                                                "Category": "Consolidated state applications & data lakes",
                                                "Summary": "Collaborating with state governments to centralize applications and create data lakes for AI analysis."
                                            }
                                        ],
                                        "Summary": "Financial and resource-based strategies for enabling large-scale AI deployment."
                                    }
                                ],
                                "Summary": "The essential role of collaborative models in driving AI innovation and adoption in India."
                            },
                            {
                                "Net": "Challenges & Solutions",
                                "Subnets": [
                                    {
                                        "Subnet": "Non-Technical Barriers",
                                        "Categories": [
                                            {
                                                "Category": "Trust (slippery concept, institutional development)",
                                                "Summary": "The inherent difficulty in defining and building societal trust, requiring strong institutional development."
                                            },
                                            {
                                                "Category": "Regional disparities (district level experience)",
                                                "Summary": "Addressing how AI initiatives are experienced differently across diverse regions and ensuring inclusivity."
                                            },
                                            {
                                                "Category": "Impact on jobs",
                                                "Summary": "The critical challenge of managing AI's potential impact on employment in a young, populous country."
                                            }
                                        ],
                                        "Summary": "Societal and systemic hurdles that impede widespread AI adoption and beneficial integration."
                                    },
                                    {
                                        "Subnet": "Enterprise Adoption Hurdles",
                                        "Categories": [
                                            {
                                                "Category": "Difficulty in imagining AI applications",
                                                "Summary": "Large organizations struggling to identify practical and impactful use cases for AI beyond basic chatbot functions."
                                            },
                                            {
                                                "Category": "Reluctance from large enterprises",
                                                "Summary": "Hesitation among larger companies to fully adopt and integrate AI, often due to perceived complexity or risk."
                                            },
                                            {
                                                "Category": "Scaling from pilots to production",
                                                "Summary": "The challenge of moving AI experiments and pilot projects into full-scale operational deployment and monetization."
                                            }
                                        ],
                                        "Summary": "Obstacles faced by businesses in implementing and scaling AI technologies."
                                    },
                                    {
                                        "Subnet": "Policy & Market Interventions",
                                        "Categories": [
                                            {
                                                "Category": "GST waiver for server imports",
                                                "Summary": "Proposal to waive Goods and Services Tax on server imports to reduce upfront infrastructure costs."
                                            },
                                            {
                                                "Category": "Income tax benefits for Indian service providers",
                                                "Summary": "Suggestion to offer tax incentives to domestic AI service providers to foster local growth."
                                            },
                                            {
                                                "Category": "Removing bottlenecks for GPU deployment",
                                                "Summary": "Addressing logistical and financial constraints to increase the availability and deployment of GPUs."
                                            }
                                        ],
                                        "Summary": "Specific policy recommendations and market adjustments to accelerate AI growth."
                                    }
                                ],
                                "Summary": "Identification of key obstacles in India's AI journey and proposed strategies to overcome them."
                            },
                            {
                                "Net": "996 Non specific TOPIC",
                                "Subnets": [
                                    {
                                        "Subnet": "Irrelevant/Uncategorized Responses",
                                        "Categories": [
                                            {
                                                "Category": "General Irrelevant Content",
                                                "Summary": "Responses that are not directly relevant to the survey questions or established themes, requiring further review for potential upcoding."
                                            }
                                        ],
                                        "Summary": "Responses that do not fit into the established thematic codeframe but are not gibberish or explicit 'don't know' type responses."
                                    }
                                ],
                                "Summary": "This category is for responses that are irrelevant to the core themes but are not outright gibberish or non-committal."
                            },
                            {
                                "Net": "997 Don’t Know Can’t Say",
                                "Subnets": [
                                    {
                                        "Subnet": "Non-Committal Responses",
                                        "Categories": [
                                            {
                                                "Category": "Vague or Undecided",
                                                "Summary": "Responses indicating uncertainty, lack of knowledge, or an inability to provide a definitive answer."
                                            }
                                        ],
                                        "Summary": "Responses that express uncertainty, indecision, or a lack of opinion."
                                    }
                                ],
                                "Summary": "This category includes all non-committal and vague responses."
                            },
                            {
                                "Net": "998 Not Applicable",
                                "Subnets": [
                                    {
                                        "Subnet": "Irrelevant to Question/Project",
                                        "Categories": [
                                            {
                                                "Category": "Not Applicable Content",
                                                "Summary": "Responses that are clearly not related to the specific question asked or the overall project scope."
                                            }
                                        ],
                                        "Summary": "Responses that are out of scope or not applicable to the context of the survey or project."
                                    }
                                ],
                                "Summary": "This category is for responses that are clearly irrelevant or not applicable to the survey question."
                            },
                            {
                                "Net": "999 None-Nothing",
                                "Subnets": [
                                    {
                                        "Subnet": "Gibberish/Meaningless Content",
                                        "Categories": [
                                            {
                                                "Category": "Non-Language Characters/Gibberish",
                                                "Summary": "Responses consisting of random characters, symbols, or incomprehensible text."
                                            },
                                            {
                                                "Category": "Empty or Null Response",
                                                "Summary": "Responses that are blank, 'none', 'nothing', or indicate no meaningful input."
                                            }
                                        ],
                                        "Summary": "Responses that are nonsensical, unintelligible, or completely empty."
                                    }
                                ],
                                "Summary": "This category is for responses that are plain gibberish, non-language characters, or contain no meaningful content."
                            }
                        ],
                        "Summary": "This document summarizes a conference discussion on India's AI strategy, highlighting a blueprint for transformation focusing on three pillars: Invest, Innovate, and Evolve. Key themes include the need for significant investment in sovereign, scalable compute and energy infrastructure, fostering innovation through a future-ready workforce and collaboration, and evolving into a responsible, agile, security-first governance structure. The discussion emphasizes the role of public-private partnerships, addressing non-technical barriers like trust and job impact, and the importance of democratizing AI access for startups and MSMEs while maintaining global competitiveness and inclusive growth."
                    }
                }
            ],
            "onlyCodeFrame": true,
            "errorList": [],
            "codeFrameGenerateType": "CODE_FRAME_ONLY",
            "requestType": "CODEFRAME",
            "requestMode": "DEVELOPMENT",
            "sourceLanguage": "English",
            "targetLanguage": "English",
            "response": {
                "status": "OK",
                "data": {
                    "status": "OK",
                    "PromptTokenCount": 11672,
                    "CandidatesTokenCount": 2781,
                    "TotalTokenCount": 15960,
                    "thoughtsTokenCount": 1507,
                    "finishReason": "STOP",
                    "data": [
                        {
                            "Net": "India's AI Vision & Strategy",
                            "Subnets": [
                                {
                                    "Subnet": "National Blueprint & Goals",
                                    "Categories": [
                                        {
                                            "Category": "Architecting India's AI Leadership Blueprint",
                                            "Summary": "Dell Technologies' blueprint for accelerating India's AI growth and leadership.",
                                            "code": 3
                                        },
                                        {
                                            "Category": "Viksit Bharat 2047 alignment",
                                            "Summary": "Positioning AI as a foundational engine to achieve India's vision of Viksit Bharat by 2047.",
                                            "code": 4
                                        },
                                        {
                                            "Category": "Human-centric AI (Manav)",
                                            "Summary": "Emphasizing AI development and deployment that prioritizes human well-being and citizen-centric approaches.",
                                            "code": 5
                                        }
                                    ],
                                    "Summary": "The overarching strategic documents and aspirations guiding India's AI journey.",
                                    "code": 2
                                },
                                {
                                    "Subnet": "Role of AI",
                                    "Categories": [
                                        {
                                            "Category": "Economic growth & productivity",
                                            "Summary": "AI drives economic growth, improves productivity, and modernizes public services.",
                                            "code": 7
                                        },
                                        {
                                            "Category": "Social empowerment & public services",
                                            "Summary": "Leveraging AI for social empowerment, improving education, healthcare, and citizen services.",
                                            "code": 8
                                        },
                                        {
                                            "Category": "Global leadership & strategic autonomy",
                                            "Summary": "AI as a driver for India's global leadership and strategic autonomy in technology.",
                                            "code": 9
                                        }
                                    ],
                                    "Summary": "The multifaceted impact and objectives of AI adoption in India.",
                                    "code": 6
                                },
                                {
                                    "Subnet": "Global Competitiveness & Autonomy",
                                    "Categories": [
                                        {
                                            "Category": "True AI creators vs. consumers",
                                            "Summary": "India's ambition to be a creator of AI value, not just a consumer of technology.",
                                            "code": 11
                                        },
                                        {
                                            "Category": "Trusted in India (vs. Made in India)",
                                            "Summary": "Emphasis on building AI systems that are not just locally developed but also trusted through robust governance.",
                                            "code": 12
                                        },
                                        {
                                            "Category": "Blueprint for other countries",
                                            "Summary": "India's AI model serving as a blueprint for other nations to avoid becoming 'digital colonies'.",
                                            "code": 13
                                        }
                                    ],
                                    "Summary": "How India aims to position itself on the global AI stage, focusing on creation, trust, and leadership.",
                                    "code": 10
                                }
                            ],
                            "Summary": "The high-level strategic direction and overarching goals for AI development and adoption in India.",
                            "code": 1
                        },
                        {
                            "Net": "AI Infrastructure & Investment",
                            "Subnets": [
                                {
                                    "Subnet": "Compute & Data Foundations",
                                    "Categories": [
                                        {
                                            "Category": "Sovereign scalable compute",
                                            "Summary": "The need for national control over large-scale computing resources for AI.",
                                            "code": 16
                                        },
                                        {
                                            "Category": "Data centers (distributed, energy-efficient)",
                                            "Summary": "Developing a network of geographically distributed, sustainable, and energy-efficient data centers across states.",
                                            "code": 17
                                        },
                                        {
                                            "Category": "GPU availability (India AI Mission)",
                                            "Summary": "Government initiatives like India AI Mission to increase access to GPUs for research and development.",
                                            "code": 18
                                        }
                                    ],
                                    "Summary": "The essential hardware and data resources required for AI development and deployment.",
                                    "code": 15
                                },
                                {
                                    "Subnet": "Energy Infrastructure",
                                    "Categories": [
                                        {
                                            "Category": "Energy requirements for compute",
                                            "Summary": "Acknowledging that compute infrastructure is reliant on robust and sufficient energy supply.",
                                            "code": 20
                                        },
                                        {
                                            "Category": "Sustainability & efficiency",
                                            "Summary": "Focus on energy-efficient data centers to manage high energy consumption and environmental impact.",
                                            "code": 21
                                        }
                                    ],
                                    "Summary": "The critical role of energy in powering AI infrastructure and the emphasis on sustainable practices.",
                                    "code": 19
                                },
                                {
                                    "Subnet": "Cost & Accessibility",
                                    "Categories": [
                                        {
                                            "Category": "Affordable AI compute for MSMEs/startups",
                                            "Summary": "Ensuring that small and medium enterprises have access to reliable and affordable AI computing power.",
                                            "code": 23
                                        },
                                        {
                                            "Category": "GST waiver & tax benefits for infrastructure",
                                            "Summary": "Industry demands for policy interventions like GST waivers and income tax benefits to reduce upfront costs for AI infrastructure.",
                                            "code": 24
                                        },
                                        {
                                            "Category": "Cheapest compute facility (65 rupees/hour)",
                                            "Summary": "India's provision of highly affordable GPU access (e.g., 65 rupees/hour) to foster innovation.",
                                            "code": 25
                                        }
                                    ],
                                    "Summary": "Strategies and policies aimed at making AI infrastructure accessible and financially viable for a broad range of users.",
                                    "code": 22
                                }
                            ],
                            "Summary": "The foundational requirements and financial considerations for building a robust AI ecosystem in India.",
                            "code": 14
                        },
                        {
                            "Net": "Talent & Workforce Development",
                            "Subnets": [
                                {
                                    "Subnet": "Skilling & Education Pipelines",
                                    "Categories": [
                                        {
                                            "Category": "Skilling from schools to workforce",
                                            "Summary": "Developing comprehensive skilling programs that span from foundational education to professional workforce training.",
                                            "code": 28
                                        },
                                        {
                                            "Category": "Academia-industry integration",
                                            "Summary": "Fostering closer collaboration between educational institutions and the industry to align curricula with market needs.",
                                            "code": 29
                                        },
                                        {
                                            "Category": "Future Skill Labs & Apprenticeships",
                                            "Summary": "Initiatives like Future Skill Labs and AI apprenticeships to accelerate practical AI expertise.",
                                            "code": 30
                                        }
                                    ],
                                    "Summary": "Developing the necessary skills across the population to support India's AI ambitions.",
                                    "code": 27
                                },
                                {
                                    "Subnet": "Developer Ecosystem",
                                    "Categories": [
                                        {
                                            "Category": "From users to developers (1 billion to 10 million)",
                                            "Summary": "Shifting focus from being a large consumer base to cultivating a massive pool of AI developers.",
                                            "code": 32
                                        },
                                        {
                                            "Category": "Leveraging open source",
                                            "Summary": "Utilizing open-source technologies to reduce development costs and foster collaboration in AI innovation.",
                                            "code": 33
                                        },
                                        {
                                            "Category": "Bringing back talent to India",
                                            "Summary": "Creating an environment that attracts and retains skilled Indian talent currently working abroad.",
                                            "code": 34
                                        }
                                    ],
                                    "Summary": "Cultivating a robust community of AI innovators and creators.",
                                    "code": 31
                                },
                                {
                                    "Subnet": "Job Impact & Creation",
                                    "Categories": [
                                        {
                                            "Category": "Post-AI jobs picture",
                                            "Summary": "Analyzing and strategizing for the evolving landscape of employment in an AI-driven economy.",
                                            "code": 36
                                        },
                                        {
                                            "Category": "Boosting jobs vs. taking jobs away",
                                            "Summary": "Addressing concerns about AI's potential to displace jobs and focusing on its capacity to create new employment opportunities.",
                                            "code": 37
                                        }
                                    ],
                                    "Summary": "Addressing the socio-economic implications of AI on employment and workforce transformation.",
                                    "code": 35
                                }
                            ],
                            "Summary": "Strategies for nurturing human capital and ensuring a skilled workforce for the AI era.",
                            "code": 26
                        },
                        {
                            "Net": "AI Governance & Trust",
                            "Subnets": [
                                {
                                    "Subnet": "Regulatory Frameworks",
                                    "Categories": [
                                        {
                                            "Category": "Balance innovation & responsibility",
                                            "Summary": "Striking a critical balance between encouraging technological advancement and ensuring ethical, responsible AI development.",
                                            "code": 40
                                        },
                                        {
                                            "Category": "Agile regulations",
                                            "Summary": "Developing regulatory policies that can adapt quickly to the fast pace of AI technological change.",
                                            "code": 41
                                        },
                                        {
                                            "Category": "Zero Trust AI architecture",
                                            "Summary": "Implementing a security model where no entity is inherently trusted, requiring verification for every interaction within AI systems.",
                                            "code": 42
                                        }
                                    ],
                                    "Summary": "The policy and legal structures required to guide responsible AI development.",
                                    "code": 39
                                },
                                {
                                    "Subnet": "Data Privacy & Security",
                                    "Categories": [
                                        {
                                            "Category": "Data governance & privacy progress",
                                            "Summary": "Efforts to establish robust frameworks for managing data, including privacy regulations like DPDP.",
                                            "code": 44
                                        },
                                        {
                                            "Category": "Cybersecurity elements",
                                            "Summary": "Integrating strong cybersecurity measures to protect AI systems and the data they process.",
                                            "code": 45
                                        },
                                        {
                                            "Category": "India's openness to data sharing",
                                            "Summary": "Acknowledging Indian citizens' willingness to share data for public good, balanced with trust and safeguards.",
                                            "code": 46
                                        }
                                    ],
                                    "Summary": "Ensuring the protection of sensitive information and the integrity of AI systems.",
                                    "code": 43
                                },
                                {
                                    "Subnet": "Societal Trust & Inclusivity",
                                    "Categories": [
                                        {
                                            "Category": "Trust infrastructure (confidence, reliability)",
                                            "Summary": "Building a foundational sense of confidence and reliability in digital and AI systems among the populace.",
                                            "code": 48
                                        },
                                        {
                                            "Category": "Redress & grievance systems",
                                            "Summary": "Establishing mechanisms for citizens to report issues, seek recourse, and resolve problems related to AI applications.",
                                            "code": 49
                                        },
                                        {
                                            "Category": "Transparency & explainability",
                                            "Summary": "Ensuring that AI models and their outcomes are understandable and transparent to users and stakeholders.",
                                            "code": 50
                                        },
                                        {
                                            "Category": "Literacy & consumer awareness",
                                            "Summary": "Educating the public on how to use AI technologies effectively and safely, including understanding potential biases.",
                                            "code": 51
                                        }
                                    ],
                                    "Summary": "Fostering public acceptance and ensuring equitable benefits from AI adoption.",
                                    "code": 47
                                }
                            ],
                            "Summary": "The ethical, legal, and social considerations for developing and deploying AI in India.",
                            "code": 38
                        },
                        {
                            "Net": "Public-Private Partnership (PPP)",
                            "Subnets": [
                                {
                                    "Subnet": "Collaboration Models",
                                    "Categories": [
                                        {
                                            "Category": "Public-private marriage of resources",
                                            "Summary": "Combining government resources with private sector innovation to maximize AI potential.",
                                            "code": 54
                                        },
                                        {
                                            "Category": "Industry-academia-government alignment",
                                            "Summary": "Fostering tighter collaboration between these three pillars for holistic AI ecosystem development.",
                                            "code": 55
                                        },
                                        {
                                            "Category": "Public resources with private innovation",
                                            "Summary": "Leveraging government-provided infrastructure and data with private sector creativity for scalable solutions.",
                                            "code": 56
                                        }
                                    ],
                                    "Summary": "Different forms of collaborative efforts between government, industry, and academia.",
                                    "code": 53
                                },
                                {
                                    "Subnet": "Scaling & Adoption Mechanisms",
                                    "Categories": [
                                        {
                                            "Category": "India AI Mission (GPUs)",
                                            "Summary": "Government-led initiative to provide access to GPUs for research and startups, accelerating AI adoption.",
                                            "code": 58
                                        },
                                        {
                                            "Category": "AI Cosh (datasets)",
                                            "Summary": "Creating platforms like AI Cosh to make thousands of datasets available for AI development and innovation.",
                                            "code": 59
                                        },
                                        {
                                            "Category": "UPI of AI (consistent API layer)",
                                            "Summary": "Vision for a standardized, accessible API layer for AI, akin to the UPI for digital payments, to democratize access.",
                                            "code": 60
                                        }
                                    ],
                                    "Summary": "Methods and initiatives to expand the reach and utilization of AI technologies across the nation.",
                                    "code": 57
                                },
                                {
                                    "Subnet": "Funding & Resource Sharing",
                                    "Categories": [
                                        {
                                            "Category": "Multi-billion dollar investments",
                                            "Summary": "Attracting substantial financial investments to fuel AI infrastructure development and innovation.",
                                            "code": 62
                                        },
                                        {
                                            "Category": "Subsidized GPU access",
                                            "Summary": "Providing affordable access to high-performance computing resources to reduce barriers for innovators.",
                                            "code": 63
                                        },
                                        {
                                            "Category": "Consolidated state applications & data lakes",
                                            "Summary": "Collaborating with state governments to centralize applications and create data lakes for AI analysis.",
                                            "code": 64
                                        }
                                    ],
                                    "Summary": "Financial and resource-based strategies for enabling large-scale AI deployment.",
                                    "code": 61
                                }
                            ],
                            "Summary": "The essential role of collaborative models in driving AI innovation and adoption in India.",
                            "code": 52
                        },
                        {
                            "Net": "Challenges & Solutions",
                            "Subnets": [
                                {
                                    "Subnet": "Non-Technical Barriers",
                                    "Categories": [
                                        {
                                            "Category": "Trust (slippery concept, institutional development)",
                                            "Summary": "The inherent difficulty in defining and building societal trust, requiring strong institutional development.",
                                            "code": 67
                                        },
                                        {
                                            "Category": "Regional disparities (district level experience)",
                                            "Summary": "Addressing how AI initiatives are experienced differently across diverse regions and ensuring inclusivity.",
                                            "code": 68
                                        },
                                        {
                                            "Category": "Impact on jobs",
                                            "Summary": "The critical challenge of managing AI's potential impact on employment in a young, populous country.",
                                            "code": 69
                                        }
                                    ],
                                    "Summary": "Societal and systemic hurdles that impede widespread AI adoption and beneficial integration.",
                                    "code": 66
                                },
                                {
                                    "Subnet": "Enterprise Adoption Hurdles",
                                    "Categories": [
                                        {
                                            "Category": "Difficulty in imagining AI applications",
                                            "Summary": "Large organizations struggling to identify practical and impactful use cases for AI beyond basic chatbot functions.",
                                            "code": 71
                                        },
                                        {
                                            "Category": "Reluctance from large enterprises",
                                            "Summary": "Hesitation among larger companies to fully adopt and integrate AI, often due to perceived complexity or risk.",
                                            "code": 72
                                        },
                                        {
                                            "Category": "Scaling from pilots to production",
                                            "Summary": "The challenge of moving AI experiments and pilot projects into full-scale operational deployment and monetization.",
                                            "code": 73
                                        }
                                    ],
                                    "Summary": "Obstacles faced by businesses in implementing and scaling AI technologies.",
                                    "code": 70
                                },
                                {
                                    "Subnet": "Policy & Market Interventions",
                                    "Categories": [
                                        {
                                            "Category": "GST waiver for server imports",
                                            "Summary": "Proposal to waive Goods and Services Tax on server imports to reduce upfront infrastructure costs.",
                                            "code": 75
                                        },
                                        {
                                            "Category": "Income tax benefits for Indian service providers",
                                            "Summary": "Suggestion to offer tax incentives to domestic AI service providers to foster local growth.",
                                            "code": 76
                                        },
                                        {
                                            "Category": "Removing bottlenecks for GPU deployment",
                                            "Summary": "Addressing logistical and financial constraints to increase the availability and deployment of GPUs.",
                                            "code": 77
                                        }
                                    ],
                                    "Summary": "Specific policy recommendations and market adjustments to accelerate AI growth.",
                                    "code": 74
                                }
                            ],
                            "Summary": "Identification of key obstacles in India's AI journey and proposed strategies to overcome them.",
                            "code": 65
                        }
                    ],
                    "Summary": "This document summarizes a conference discussion on India's AI strategy, highlighting a blueprint for transformation focusing on three pillars: Invest, Innovate, and Evolve. Key themes include the need for significant investment in sovereign, scalable compute and energy infrastructure, fostering innovation through a future-ready workforce and collaboration, and evolving into a responsible, agile, security-first governance structure. The discussion emphasizes the role of public-private partnerships, addressing non-technical barriers like trust and job impact, and the importance of democratizing AI access for startups and MSMEs while maintaining global competitiveness and inclusive growth."
                }
            },
            "summaryList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 11333,
                        "CandidatesTokenCount": 1546,
                        "TotalTokenCount": 14144,
                        "thoughtsTokenCount": 1265,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Summary": "<h2>Comprehensive Synthesis Report: Architecting India's AI Leadership</h2><h3>1. Introduction and Vision for India's AI Leadership</h3><p>The overarching goal of this discourse, aligning with the summit's objective, is to bridge the global AI divide, positioning India for significant economic growth, social empowerment, and global leadership through AI. This initiative is not merely a presentation but a concerted call to action, aiming to frame the execution pathway for AI adoption and scaling from an industry vantage point. India stands at the precipice of profound transformation and progress driven by AI, with aspirations that are not only bold for the nation but also globally ambitious. Dell Technologies, with over 30 years of presence and partnership with Indian government agencies and the broader ecosystem, has leveraged its global expertise and deep understanding of India to craft an 'AI blueprint'. This blueprint serves as a practical guide for both the country and individual companies to harness this immense opportunity, advancing India's vision of Viksit Bharat 2047 by establishing AI as a foundational engine for productivity, modernized public services, opportunity expansion, and strategic autonomy.</p><p>The expected growth in compute on the back of AI in India is projected to exceed 10 exaflops, with AI workloads growing at a compound annual growth rate of over 30% in the coming years. This growth underscores the urgent need for a structured approach to AI adoption, emphasizing a human-centric philosophy, as articulated through principles like 'Manav', to ensure technology serves the people.</p><h3>2. Dell Technologies AI Blueprint: Pillars of Transformation</h3><p>The Dell Technologies AI Blueprint for accelerating India's AI growth is structured around three critical pillars: <strong>Invest, Innovate, and Evolve</strong>, all underpinned by the necessity of Public-Private Partnerships (PPP).</p><h4>2.1. Invest: Sovereign, Scalable Compute and Data Foundations</h4><p>The first pillar emphasizes substantial investment in core infrastructure. This includes:</p><ul><li><strong>Compute Infrastructure:</strong> A country needs to establish robust compute infrastructure, ensuring accessibility for all, including MSMEs, which often lack the capacity to build their own. There is a recognized need for significantly more GPU infrastructure, with targets for the India AI Mission far exceeding initial estimates, and an ultimate goal of achieving one lakh (100,000) GPUs threefold by the end of the year.</li><li><strong>Energy Infrastructure:</strong> Integral to compute, sufficient energy infrastructure is crucial to power the burgeoning AI ecosystem.</li><li><strong>Financial Investment:</strong> Multiple billions of dollars of investment are required to build this infrastructure, necessitating facilitation of capital inflow into the country.</li><li><strong>Distributed Data Centers:</strong> A model involving distributed data centers across multiple states (e.g., NextGen Cloud Technologies planning 100 megawatts across six states) is proposed to bring compute closer to users, particularly in areas like education, healthcare, and citizen services. This distribution, coupled with strong interconnection networks (leveraging existing railway and power networks), will democratize access and enable wide-scale AI application.</li></ul><h4>2.2. Innovate: Future-Ready Workforce and Collaboration</h4><p>The second pillar focuses on fostering an environment of innovation:</p><ul><li><strong>Skilling:</strong> A key aspect is developing a future-ready workforce through comprehensive skilling initiatives across all educational levels – from schools to colleges and into the existing workforce. This requires innovative approaches to skill delivery through various mediums.</li><li><strong>Collaboration:</strong> Tighter alignment and collaboration between policymakers, industry, academia, and institutional capacities are essential to convert ambition into nation-scale execution.</li><li><strong>Leveraging Open Source:</strong> Utilizing open-source technologies can significantly reduce the cost of compute, making AI more accessible and palatable for a broader segment of Indian citizens, beyond the urban elite.</li><li><strong>Developer-Centric Approach:</strong> The transition from 1 billion users to 1-10 million developers is crucial for creating value, emphasizing the need to nurture the largest talent pool residing within the country.</li></ul><h4>2.3. Evolve: Responsible, Agile, Security-First Governance</h4><p>The third pillar centers on establishing a robust and trustworthy governance framework:</p><ul><li><strong>Regulatory Framework:</strong> Governments must implement regulatory frameworks that strike a fundamental balance between innovation and responsibility, with responsibility as the anchor. These regulations must be agile to keep pace with rapidly advancing AI technologies, avoiding frameworks tied to outdated technologies.</li><li><strong>Trust Infrastructure:</strong> This non-technical yet critical element involves building confidence among users. Key components include data governance, privacy, security, transparency (especially challenging with AI), redress and grievance systems, and digital literacy. India, despite its high inherent trust in digital systems, needs to strengthen its institutional side of trust.</li><li><strong>Zero-Trust AI Architecture:</strong> Practically, this entails verifying every protocol in design, extending to data, AI models, usability, cybersecurity elements, and identity and access management at a national level. This includes segmented and anonymized data sets for researchers and innovators, audit trails for AI models, and consumer awareness campaigns to ensure verifiable and legal content.</li><li><strong>Sustainability:</strong> Energy efficiency and sustainability in data centers and new architectural models are becoming paramount to optimize energy resource usage.</li></ul><h3>3. Challenges, Bottlenecks, and Actionable Recommendations</h3><h4>3.1. Technical and Infrastructure Bottlenecks</h4><p><strong>Challenge:</strong> Access to reliable, affordable AI compute and cloud at scale remains a significant barrier for MSMEs and startups, despite government initiatives like India AI Mission offering subsidized GPU access. The sheer demand outstrips current supply, and financial investment remains a hurdle (Raj Gopal).</p><p><strong>Recommendations:</strong></p><ul><li><strong>Policy Interventions:</strong> Consider waiving GST upfront on server imports, collecting it only when services are delivered, to reduce initial infrastructure costs (by about 18%).</li><li><strong>Tax Benefits:</strong> Extend income tax benefits to Indian service providers for domestic services, akin to those offered for global services hosted in India, to incentivize in-country development.</li><li><strong>Increased Investment:</strong> Facilitate multi-billion dollar investments into the country to rapidly expand GPU and data center infrastructure.</li></ul><h4>3.2. Non-Technical Barriers and Trust Infrastructure</h4><p><strong>Challenge:</strong> Beyond technical aspects, non-technical bottlenecks, particularly regarding trust, can limit AI's momentum and societal good. While India has high innate trust in digital systems, institutional aspects like data governance, privacy, and security are still developing. The regional disparity in trust exercise (e.g., district level vs. union level policies) and a lack of transparency and redressal mechanisms hinder broader adoption (Professor Bhaskar Chakravarthy).</p><p><strong>Recommendations:</strong></p><ul><li><strong>Strengthen Institutional Capacity:</strong> Develop institutional infrastructure specifically tailored for key Indian use cases like agriculture, skilling, and healthcare, where the impact is societal rather than purely revenue-driven.</li><li><strong>Farmer-Centric AI:</strong> For agriculture, build trust by demonstrating practical outcomes (e.g., pest reduction using simple AI on phones) and providing simple, explainable interfaces.</li><li><strong>Healthcare Trust:</strong> Address concerns about data privacy and system reliability for healthcare applications, building foundational trust elements to encourage adoption.</li><li><strong>Literacy and Understanding:</strong> Enhance digital literacy to enable people to understand and use technology effectively while protecting themselves.</li><li><strong>"
                            }
                        ]
                    }
                }
            ]
        }
    ],
    "headerCols": {
        "col1": "Part Id",
        "col2": "FileName",
        "col3": "Content",
        "col4": "Status"
    },
    "fileName": "",
    "projectType": "VIDEO",
    "sourceLanguage": "English",
    "targetLanguage": "English",
    "__v": 0
},
    {
    "_id": "6998404a6c61c5bef9ef7f11",
    "projectId": "110743",
    "projectName": "AI Impact Summit 2026 l Keynote Speakers - English",
    "projectDescription": "AI Impact Summit 2026 l Keynote Speakers - English",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-20T11:06:44.322Z",
    "projectUpdatedDate": "2026-02-20T11:06:44.322Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-20T11:46:47.151Z",
    "projectDeleted": false,
    "projectDataLength": 9,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_3",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_4",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_5",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_6",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_7",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_8",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
    {
    "_id": "699842606c61c5bef9ef7f49",
    "projectId": "880679",
    "projectName": "AI Impact Summit 2026 l Keynote Speakers - Spanish",
    "projectDescription": "AI Impact Summit 2026 l Keynote Speakers - Spanish ",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-20T11:15:41.717Z",
    "projectUpdatedDate": "2026-02-20T11:15:41.717Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-20T11:49:38.871Z",
    "projectDeleted": false,
    "projectDataLength": 9,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_3",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_4",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_5",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_6",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_7",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_8",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Spanish"
},
   {
    "_id": "6998414c6c61c5bef9ef7f27",
    "projectId": "228050",
    "projectName": "AI Impact Summit 2026 l Keynote Speakers - Hindi",
    "projectDescription": "AI Impact Summit 2026 l Keynote Speakers - Hindi",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-20T11:11:06.400Z",
    "projectUpdatedDate": "2026-02-20T11:11:06.400Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-20T12:29:34.078Z",
    "projectDeleted": false,
    "projectDataLength": 9,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_3",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_4",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_5",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_6",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_7",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_8",
            "col2": "AI Impact Summit 2026 l Keynote Speakers - English.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Hindi"
},
 {
    "_id": "69982d8a6c61c5bef9ef7e27",
    "projectId": "448244",
    "projectName": "How Multilingual AI is Making Information Accessible to Everyone",
    "projectDescription": "How Multilingual AI is Making Information Accessible to Everyone",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-20T09:46:49.718Z",
    "projectUpdatedDate": "2026-02-20T09:46:49.718Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-20T10:01:14.192Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "How Multilingual AI Bridges the Gap to Inclusive Access(MP3_128K).mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771580798145.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:18",
                    "text": " as a bridge to democratic access. Switzerland is very pleased to contribute to this global conversation at a pivotal time, a pivotal moment for responsible AI. Our message, which was supposed to be delivered by our president, is very clear."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:18",
                    "end": "00:00:37",
                    "text": " AI can only serve the public good if it serves all languages and all cultures. Today, linguistic exclusion remains one of the most persistent barriers to digital participation, ensuring multilingual access is therefore not only a technical challenge, it's a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:37",
                    "end": "00:01:01",
                    "text": " democratic imperative. This discussion forms part of the international arc that began with the Paris 2025 public interest AI process, continues here at India AI Summit 2026, and will advance further when Switzerland will happily host the Geneva AI Summit in 2027."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:01",
                    "end": "00:01:22",
                    "text": " Our shared objective is continuity, cooperation, and genuinely global approach to AI governance. Switzerland is proud that this session brings together partners who embody open and collaborative innovation. India's Parshini Initiative, current AI,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:22",
                    "end": "00:01:44",
                    "text": " that emerged from the French AI Summit, and then many partners from the broader network of academic and policy institutions of ICANN, the International Computation and AI Network, such partners as ELIS, NTU, Singapore, and of course the Swiss partners ETH and EPFL."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:44",
                    "end": "00:02:11",
                    "text": " ICANN really reflects Switzerland's commitment to equitable access to compute data and multilingual models. A notable example is Apertus, which maybe many of you have heard of. It was developed by ETH Zurich and EPFL, fully open and transparent multilingual model designed to support public interest applications across diverse linguistic communities."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:11",
                    "end": "00:02:36",
                    "text": " As we prepare for Geneva 2027, Switzerland views multilingual AI as a foundation for inclusive digital public services and for strengthening participation across societies. Allow me to briefly, just very briefly outline today's agenda. We will begin with the announcement of the launch of the three new joint calls under the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:36",
                    "end": "00:03:05",
                    "text": " lead of the Indo-Swiss Joint Research Program, JRP, which is making a further strengthening of our bilateral ties in science, innovation, and research between Switzerland and India. This will be followed by a panel discussion with distinguished international guests, and I'm very happy to announce that this will be moderated by my colleague, Nina Frey, the executive director of ICAIN."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:05",
                    "end": "00:03:40",
                    "text": " Thank you so much for attending. I will hand over the floor to the next speaker, Professor Thorsten Svede, president of the Swiss National Science Foundation. Very warm welcome. Thank you. Your excellencies, ladies and gentlemen. Namaste. It's my great pleasure to be here today with you. It's a moment to highlight a particularly exciting moment in the Indo-Swiss research collaboration. As many of you know,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:40",
                    "end": "00:04:05",
                    "text": " Switzerland and India have long-standing trusted partnership in research built on reciprocity, on joint excellence and on shared priorities. Today this collaboration is stronger than ever and I'm delighted to announce three new calls for joint research projects as well as the launch of our new Indo-Swiss research framework program between the Swiss National Science Foundation and our Indian partner organizations."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:05",
                    "end": "00:04:34",
                    "text": " This is a really remarkable convergence that underscores both the depths and the breadth of our bilateral engagements. The three calls for joint research programs span a very diverse range of disciplines and are designed to foster cutting-edge, high-impact research. The first two calls that we launched earlier this year are in the geosciences and in the social sciences. Together with the Indian Ministry of Earth Sciences, we're inviting proposals"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:34",
                    "end": "00:05:05",
                    "text": " on natural hazards in mountain regions. A field of great relevance for both our countries as we are each facing very unique geological challenges. In parallel, our call with the Indian Council of Social Science Research opens the door for joint projects on pressing social and societal questions. Again, strengthening our collaboration in a domain where cross-cultural perspectives are significantly enriching the research outcomes."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:05",
                    "end": "00:05:35",
                    "text": " Two weeks ago, the Swiss National Science Foundation together with the Indian Department of Biotechnology and the Indian Council of Medical Research launched a third call focused on One Health, a topic of real global urgency. This One Health call is particularly important for us. It reflects many months of preparations and close coordination with our Indian partners and embodies a holistic approach needed to understand"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:35",
                    "end": "00:06:02",
                    "text": " interconnected health of humans, animals and the environment. The challenges we face in this area know no borders and international collaboration is indispensable. We therefore anticipate a very high uptake and interest and participation of researchers in both our communities. Taken together, these three simultaneous calls represent an exceptional moment in industry's research cooperation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:03",
                    "end": "00:06:31",
                    "text": " They showcase our commitment enabling ambitious science from fundamental research questions in the natural and the life sciences to complex issues shaped by society, geography or technology. And with each call we reaffirm our shared belief that long-term co-created research is the key to addressing the major challenges of our times. So building on these strong foundations now is the right moment."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:31",
                    "end": "00:06:58",
                    "text": " to announce a new strategic long-term collaboration, the Indo-Swiss Research Framework Program between the SNSF and our Indian partner organizations. We aim to create a program in which all researchers wishing to contribute to the Indo-Swiss cooperation can find appropriate support. Thematic calls on strategic areas will be launched together with our Indian partnering partners and remain at the core of this program."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:58",
                    "end": "00:07:28",
                    "text": " And to this audience, it might not come as a real surprise that one of the high priority topics we are currently considering is artificial intelligence. In addition to these bilateral and multilateral calls, I'm also pleased to announce that we are launching several new measures and funding schemes to support collaborative research. With our brand new Explore, Experiment and Expand grants, we want to give consortia the opportunities to explore"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:28",
                    "end": "00:07:48",
                    "text": " new collaborations, new networks, new partnerships. We want to allow them to experiment with blue sky thinking topics and methodologies that haven't been tried before. But we also want to allow them to expand on already established functional collaborations and build them in an innovative way into the future."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:48",
                    "end": "00:08:17",
                    "text": " We're also increasing mobility funding for existing consortia to make sure that every project we fund by our program can lead to a durable collaboration, impactful events that connect with the wider society, and early career researchers can truly benefit from the mobility and the capacity building. We plan to hold frequent flagship events, both in Switzerland and in India, to keep connecting our various partners of this program from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:17",
                    "end": "00:08:43",
                    "text": " funding actors, beneficiaries of the calls, policymakers and prospective applicants and early career researchers. So make sure you follow our website and social media and there's more updates coming soon. I want to extend my sincere thanks to all our partner organizations here in India for their continued trust and collaboration and the two research communities in both our countries that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:43",
                    "end": "00:09:10",
                    "text": " show a lot of enthusiasm and engagement in these programs. So I encourage all interested researchers here in the room and out there to take advantage of these new opportunities and continue building the bridges that make our partnership so successful and meaningful. Thank you very much for your attention."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:11",
                    "end": "00:09:41",
                    "text": " Thank you so much also from my side. My name is Nina Frey or Katarina Frey as my colleague or former colleague and Marcus Roewe said introduced me. I am the executive director of ICANN, which is this network linking already academic partners from Europe, Africa and Singapore. And I'm very glad that I have many representatives from the network that will be on the panel and actually also one of the board members sitting in the second row from the finished"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:41",
                    "end": "00:11:57",
                    "text": " Supercomputing Center. Thank you, Damian, for coming. So we have such a big panel representing ICANN that there's not even a space for me, so I will be standing here. And I would like to invite my panelists to take seats on the different names. I will introduce you and hand over the mic to you in a minute. Please, please have a seat. Turned out there was a seat for me. Yeah, I know. Thank you so much. Thank you so much for..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:59",
                    "end": "00:12:17",
                    "text": " preparing with us for taking pictures. We actually talk about language, but let me think about an analogy to pictures, but we'll dive right into the importance of, I would say adds to the language question, obviously also the cultural and contextual embedment of different AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:17",
                    "end": "00:12:35",
                    "text": " in the different settings. So again, allow me to extend my thanks to all my distinguished panelists for coming, for also allowing us to show how this ICANN collaboration works from very different angles. The idea of this next 40 minutes is really to..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:35",
                    "end": "00:12:55",
                    "text": " to try to give a linee rouge, like a red line, I think you say, between the different summits. Actually it started obviously in Bletchley's and I hope we can then showcase how this topic of language and cultural diversity was somehow present in all the different summits and unites us all."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:55",
                    "end": "00:13:25",
                    "text": " Since we're here in your host country, allow me to hand over the mic to you to share with us why did Bajini been funded? You had presented your work this morning to me and Alex. It was very impressive how it translated immediately live from Hindi to German to English. But please share with us maybe the next five minutes what your work is, what has it been, and where you're going. Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:25",
                    "end": "00:13:53",
                    "text": " Yeah, thank you very much Nina, and thanks for inviting me here. Bhashani stands for Bhasha Interface for India. So it is basically looking at 22 languages which are enshrined in our 8th schedule of constitution, which basically says that we will conduct or we will have these languages as the languages to start off with for our work in the regions."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:53",
                    "end": "00:14:22",
                    "text": " We started off as a program for transcending the language barrier using artificial intelligence in these 22 languages at the first place. We had our own challenges but the methodology which we followed was to collaborate with 70 research institutes across the country and the problem statement was actually divided in between all the 70 research institutes."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:22",
                    "end": "00:14:50",
                    "text": " We were solving five problems to work on. First was automatic speech recognition. That means the digital systems should be able to understand what we are speaking in all 22 languages. Then we are looking at the second piece of it which is text-to-text translation in again bi-directionally in all 22 languages. Third was text-to-speech which was basically again that the digital system should be able to speak to you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:50",
                    "end": "00:15:20",
                    "text": " That is again in 22 languages. And then we are looking at optical character recognition in 22 languages. And also our digital dictionary, which is basically the vocabulary in all 22 languages are not digitized. So there was an attempt to digitize all the vocabulary which is around. That includes names of places, people, companies, et cetera, et cetera. We have till now achieved"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:20",
                    "end": "00:15:44",
                    "text": " 22 languages in all the modalities. We also have increased the number of languages. Incidentally, in India, there are 100 languages which are spoken or written by at least 100,000 plus people. So our journey is not complete when we do 22 languages. We are moving ahead with more languages. So we now have 36 languages on text."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:44",
                    "end": "00:16:13",
                    "text": " and we are going to add more languages to move forward. We also have languages which don't have script and those are basically in the tribal area. So we are attempting to digitize that also and that is being digitized. One of them has been digitized and will be launched in next few days. So in all of this, we had one basic challenge which was non-availability of digital data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:13",
                    "end": "00:16:42",
                    "text": " So the non-availability of digital data, which is oil to the AI models, was basically done for the first time in the world as a brute force data, digital data collection. So what we had done was that we had about 200 odd people who would go down on the field and speak to the people on a certain subject, pick up a picture or any other things so that it becomes the topic of discussion."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:42",
                    "end": "00:17:11",
                    "text": " We will create the monolingual corpus by requesting them to write the same thing or bilingual corpus if they are, you know, having two languages. And that is how we build the bare minimal digital data. Obviously, when we have done these things, the model is like a child, it only read 100 books. So it will be as intelligent as those 100 books. So we realize that over a period of time, we need to collect more data. That means give the child 1000 books so he is more intelligent and more."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:11",
                    "end": "00:17:41",
                    "text": " And that journey continues. So we have taken AI as a journey, but we haven't waited for some things to become perfect so that we are in a position to launch them as a product. We launched them and built narrow use cases. Narrow use cases in the sense that, okay, let's build something for the farmers. I will try to give two examples for want of time. One is that we have built up an interface for the farmers where farmers in their own language can ask a question about"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:41",
                    "end": "00:18:11",
                    "text": " agriculture advisory and he or she is answered in that particular language. So it's a voice first and voice journey. So that means I will be talking in voice and you know the answers will be coming in voice. So that's a voice journey sequence. The other thing which we had actually experimented on or are working is this is a deployed system. So it is actually a very large system. We are now, we are working which is one of the things which have been displayed here is a project called Gyan Bharatham where"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:11",
                    "end": "00:18:39",
                    "text": " You know, the manuscripts have been made interactive. Plus, we have multiple other use cases. Perhaps I'll come to them during the discussion. That means we have about 20 ought of them displayed in our stall. But we have been focusing, the clear point is that we have been focusing on narrow use cases so that we are able to control the vocabulary and the purpose and deliver value. Because at the end of the day, everybody has to contribute to the AI system."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:39",
                    "end": "00:19:10",
                    "text": " especially in the languages, for it to improve and become perfect. Thank you, Amitabh. Thank you so much. And I somehow assumed everyone knows you, but obviously I should introduce you as well. So apologies for that. Mr. Amitabh Nag, he's the CEO of Baljini, the National Language Initiative. And we will be collaborating. Alex will be mentioning more later on on that. But before that, I would turn a year back to Paris."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:10",
                    "end": "00:19:21",
                    "text": " where obviously current AI was started and came out of the public interest AI working group, if I think."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:21",
                    "end": "00:19:48",
                    "text": " Baddiel, you say? Baddiel, sorry for that. She's the CEO, quite recent CEO of current AI. Very, very important initiative that amongst others also wants to thanks in the topic that we're talking about. But please, I know you come from a wrong background also in hardware. You are launching, I think, this afternoon something very impressive that also helps."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:48",
                    "end": "00:20:12",
                    "text": " the importance of language diversity. Could you share with us some of your key focus interests and also why you also focus on hardware? Thank you. Thank you so much for having me. So my name is Aya Bader. I, yes, did join recently, about a month and a half ago exactly. So really feeling the very warm welcome in India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:12",
                    "end": "00:20:41",
                    "text": " Current AI was a initiative that came out of the French AI Summit. The founder, Martin Tisney, was the special envoy of President Macron at the summit. And the initiative essentially has a vision for AI that is global, that is collaborative, and that is collective. And so the idea is that if we acknowledge that some of the biggest"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:41",
                    "end": "00:21:07",
                    "text": " tech companies that are really governing our lives and really governing AI and the way we consume it in day to day. They are a handful of these companies. They are big, they have scale, they have a lot of financial resources and they are very ambitious. And so the initiative acknowledges that to be able to stand the chance to be an alternative and a counterpart to these..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:07",
                    "end": "00:21:32",
                    "text": " large companies we must fight scale with scale and so obviously there is lots of Interesting work happening in public interest AI around the world but oftentimes the work is distributed the work is decentralized and Sometimes it's duplicative and it's not always additive and so as a result current AI has this vision that we need to sort of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:32",
                    "end": "00:21:45",
                    "text": " bring together and bring more collaboration into the space, but also raise the level of ambition and of financial scale that is taken on."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:45",
                    "end": "00:22:03",
                    "text": " Current AI is a public-private partnership between philanthropy, between the private sector and governments. It has initial commitments of about $400 million, but the ambition is to get to $2.5 billion and hopefully more."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:03",
                    "end": "00:22:33",
                    "text": " The initial commitments are from the French government. There are also partners, multiple other governments, including the Indian government, the Kenyan government, Moroccan, and many others, as well as from MacArthur Foundation, Ford Foundation, McGovern, and a few others, and the private sector, so Google DeepMind, Salesforce, and others. So it really is a public-private partnership with the intention of kind of bringing everybody around the table that has sort of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:33",
                    "end": "00:22:48",
                    "text": " of the same commitment to public interest AI, to AI that works for individuals and for the public good. And one of the main vehicles of doing that is really investing in open source."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:48",
                    "end": "00:23:06",
                    "text": " Language has been a priority for current AI ever since its inception. The initiative was called multilingual diversity, which I know is something everybody here is committed to and we've been hearing a lot about over the past few days. I joined about a month ago and I'm..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:06",
                    "end": "00:23:35",
                    "text": " myself very passionate about the topic and I sort of expanded the topic to be about culture, diversity and culture preservation. So it's really not just about language, it's also about acknowledging that culture exists in many facets. Language is one of them, but there are also behaviors, there are norms, there are also artifacts, physical and digital artifacts, and there are many things that are digitized and non-digitized. And so we now..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:35",
                    "end": "00:23:41",
                    "text": " talk about culture preservation as one of our big priorities."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:41",
                    "end": "00:24:10",
                    "text": " and it's something that we'll be doing a lot of work in. As part of the cultural preservation work, also when I came in, there had already been conversations between current AI and Bashni about doing a collaboration together for the summit. And to be honest, I fell in love with the work that Amitabh and his team were doing and the care that they were taking with their dialects and languages and really the fact that they were going to sort of the source."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:10",
                    "end": "00:24:24",
                    "text": " and getting a lot of this knowledge, not just data, this knowledge about the language from individuals and from communities themselves, no matter how small they were."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:24",
                    "end": "00:24:52",
                    "text": " And so we ended up collaborating on a device that will launch later today at 3.30 in room 10. Hope you all can attend. I'm not going to say much about it because there's a drum roll situation that will happen. So you all can come see for yourselves. But the intention of the device is to really get as far, as close as possible to the individuals and the communities themselves."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:52",
                    "end": "00:25:04",
                    "text": " There is one concern I have that could be kind of a negative repercussion, I think, of having so much attention on multilingual diversity in a summit like this, which is..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:04",
                    "end": "00:25:30",
                    "text": " this idea that a lot of the big companies and big players have to do all the work. It's interesting and positive that the big tech companies are saying, we're going to make commitments to more multilingual diversity and more languages. That's good, but oftentimes when they are in the leadership taking these positions, there's a brute force."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:30",
                    "end": "00:25:40",
                    "text": " methodology that they deploy because of the scale at which they operate in and so oftentimes it's about scraping data, oftentimes it's about taking data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:40",
                    "end": "00:26:01",
                    "text": " without licensing it, it's about treating individuals and communities as data, whereas they are people, they are not data. And so that's sort of my concern in this area. And I believe that we have to get as close as possible to the communities themselves and invite them and support them in doing that kind of work."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:01",
                    "end": "00:26:31",
                    "text": " themselves. So it's really about them preserving their own cultures and languages and not about us doing it for them in this sort of like somewhat condescending way. I'll also say one last thing, which is I myself grew up in Beirut in Lebanon, very tiny country, but that everybody has heard of sometimes for good and not good reasons. But, you know, Arab, Arab language is also"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:31",
                    "end": "00:26:52",
                    "text": " very concerned about AI and representation in AI. And we have thousands of different cultures and dialects within Arab culture. And we also have varying degrees of resource availability across Arab countries. Some countries are very resourced financially."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:52",
                    "end": "00:27:15",
                    "text": " from a government perspective, others have very scarce access to resources. So I'm also very concerned about thinking about AI that is more resilient, that operates from scarcity, and operates from frugality, and operates from a limited amount of resources, and looking at that as a positive as opposed to a negative."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:15",
                    "end": "00:27:28",
                    "text": " So that's something that current AI will be prioritizing in a big way, and we'll hope to do more of. So hope to see you all at 3.30, and hope you will drum roll with us when we call for it. So thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:28",
                    "end": "00:27:56",
                    "text": " Thank you so much Aya. Let me hand over because you mentioned like the obviously the many announcements that were made as well as private companies to start collecting data. I think it's fantastic to see that governments can do that as well and that you also invest from in this PPP so far and allow me to hand over to my colleagues sitting to my left because I think you can also showcase how also public institutions like universities can also"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:56",
                    "end": "00:28:26",
                    "text": " train a model multilingual from scratch. Scratch, not stretch. It was probably a stretch sometimes, but let me introduce you to Dr. Alex Illich. He founded and is the Executive Director of the AI Center, a co-founder also of ICANN. And please, could you share your experiences with Apertus, which is this multilingual model, and maybe also mention something on Swiss AI and how"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:26",
                    "end": "00:28:52",
                    "text": " like the Indian languages we can maybe then present next year in Apertus. Alex, please. Thank you very much for the warm introduction and also I think like expanding maybe on the point that Aya made like, you know, Big Tech won't solve it. I think there's a big role to play for academia and always there was the question when can academia step in?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:52",
                    "end": "00:29:20",
                    "text": " So in Switzerland, we're in a very fortunate situation that we have a supercomputer that allows us to do frontier scale models. So over 11,000 GPUs of the newest generation, we were the first one to get those GPUs. And we already released the world's first fully open language model, Apertus, at 70 billion per meter scale. It's basically a team up between ETH Zurich and ETHL."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:20",
                    "end": "00:29:40",
                    "text": " We learned a lot in this process. First of all, we achieved a performance that's comparable to the open weight models from Meta and others as well, so performance-wise. But then we also put a deliberate focus that we don't do English first and then let's tune it for other languages, but we build the multilingual part in from the beginning."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:40",
                    "end": "00:30:07",
                    "text": " In Switzerland we have already four languages, many dialects as well, so I think we faced the same challenge. And it was also quite interesting to see that in some areas you don't actually need so much data to increase the capabilities of the model. I will come back to a little bit on how far are we off for the goal. But basically we were able to build this model and one of the key bottlenecks that we also identified is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:07",
                    "end": "00:30:35",
                    "text": " It's not just the infrastructure where currently a lot of money is going in, but also the talent. Outside of big tech, you have maybe 100 people on the planet who have the experience and capabilities to build such foundation models. And that's not enough. And I think that's something where academia can change it. And I think that's why it's important. We not just need supercomputer and data centers for the companies, we needed to empower academia. This is very, very critical that we also push this very, very strongly."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:35",
                    "end": "00:31:02",
                    "text": " We named the model Apertus, Latin for open, because we wanted to be a foundation where everyone can take it and build on top of it. So it's not something that, you know, we force up on someone, but it's something that can be a thriving community where each university, each project, each country gets a step further. And I think we will hear later a little bit from the perspective also from Singapore, from India, we already heard."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:02",
                    "end": "00:31:23",
                    "text": " There are not many countries that recognize how important that is as a public infrastructure, that you really take it serious to develop your own benchmarks and your own data sources as well. Because today, still, if you read LinkedIn, the majority is driven by benchmarks that the big companies are publishing."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:23",
                    "end": "00:31:53",
                    "text": " And surprise, in every benchmark they publish, they are of course the best because they pick whatever their metric is usable. And I think this metric should be driven by what do we want it to be in the cultures and the regions to empower this. And so we have a thousand languages included because we train it with data on the internet. As you know, the internet is not the most diverse data source there is. 60% of the data in our training set is English."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:53",
                    "end": "00:32:23",
                    "text": " 40% is non-English and so what we're thinking about now strategically is how can we increase the number of languages that are close to the performance we see in English step by step for the next hundred languages and so on. And this is I think like important because many companies that are going in that area and say oh we sponsor a data collection effort they just do it on best effort like you let's do something and you don't know does it actually move the needle. So the next step for us is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:23",
                    "end": "00:32:47",
                    "text": " that with all the experiences in Baschini and other parts, I think we can find out now very strategically how much does it cost us to raise the bar significantly, not just make a check mark out of that. So that will be also the hope for connecting forward through the mission of ICANN and also Geneva next year that we can present also how far of a progress"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:47",
                    "end": "00:33:15",
                    "text": " could we make like why do we stand today that it's really usable and economically usable and to elevate this. I think that's super critical on that side. And yeah, we're also very happy that from the side from Singapore already with the experience that's the first version of Sea Lion that's already built on Apertus and we want to extend the collaborations now globally. For the researchers, we also have a very strong international"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:15",
                    "end": "00:33:32",
                    "text": " a program like what we share basically our compute infrastructure. So that's also very very unique and we would like to see also other countries to do that because we know that you know for where we stand with AI today we're maybe at five or ten percent of the potential to train the next models that are you know"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:32",
                    "end": "00:34:01",
                    "text": " including more data, becoming more aware of the physical world and so on. We need more compute. We need to team up. And I think that's also a question of how can we collaborate more and share more. And with ICANN, I think like in the beginning we had this like the bottlenecks, you know, a compute. That's why we have strong compute representation. It's data and benchmarks and it's talent. And I think on these three capabilities, we need to jointly increase. And whoever doesn't have it,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:01",
                    "end": "00:34:27",
                    "text": " should be able to easily get it from other sides to do it themselves basically. Thank you. Thank you so much Alex and also for emphasizing the talent need. And I think if I can just add that, you mentioned the talent or the capabilities in knowing how to train a model, something that and again I'm looking at the supercomputing representatives, I mean it's also a talent in knowing how to build up"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:27",
                    "end": "00:34:36",
                    "text": " such an HPC, so I think that's something we could also add to the table. But allow me to come back to the models themselves and the really very concrete..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:36",
                    "end": "00:35:00",
                    "text": " applications. Now I'm turning first left to the north, to the Finns. Petri, you're here as obviously as a founding member of ICANN, but those representing Alice Network, but also to Team Finland, if I can say that. But you were also a member of the H-Lab of the UN Secretary General, so were you. One of the recommendations was also exactly this, that we collaborate."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:00",
                    "end": "00:35:28",
                    "text": " Could you mention maybe more from the perspective of the Nordics? You had already your own language models, but maybe also you can share some thoughts on why you recommended that to the world, if I can say that. Yes, thank you. So happy to be here. So, indeed, I mean, as you all know, Nordic languages are not the biggest major languages in the world. So obviously we take the kind of the preservation of..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:28",
                    "end": "00:35:54",
                    "text": " languages and cultures very seriously. Talking about the H-Lab of the UN, there was just upstairs a kind of a handover to the new international independent scientific panel on AI. So maybe one thing I learned in this UN advisory body was that, I didn't know this, but like one thing I learned was that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:54",
                    "end": "00:36:21",
                    "text": " access to language and culture is a human right. One of the human rights that all the countries in the world have accepted. So to me this was a surprise and pleasant surprise because as also like language is already important because we operate with language but like what Aya was saying like even more important is the culture behind the language. We have different value frameworks and norms in different countries. So if there's one size fits all"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:21",
                    "end": "00:36:47",
                    "text": " English version AI that we all start to use. What is the value framework behind that? So that's kind of I think this is a very critical issue. Another thing I learned in UN was that I mean like and that's why is that there are several global initiatives towards like making this more accessible to all countries."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:47",
                    "end": "00:37:14",
                    "text": " Seven of the UN member states of 193 countries are included in all these initiatives. 119 countries are included in none. So initiatives like ICANN or current AI or this summit are very important to make this more inclusive. So I now shamelessly steal a quote from Joshua Benjio who was just upstairs saying like we need to..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:14",
                    "end": "00:37:33",
                    "text": " make sure that all the countries in the world are invited to the dinner table, but not part of the menu, but they are dining guests. So I think this was hilarious. Thank you so much for sharing. I didn't hear, but it's..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:33",
                    "end": "00:38:03",
                    "text": " I think it's a good thing that we can take up. Because also, obviously, food is very culturally diverse. Thank you. So let me turn from the north first, more to the south. And to Singapore, you were also quite recently, I think, at NTU Singapore, which is also the newest member of ICANN. You had already developed, and I think you will share something on the sea lion model, which is obviously for the ASEAN region, the famous language model, but you also had..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:03",
                    "end": "00:38:33",
                    "text": " already collaboration with Apetus and if time allows, because I would also allow any to speak, but if time allows, you could also mention something on the importance of sovereignty and language. Please. So it's wonderful. NTU Singapore, we're the newest members of ICANN, but it's fantastic that the, and I've only been NTU Singapore for six months, but it's, you know, the conversation that we're having here is the same conversation that we're having."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:33",
                    "end": "00:39:03",
                    "text": " about the importance of kind of multilingual diversity, the importance of getting close to the ground, the importance of culture as well as tech. And I'm the Dean of a College of Humanities, Arts and Social Sciences. I'm a historian, you know, sort of. And it's where my college is, you know, in the lead collaborating with computer science into you in kind of in developing, in engaging with sea lion and thinking about kind of AI in the context we're talking about. So, you know."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:03",
                    "end": "00:39:31",
                    "text": " this point is very important, that it's about culture and thinking about cultural diversity and how AI models, et cetera, reflect culture, how we engage with culture and history, et cetera, as well as simply technology. And I think that's something that's very evident in this conversation. So it's great to be part of this club. So C-Line is a language model that reflects"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:31",
                    "end": "00:39:60",
                    "text": " 13 languages across Southeast Asia and in fact includes Tamil because Tamil is a Southeast Asian language because it's an important national language within Singapore with aspirations actually to kind of expand, potentially connect beyond to other parts of Asia. It's a nationally funded initiative. I mean, but it connects from Singapore, so it's part of Singapore's public infrastructure, but then it connects regionally and is used by"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:60",
                    "end": "00:40:24",
                    "text": " has good connections with private sector providers across Southeast Asia, platforms in Indonesia on various things, et cetera. And as we're hearing a moment ago, I mean, there's a number of different versions of it. One of them is built on apparatus. So there's a real synergy here. And I think I just want to flag the connection between Singapore and Switzerland. They're both multilingual, multicultural kind of."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:24",
                    "end": "00:40:49",
                    "text": " relatively small societies. So there's a very obvious collaboration there. And I think another echo again, something that I was saying is when we're thinking about AI and we're thinking about the relationship between culture and language, we're also interested in frugality. We're also interested in using resources effectively and in thinking about how we can draw on sort of..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:49",
                    "end": "00:41:19",
                    "text": " deep truths about language and culture without vast amounts of data, you know, kind of with relatively small amounts of data. I mean, one of the, you know, we have languages like Laotian, Khmer, et cetera, within Sea Lion, and so you're, you know, the colleagues are really thinking hard about, you know, how you leverage relatively small amounts of language to kind of, to then produce an effective model in the discussions. Just a couple of additional points, and I'm looking at the clock."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:19",
                    "end": "00:41:45",
                    "text": " Sovereignty is the big word within the AI summit. I'm a historian actually in some ways of sovereignty at the moment. And kind of sovereignty means power. It's a power that we want for ourselves, for our communities, for our nation states. But in a sense it's also about individuals as well. And there's a kind of complicated relationship between those two things. And so I just wanted to kind of..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:45",
                    "end": "00:42:13",
                    "text": " just reflect on, I suppose, the importance of sovereignty and that we're talking about the sovereignty of societies that are, I guess, neither the US or China in this discussion, the two big superpowers maybe. And this discussion is about how we can think about a world that is multipolar and is multicultural and reflects the fact that sovereignty actually is dispersed in the world in which we live in. And that's very important. And that's Indian."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:13",
                    "end": "00:42:41",
                    "text": " the principles of non-alignment that go back to the 1940s, you know, and so on. So I don't know if I'm allowed to use that phrase in today's India, but anyway, it's a similar set of principles that we're talking about. So the dispersal of sovereignty that we're talking about here and power is important, but it's also as part of that, I think, to reflect the limits of the nation state, I suppose, and the limits of kind of national approaches to language. In that, we all live in..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:41",
                    "end": "00:43:10",
                    "text": " environments in which people speak complicated, they're multilingual societies in a minute by minute way. People code switch. They're speaking Hindi in one minute, they're speaking English the next, Swiss, German, etc. Similarly in Singapore people speak Mandarin Chinese, then they speak a Chinese dialect and then they'll speak English. So sovereignty is crucial, but if we're interested in the sovereignty of the individual and the power of individuals, then we need to have a more nuanced account of..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:10",
                    "end": "00:43:40",
                    "text": " of language that allows for things like code switching and dialects, etc. And that's something that we're very much interested in in NTU. Thank you so much and thank you to all the speakers. Also allow Annie to speak because I think you're obviously from South Africa but now living in Switzerland and the US and you lead this linkages between medicine and the AI and I always think you explain very concretely in your work."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:40",
                    "end": "00:44:10",
                    "text": " What if you just take an English-speaking language and train it in a tiny set of local data, like how you experience that in the medical field in reality. So if you could share something on that, and obviously also your role in ICANN. Thank you, Annie. Professor Annie, she's at EPFL at the moment, and then Yale. Thank you. Yeah, so thank you very much, and I think I'll take it down to the ground, then, about the consequences, about what happens really when you are at a patient's bedside."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:10",
                    "end": "00:44:39",
                    "text": " and you ask questions that are high stakes. And something that I do to just test these models in different places, because we are rolling out these tools in different hospitals around the world, I ask the same question, which is a very high stakes question, how to treat diabetic ketoacidosis, which is a diabetic crisis in a child, right? And I did this recently in Ethiopia, in a language that's not very well known, Afanaromo, and it's responded to me, thou..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:39",
                    "end": "00:44:60",
                    "text": " Shelt not eat insulin on a Tuesday. And I did share this advice with the, because I thought it was actually very good advice. You should not eat insulin any day, actually. But I did share this advice. But it comes to something that's really, really important. I'm stating the obvious."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:60",
                    "end": "00:45:26",
                    "text": " It means that if you do not like, because it obviously only trained on the Bible, right? That's something that is, that's very available. That's the one book that is available in every single language in the world. And so you have these biblical kind of terms, but the Bible isn't like very necessarily very accurate in medicine or other things, but depending on where you're coming from. But the thing is that you can't rely on these models to make these decisions."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:26",
                    "end": "00:45:56",
                    "text": " because they are inequitably inaccurate in the places that need it most. So we know that they'll be inaccurate, but the point is that we actually have to, if we're collecting this kind of information, we have to make an effort to collect it in the highest stakes environments and in those contexts. So if you have use cases for collecting language, it's interesting to collect it in maybe like historical texts or to represent culture, of course, but I think something that has a much bigger urgency"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:56",
                    "end": "00:46:24",
                    "text": " are the urgent questions. These are high stakes decisions that we are making, and people will believe that the model performs well if it only speaks a language, but they might get a false sense of security if we don't really train it to be accurate in the questions that people are relying on these tools for the most. And so this is why we actually have to, when we collect languages, and when we are trying to test these tools in reality, we have to make sure that we represent those kinds of contexts."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:24",
                    "end": "00:46:53",
                    "text": " And that's what we are doing. So I lead a lab called Lights. It's the laboratory for intelligent global health and humanitarian response technology. So obviously I'm interested in these high stakes environments and these cultures that are so underrepresented that they will never be represented with any kind of large commercial enterprise, right? No commercial entity has ever said there's a great place to make money and it's that war zone. Okay, unfortunately they have, I suppose. But the point is that people don't want to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:53",
                    "end": "00:47:21",
                    "text": " represents that kind of place because it's not in their interests. And this means that it is so important for academia to play a role. We don't just play a role because we've got expertise. We play a role because we do something that commercial entities cannot provide. We are neutral and we create a neutral space for this kind of collection of data to represent the needs of people and also to make sure that we can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:21",
                    "end": "00:47:51",
                    "text": " test it in reality, right? This is why we can do open science. It's because we don't have like any money in the game to lose, right? And the most important thing to do is actually to see when we do represent these languages, not just to represent them and be happy about it, which is the first step, but to go the extra mile to actually test whether the languages are being represented as you expect them to be."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:51",
                    "end": "00:48:18",
                    "text": " So some of my patients, for example, it might speak their language, but does it speak their language in the way that they expect? And do they follow the advice or don't they? And this is a really important thing to test in these high stakes environments. My patients will come to me in South Africa. We speak 11 official languages. And in Plaza, a way of explaining certain things is it's very..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:18",
                    "end": "00:48:47",
                    "text": " Different and it gets translated into English in a strange way sometimes and and so one of my patients came to tell me You know, I've got I've got elephants running in my head, right? I know exactly how to respond because I'm that's my culture. I'm South African So but what would what wouldn't AI respond right and I have a pregnancy in my knee Right. I'm pregnant in my knee. That's that's a patient came to tell and that's that's actually it doesn't come from a mistranslation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:47",
                    "end": "00:49:16",
                    "text": " It comes from the way that people understand how their bodies work. And this is very, very cultural. What is the next most likely word after pregnancy in my knee, right? So it's really important that we understand how it works when it's on the ground and taking these models on the ground, right? And testing it with clinical trials and making sure that we get feedback from reality. This is what we're trying to do. So we have..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:16",
                    "end": "00:49:45",
                    "text": " starting with ICANE, a flagship project that we made. It's called MOVE. It stands for Massive Open Online Validation and Evaluation. And it's about getting these real-world signals from real people in high-stakes decision-making processes, from our doctors, from the people on the ground in different countries around the world, and to get that information from how they are using any tool, because we are neutral. If any tool comes out, any new model,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:45",
                    "end": "00:50:13",
                    "text": " we can test it and then we get how it works and then when it breaks we don't just say oh this model is bad in the setting and this model is good we really try to get that information and put it back into the model to continuously improve it and so learning from reality learning from the real workflows of how people use models and I think that's important to to represent reality and not just the language but the reality that the language functions in so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:13",
                    "end": "00:50:40",
                    "text": " Last thing I'd like to say about this is this does cost a little bit more money and it's not the traditional kind of way of working in science. And people don't appreciate that implementation science is science. And it's such a fantastic opportunity where we can actually do impact, like actually measure it, the impact of the models that we are making. We can measure it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:40",
                    "end": "00:51:10",
                    "text": " and feed that back into our models and really create impact-driven models. And to run these trials, it's ambitious. But we do need to start asking different kinds of funding and being more ambitious. And I think academia does need to be more ambitious because we are representing something that's actually very important these days and very rare, which is this neutrality. When OpenAI updates it from 4.5 to 5 or 5 to 5.1,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:10",
                    "end": "00:51:40",
                    "text": " Did they ask your permission? No. Right? Did they ask the doctors who had validated those models for their context? No. We need control. We need to know how these tools work in reality and we need to be able to control the tools and so sovereignty for me is control of tools and control of the environment and to understand how these models work in reality so that we know when we want to update it and what we update it with."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:40",
                    "end": "00:52:08",
                    "text": " So yeah, thank you. Thank you so much. Thank you, honestly. Thank you, everyone. Thank you, everyone, for keeping the time and for making sure that we are actually creating the menu and controlling the menu to also steal Professor Bencho's words and for contributing here. And I think we will be more than happy to update you, hopefully, next year on our more joint collaborative work."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:08",
                    "end": "00:52:25",
                    "text": " Thank you everyone for coming, for staying with us and on the speakers. Thank you."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "69982c3c6c61c5bef9ef7e13",
    "projectId": "603613",
    "projectName": "The Agent Universe: Exploring the Spectrum from Automation to Full Autonomy",
    "projectDescription": "The Agent Universe From Automation to autonomy",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-20T09:41:15.114Z",
    "projectUpdatedDate": "2026-02-20T09:54:48.258Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-20T10:01:16.418Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "The Agent Universe_ From Automation to Autonomy(MP3_128K).mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771580428949.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:02:35",
                    "end": "00:03:09",
                    "text": " Morning while we still wait for one of our panelists to arrive, Sanjeev. Sanjeev from the Bajaj house. He is like all of us has been stuck in the mega traffic that we all have been observing for the last five days. Who all has been stuck in traffic since? Yeah, just three, four people. That's a big smile about traffic. So when I was returning home the day before, it took me three and a half hours to get back to Goodgaon."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:09",
                    "end": "00:03:41",
                    "text": " So today I started like very early. I said, I'm going to beat the traffic. I'll come early. So I started 7 a.m. and there was no traffic. So I was here far faster than I anticipated. Okay. So we'll be discussing hyper automation in BFSI, in banking financial services, insurance companies. Yeah. Who all here do banking and financial services?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:42",
                    "end": "00:04:16",
                    "text": " Everyone, that's good. Ruchika didn't raise her hand. So we'll be talking about how we will move away from what we used to do with automation to getting to an end-to-end flavor of doing transformation or automation as I call it, using agentic or AI as we call it. We've got a good panel today. We've got Dr. Pankaj Dixit. He's part of Cignet One. He's worked with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:16",
                    "end": "00:04:52",
                    "text": " He set up national level AI solutions, national level tech solutions, and he is now at the forefront of SIGNET. And we have Ruchika. Ruchika works for Netwest. Netwest is a mega bank, global bank, and she has some of the most challenging roles in managing the tech infrastructure for Netwest. We'll start our conversation while..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:52",
                    "end": "00:05:38",
                    "text": " Sanjeev arrives. Sanjeev is here somewhere. He's finding his way through the security and then walking past all the gates to come here. So he should be here any moment. Once he comes, he'll take the last seat, but we'll start a conversation. Yeah. Thank you. So you have to do a group picture, but Sanjeev is still not here. When Sanjeev comes, then you'll do the group picture. So today, what we'll do is we'll focus on three key themes. One is enterprise reality of being able to bring"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:38",
                    "end": "00:06:08",
                    "text": " hyper automation and AI to life? What is the architecture, governance and trust we would love to infuse in the hyper automation or the agentic solutions we're going to put to place? And what is the future readiness of the workforce we need to establish? Make sense? So if we'll just focus on these three that should give us the right flavor for the conversation today. Okay. Ruchika, can I start with you if you don't mind? Yeah. So Ruchika leads."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:12",
                    "end": "00:06:42",
                    "text": " I think microphone will be better. So Ruchka, as I said, is part of Netwest, massive scale. Netwest is all about scale. I'd ask you, from your experience, AI-led hyperautomation, how do you see we'll deliver business impact? Especially while you're delivering the business impact, how do you ensure risk is managed and you are able to provide the right customer engagement?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:43",
                    "end": "00:07:12",
                    "text": " Look, first of all, thank you for having me. Really glad to be here. It's a very exciting summit. And there's obviously, you know, we'll be waiting on Sanjeev here, given the crowds, et cetera. So really glad to be here. At NatWist Group, you know, when you look at the big picture for us, AI is really helping us build deeper."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:13",
                    "end": "00:07:42",
                    "text": " and more trusted relationships with our customers, really. And when you think about the places where we want to adopt more AI, it is in that deepening of the relationship. It is in getting to know our customers better. It is to meet them in the channels of their choice. It is to give services to them that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:42",
                    "end": "00:08:11",
                    "text": " are going to help transform banking for them today and for people and customers to come. That's really at the heart of why, right? And for us, you know, through the last couple of years, especially in 2025, AI has been really about scaling. We have at this point given tools like Microsoft Copilot,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:11",
                    "end": "00:08:41",
                    "text": " and our internal LLM tools to 60,000 of our colleagues across the bank. And when you look across our businesses, whether it's in the retail business, where we are, leveraging Cora, that is our intelligent assistant for customers, for 20 million customers, it is taking on almost 12.9 million customer interactions."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:41",
                    "end": "00:09:09",
                    "text": " and we've taken from four to about 11 customer journeys from AI to GenAI in the last year. When you look at what we're doing in commercial and institutional, all our complaints, 100% of them today are summarized and supported by AI. When you look at our private banking and wealth,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:09",
                    "end": "00:09:39",
                    "text": " We've supercharged our relationship managers to have access to AI and therefore making them 30% more productive. When you look at the expert roles across the bank and you think about people who code, over 12,000 of our engineers have access to AI tools and think of it as about 35% of our code is actually generated by AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:39",
                    "end": "00:10:08",
                    "text": " really across the customer experience, across the customer life cycle and AI led hyper automation for us as you've asked is really about making our colleagues more productive. It is about generating more time, removing the manual tasks and allowing them to spend time in places where the human connection"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:08",
                    "end": "00:10:37",
                    "text": " with our customers is more important. So they are really now spending time with our customers in money management, in their life moments, in their deeper needs and understanding the customers versus AI is taking over more of the manual tasks. I hope that helps you bring to life what this means for us. Thanks, Ruchika. I'll just extend this to ask in."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:37",
                    "end": "00:11:07",
                    "text": " Especially for banks, regulator increasingly ask, why did a model maker decision not what just decided? So how do you balance the speed of innovation with getting the right governance and explainability in your systems? I, as a consultant, go about trying to solve this for my customer, which is not a easy ask. But from your lens, we would like to understand how do you manage speed,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:07",
                    "end": "00:11:35",
                    "text": " of innovation while providing the right governance and explainability? Yeah, of course. It's a very pertinent topic, of course, for us in banking, in any regulated industry. So first of all, I will say that the AI opportunity brings with it a level of responsibility, doesn't it? And, you know, for us, we want to be, as I said, a deep and a trusted partner with our customers. So it is absolutely important that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:35",
                    "end": "00:12:05",
                    "text": " We are embedding responsible AI in everything that we do. Every single colleague across the bank is accountable to make sure that AI is implemented responsibly. We have an AI and data ethics code of conduct and training that we've rolled out across the bank to all our colleagues. And I'd say 58,000 or so of them have actually completed that training, right?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:05",
                    "end": "00:12:32",
                    "text": " And you have to bring every leader into the development of AI. I think even more than broad tech implementations, when you think about AI implementations, you have to bring all leaders into the conversation. You have to think about governance and actual controls at all levels and all layers. And that's what we have done. We have..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:32",
                    "end": "00:13:02",
                    "text": " In 2024, we hired a head of responsible AI. Last year, we also hired a chief AI research officer and several of these roles along with both the responsible AI trainings, the policies and the governance that we use is to make sure that anything we might implement and put in the hand of the customers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:02",
                    "end": "00:13:31",
                    "text": " must absolutely be going through those layers of governance and make sure that they are safe and that our customers can trust this, trust them, yeah. That means having a bank account with Netverse is the safest place to keep your money, yeah? Thank you, Ruchika. You're making our lives safer. Let's move to Dr. Pankaj. Pankaj, you're an executive director and a chief AI officer for Signet, but you've also built national-scale digital systems."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:31",
                    "end": "00:14:03",
                    "text": " Do you want to tell everyone, because everyone would love to know what you have done in your past life before SIGNET building these national scale solutions? Thank you, Prashant. I would like to start with thanking the MITEI for doing such a great event here. There's a churn happening in technology. And in the last five days, we've actually been a physical witness to the churn in the number of people coming from all over the world, right?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:03",
                    "end": "00:14:32",
                    "text": " And of course, EY is the executing agency on behalf of MIT. Thank you very much. And thank you all very much for being here to be part of the channel. Coming to your question, Prashant, what I would like to say is, you see, I've had 28 years worth of as a colonel in the army. I then joined IBM. I have been eight years in GST and where we built the GST system and then as a CTO of GEM."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:32",
                    "end": "00:15:00",
                    "text": " I have seen technology evolve in the last 30 plus years from what it was most of you probably the younger generation would not even imagine the kind of phones and the communication that used to exist and what we have today. But being part of this journey, building systems like GST and at GEM, what has remained the constant threat throughout has been the respect for the end user."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:01",
                    "end": "00:15:26",
                    "text": " My encounter with JNI happened in November 22 when I was authoring a paper for the Prime Minister's Economic Advisory Council with other members. And I used that to look up some references for my text in which I wanted to quote some relevant documents or papers from either the World Bank or the IMF."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:26",
                    "end": "00:15:49",
                    "text": " it just churned out documents which were completely non-existent. And I spent at least a good amount of time and days to get the right references. I ultimately got them. But I found that the engine was just cooking up those references. Finally, of course, I did do that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:49",
                    "end": "00:16:16",
                    "text": " Going on to my experience in Jam where I built the first public sector AI engine for the Jam portal, what I realized was that the information has to be curated for the user. Again, the user is at the forefront and we cannot have hallucinations because there were some instances where some political names were taken and thrown to an AI engine and it just throws up unpalatable answers."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:16",
                    "end": "00:16:42",
                    "text": " So what happens is that when you are in a financial industry, numbers are very important. The mathematics is very important. When it's two and two, it has to come to four, it cannot be five, it cannot be three. That's the principle that I keep in mind always when I try and curate something for the customer and for the company to be used by the end client."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:42",
                    "end": "00:17:12",
                    "text": " Now that has been my journey to put it shortly, Prashant. And across the 30 years, having faced every kind of technology, along with national scale systems we built where we collect almost $10 to $12 billion per month, and GEM where we have $60 billion worth of GMV being transacted every year. I can say with confidence that yes, technology works, but you have to have the right people driving it at the helm and the right teams."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:12",
                    "end": "00:17:38",
                    "text": " Thanks, Dr. Dushyat. So two words, AI and compliance seem to not sit in the same box together very easily. In your view, what are the non-negotiable principles that we need to follow to ensure that we can make AI-led automation and compliance work in tandem with each other?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:39",
                    "end": "00:17:57",
                    "text": " So what happens now with hyper automation? You have to have the architectural principles that make these systems very, very solid, primarily being the AI, the RP, and ML. And we have to make compliant and scalable systems."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:57",
                    "end": "00:18:21",
                    "text": " At the chief AI officer at SIGNET, what I'm building is treasury management systems. We are working with the e-commerce companies, the biggest e-commerce companies of India. We are doing the e-invoicing for them. We are bringing account payables and account receivable systems. We are doing the latest, what we are doing is litigation management. So if you receive a notice from the GST authorities, we also create a response for you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:21",
                    "end": "00:18:48",
                    "text": " after evaluating your complete returns that you've filed. In addition to that, we have the material management based systems and the tax compliance system. So what we have to ensure is data fabric and security is primary. The unified structures, we have to have zero trust security and previously based design at the forefront."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:48",
                    "end": "00:19:07",
                    "text": " Then compliance by design, we have to have governance into every STLC layer and modularity and interoperability. So we try to build very strongly and tightly defined systems, but loosely coupled through API modules."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:07",
                    "end": "00:19:31",
                    "text": " Governance and explainability, as I mentioned, you have to have audit trails, you have to have explainability, you have to have the understanding when an answer or a response goes wrong for a customer, you should be able to trace it back as to what particular layer, what went wrong, and then work upon it. And once we have all this in place, we have to have the customer feedback to understand."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:31",
                    "end": "00:19:50",
                    "text": " regularly as to where the systems are working fine and where they're not working so fine. And take that feedback and build it back into the systems. And like I mentioned, the number crunching systems, I tend to use a hybrid approach. I have full trust, 100% trust on an Excel."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:50",
                    "end": "00:20:19",
                    "text": " I may not have that 100 percent trust on a AI engine to do the maths. So, what I do is I put an AI layer on top of the Excels. So, the numbers are crunched by the Excels and the context understanding and the human interaction happens through the AI engine because these are the two best of the world's approach, right? Somebody does something right, use that. Interaction, conversation, interaction, UI layer with AI is good, use that. Mathematics is better by Excel, use that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:19",
                    "end": "00:20:55",
                    "text": " and couple them through tightly defined systems and integrated through the APIs. So that's how I go about building AI systems for Numbers and for you by customers. Thank you. So we'll delve into the next question for you. So we just got Sanjeev walk in. Thank you, Sanjeev. We're sorry, Delhi is not hospitable in terms of traffic right now."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:55",
                    "end": "00:21:44",
                    "text": " So we'll ask you to all stand up. There's this gentleman waiting to take a picture of us. So our friend has been waiting for a while, Sanjeev, just for you to walk in so that a picture could be done. So we couldn't have made him wait any further. Well, I'll get back to Dr. Pankaj. So we'll talk about regulated workflows. So most of the work in banking industry requires quite a bit of..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:44",
                    "end": "00:22:15",
                    "text": " workflows and they need to be regulated. How do you bring human in the loop in play to ensure trust without killing productivity? So the challenge that I experienced when I started doing automation back like 10, 12 years ago was if we did too much of automation, the trust was killed and if we brought too much of human in the loop, the benefits of automation got killed. Now with AI, I think some of this is getting solved."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:15",
                    "end": "00:22:45",
                    "text": " How do you see this transitioning and how do we build trust with humans without killing productivity? Yeah, thank you Prashant. So like I was mentioning in my earlier part of the response, best of the best have to be coupled together. Now what an AI engine can do best, I will leave it to that. But I would have to have the human in the loop to approve certain things. So you have to create a threshold."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:45",
                    "end": "00:23:12",
                    "text": " A threshold-based risk delegation is important. Where I receive inputs from the AI engine, where I expect it to be about 80 percent, 90 percent compliant. For the remaining 10 percent, I'll put a human in the loop. For example, I was talking about the litigation management system that we are creating for responding to regulatory notices. The AI engine creates the complete notice response, but."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:12",
                    "end": "00:23:40",
                    "text": " who will approve the text and the content of the notice. That's where the human comes in. He approves it. After validating it, he might edit it. In addition to that, what I do is we take the feedback from each of the responses that I created. Was this 90% good, 95% good? And based on that feedback, we will try to curate so that the next response is better. There are certain tasks like KYC."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:40",
                    "end": "00:24:03",
                    "text": " where you have some documents which are read and validated. This is something which can be done by automating it and completely delegating it to the AI engines based on the data collected. But wherever risk factor is high, for example, to a law-based activity, or for example, responding to tax notices or..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:03",
                    "end": "00:24:33",
                    "text": " preparing returns, right? If you have to file a return which has to be 100% accurate, you will prepare it through the AI engines up to the level of approval and dispatch and submission where a human will just have a final look at it and be able to respond to that. So, like I mentioned, we use a threshold-based mechanism where certain activities based on the accuracy and that's how I increase the productivity. If everything were left to a human, we would not achieve that productivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:33",
                    "end": "00:25:09",
                    "text": " So the human in the loop comes in only at the last stages. So just about five or 10 percent of the complete work is then done by the human. Thank you, Dr. Dixit. Sanjeev sir, can we move to you? Thank you. We live in a city of VIPs, so coming to Delhi and finding traffic is normal. Your business is all about customers, more customers, and it is very much all about engaging with customers at scale and at mass."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:09",
                    "end": "00:25:39",
                    "text": " From a wealth management advisory perspective, how is AI led hyper automation changing the advisor come customer relationship in your world? Where do you see automation generally enhancing trust? Where must humans remain central? See, I think AI is a very exciting frontier we are at. So AI can be your biggest friend. It can be your biggest enemy. And the best part is the choice is yours."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:40",
                    "end": "00:26:06",
                    "text": " You know so what people don't understand AI is not a tool like the computer softwares or the tools or excel it is not a tool It is it is something to enhance you why has Indian IT industry stocks crashed or even the global IT industry Because what it has been realized is that earlier the USP of India used to be that that Indian labor was cheap"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:06",
                    "end": "00:26:33",
                    "text": " it was one-fourth, one-tenth, one-eighth the cost of what you would get in US. But if you look at the ability to do work on your own, a typical guy outside is 2-3x an Indian because there he has to do everything himself. He has to wash his clothes, wake up, drop his children and do 10 more activities. This is just a very basic, but even I am talking about when you go and start working,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:33",
                    "end": "00:27:02",
                    "text": " They are used to doing the Excel themselves. They're used to doing the formulas. They're used to working much harder, you know, on those things. Whereas we like to delegate. We want a model made, we'll call a guy, okay, make a model for me. We'll call a research analyst, do this research for me. We'll call a third guy and do this. And that worked for India. And that was the reason for India's boom. But what AI has done is that this 2X guy today can become 10X"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:02",
                    "end": "00:27:26",
                    "text": " 20x, 30x, by using AI to empower. So what we have to understand, AI is not going to replace you. AI is going to replace those who don't adopt AI. So you have to be the AI expert. Because AI is not an expert. AI is like a, you know, today AI is like a servant, it is an assistant. It is doing..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:26",
                    "end": "00:27:55",
                    "text": " Like earlier, if I wanted to crunch my data, I wanted some ratio analysis and all, I would have to ask my EA, okay, do a research on this sector, compare this, compare this, compare this, and make this thing. Today, I can ask it a question. Yes, does it make a mistake? Yes, maybe one in a 10, but I am the expert who's able to understand, look at it, and find the mistake. If somebody else was doing it, maybe the mistake would never get found."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:55",
                    "end": "00:28:18",
                    "text": " So, but today as an expert, each and every one of you here has to become AI expert. Okay? You have to be expert at what you do and you have to use AI to enable you to be a 10x. Now, when you talk about our financial services, wealth management, it was a very people's oriented world where people were using their brains."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:18",
                    "end": "00:28:45",
                    "text": " But what was restricting them is that I don't have research access to this company. I don't have this data. I don't have that. I have asked my research guys, they are not responding in five days. I want a portfolio uploaded. The data entry guy is going to take two hours. Today, AI can do everything. So it's a combination of AI, RPS. So robotic process automations are also huge because there are a lot of connections to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:45",
                    "end": "00:29:15",
                    "text": " pull in that data. But if you do that and you build that yourself, you will be 10 X off. So what we see our biggest challenge used to be ability to deliver service, the ability to manage the number of people. And that's where AI is rocking for financial services. It is the biggest boon. I think the financial services is going to see the massive, greatest increase in their profits because they are people who understand this."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:15",
                    "end": "00:29:44",
                    "text": " will become much, much more productive. You know, I very personally, there is a recording. I tried to build processes where humans were not involved. And frankly, today, AI is not ready for that. But processes where AI enables a human, it is very much ready. But there needs to be a human eye to see whether it's doing. And that happens with my EA also. I was joking with them. You make mistakes 20% of the time. Today, AI makes only 10%."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:44",
                    "end": "00:30:14",
                    "text": " Take it, but we take that 10% mistake and we hype it up. Oh my god. There's a mistake You have to figure out why was the mistake why is it happening as he said you have to figure it out and eventually You know you can just write a prompt check what this guy has done So I have made a maker checker in my own prompt that okay Everything is done should get rechecked and today mistakes have come down to maybe point zero zero one reason I'm saying point zero zero one because I've not seen a mistake in the last two months"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:14",
                    "end": "00:30:42",
                    "text": " I have not seen a mistake. I have not seen numbers getting crunched wrong or some fake numbers coming in or something. So today, AI has become so exciting. Earlier, you had to even do the prompting. Today, you can just simply ask, how do you expect me to make a prompt to get this thing done? So it will make a prompt for you also. It will first explain to you, do this, this, this. This is how you make a prompt. Then it will make a prompt."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:42",
                    "end": "00:31:09",
                    "text": " So today it's becoming so easy. So guys, use it to enhance yourself. Use it to empower the self. You know, I keep giving these lectures to people. I spent in Dubai with 150 CEOs. We did a session. But one thing that really irked me there, that most of the people who are not Indian origin were actually doing it themselves. Their level of AI, NA 10s or the mix and this, the..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:09",
                    "end": "00:31:35",
                    "text": " Automation, they all knew it by heart. They were doing it themselves. Whereas a typical Indian guy, okay, can I find an EA who can do it for me? Can I find an AI expert? I want to hire this guy. Now point is even if you hire, he's not going to be able to do it. But for the youngsters in here, it is the biggest opportunity, the people who are not ready to change. You don't get old, you get rigid."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:35",
                    "end": "00:32:04",
                    "text": " The people who have got rigid, today you have the flexibility to replace them. Thank you. What you said about EA is what I experienced. So I was asking my EA yesterday to book my tickets for Monday. And she said, Prashant, I'm not getting the rightly priced tickets for you. What should I do? So instead, I became her EA. And I went to the Gemini. And I said, find me the best priced tickets which fly at this time. I got seven options. They're all."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:04",
                    "end": "00:32:33",
                    "text": " within the price range that I wanted it. And I sent it to her, see these are the tickets, go and book now. So. So I've actually built myself an EA, AI EA, which automatically will enter every meeting, summarize every meeting, find out who's there, what needs to be done with them, what is the follow-up required with them. I've created a separate email so that people know that's my AI avatar emailing them. But it'll keep following them, hounding them."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:33",
                    "end": "00:33:23",
                    "text": " So a question more grounded in on return of investments. So how do you see, how do you plan for return on investments with AI, which is grounded in business outcomes? Because I am in consult. This is very, very important. You know, typically I am, we made a committee. We chose 15 AI projects and then I got get a report. Seven are done."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:23",
                    "end": "00:33:53",
                    "text": " I say, friend, what does done mean? AI is never done. Once you think you have completed, it has to go into monitoring. You have to see cost-benefit analysis. When I said I tried to replace human beings, what I realized, AI wasn't reducing my cost. Today, some processes we have put on AI, like verification calls, because there, some human beings used to get compromised. So those things we have put on AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:53",
                    "end": "00:34:14",
                    "text": " But honestly, they are costing me 7x of what a human being was doing. We have put AI on monitoring every camera in the country and immediately flagging if something goes wrong. Today, it's costing me 4x that my human team used to be doing. So point is that, look, AI is not able to become cheaper than us."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:14",
                    "end": "00:34:40",
                    "text": " But what AI is doing, it can do things where people used to make compromises or people used to miss out. It will not miss out on those things. So today, I feel in financial services, it is going to be where I am investing in our processes, which are very, compliance is very required or those processes cannot be compromised. You know, it is checking something where I'm willing to pay an extra price to make sure I get"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:40",
                    "end": "00:35:09",
                    "text": " So this is where I am investing in today. Second thing I'm investing is capability building of my team. Those people, I have launched a program called Beyond. Those people who are willing to change with AI. So I am investing in them, making them AI enabled, taking them to the next level and making it clear that you may even replace your boss if he doesn't agree to change with the time. So that's the second place. The AI training, I think, is a big thing."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:09",
                    "end": "00:35:36",
                    "text": " and in financial services see our data as I said used to be so I think overall the benefit side I think returns are going to get much better we are going to deliver much better service earlier some people used to get sold a wrong insurance or some people would buy a product which was not right for them today that's not going to happen with AI AI is going to explain to them AI is going to tell please listen to these five points you must know them take care or"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:36",
                    "end": "00:36:04",
                    "text": " Today, we are able to quicken our service. 24 by 7, we are able to give him a response then and there. We have built an AI bot to explain motor insurance, do quotes. Earlier, making a quote used to take half an hour. Today, that quote can be made in 30 seconds. And customer can be told, comparisons can be done. So I think the experience is going to go up. The portfolio returns are going to go up because a lot of people had mistakes in the portfolio."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:04",
                    "end": "00:36:26",
                    "text": " For us to analyze two lakh portfolios was a crazy task. Today my AI analyzes 500 portfolios every day and I'm trying to save on the cost of compute. But I can actually do even two lakh portfolios. It is computing, it is finding out the mistakes, it's finding out how the portfolio can be optimized. It is making a lead."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:26",
                    "end": "00:36:56",
                    "text": " for the wealth manager, okay, talk to your client as per us. This is a discussion you should have. This is what is causing or these are the things. And what we are seeing is that where we have implemented this in the last seven, eight months, the portfolio returns are on average at least five to 10% higher than non-optimized portfolios. So that's a very exciting part. Thank you, Sanjeev. We've got about 10 minutes to go."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:56",
                    "end": "00:37:36",
                    "text": " what I would do is ask one more question to all three of you, and then we'll open it for our audience to ask questions. So skill versus mindset. What do you see in this new world of AI-led automation? Is it re-skilling people or rewiring decision-making more important, and how would you balance it? Ladies always. What I would say to that is I think both are very important."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:36",
                    "end": "00:38:04",
                    "text": " I did speak a little bit about the work that we are doing on skilling. The example I gave is across our 60,000 colleagues, we have extended tools like Microsoft Copilot. We have trained 100% of our colleagues and about 50% of them have actually gone back themselves and chosen to do more deeper re-skilling."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:04",
                    "end": "00:38:33",
                    "text": " I also shared on the earlier question that our AI and data ethics code of conduct and the trainings that we've developed in partnership with University of Edinburgh, as well as of course, we work with several Indian tier one institutes, et cetera. We have developed a program that we've actually all of our colleagues across the bank have."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:33",
                    "end": "00:39:02",
                    "text": " have attended and certified. But it is about engaging all levels of leadership. I think in a prior question, I said that more than digitization projects or tech projects, I think it's really important to engage leaders at all levels when you're doing AI deployment at scale, especially, as well as help them think about"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:02",
                    "end": "00:39:31",
                    "text": " responsible AI at every layer of the implementation, every layer of the stack. So it is going to be about both. I would also give in that this group is largely, I think, banking and financial services folks. I will reiterate something I said in the beginning. For us, it is about building deeper and more trusted customer relationships."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:31",
                    "end": "00:39:57",
                    "text": " And AI is helping us deepen those relationships, understand our customers better, know and meet them in the places and choices and moments of matter for them, but also giving them services that are going to help transform banking for not just our customers, but for banking more broadly. And I think to do that,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:57",
                    "end": "00:40:25",
                    "text": " having leaders really think about customer experiences first and think about how do we deepen and leverage AI to develop more trust and develop more deeper understanding of the needs of our customers, I think is the biggest important skill to have. In addition to, of course, all of the others that we may be building"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:25",
                    "end": "00:40:55",
                    "text": " in areas of expertise that I've already talked about. So as an example, I think I shared earlier that about 12,000 of our engineers across the bank actually use AI for coding. And about 35% of the code is actually generated by AI. But they have to be, we want to encourage human agency. We want to make sure that we are building systems that our customers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:55",
                    "end": "00:41:24",
                    "text": " continue to build deep trust with. So therefore, engaging our colleagues, making sure each one of them understands and understands their responsibility in it is extremely important for us. Yeah. Thank you, Ruchika. Sanjeev, sir. For me, there is no re-skilling. It is only rewiring. Because as I've been trying to make you understand, nobody can teach you AI because it is so dynamic. Every day, there's a new tool."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:24",
                    "end": "00:41:53",
                    "text": " The only way you can learn AI is do it yourself or go to YouTube. At the end of the day, you can watch some video. Somebody, somewhere in the world would have made something. That's the only way. But you have to rewire yourself. Is that how do I increase my productivity? I understand. See, I'm a business owner. And AI is all about risk, benefit. What can I share? What data do I put in? I'm able to take all the risk. So I'm able to take that call."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:53",
                    "end": "00:42:21",
                    "text": " The only person who can out-vote me is my compliance. So at the end of the day, no, I am not going to allow that. That's fine. That's the role I've given to him. But the point is that I'm able to do it. And that's what I see. The AI is actually getting used by the leaders. But where we have to rewire ourselves is that we have to understand that we now will have to work much harder. You know, future offices are not going to be 100 people sitting."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:21",
                    "end": "00:42:49",
                    "text": " It is going to be 15 people sitting surrounded by 5, 10 screens. And those 5, 10 screens each has 10, 15, 20 bots which are working. And which are doing the work for them. Today, but we have to rewire ourselves to accept that the world has changed. This is a moment like when internet came. At that time, a lot of people said, I will never adopt to internet."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:49",
                    "end": "00:43:15",
                    "text": " or when the typewriters, people had to give up typewriters and move from typewriters to computers. And the kind of rewiring that was required at that time. So a lot of, but those time, that time we had time. Now those same changes are happening at 50x the speed. You know, what I learned today becomes outdated. The bot you make today will not work after one month because the model would have changed."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:15",
                    "end": "00:43:40",
                    "text": " So you have to now start doing it yourself. You have to put your legs up. For me, my most productive days are the days I'm sitting alone and I'm working on training my AI, training my agent, training my AI EA, training my legal compliance guy, what needs to be done, how I need it to be answered, what are the things that he needs to look after. So today, it's become very, very important."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:40",
                    "end": "00:44:00",
                    "text": " Like I shared my own example, some of the bots I was trying to make, I was trying to explain to others, I want this, I want this. Maybe I had 50 meetings, I was not able to explain, but when I started making it myself, it took me six, seven hours, I had to learn those integrations and this thing. But I was able to make it and then I'm able to tweak it in one hour."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:00",
                    "end": "00:44:28",
                    "text": " So I think it is all about rewiring ourselves, that the world is changing, and let us change with this time. Let us be flexible, let us not be rigid, and let us change, let us adopt. And this is such an exciting opportunity for those who are willing to rewire. Companies are dying to hire you. You know, I am today hiring people. You know, I sometimes put a post, if you are innovative, you are this thing. And I've actually ended up hiring six, seven people who are working closely with me."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:28",
                    "end": "00:44:57",
                    "text": " and they're doing a fantastic job. Not the head, but one of my marketing guys, he sent me a nice presentation, young MBA grad, and I sent message to my marketing, I said, hire this guy. Today, he's doing such a fantastic job, what five people used to do, he's doing it himself because he knows how to master AI. He's doing SEO optimization, he's doing this, he's doing this, he's doing that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:57",
                    "end": "00:45:33",
                    "text": " because he's been able to make bots. And that was his USP. And I asked him, how long did it take you to learn, master these things? He says, sir, three weeks I invested in myself. Imagine the power. Three weeks you invest will change your life. Thanks, Sanjeev. I just say before Dr. Dixit speaks, I want to be one of those 15 people that are left in the room. I think I completely agree with both Sanjeev and Ruchika."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:33",
                    "end": "00:46:00",
                    "text": " And I'm experiencing this right, you know, firsthand. So we are hiring, we are also re-skilling, we are also re-learning in our teams. And the productivity increase is tremendous, right? Right now there might not be a percentage to put to that. But in case I want to hire a new person, it would rather be an intern who has the will to learn something new and who can sit with us throughout and..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:00",
                    "end": "00:46:28",
                    "text": " understand what we need to do and do that through AI. And the example that Sanjeev gave, right, 20 years back, 25 years back when the computers first came, there was so much resistance and there was so much apprehensions and there was so much thoughts that yeh kaise hoga, kahan hoga. But look, today the world is different. And like one of the leaders yesterday was mentioning in the speeches given with the PM, industrial revolution which took 200 years to take place."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:28",
                    "end": "00:46:55",
                    "text": " We are going through that same change. Only thing is that the complete change and impact and momentum that happened in 200 years is now going to be compressed in just a few years. And that is the power of technology that's happening. Anybody who's sitting in this room or anybody who's outside who says that I will not take part in this is only, you know, forcibly keeping himself out of the playing field."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:55",
                    "end": "00:47:23",
                    "text": " and he's going to watch the world go by. All companies everywhere, whether it be directly involved in technology, whether it be BFSI or media, everyone is touched by this power of AI. The only thing that government is doing today is enabling the complete stacks right from the chips right up to the application layer. Barring the semiconductors, we have the capability to do all the six layers of the AI stack today."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:23",
                    "end": "00:47:45",
                    "text": " And the change that will be brought about within the country and in every institution, every enterprise is going to happen from each one of you. And you have to learn it. And today, learning is so easy. You just have to go to any of the mediums and you can re-skill and rewire yourself. You just have to have the will. So the people that we hire, the people that we have inside,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:45",
                    "end": "00:48:10",
                    "text": " We are looking at how they can re-engineer and produce the same kind of products, maybe better, more secure, more proven and trustworthy, but with the capability of AI that's built in. And being in the FinTech industry, BFSI, this is the most important thing that you can do that today you reskill and rewire and understand the AI power of AI and do that for your team, for your function, for your company. And that is the most important."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:17",
                    "end": "00:48:51",
                    "text": " I truly believe financial services stocks are going to deliver fantastic returns over the next few years, especially in the companies who are able to know how to tame the AI. I have invested in New Equity. Other than Nvidia and Broadcom, I also have Bajaj in my portfolio. Thank you. Last four minutes. If there are any questions, please ask. We may not be able to take all, but quick one question at a time. Go ahead."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:07",
                    "end": "00:49:35",
                    "text": " issue that is due to we don't understand the thing that is data breaches in the fintech segment. So the data in the fintech is wide. So I wanted to ask you all that how the companies, how are your companies maintaining the data breach segment that the data is protected and cyber security is ensured in all the segments. I think Ruchika will answer this best. You know I'll answer the question really talking about data strategy. You know clearly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:35",
                    "end": "00:50:03",
                    "text": " AI without data would be fairly artificial, right? And so for us at NatWest Group, we have really doubled down on our data strategy, and are in fact, there's a public partnership we announced with AWS and Accenture last year, where they're working with them across the data that we have."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:03",
                    "end": "00:50:32",
                    "text": " to make sure that we have the right guardrails, have the right democratization and therefore are able to power AI through it. So the only thing I will say on that is that absolutely your data strategy and how you are organizing the data and how you're democratizing data across the organization and everything that you do when making sure that it is accessible to AI and therefore"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:32",
                    "end": "00:51:02",
                    "text": " is able to build the right customer experiences is critical. Thank you ma'am. I would like to add because he has a specific query and see having built systems from ground upwards and we are also building systems for hundreds of our clients and we have client data, right? What we must ensure is the client's data is secured. So you might have heard of the term multi-tenant, right? So when you create a product, the product could be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:02",
                    "end": "00:51:13",
                    "text": " unique and it could be universal, but the data that you use of the client has to be siloed. So you have to have zero trust security and show that when the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:13",
                    "end": "00:51:45",
                    "text": " product talks to that particular data, it happened only for that client. I cannot be providing you your answers based on somebody else's data, right? So ensure this zero trust data, the data are siloed, protected, whether by encryption, by other means, but I cannot have this happening because then the client will lose trust. Accessibility and protect the client's data and keep it separate from other client's data. Methods you use would be unique to you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:45",
                    "end": "00:52:15",
                    "text": " One more question. Yeah. Hi, so I sit at an intersection of private credit and tech. I want to know your perspective. I mean across BFSI we have seen AI is taking over fraud detection, support and the mainstream tasks. But when we talk of the other side, which is credit, you know, I do not have an agent yet that can handle my disbursements. What are the bottlenecks, you know, that's preventing this? I mean, sir, you mentioned that N8N and make is being used extensively outside of India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:15",
                    "end": "00:52:44",
                    "text": " Is it these social factors? I mean, what exactly are the bottlenecks that stop us doing that? So you know, one very interesting point that when we are moving to a virtual world, what is going to become the most important currency? Let me ask you that question. Data is definitely going to be there. I'm talking about beyond that. For to build a trust, you need a physical connect. Yes or no? You want to see, you get an answer."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:44",
                    "end": "00:53:13",
                    "text": " that you don't know whether AI has answered it or a human has answered it and who's behind it. So you want a physical connect. So what is happening in this AI-based data rush is also that a lot of the forward-thinking organizations have started investing in physical branches also, so that where people can go, visit, touch, and validate. So that also has happened because today, physical presence will have to be done."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:13",
                    "end": "00:53:41",
                    "text": " you know, people moving from compliance tasks, checking operations, it's going to move to sales. I think sales guys are going to and relationship guys are going to become much, much more important. Their role is because people will want when you when you're talking about your money, you don't want to give your money only to a computer which you can never see. You want to give your money to a computer which is helping a person who you can see who you can catch who you can talk to. So I think that physical presence"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:41",
                    "end": "00:54:08",
                    "text": " is going to be definitely, there are some roles which are going to get enhanced and especially sales or any expert role. I think it's a very exciting place that we are living in. There are a lot of exciting things happening. When you talk about data, see, as I said, I will answer that little bit question. You don't have to put everything in AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:08",
                    "end": "00:54:18",
                    "text": " You have to be very careful. AI is very, very expensive. You use AI only to do things which only AI can do. Rest you use an RPA."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:18",
                    "end": "00:54:46",
                    "text": " An RPA or data is with you, it's a robotic process, it's a software written, do this calculation, do this, do this, do this. Why would you want to expose that to AI? It's not required. So it needs to be the right combination and the way we are doing it is we are protecting it by only giving very limited amount of data and not complete data to one AI agent also. So like that's what we do with a human being, that you get only very limited part of the information on which you are supposed to act."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:46",
                    "end": "00:55:16",
                    "text": " not give you the whole information. We've run out of time and we'll have the next session starting. You can find me on LinkedIn, send me your questions, I'll get it answered. If there's a question needs to be answered by Sanjeev, Ruchika or Dr. Dixit, I will reach out to them, get the answers and send it back to you. So you can send questions on LinkedIn to me, I'll respond to all of them. Thank you for patience and listening to us."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
    {
    "_id": "6996fc126c61c5bef9ef788b",
    "projectId": "894303",
    "projectName": "Sam altan at the  AI Summit 2026 - Bengali",
    "projectDescription": "Sam altan at the  AI Summit 2026",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T12:03:28.187Z",
    "projectUpdatedDate": "2026-02-19T12:03:28.187Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-19T12:04:23.147Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Sam altan at the  AI Summit 2026.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Bengali"
},
    {
    "_id": "6996fe176c61c5bef9ef793d",
    "projectId": "886032",
    "projectName": "Vishal Sikka, Rajesh Subramaniam (FedEx CEO), Rishi Sunak (ex-UK PM) - German",
    "projectDescription": "Vishal Sikka, Rajesh Subramaniam (FedEx CEO), Rishi Sunak (ex-UK PM),",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T12:12:05.441Z",
    "projectUpdatedDate": "2026-02-19T12:12:05.441Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-19T12:13:00.482Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Vishal Sikka, Rajesh Subramaniam (FedEx CEO), Rishi Sunak (ex-UK PM),.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "German"
},
{
    "_id": "6996f63c565455da1fc761e5",
    "projectId": "404208",
    "projectName": "Vishal Sikka, Rajesh Subramaniam (FedEx CEO), Rishi Sunak (ex-UK PM), - Gujrati",
    "projectDescription": "Vishal Sikka, Rajesh Subramaniam (FedEx CEO), Rishi Sunak (ex-UK PM),",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T11:38:35.372Z",
    "projectUpdatedDate": "2026-02-19T11:38:35.372Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-19T12:14:38.270Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Vishal Sikka,....docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Gujarati"
},
{
    "_id": "6996fe796c61c5bef9ef795b",
    "projectId": "473834",
    "projectName": "Sam altan at the  AI Summit 2026 - Italian",
    "projectDescription": "Sam altan at the  AI Summit 2026",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T12:13:43.425Z",
    "projectUpdatedDate": "2026-02-19T12:13:43.425Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-19T12:14:42.386Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Sam altan at the  AI Summit 2026.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Italian"
},
    {
    "_id": "6996b1c2565455da1fc75c60",
    "projectId": "772812",
    "projectName": "sundar Pichai speech",
    "projectDescription": "sundar Pichai speech",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T06:46:26.146Z",
    "projectUpdatedDate": "2026-02-19T06:46:26.146Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-19T09:59:39.629Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Sundar pichai Speech.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
    {
    "_id": "6996e288565455da1fc7600c",
    "projectId": "897493",
    "projectName": "Rishi Sunak",
    "projectDescription": "Rishi Sunak",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T10:14:30.357Z",
    "projectUpdatedDate": "2026-02-19T10:14:30.357Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-19T11:22:18.691Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Rishi Sunak Trans.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
    {
    "_id": "6996e524565455da1fc76063",
    "projectId": "723829",
    "projectName": "Fedex CEO Rajesh Subramanian",
    "projectDescription": "Fedex CEO Rajesh Subramanian",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T10:25:38.505Z",
    "projectUpdatedDate": "2026-02-19T10:25:38.505Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-19T11:20:54.106Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "nscript Fedex CEO.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
    {
    "_id": "6996e41e565455da1fc76051",
    "projectId": "831257",
    "projectName": "Arthur Mensch",
    "projectDescription": "Arthur Mensch",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T10:21:17.107Z",
    "projectUpdatedDate": "2026-02-19T10:21:17.107Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-19T11:15:07.397Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "transcript Arthur Mensch.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6996deab565455da1fc75f96",
    "projectId": "148102",
    "projectName": "Schneider Electric CEO",
    "projectDescription": "Schneider Electric CEO",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T09:58:01.490Z",
    "projectUpdatedDate": "2026-02-19T09:58:01.490Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-19T09:58:43.504Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "If We Crack The Code In India, We Crack It Everywhere,” Schneider Electric CEO at AI Summit.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
    {
    "_id": "6996f050565455da1fc7616e",
    "projectId": "796808",
    "projectName": "Jeet Adani, Adani digital Labs",
    "projectDescription": "Jeet Adani, Adani digital Labs",
     "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T11:13:18.739Z",
    "projectUpdatedDate": "2026-02-19T11:13:18.739Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-19T11:59:27.267Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Jeet Adani Adani Digital labs.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6996dff5565455da1fc75fdc",
    "projectId": "948436",
    "projectName": "Roy Jecobs Philips",
    "projectDescription": "Roy Jecobs Philips",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T10:03:31.449Z",
    "projectUpdatedDate": "2026-02-19T10:03:31.449Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-19T10:04:22.240Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Roy Jecobs Philips Trans.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6996e9e4565455da1fc760de",
    "projectId": "990530",
    "projectName": "Alexendr Wang",
    "projectDescription": "Alexendr Wang",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T10:45:53.997Z",
    "projectUpdatedDate": "2026-02-19T10:45:53.997Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-19T10:59:59.569Z",
    "projectDeleted": false,
    "projectDataLength": 3,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "Alexendr wang CEO Meta trans.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "Alexendr wang CEO Meta trans.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6996dd16565455da1fc75f6b",
    "projectId": "885286",
    "projectName": "Julie Sweet accenture",
    "projectDescription": "Julie Sweet accenture",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T09:51:17.604Z",
    "projectUpdatedDate": "2026-02-19T09:51:17.604Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-19T11:00:58.094Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "Accenture_CEO_Julie_Sweet_speaks_at_India_AI_Impact_summit___AI_Growth_in_India__PM_Modi__AI_Use__IT(M4A_128K)[1].mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771494666171.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:28",
                    "text": " and private partnerships and the urgency of what we must do in order for AI to drive growth. So companies, companies must be willing to reinvent how they operate, their processes, how they've been doing work for the last decades. Underneath the headlines of a failure of AI is mostly a failure to reinvent."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:28",
                    "end": "00:00:54",
                    "text": " Companies have to invest to reshape their workforces, and companies must commit to creating sustained entry-level jobs. Now, entry-level jobs makes economic sense. They're the only way to create future leaders, and they bring needed, truly AI-native talent to each of our organizations."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:54",
                    "end": "00:01:22",
                    "text": " But AI fundamentally is changing what an entry-level job looks like. And so a commitment means we have to be intentional about changing the rules, investing in training, which is exactly what Accenture is doing. We will hire more entry-level into more entry-level jobs this year than last year. But the skills we require and the way we're onboarding those individuals is fundamentally different."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:22",
                    "end": "00:01:51",
                    "text": " Now, countries must also reinvent. They must reinvent their role and how they work with the private sector. They have to, themselves as governments, become the best credential for why AI matters. They must work with the private sector to help create lifelong learning because education is no longer a destination. We have to have lifelong learning."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:51",
                    "end": "00:02:22",
                    "text": " India is doing a great job of embedding AI into the educational system, starting in primary school, and governments across the world will need to do so. At the same time, as countries are thinking differently, individuals have to think differently and recognize that formal education is no longer the destination. But perhaps the biggest fundamental change that must be made is that companies and countries need to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:22",
                    "end": "00:02:46",
                    "text": " hound the table for global standards. These standards should apply to safety, but also to the industries where AI can make the greatest impact. For example, in pharma, if one country is allowing pharma companies to use the latest technologies to discover drugs and then test drugs, but other countries don't follow suit,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:46",
                    "end": "00:03:17",
                    "text": " It means that you won't be able to scale. You won't be able to bring it. And we know that most often that impacts the most vulnerable. Now, we have a view, of course, that we have to reinvent. But as we think about that reinvention or to our future is the fundamental belief that it is humans in the lead, not humans in the loop that will shape that future."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:17",
                    "end": "00:03:46",
                    "text": " We should not confuse how you deploy AI responsibly. Of course, all of our compliance programs have humans, they have technology. That doesn't change the critical lesson that we've learned over and over again. Technology, no matter how powerful, is only a tool. It is simply a tool. It is leaders"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:46",
                    "end": "00:04:16",
                    "text": " who decide how to use those tools. It is leaders who decide to commit to reinvent, who dedicate their time to making sure that people come along the journey. And it is leaders who must choose to work together to ensure the safe, widespread adoption of AI. There are lots of headlines today that predict less."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:16",
                    "end": "00:04:47",
                    "text": " less jobs, less opportunity, less human relevance. We are here because we see a future of more. At Accenture, we live by eight leadership essentials, the qualities we believe we need to run our company. And one of them is particularly important. We expect leaders to lead with excellence, confidence, and humility."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:48",
                    "end": "00:05:13",
                    "text": " As we look to our collective future, we should have the confidence to have the unwavering belief that together we can make a future that is better for all. We also must hold ourselves individually and collectively accountable for executing on that belief"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:13",
                    "end": "00:05:52",
                    "text": " with a high bar of excellence because our people around the world are counting on that excellence. And finally, we must all have the humility to know that we cannot do this alone. Thank you very much. Thank you so much, Ms. Julie Sweet. I think I can take a tagline out of an address which says that AI should make the impossible possible, but"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:52",
                    "end": "00:06:18",
                    "text": " Thank you for your thoughtful articulation of AI's impact on industry and on society. Ladies and gentlemen, our next speaker is the youngest billionaire and growing reinventers here in India. We also have one of the largest AI workforces in the world, tightly integrated with our growing AI hubs in the US, Europe, the Middle East and Japan."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:18",
                    "end": "00:06:49",
                    "text": " And I want to take this time to thank all of our people in India for your incredible commitment to value and to our clients. Today, I want to leave you with three perspectives that we believe will help us ensure that AI's immense potential is captured for the benefit of all. First, using AI as an engine for growth is the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:49",
                    "end": "00:07:21",
                    "text": " only path for global prosperity for all. Second, the agenda ahead of us is unprecedented. Companies, countries, and individuals must reinvent how they work, how they work together, and how they learn. And finally, it is humans in the lead, not humans in the loop that will determine our future. As we turn to the imperative for growth,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:21",
                    "end": "00:07:51",
                    "text": " I want to take you back for a moment to 2013. Oxford University had just published a widely read study that said based on technology progress at that time, 47% of US jobs would be automatable. Dire headlines and predictions soon followed. One of those technologies was robotic process automation or RPA."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:51",
                    "end": "00:08:19",
                    "text": " And there were predictions that IT services would be badly damaged because it would automate so many jobs. And in fact, we used RPA to automate thousands of jobs. And we also, as an industry, embraced the new technologies of digital and classical AI, and we created many, many more jobs. We helped our clients adopt RPA and those who did,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:19",
                    "end": "00:08:38",
                    "text": " created investment capacity to invest in new technologies and to grow. And in fact, the IT services industry has thrived over the last decade, including many of India's most successful companies that you've heard from today. At Accenture alone in 2013,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:38",
                    "end": "00:09:09",
                    "text": " We were roughly 275,000 people and $29 billion in revenue. And today we're over 750,000 and growing and $70 billion in revenue. What the last decade has taught us is a critical lesson. When companies and countries embrace new technologies and then use them to drive growth and productivity, they prosper."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:09",
                    "end": "00:09:40",
                    "text": " Advanced AI should be the same. In fact, in our latest quarterly survey of C suites across 20 countries, they agree. 78% say AI's greatest value is in growth. Now, as we think about what growth should look like, there's two important considerations. First, AI should make the impossible possible."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:40",
                    "end": "00:10:07",
                    "text": " AI should make the impossible possible. If in a few years as a CEO, you cannot point to new products and services, new levels of performance that were not possible before, then you have not captured potential of AI. Think about the consumer and retail industries. LLMs are about to become the new mall."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:07",
                    "end": "00:10:35",
                    "text": " This is an entirely new way to engage customers and to engage in commerce that did not exist in 2022. If you think about pharma, we see a path toward bringing drugs to market much faster than the average of nine years. Not possible before, which means that life-saving drugs will get to people faster and pharma will have accelerated growth."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:35",
                    "end": "00:11:05",
                    "text": " And we are just beginning to understand how AI will create new drugs, new materials, new products across industries. A second consideration around growth is that we must commit to providing access to the technology and the talent for small and medium-sized enterprises. If we are to use AI as an engine for growth, we need to make sure that the engine for growth"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:05",
                    "end": "00:11:33",
                    "text": " These types, these size enterprises have access. 50% of the world's GDP are small and medium sized enterprises. And in the global south, it's 70% of employment. To do so, there will be lots of business opportunities. So many industries will serve small and medium sized enterprises. But that will not be enough. Private and public partnerships will be critical to making sure there's access."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:33",
                    "end": "00:12:03",
                    "text": " For example, we're working with the U.S. college system where we're funding internships of college students at small and medium-sized enterprises. It's a win-win. Statistically, if you have an internship, you have a better chance of getting a job, and it's providing these enterprises access to some of the cutting-edge talent. And so we must make sure that we're continuing to focus on the small and medium-sized enterprises. Now, I know, and we all know,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:03",
                    "end": "00:12:32",
                    "text": " that advanced AI is much more powerful than the technology advancements of the last decade. And of course, that means that the impact is more profound. But that doesn't change the critical lesson that AI must be used for growth and productivity. What it does change are the sets of actions."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:32",
                    "end": "00:13:01",
                    "text": " the time frame, the need for global collaboration, the need for more public and private partnerships, and the urgency of what we must do in order for AI to drive growth. So companies. Companies must be willing to reinvent how they operate, their processes, how they've been doing work for the last decades. Underneath the headlines of a failure of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:01",
                    "end": "00:13:31",
                    "text": " AI is mostly a failure to reinvent. Companies have to invest to reshape their workforces. And companies must commit to creating sustained entry-level jobs. Now, entry-level jobs makes economic sense. They're the only way to create future leaders. And they bring needed, truly AI-native talent to each of our organizations."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:31",
                    "end": "00:13:59",
                    "text": " But AI fundamentally is changing what an entry-level job looks like. And so a commitment means we have to be intentional about changing the rules, investing in training, which is exactly what Accenture is doing. We will hire more entry-level into more entry-level jobs this year than last year. But the skills we require and the way we're onboarding those individuals is fundamentally different."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:59",
                    "end": "00:14:28",
                    "text": " Now, countries must also reinvent. They must reinvent their role and how they work with the private sector. They have to, themselves as governments, become the best credential for why AI matters. They must work with the private sector to help create lifelong learning because education is no longer a destination. We have to have lifelong learning."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:28",
                    "end": "00:14:59",
                    "text": " India is doing a great job of embedding AI into the educational system starting in primary school and governments across the world will need to do so. At the same time, as countries are thinking differently, individuals have to think differently and recognize that formal education is no longer the destination. But perhaps the biggest fundamental change that must be made is that companies and countries need to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:59",
                    "end": "00:15:23",
                    "text": " hound the table for global standards. These standards should apply to safety, but also to the industries where AI can make the greatest impact. For example, in pharma, if one country is allowing pharma companies to use the latest technologies to discover drugs and then test drugs, but other countries don't follow suit,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:23",
                    "end": "00:15:54",
                    "text": " It means that you won't be able to scale. You won't be able to bring it. And we know that most often that impacts the most vulnerable. Now, we have a view, of course, that we have to reinvent. But as we think about that reinvention, or to our future, is the fundamental belief that it is humans in the lead, not humans in the loop, that will shape that future."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:54",
                    "end": "00:16:23",
                    "text": " We should not confuse how you deploy AI responsibly. Of course, all of our compliance programs have humans, they have technology. That doesn't change the critical lesson that we've learned over and over again. Technology, no matter how powerful, is only a tool. It is simply a tool. It is leaders"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:23",
                    "end": "00:16:53",
                    "text": " who decide how to use those tools. It is leaders who decide to commit to reinvent, who dedicate their time to making sure that people come along the journey. And it is leaders who must choose to work together to ensure the safe, widespread adoption of AI. There are lots of headlines today that predict less."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:53",
                    "end": "00:17:24",
                    "text": " less jobs, less opportunity, less human relevance. We are here because we see a future of more. At Accenture, we live by eight leadership essentials, the qualities we believe we need to run our company. And one of them is particularly important. We expect leaders to lead with excellence, confidence, and humility."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:24",
                    "end": "00:17:53",
                    "text": " As we look to our collective future, we should have the confidence to have the unwavering belief that together we can make a future that is better for all. We also must hold ourselves individually and collectively accountable for executing on that belief with a high bar of excellence because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:53",
                    "end": "00:18:36",
                    "text": " our people around the world are counting on that excellence. And finally, we must all have the humility to know that we cannot do this alone. Thank you very much. Thank you, Prime Minister Modi, Minister Vaishnav and your outstanding teams for convening us for this critical summit around AI. The breadth"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:36",
                    "end": "00:19:05",
                    "text": " distinguished guests from around the world is a recognition of the importance of broad global partnerships to capture the incredible potential of AI and address the risks. It is also a recognition of the importance of India in our AI enabled future. At Accenture, we're incredibly proud to have over 350,000 and growing reinventers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:05",
                    "end": "00:19:28",
                    "text": " here in India. We also have one of the largest AI workforces in the world tightly integrated with our growing AI hubs in the US, Europe, the Middle East, and Japan. And I want to take this time to thank all of our people in India for your incredible commitment to value and to our clients."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:29",
                    "end": "00:19:59",
                    "text": " Today I want to leave you with three perspectives that we believe will help us ensure that AI's immense potential is captured for the benefit of all. First, using AI as an engine for growth is the only path for global prosperity for all. Second, the agenda ahead of us is unprecedented."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:59",
                    "end": "00:20:30",
                    "text": " Companies, countries, and individuals must reinvent how they work, how they work together, and how they learn. And finally, it is humans in the lead, not humans in the loop, that will determine our future. As we turn to the imperative for growth, I want to take you back for a moment to 2013. Oxford University had just published"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:30",
                    "end": "00:20:59",
                    "text": " a widely read study that said based on technology progress at that time, 47% of US jobs would be automatable. Dire headlines and predictions soon followed. One of those technologies was robotic process automation or RPA. And there were predictions that IT services would be badly damaged because it would automate so many jobs. And in fact,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:59",
                    "end": "00:21:24",
                    "text": " We used RPA to automate thousands of jobs. And we also, as an industry, embraced the new technologies of digital and classical AI, and we created many, many more jobs. We helped our clients adopt RPA, and those who did created investment capacity to invest in new technologies and to grow."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:24",
                    "end": "00:21:51",
                    "text": " And in fact, the IT services industry has thrived over the last decade, including many of India's most successful companies that you've heard from today. At Accenture alone in 2013, we were roughly 275,000 people and $29 billion in revenue. And today we're over 750,000 and growing and $70 billion in revenue."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:51",
                    "end": "00:22:18",
                    "text": " What the last decade has taught us is a critical lesson. When companies and countries embrace new technologies and then use them to drive growth and productivity, they prosper. Advanced AI should be the same. In fact, in our latest quarterly survey of C suites across 20 countries, they agree."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:18",
                    "end": "00:22:45",
                    "text": " 78% say AI's greatest value is in growth. Now as we think about what growth should look like, there's two important considerations. First, AI should make the impossible possible. AI should make the impossible possible."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:45",
                    "end": "00:23:15",
                    "text": " If in a few years as a CEO, you cannot point to new products and services, new levels of performance that were not possible before, then you have not captured potential of AI. Think about the consumer and retail industries. LLMs are about to become the new mall. This is an entirely new way to engage customers and to engage in commerce that did not exist."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:15",
                    "end": "00:23:44",
                    "text": " in 2022. If you think about pharma, we see a path toward bringing drugs to market much faster than the average of nine years. Not possible before, which means that life-saving drugs will get to people faster and pharma will have accelerated growth. And we are just beginning to understand how AI will create new drugs, new materials, new products across industries."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:45",
                    "end": "00:24:13",
                    "text": " A second consideration around growth is that we must commit to providing access to the technology and the talent for small and medium sized enterprises. If we are to use AI as an engine for growth, we need to make sure that the engine for growth, these types, these size enterprises have access. 50% of the world's GDP are small and medium sized enterprises."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:13",
                    "end": "00:24:43",
                    "text": " And in the global south, it's 70% of employment. To do so, there will be lots of business opportunities. So many industries will serve small and medium-sized enterprises. But that will not be enough. Private and public partnerships will be critical to making sure there's access. For example, we're working with the US college system where we're funding internships of college students at small and medium-sized enterprises. It's a win-win."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:43",
                    "end": "00:25:13",
                    "text": " Statistically, if you have an internship, you have a better chance of getting a job and is providing these enterprises access to some of the cutting edge talent. And so we must make sure that we're continuing to focus on the small and medium sized enterprises. Now I know, and we all know, that advanced AI is much more powerful than the technology advancements of the last decade. And of course,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:13",
                    "end": "00:25:39",
                    "text": " That means that the impact is more profound. But that doesn't change the critical lesson that AI must be used for growth and productivity. What it does change are the sets of actions, the time frame, the need for global collaboration, the need for more public and private partnerships, and the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:39",
                    "end": "00:26:09",
                    "text": " urgency of what we must do in order for AI to drive growth. So companies, companies must be willing to reinvent how they operate, their processes, how they've been doing work for the last decades. Underneath the headlines of a failure of AI is mostly a failure to reinvent. Companies have to invest to reshape their workforces."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:09",
                    "end": "00:26:32",
                    "text": " And companies must commit to creating sustained entry-level jobs. Now, entry-level jobs makes economic sense. They're the only way to create future leaders. And they bring needed, truly AI-native talent to each of our organizations."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:32",
                    "end": "00:27:02",
                    "text": " fundamentally is changing what an entry-level job looks like. And so a commitment means we have to be intentional about changing the rules, investing in training, which is exactly what Accenture is doing. We will hire more entry-level into more entry-level jobs this year than last year. But the skills we require and the way we're onboarding those individuals is fundamentally different. Now, countries must also reinvent."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:02",
                    "end": "00:27:27",
                    "text": " They must reinvent their role and how they work with the private sector. They have to, themselves as governments, become the best credential for why AI matters. They must work with the private sector to help create lifelong learning because education is no longer a destination. We have to have lifelong learning."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:28",
                    "end": "00:27:59",
                    "text": " India is doing a great job of embedding AI into the educational system, starting in primary school, and governments across the world will need to do so. At the same time, as countries are thinking differently, individuals have to think differently and recognize that formal education is no longer the destination. But perhaps the biggest fundamental change that must be made is that companies and countries need to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:59",
                    "end": "00:28:23",
                    "text": " pound the table for global standards. These standards should apply to safety, but also to the industries where AI can make the greatest impact. For example, in pharma, if one country is allowing pharma companies to use the latest technologies to discover drugs and then test drugs, but other countries don't follow suit,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:23",
                    "end": "00:28:54",
                    "text": " It means that you won't be able to scale, you won't be able to bring it, and we know that most often that impacts the most vulnerable. Now, we have a view, of course, that we have to reinvent. But as we think about that reinvention, or to our future, is the fundamental belief that it is humans in the lead, not humans in the loop, that will shape that future."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:54",
                    "end": "00:29:22",
                    "text": " We should not confuse how you deploy AI responsibly. Of course, all of our compliance programs have humans, they have technology. That doesn't change the critical lesson that we've learned over and over again. Technology, no matter how powerful, is only a tool. It is simply a tool. It is leaders"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:22",
                    "end": "00:29:52",
                    "text": " who decide how to use those tools. It is leaders who decide to commit to reinvent, who dedicate their time to making sure that people come along the journey. And it is leaders who must choose to work together to ensure the safe, widespread adoption of AI. There are lots of headlines today that predict less."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:52",
                    "end": "00:30:24",
                    "text": " less jobs, less opportunity, less human relevance. We are here because we see a future of more. At Accenture, we live by eight leadership essentials, the qualities we believe we need to run our company. And one of them is particularly important. We expect leaders to lead with excellence, confidence, and humility."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:24",
                    "end": "00:30:53",
                    "text": " As we look to our collective future, we should have the confidence to have the unwavering belief that together we can make a future that is better for all. We also must hold ourselves individually and collectively accountable for executing on that belief with a high bar of excellence because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:53",
                    "end": "00:31:36",
                    "text": " our people around the world are counting on that excellence. And finally, we must all have the humility to know that we cannot do this alone. Thank you very much. Ms. Julie Sweet. Thank you, Prime Minister Modi, Minister Vaishnav and your outstanding teams for convening us for this critical summit around AI. The breadth"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:36",
                    "end": "00:32:05",
                    "text": " distinguished guests from around the world is a recognition of the importance of broad global partnerships to capture the incredible potential of AI and address the risks. It is also a recognition of the importance of India in our AI enabled future. At Accenture, we're incredibly proud to have over 350,000 and growing reinventers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:05",
                    "end": "00:32:28",
                    "text": " here in India. We also have one of the largest AI workforces in the world tightly integrated with our growing AI hubs in the US, Europe, the Middle East, and Japan. And I want to take this time to thank all of our people in India for your incredible commitment to value and to our clients."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:28",
                    "end": "00:32:58",
                    "text": " Today, I want to leave you with three perspectives that we believe will help us ensure that AI's immense potential is captured for the benefit of all. First, using AI as an engine for growth is the only path for global prosperity for all. Second, the agenda ahead of us is unprecedented."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:59",
                    "end": "00:33:29",
                    "text": " Companies, countries, and individuals must reinvent how they work, how they work together, and how they learn. And finally, it is humans in the lead, not humans in the loop, that will determine our future. As we turn to the imperative for growth, I want to take you back for a moment to 2013. Oxford University had just published"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:29",
                    "end": "00:33:59",
                    "text": " a widely read study that said based on technology progress at that time, 47% of US jobs would be automatable. Dire headlines and predictions soon followed. One of those technologies was robotic process automation or RPA. And there were predictions that IT services would be badly damaged because it would automate so many jobs. And in fact,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:59",
                    "end": "00:34:24",
                    "text": " We used RPA to automate thousands of jobs. And we also, as an industry, embraced the new technologies of digital and classical AI, and we created many, many more jobs. We helped our clients adopt RPA, and those who did created investment capacity to invest in new technologies and to grow."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:24",
                    "end": "00:34:50",
                    "text": " And in fact, the IT services industry has thrived over the last decade, including many of India's most successful companies that you've heard from today. At Accenture alone in 2013, we were roughly $275,000 people and $29 billion in revenue. And today we're over $750,000 and growing and $70 billion in revenue."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:51",
                    "end": "00:35:20",
                    "text": " What the last decade has taught us is a critical lesson. When companies and countries embrace new technologies and then use them to drive growth and productivity, they prosper. Advanced AI should be the same. In fact, in our latest quarterly survey of C suites across 20 countries, they agree. 78% say"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:20",
                    "end": "00:35:45",
                    "text": " AI's greatest value is in growth. Now as we think about what growth should look like, there's two important considerations. First, AI should make the impossible possible. AI should make the impossible possible."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:45",
                    "end": "00:36:15",
                    "text": " If in a few years as a CEO, you cannot point to new products and services, new levels of performance that were not possible before, then you have not captured potential of AI. Think about the consumer and retail industries. LLMs are about to become the new mall. This is an entirely new way to engage customers and to engage in commerce that did not exist."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:15",
                    "end": "00:36:44",
                    "text": " in 2022. If you think about pharma, we see a path toward bringing drugs to market much faster than the average of nine years. Not possible before, which means that life-saving drugs will get to people faster and pharma will have accelerated growth. And we are just beginning to understand how AI will create new drugs, new materials, new products across industries."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:45",
                    "end": "00:37:09",
                    "text": " A second consideration around growth is that we must commit to providing access to the technology and the talent for small and medium-sized enterprises. If we are to use AI as an engine for growth, we need to make sure that the engine for growth, these types of these size enterprises have access."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:09",
                    "end": "00:37:32",
                    "text": " 50% of the world's GDP are small and medium-sized enterprises, and in the global south, it's 70% of employment. To do so, there will be lots of business opportunities, so many industries will serve small and medium-sized enterprises. But that will not be enough. Private and public partnerships will be critical to making sure there's access."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:32",
                    "end": "00:38:02",
                    "text": " For example, we're working with the U.S. college system where we're funding internships of college students at small and medium-sized enterprises. It's a win-win. Statistically, if you have an internship, you have a better chance of getting a job. And it's providing these enterprises access to some of the cutting-edge talent. And so we must make sure that we're continuing to focus on the small and medium-sized enterprises. Now I know, and we all know,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:02",
                    "end": "00:38:33",
                    "text": " that advanced AI is much more powerful than the technology advancements of the last decade. And of course, that means that the impact is more profound. But that doesn't change the critical lesson that AI must be used for growth and productivity. What it does change are the sets of actions, the timeframe,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:33",
                    "end": "00:39:01",
                    "text": " the need for global collaboration, the need for more public and private partnerships, and the urgency of what we must do in order for AI to drive growth. So companies, companies must be willing to reinvent how they operate, their processes, how they've been doing work for the last decades. Underneath the headlines of a failure of AI,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:01",
                    "end": "00:39:32",
                    "text": " AI is mostly a failure to reinvent. Companies have to invest to reshape their workforces. And companies must commit to creating sustained entry-level jobs. Now, entry-level jobs makes economic sense. They're the only way to create future leaders. And they bring needed, truly AI-native talent to each of our organizations."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:32",
                    "end": "00:40:02",
                    "text": " fundamentally is changing what an entry-level job looks like. And so a commitment means we have to be intentional about changing the rules, investing in training, which is exactly what Accenture is doing. We will hire into more entry-level jobs this year than last year, but the skills we require and the way we're onboarding those individuals is fundamentally different. Now, countries must also reinvent."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:02",
                    "end": "00:40:27",
                    "text": " They must reinvent their role and how they work with the private sector. They have to, themselves as governments, become the best credential for why AI matters. They must work with the private sector to help create lifelong learning because education is no longer a destination. We have to have lifelong learning."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:27",
                    "end": "00:40:58",
                    "text": " India is doing a great job of embedding AI into the educational system starting in primary school and governments across the world will need to do so. At the same time, as countries are thinking differently, individuals have to think differently and recognize that formal education is no longer the destination. But perhaps the biggest fundamental change that must be made is that companies and countries need to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:58",
                    "end": "00:41:22",
                    "text": " hound the table for global standards. These standards should apply to safety, but also to the industries where AI can make the greatest impact. For example, in pharma, if one country is allowing pharma companies to use the latest technologies to discover drugs and then test drugs, but other countries don't follow suit,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:22",
                    "end": "00:41:54",
                    "text": " It means that you won't be able to scale. You won't be able to bring it. And we know that most often that impacts the most vulnerable. Now, we have a view, of course, that we have to reinvent. But as we think about that reinvention or to our future is the fundamental belief that it is humans in the lead, not humans in the loop that will shape that future."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:54",
                    "end": "00:42:22",
                    "text": " We should not confuse how you deploy AI responsibly. Of course, all of our compliance programs have humans, they have technology. That doesn't change the critical lesson that we've learned over and over again. Technology, no matter how powerful, is only a tool. It is simply a tool. It is leaders"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:22",
                    "end": "00:42:52",
                    "text": " who decide how to use those tools. It is leaders who decide to commit to reinvent, who dedicate their time to making sure that people come along the journey. And it is leaders who must choose to work together to ensure the safe, widespread adoption of AI. There are lots of headlines today that predict less."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:52",
                    "end": "00:43:23",
                    "text": " less jobs, less opportunity, less human relevance. We are here because we see a future of more. At Accenture, we live by eight leadership essentials, the qualities we believe we need to run our company. And one of them is particularly important. We expect leaders to lead with excellence, confidence, and humility."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:24",
                    "end": "00:43:52",
                    "text": " As we look to our collective future, we should have the confidence to have the unwavering belief that together we can make a future that is better for all. We also must hold ourselves individually and collectively accountable for executing on that belief with a high bar of excellence"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:52",
                    "end": "00:44:34",
                    "text": " because our people around the world are counting on that excellence. And finally, we must all have the humility to know that we cannot do this alone. Thank you very much."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:34",
                    "end": "00:44:35",
                    "text": " The."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
   {
    "_id": "6996b6df565455da1fc75ca2",
    "projectId": "835151",
    "projectName": "Demis Hassabis Google Deep Mind",
    "projectDescription": "Demis Hassabis Google Deep Mind",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T07:08:13.862Z",
    "projectUpdatedDate": "2026-02-19T07:08:13.862Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-19T10:33:45.493Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Demis Hassabis Google Deep Mind.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6996d244565455da1fc75ec9",
    "projectId": "402491",
    "projectName": "Microsoft Brad Smith",
    "projectDescription": "Microsoft Brad Smith",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T09:05:07.174Z",
    "projectUpdatedDate": "2026-02-19T09:05:07.174Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-19T10:56:18.381Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO_LINK",
    "cfDataLength": 2,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames"
        },
        {
            "col1": "P1",
            "col2": "https://www.youtube.com/watch?v=VIAFCJeKh3A",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEOLINK",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:01",
                    "end": "00:00:06",
                    "text": "We live in a tumultuous time and in a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:03",
                    "end": "00:00:10",
                    "text": "fragmented world. But I think in so many"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:06",
                    "end": "00:00:13",
                    "text": "ways the deepest and most enduring"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:10",
                    "end": "00:00:16",
                    "text": "divide has been the economic divide"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:13",
                    "end": "00:00:18",
                    "text": "between the global north and south. And"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:16",
                    "end": "00:00:21",
                    "text": "what I believe we need to recognize is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:18",
                    "end": "00:00:26",
                    "text": "that this economic divide"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:21",
                    "end": "00:00:30",
                    "text": "is a result more than anything else of a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:26",
                    "end": "00:00:32",
                    "text": "technology divide. the technology divide"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:30",
                    "end": "00:00:34",
                    "text": "created by unequal access to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:32",
                    "end": "00:00:36",
                    "text": "electricity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:34",
                    "end": "00:00:38",
                    "text": "Electricity became one of humanity's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:36",
                    "end": "00:00:41",
                    "text": "most important generalpurpose"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:38",
                    "end": "00:00:44",
                    "text": "technologies, meaning it spread across"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:41",
                    "end": "00:00:47",
                    "text": "economies. It was applied in every"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:44",
                    "end": "00:00:51",
                    "text": "industry. It boosted productivity. Where"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:47",
                    "end": "00:00:54",
                    "text": "electricity went, economic development"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:51",
                    "end": "00:00:56",
                    "text": "and prosperity followed."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:54",
                    "end": "00:00:60",
                    "text": "But as we all know, electricity did not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:56",
                    "end": "00:01:04",
                    "text": "spread everywhere at the same pace. It"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:60",
                    "end": "00:01:06",
                    "text": "was literally 144 years ago that the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:04",
                    "end": "00:01:09",
                    "text": "first electrical power plant started"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:06",
                    "end": "00:01:12",
                    "text": "operating in lower Manhattan. And yet we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:09",
                    "end": "00:01:16",
                    "text": "come together today and we still live in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:12",
                    "end": "00:01:19",
                    "text": "a world where 700 million people lack"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:16",
                    "end": "00:01:23",
                    "text": "access to electricity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:19",
                    "end": "00:01:26",
                    "text": "Now comes AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:23",
                    "end": "00:01:28",
                    "text": "AI perhaps more than any other"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:26",
                    "end": "00:01:31",
                    "text": "technology this century will play a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:28",
                    "end": "00:01:35",
                    "text": "bigger role either in closing this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:31",
                    "end": "00:01:39",
                    "text": "economic divide or in exacerbating it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:35",
                    "end": "00:01:41",
                    "text": "and making it even wider. That is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:39",
                    "end": "00:01:44",
                    "text": "perhaps the single most important"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:41",
                    "end": "00:01:46",
                    "text": "question for us I would suggest today as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:44",
                    "end": "00:01:48",
                    "text": "we think about the role of AI in the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:46",
                    "end": "00:01:53",
                    "text": "global south. How can we do better?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:48",
                    "end": "00:01:56",
                    "text": "Because we need to do better."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:53",
                    "end": "00:01:58",
                    "text": "What will it take? I think it's going to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:56",
                    "end": "00:02:01",
                    "text": "take a few things that will require that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:58",
                    "end": "00:02:04",
                    "text": "we all come together and work together."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:01",
                    "end": "00:02:06",
                    "text": "First, the obvious. We need to bring"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:04",
                    "end": "00:02:09",
                    "text": "infrastructure to the global south. That"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:06",
                    "end": "00:02:12",
                    "text": "means data centers and commute compute."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:09",
                    "end": "00:02:16",
                    "text": "It also means more connectivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:12",
                    "end": "00:02:18",
                    "text": "It me it means more electricity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:16",
                    "end": "00:02:21",
                    "text": "That is going to take not only the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:18",
                    "end": "00:02:23",
                    "text": "world's best technology, it's going to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:21",
                    "end": "00:02:25",
                    "text": "require an enormous amount of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:23",
                    "end": "00:02:27",
                    "text": "investment."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:25",
                    "end": "00:02:29",
                    "text": "That's why we at Microsoft announced"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:27",
                    "end": "00:02:31",
                    "text": "yesterday morning that we're on pace to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:29",
                    "end": "00:02:35",
                    "text": "spend $50 billion by the end of this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:31",
                    "end": "00:02:38",
                    "text": "decade to bring AI to the global south."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:35",
                    "end": "00:02:41",
                    "text": "And of all the countries in which we are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:38",
                    "end": "00:02:45",
                    "text": "investing, India, not surprisingly, is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:41",
                    "end": "00:02:45",
                    "text": "one of the largest."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:45",
                    "end": "00:02:49",
                    "text": "The"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:46",
                    "end": "00:02:51",
                    "text": "But we'll need to harness private"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:49",
                    "end": "00:02:54",
                    "text": "capital, investments from tech"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:51",
                    "end": "00:02:56",
                    "text": "companies, other sources of private"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:54",
                    "end": "00:02:58",
                    "text": "capital, government funding. We'll need"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:56",
                    "end": "00:03:00",
                    "text": "governments and others to generate"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:58",
                    "end": "00:03:03",
                    "text": "demand for the use of AI in the global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:00",
                    "end": "00:03:05",
                    "text": "south because that is the only way to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:03",
                    "end": "00:03:09",
                    "text": "get the wheels of the market spinning"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:05",
                    "end": "00:03:11",
                    "text": "and to do what we need to do together."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:09",
                    "end": "00:03:12",
                    "text": "That's the first thing we need to do"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:11",
                    "end": "00:03:14",
                    "text": "together."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:12",
                    "end": "00:03:16",
                    "text": "There's a second thing we need to do."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:14",
                    "end": "00:03:18",
                    "text": "We've talked about it here already today"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:16",
                    "end": "00:03:21",
                    "text": "and it is so clear when you study the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:18",
                    "end": "00:03:24",
                    "text": "history of technology. Infrastructure is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:21",
                    "end": "00:03:28",
                    "text": "not only hardware. It's not only wires"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:24",
                    "end": "00:03:30",
                    "text": "and grids. It's skilling for people."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:28",
                    "end": "00:03:33",
                    "text": "Because the key to enabling a country"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:30",
                    "end": "00:03:36",
                    "text": "and a population to use a general"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:33",
                    "end": "00:03:39",
                    "text": "purpose technology at scale is to give"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:36",
                    "end": "00:03:41",
                    "text": "people across the country access to the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:39",
                    "end": "00:03:43",
                    "text": "skills they need to put it to work. And"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:41",
                    "end": "00:03:46",
                    "text": "that's why it's such good news that we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:43",
                    "end": "00:03:48",
                    "text": "seem so many governments supported by"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:46",
                    "end": "00:03:51",
                    "text": "the United Nations and supported by"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:48",
                    "end": "00:03:53",
                    "text": "private companies investing in more"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:51",
                    "end": "00:03:54",
                    "text": "skilling."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:53",
                    "end": "00:03:57",
                    "text": "It's actually something that should"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:54",
                    "end": "00:03:60",
                    "text": "speak to all of us. Certainly, as a tech"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:57",
                    "end": "00:04:02",
                    "text": "company, we're committed. We've launched"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:60",
                    "end": "00:04:04",
                    "text": "through Microsoft Elevate new"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:02",
                    "end": "00:04:06",
                    "text": "initiatives, including one we're"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:04",
                    "end": "00:04:09",
                    "text": "announcing this week, Microsoft Elevate"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:06",
                    "end": "00:04:13",
                    "text": "for educators to equip teachers with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:09",
                    "end": "00:04:16",
                    "text": "access to help their students learn how"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:13",
                    "end": "00:04:19",
                    "text": "to use AI. But in the truth, it doesn't"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:16",
                    "end": "00:04:22",
                    "text": "matter where you work, you have a role"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:19",
                    "end": "00:04:26",
                    "text": "to play. Because the lesson of digital"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:22",
                    "end": "00:04:28",
                    "text": "technology, I was an example of this was"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:26",
                    "end": "00:04:32",
                    "text": "that it took employers to open their"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:28",
                    "end": "00:04:35",
                    "text": "doors to computing. It takes employers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:32",
                    "end": "00:04:37",
                    "text": "today to open their doors to new AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:35",
                    "end": "00:04:41",
                    "text": "tools. It will take employers to invest"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:37",
                    "end": "00:04:44",
                    "text": "in the skilling of their employees."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:41",
                    "end": "00:04:46",
                    "text": "It's not just for the next generation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:44",
                    "end": "00:04:48",
                    "text": "It's for every generation that this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:46",
                    "end": "00:04:50",
                    "text": "fully matters."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:48",
                    "end": "00:04:53",
                    "text": "Then there's a third challenge for the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:50",
                    "end": "00:04:55",
                    "text": "global south. We need to make AI work"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:53",
                    "end": "00:04:57",
                    "text": "effectively for the global south and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:55",
                    "end": "00:05:01",
                    "text": "that requires some special initiatives"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:57",
                    "end": "00:05:04",
                    "text": "at least two. First, we need to make AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:01",
                    "end": "00:05:07",
                    "text": "as effective in every language as it is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:04",
                    "end": "00:05:10",
                    "text": "in English. And today it is not."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:07",
                    "end": "00:05:12",
                    "text": "Performance tests show that's the case."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:10",
                    "end": "00:05:13",
                    "text": "That's why one of the good things to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:12",
                    "end": "00:05:16",
                    "text": "come out of this week is new"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:13",
                    "end": "00:05:19",
                    "text": "announcements to invest upstream in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:16",
                    "end": "00:05:21",
                    "text": "better data in other languages to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:19",
                    "end": "00:05:24",
                    "text": "provide better tools and measurement"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:21",
                    "end": "00:05:27",
                    "text": "systems for AI that is built in other"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:24",
                    "end": "00:05:31",
                    "text": "languages to build out data providence"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:27",
                    "end": "00:05:34",
                    "text": "with a view to linguistic diversity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:31",
                    "end": "00:05:37",
                    "text": "diversity that we need to advance around"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:34",
                    "end": "00:05:39",
                    "text": "the world and we need to use AI in the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:37",
                    "end": "00:05:42",
                    "text": "global south to solve the problems that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:39",
                    "end": "00:05:44",
                    "text": "matter to the global south oftent times"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:42",
                    "end": "00:05:47",
                    "text": "as we're doing here in India that's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:44",
                    "end": "00:05:49",
                    "text": "about improvements in agriculture or as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:47",
                    "end": "00:05:52",
                    "text": "a number of partners including Microsoft"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:49",
                    "end": "00:05:54",
                    "text": "are doing this week launching a new"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:52",
                    "end": "00:05:58",
                    "text": "initiative to address food security"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:54",
                    "end": "00:06:01",
                    "text": "across Africa. These are just two of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:58",
                    "end": "00:06:04",
                    "text": "myriad of opportunities we have to put"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:01",
                    "end": "00:06:07",
                    "text": "AI to work in ways that will bring"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:04",
                    "end": "00:06:11",
                    "text": "faster benefits to countries in most of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:07",
                    "end": "00:06:13",
                    "text": "the world. If we do those three things"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:11",
                    "end": "00:06:16",
                    "text": "well, build infrastructure, invest in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:13",
                    "end": "00:06:18",
                    "text": "skilling, address real world problems,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:16",
                    "end": "00:06:19",
                    "text": "then I think it may create the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:18",
                    "end": "00:06:21",
                    "text": "foundation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:19",
                    "end": "00:06:24",
                    "text": "to think more and do more about the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:21",
                    "end": "00:06:27",
                    "text": "question I am willing to bet will be a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:24",
                    "end": "00:06:32",
                    "text": "bigger part of the conversation in the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:27",
                    "end": "00:06:35",
                    "text": "next few AI summits. What will AI mean"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:32",
                    "end": "00:06:38",
                    "text": "for the future of work and jobs?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:35",
                    "end": "00:06:41",
                    "text": "Within these walls, I think we're mostly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:38",
                    "end": "00:06:44",
                    "text": "enthusiastic about the future of AI. But"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:41",
                    "end": "00:06:46",
                    "text": "outside these walls, I think we need to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:44",
                    "end": "00:06:49",
                    "text": "recognize that increasingly around the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:46",
                    "end": "00:06:51",
                    "text": "world and especially in some countries,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:49",
                    "end": "00:06:56",
                    "text": "many parents are asking a common"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:51",
                    "end": "00:06:59",
                    "text": "question. What will AI mean for my kids?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:56",
                    "end": "00:07:03",
                    "text": "What will AI mean for my family? What"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:59",
                    "end": "00:07:06",
                    "text": "will AI mean for our future? I get it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:03",
                    "end": "00:07:10",
                    "text": "that some people are excited and they'll"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:06",
                    "end": "00:07:13",
                    "text": "do well, but what about us? Us meaning"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:10",
                    "end": "00:07:16",
                    "text": "most of the people who live on this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:13",
                    "end": "00:07:18",
                    "text": "planet. I think we have something to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:16",
                    "end": "00:07:20",
                    "text": "prove. I think we have something to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:18",
                    "end": "00:07:22",
                    "text": "prove not only to communities and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:20",
                    "end": "00:07:25",
                    "text": "countries and our customers, but to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:22",
                    "end": "00:07:30",
                    "text": "ourselves that we can not only embrace"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:25",
                    "end": "00:07:32",
                    "text": "but pursue a brighter future for people."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:30",
                    "end": "00:07:34",
                    "text": "It's great to come to conferences like"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:32",
                    "end": "00:07:36",
                    "text": "this and hear people talk about all of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:34",
                    "end": "00:07:39",
                    "text": "the advances in technology, but let's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:36",
                    "end": "00:07:44",
                    "text": "remember one other thing as well. Human"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:39",
                    "end": "00:07:46",
                    "text": "capability is neither fixed nor finite."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:44",
                    "end": "00:07:48",
                    "text": "It's great to think about what it would"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:46",
                    "end": "00:07:49",
                    "text": "mean to have computers in a data center"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:48",
                    "end": "00:07:53",
                    "text": "that would be like an country of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:49",
                    "end": "00:07:55",
                    "text": "geniuses. But let's also recognize this."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:53",
                    "end": "00:07:59",
                    "text": "Compared to the people who lived in the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:55",
                    "end": "00:08:02",
                    "text": "bronze age, all of you, all of us are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:59",
                    "end": "00:08:05",
                    "text": "already geniuses."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:02",
                    "end": "00:08:09",
                    "text": "Whenever technology advances, it creates"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:05",
                    "end": "00:08:12",
                    "text": "a new platform, a new foundation that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:09",
                    "end": "00:08:16",
                    "text": "enables people to stand taller and reach"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:12",
                    "end": "00:08:19",
                    "text": "higher if, and only if we're committed"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:16",
                    "end": "00:08:22",
                    "text": "to using that technology well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:19",
                    "end": "00:08:24",
                    "text": "As AI makes it possible to cure more"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:22",
                    "end": "00:08:25",
                    "text": "diseases, then it is right that we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:24",
                    "end": "00:08:29",
                    "text": "expect that it will improve human"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:25",
                    "end": "00:08:32",
                    "text": "health. As we use AI already every day"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:29",
                    "end": "00:08:36",
                    "text": "to find faster answers,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:32",
                    "end": "00:08:37",
                    "text": "it gives us the opportunity to ask more"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:36",
                    "end": "00:08:41",
                    "text": "questions."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:37",
                    "end": "00:08:45",
                    "text": "The fundamental fuel of human capability"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:41",
                    "end": "00:08:48",
                    "text": "has always been the same, curiosity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:45",
                    "end": "00:08:52",
                    "text": "We need to look at AI as the next great"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:48",
                    "end": "00:08:56",
                    "text": "generator for human curiosity and we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:52",
                    "end": "00:08:58",
                    "text": "need to take some inspiration because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:56",
                    "end": "00:08:60",
                    "text": "we all know the world could use a little"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:58",
                    "end": "00:09:01",
                    "text": "more inspiration."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:60",
                    "end": "00:09:04",
                    "text": "I often think about the following."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:01",
                    "end": "00:09:07",
                    "text": "Before the invention of the washing"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:04",
                    "end": "00:09:11",
                    "text": "machine, it took someone almost always a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:07",
                    "end": "00:09:14",
                    "text": "woman between 6 and 8 hours to wash a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:11",
                    "end": "00:09:15",
                    "text": "load of laundry. But as the washing"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:14",
                    "end": "00:09:17",
                    "text": "machine improved, that was compressed to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:15",
                    "end": "00:09:19",
                    "text": "30 minutes. But do you know what"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:17",
                    "end": "00:09:23",
                    "text": "happened? One thing happened more than"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:19",
                    "end": "00:09:25",
                    "text": "anything else. Everyone wanted to wear"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:23",
                    "end": "00:09:28",
                    "text": "cleaner clothes."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:25",
                    "end": "00:09:30",
                    "text": "Everyone expected to wear cleaner"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:28",
                    "end": "00:09:33",
                    "text": "clothes. People did their laundry a lot"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:30",
                    "end": "00:09:35",
                    "text": "more often. They had better clothes and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:33",
                    "end": "00:09:39",
                    "text": "they had more time and they put that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:35",
                    "end": "00:09:40",
                    "text": "time to work to do more with their"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:39",
                    "end": "00:09:43",
                    "text": "lives."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:40",
                    "end": "00:09:45",
                    "text": "Ultimately that is the question for us"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:43",
                    "end": "00:09:48",
                    "text": "and this too is not just a question for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:45",
                    "end": "00:09:51",
                    "text": "tech companies. This is what tech will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:48",
                    "end": "00:09:54",
                    "text": "do to people is people will use"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:51",
                    "end": "00:09:56",
                    "text": "technology to do for people and it's not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:54",
                    "end": "00:09:58",
                    "text": "people who create products. It's every"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:56",
                    "end": "00:10:02",
                    "text": "government. It's every company. It's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:58",
                    "end": "00:10:04",
                    "text": "every nonprofit. every employer"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:02",
                    "end": "00:10:07",
                    "text": "because we all have the opportunity to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:04",
                    "end": "00:10:10",
                    "text": "work with our people to manage through"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:07",
                    "end": "00:10:12",
                    "text": "the change that is coming"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:10",
                    "end": "00:10:18",
                    "text": "to show people how with the right AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:12",
                    "end": "00:10:21",
                    "text": "skills, they can create jobs and careers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:18",
                    "end": "00:10:23",
                    "text": "that will be brighter for their future."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:21",
                    "end": "00:10:26",
                    "text": "That will not be easy. But if that's not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:23",
                    "end": "00:10:27",
                    "text": "our goal, then we're missing the big"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:26",
                    "end": "00:10:30",
                    "text": "picture."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:27",
                    "end": "00:10:32",
                    "text": "As we come away from this AI summit, it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:30",
                    "end": "00:10:35",
                    "text": "reflects so much progress. But I would"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:32",
                    "end": "00:10:38",
                    "text": "say one thing as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:35",
                    "end": "00:10:41",
                    "text": "Each of these AI summits is a proud"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:38",
                    "end": "00:10:43",
                    "text": "moment for a great nation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:41",
                    "end": "00:10:45",
                    "text": "But we have an opportunity"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:43",
                    "end": "00:10:47",
                    "text": "rather than have summits that are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:45",
                    "end": "00:10:50",
                    "text": "islands that are disconnected from the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:47",
                    "end": "00:10:53",
                    "text": "summits before or that follow. We need"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:50",
                    "end": "00:10:55",
                    "text": "to build bridges. We need to be build"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:53",
                    "end": "00:10:58",
                    "text": "bridges between these summits. We need"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:55",
                    "end": "00:11:00",
                    "text": "to define clear goals. We need to have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:58",
                    "end": "00:11:04",
                    "text": "common measurement systems. And every"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:00",
                    "end": "00:11:07",
                    "text": "year we need to ask the same question."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:04",
                    "end": "00:11:12",
                    "text": "Did we make 12 months of progress in the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:07",
                    "end": "00:11:14",
                    "text": "year that just preceded our meeting? How"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:12",
                    "end": "00:11:18",
                    "text": "can we build on that progress to do more"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:14",
                    "end": "00:11:20",
                    "text": "and move faster in the year ahead?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:18",
                    "end": "00:11:22",
                    "text": "For those of us who come to these"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:20",
                    "end": "00:11:25",
                    "text": "meetings in different countries every"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:22",
                    "end": "00:11:29",
                    "text": "year, I hope we will take that away."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:25",
                    "end": "00:11:32",
                    "text": "Let's aim higher, not just for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:29",
                    "end": "00:11:35",
                    "text": "technology, but for what technology can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:32",
                    "end": "00:11:38",
                    "text": "do for people. Let's be clear in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:35",
                    "end": "00:11:42",
                    "text": "defining what we want to accomplish."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:38",
                    "end": "00:11:46",
                    "text": "Let's put in place the ability for us to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:42",
                    "end": "00:11:49",
                    "text": "measure our progress and all hold"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:46",
                    "end": "00:11:53",
                    "text": "ourselves accountable as a global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:49",
                    "end": "00:11:56",
                    "text": "community. If we can do these things,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:53",
                    "end": "00:11:58",
                    "text": "then we can use these summits and we can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:56",
                    "end": "00:12:01",
                    "text": "use this next generation of technology"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:58",
                    "end": "00:12:04",
                    "text": "to build a better world. I know that's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:01",
                    "end": "00:12:07",
                    "text": "what we within these walls want us"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:04",
                    "end": "00:12:09",
                    "text": "ourselves to do. But even more than"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:07",
                    "end": "00:12:14",
                    "text": "that, I know that the people outside"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:09",
                    "end": "00:12:19",
                    "text": "these walls are hoping and expecting us"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:14",
                    "end": "00:12:19",
                    "text": "to do just that. Thank you very much."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:24",
                    "end": "00:12:27",
                    "text": ">> Thank you so much, Mr. Bradith, for that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:26",
                    "end": "00:12:30",
                    "text": "very energizing address. Ladies and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:27",
                    "end": "00:12:33",
                    "text": "gentlemen, I think he really deserves an"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:30",
                    "end": "00:12:34",
                    "text": "energetic applause from you all. His"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:33",
                    "end": "00:12:36",
                    "text": "address has actually given a very"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:34",
                    "end": "00:12:39",
                    "text": "constructive direction to the discourse"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:36",
                    "end": "00:12:41",
                    "text": "on artificial intelligence."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:39",
                    "end": "00:12:44",
                    "text": ">> Important day when the world comes"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:41",
                    "end": "00:12:47",
                    "text": "together under one roof as we have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:44",
                    "end": "00:12:50",
                    "text": "today. It gives us an opportunity"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:47",
                    "end": "00:12:53",
                    "text": "together to ask important questions even"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:50",
                    "end": "00:12:56",
                    "text": "hard questions and think about how we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:53",
                    "end": "00:12:59",
                    "text": "want to answer them. As we think about"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:56",
                    "end": "00:13:01",
                    "text": "this summit, the first AI summit in the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:59",
                    "end": "00:13:04",
                    "text": "global south, it's only right that we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:01",
                    "end": "00:13:07",
                    "text": "start, I believe, by focusing on AI and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:04",
                    "end": "00:13:09",
                    "text": "what it means for the global south. In"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:07",
                    "end": "00:13:12",
                    "text": "some ways, I think the best way to start"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:09",
                    "end": "00:13:14",
                    "text": "thinking about AI is to look more"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:12",
                    "end": "00:13:17",
                    "text": "broadly and think about the state of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:14",
                    "end": "00:13:20",
                    "text": "world in which we live."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:17",
                    "end": "00:13:23",
                    "text": "We live in a tumultuous time and in a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:20",
                    "end": "00:13:26",
                    "text": "fragmented world. But I think in so many"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:23",
                    "end": "00:13:30",
                    "text": "ways the deepest and most enduring"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:26",
                    "end": "00:13:33",
                    "text": "divide has been the economic divide"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:30",
                    "end": "00:13:35",
                    "text": "between the global north and south. And"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:33",
                    "end": "00:13:38",
                    "text": "what I believe we need to recognize is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:35",
                    "end": "00:13:42",
                    "text": "that this economic divide"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:38",
                    "end": "00:13:46",
                    "text": "is a result more than anything else of a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:42",
                    "end": "00:13:48",
                    "text": "technology divide. the technology divide"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:46",
                    "end": "00:13:50",
                    "text": "created by unequal access to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:48",
                    "end": "00:13:52",
                    "text": "electricity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:50",
                    "end": "00:13:54",
                    "text": "Electricity became one of humanity's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:52",
                    "end": "00:13:57",
                    "text": "most important generalpurpose"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:54",
                    "end": "00:14:00",
                    "text": "technologies, meaning it spread across"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:57",
                    "end": "00:14:03",
                    "text": "economies. It was applied in every"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:00",
                    "end": "00:14:06",
                    "text": "industry. It boosted productivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:03",
                    "end": "00:14:10",
                    "text": "Where electricity went, economic"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:06",
                    "end": "00:14:13",
                    "text": "development and prosperity followed."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:10",
                    "end": "00:14:16",
                    "text": "But as we all know, electricity did not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:13",
                    "end": "00:14:20",
                    "text": "spread everywhere at the same pace. It"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:16",
                    "end": "00:14:22",
                    "text": "was literally 144 years ago that the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:20",
                    "end": "00:14:26",
                    "text": "first electrical power plant started"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:22",
                    "end": "00:14:28",
                    "text": "operating in lower Manhattan. And yet we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:26",
                    "end": "00:14:33",
                    "text": "come together today. And we still live"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:28",
                    "end": "00:14:36",
                    "text": "in a world where 700 million people lack"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:33",
                    "end": "00:14:39",
                    "text": "access to electricity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:36",
                    "end": "00:14:42",
                    "text": "Now comes AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:39",
                    "end": "00:14:45",
                    "text": "AI perhaps more than any other"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:42",
                    "end": "00:14:47",
                    "text": "technology this century will play a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:45",
                    "end": "00:14:51",
                    "text": "bigger role either in closing this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:47",
                    "end": "00:14:55",
                    "text": "economic divide or in exacerbating it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:51",
                    "end": "00:14:57",
                    "text": "and making it even wider."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:55",
                    "end": "00:14:59",
                    "text": "That is perhaps the single most"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:57",
                    "end": "00:15:01",
                    "text": "important question for us I would"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:59",
                    "end": "00:15:04",
                    "text": "suggest today as we think about the role"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:01",
                    "end": "00:15:09",
                    "text": "of AI in the global south. How can we do"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:04",
                    "end": "00:15:12",
                    "text": "better? Because we need to do better."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:09",
                    "end": "00:15:14",
                    "text": "What will it take? I think it's going to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:12",
                    "end": "00:15:18",
                    "text": "take a few things that will require that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:14",
                    "end": "00:15:20",
                    "text": "we all come together and work together."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:18",
                    "end": "00:15:23",
                    "text": "First, the obvious. We need to bring"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:20",
                    "end": "00:15:26",
                    "text": "infrastructure to the global south. That"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:23",
                    "end": "00:15:28",
                    "text": "means data centers and commute compute."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:26",
                    "end": "00:15:32",
                    "text": "It also means more connectivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:28",
                    "end": "00:15:35",
                    "text": "It me it means more electricity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:32",
                    "end": "00:15:37",
                    "text": "That is going to take not only the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:35",
                    "end": "00:15:40",
                    "text": "world's best technology, it's going to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:37",
                    "end": "00:15:41",
                    "text": "require an enormous amount of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:40",
                    "end": "00:15:43",
                    "text": "investment."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:41",
                    "end": "00:15:45",
                    "text": "That's why we at Microsoft announced"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:43",
                    "end": "00:15:48",
                    "text": "yesterday morning that we're on pace to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:45",
                    "end": "00:15:52",
                    "text": "spend $50 billion by the end of this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:48",
                    "end": "00:15:54",
                    "text": "decade to bring AI to the global south."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:52",
                    "end": "00:15:58",
                    "text": "And of all the countries in which we are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:54",
                    "end": "00:16:01",
                    "text": "investing, India, not surprisingly, is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:58",
                    "end": "00:16:01",
                    "text": "one of the largest."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:01",
                    "end": "00:16:06",
                    "text": "The"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:03",
                    "end": "00:16:08",
                    "text": "But we'll need to harness private"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:06",
                    "end": "00:16:10",
                    "text": "capital, investments from tech"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:08",
                    "end": "00:16:13",
                    "text": "companies, other sources of private"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:10",
                    "end": "00:16:15",
                    "text": "capital, government funding. We'll need"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:13",
                    "end": "00:16:17",
                    "text": "governments and others to generate"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:15",
                    "end": "00:16:19",
                    "text": "demand for the use of AI in the global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:17",
                    "end": "00:16:22",
                    "text": "south because that is the only way to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:19",
                    "end": "00:16:26",
                    "text": "get the wheels of the market spinning"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:22",
                    "end": "00:16:27",
                    "text": "and to do what we need to do together."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:26",
                    "end": "00:16:29",
                    "text": "That's the first thing we need to do"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:27",
                    "end": "00:16:31",
                    "text": "together."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:29",
                    "end": "00:16:32",
                    "text": "There's a second thing we need to do."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:31",
                    "end": "00:16:34",
                    "text": "We've talked about it here already today"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:32",
                    "end": "00:16:37",
                    "text": "and it is so clear when you study the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:34",
                    "end": "00:16:39",
                    "text": "history of technology."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:37",
                    "end": "00:16:42",
                    "text": "Infrastructure is not only hardware."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:39",
                    "end": "00:16:45",
                    "text": "It's not only wires and grids."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:42",
                    "end": "00:16:47",
                    "text": "It's skilling for people. Because the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:45",
                    "end": "00:16:50",
                    "text": "key to enabling a country and a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:47",
                    "end": "00:16:53",
                    "text": "population to use a general purpose"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:50",
                    "end": "00:16:55",
                    "text": "technology at scale is to give people"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:53",
                    "end": "00:16:58",
                    "text": "across the country access to the skills"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:55",
                    "end": "00:17:00",
                    "text": "they need to put it to work. And that's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:58",
                    "end": "00:17:03",
                    "text": "why it's such good news that we seem so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:00",
                    "end": "00:17:05",
                    "text": "many governments supported by the United"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:03",
                    "end": "00:17:09",
                    "text": "Nations and supported by private"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:05",
                    "end": "00:17:11",
                    "text": "companies investing in more skilling."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:09",
                    "end": "00:17:13",
                    "text": "It's actually something that should"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:11",
                    "end": "00:17:16",
                    "text": "speak to all of us. Certainly, as a tech"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:13",
                    "end": "00:17:19",
                    "text": "company, we're committed. We've launched"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:16",
                    "end": "00:17:20",
                    "text": "through Microsoft Elevate new"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:19",
                    "end": "00:17:22",
                    "text": "initiatives, including one we're"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:20",
                    "end": "00:17:26",
                    "text": "announcing this week, Microsoft Elevate"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:22",
                    "end": "00:17:29",
                    "text": "for Educators to equip teachers with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:26",
                    "end": "00:17:32",
                    "text": "access to help their students learn how"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:29",
                    "end": "00:17:35",
                    "text": "to use AI. But the truth, it doesn't"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:32",
                    "end": "00:17:38",
                    "text": "matter where you work, you have a role"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:35",
                    "end": "00:17:42",
                    "text": "to play. Because the lesson of digital"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:38",
                    "end": "00:17:45",
                    "text": "technology, I was an example of this was"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:42",
                    "end": "00:17:48",
                    "text": "that it took employers to open their"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:45",
                    "end": "00:17:51",
                    "text": "doors to computing. It takes employers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:48",
                    "end": "00:17:54",
                    "text": "today to open their doors to new AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:51",
                    "end": "00:17:57",
                    "text": "tools. It will take employers to invest"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:54",
                    "end": "00:18:00",
                    "text": "in the skilling of their employees."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:57",
                    "end": "00:18:03",
                    "text": "It's not just for the next generation,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:00",
                    "end": "00:18:05",
                    "text": "it's for every generation that this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:03",
                    "end": "00:18:07",
                    "text": "fully matters."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:05",
                    "end": "00:18:09",
                    "text": "Then there's a third challenge for the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:07",
                    "end": "00:18:11",
                    "text": "global south. We need to make AI work"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:09",
                    "end": "00:18:13",
                    "text": "effectively for the global south and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:11",
                    "end": "00:18:18",
                    "text": "that requires some special initiatives"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:13",
                    "end": "00:18:20",
                    "text": "at least two. First, we need to make AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:18",
                    "end": "00:18:23",
                    "text": "as effective in every language as it is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:20",
                    "end": "00:18:26",
                    "text": "in English. And today it is not."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:23",
                    "end": "00:18:28",
                    "text": "Performance tests show that's the case."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:26",
                    "end": "00:18:30",
                    "text": "That's why one of the good things to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:28",
                    "end": "00:18:32",
                    "text": "come out of this week is new"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:30",
                    "end": "00:18:36",
                    "text": "announcements to invest upstream in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:32",
                    "end": "00:18:38",
                    "text": "better data in other languages to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:36",
                    "end": "00:18:40",
                    "text": "provide better tools and measurement"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:38",
                    "end": "00:18:44",
                    "text": "systems for AI that is built in other"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:40",
                    "end": "00:18:48",
                    "text": "languages to build out data providence"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:44",
                    "end": "00:18:51",
                    "text": "with a view to linguistic diversity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:48",
                    "end": "00:18:54",
                    "text": "Diversity that we need to advance around"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:51",
                    "end": "00:18:56",
                    "text": "the world. And we need to use AI in the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:54",
                    "end": "00:18:59",
                    "text": "global south to solve the problems that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:56",
                    "end": "00:19:01",
                    "text": "matter to the global south. Oftentimes,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:59",
                    "end": "00:19:03",
                    "text": "as we're doing here in India, that's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:01",
                    "end": "00:19:06",
                    "text": "about improvements in agriculture or as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:03",
                    "end": "00:19:08",
                    "text": "a number of partners including Microsoft"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:06",
                    "end": "00:19:11",
                    "text": "are doing this week, launching a new"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:08",
                    "end": "00:19:15",
                    "text": "initiative to address food security"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:11",
                    "end": "00:19:18",
                    "text": "across Africa. These are just two of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:15",
                    "end": "00:19:20",
                    "text": "myriad of opportunities we have to put"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:18",
                    "end": "00:19:24",
                    "text": "AI to work in ways that will bring"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:20",
                    "end": "00:19:25",
                    "text": "faster benefits to countries in most of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:24",
                    "end": "00:19:28",
                    "text": "the world."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:25",
                    "end": "00:19:31",
                    "text": "If we do those three things well, build"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:28",
                    "end": "00:19:33",
                    "text": "infrastructure, invest in skilling,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:31",
                    "end": "00:19:36",
                    "text": "address real world problems, then I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:33",
                    "end": "00:19:38",
                    "text": "think it may create the foundation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:36",
                    "end": "00:19:41",
                    "text": "to think more and do more about the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:38",
                    "end": "00:19:44",
                    "text": "question I am willing to bet will be a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:41",
                    "end": "00:19:48",
                    "text": "bigger part of the conversation in the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:44",
                    "end": "00:19:51",
                    "text": "next few AI summits. What will AI mean"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:48",
                    "end": "00:19:55",
                    "text": "for the future of work and jobs?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:51",
                    "end": "00:19:58",
                    "text": "Within these walls, I think we're mostly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:55",
                    "end": "00:20:00",
                    "text": "enthusiastic about the future of AI. But"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:58",
                    "end": "00:20:03",
                    "text": "outside these walls, I think we need to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:00",
                    "end": "00:20:06",
                    "text": "recognize that increasingly around the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:03",
                    "end": "00:20:08",
                    "text": "world and especially in some countries,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:06",
                    "end": "00:20:09",
                    "text": "many parents are asking a common"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:08",
                    "end": "00:20:13",
                    "text": "question."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:09",
                    "end": "00:20:17",
                    "text": "What will AI mean for my kids? What will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:13",
                    "end": "00:20:21",
                    "text": "AI mean for my family? What will AI mean"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:17",
                    "end": "00:20:23",
                    "text": "for our future? I get it that some"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:21",
                    "end": "00:20:28",
                    "text": "people are excited and they'll do well,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:23",
                    "end": "00:20:31",
                    "text": "but what about us? Us meaning most of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:28",
                    "end": "00:20:34",
                    "text": "the people who live on this planet. I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:31",
                    "end": "00:20:36",
                    "text": "think we have something to prove. I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:34",
                    "end": "00:20:38",
                    "text": "think we have something to prove not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:36",
                    "end": "00:20:41",
                    "text": "only to communities and countries and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:38",
                    "end": "00:20:44",
                    "text": "our customers, but to ourselves that we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:41",
                    "end": "00:20:47",
                    "text": "can not only embrace but pursue a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:44",
                    "end": "00:20:49",
                    "text": "brighter future for people. It's great"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:47",
                    "end": "00:20:51",
                    "text": "to come to conferences like this and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:49",
                    "end": "00:20:53",
                    "text": "hear people talk about all of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:51",
                    "end": "00:20:56",
                    "text": "advances in technology, but let's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:53",
                    "end": "00:21:00",
                    "text": "remember one other thing as well. Human"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:56",
                    "end": "00:21:02",
                    "text": "capability is neither fixed nor finite."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:00",
                    "end": "00:21:04",
                    "text": "It's great to think about what it would"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:02",
                    "end": "00:21:06",
                    "text": "mean to have computers in a data center"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:04",
                    "end": "00:21:10",
                    "text": "that would be like an country of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:06",
                    "end": "00:21:11",
                    "text": "geniuses. But let's also recognize this."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:10",
                    "end": "00:21:16",
                    "text": "Compared to the people who lived in the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:11",
                    "end": "00:21:18",
                    "text": "bronze age, all of you, all of us are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:16",
                    "end": "00:21:21",
                    "text": "already geniuses."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:18",
                    "end": "00:21:25",
                    "text": "Whenever technology advances, it creates"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:21",
                    "end": "00:21:28",
                    "text": "a new platform, a new foundation that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:25",
                    "end": "00:21:33",
                    "text": "enables people to stand taller and reach"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:28",
                    "end": "00:21:37",
                    "text": "higher if and only if we're committed to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:33",
                    "end": "00:21:39",
                    "text": "using that technology. Well, as AI makes"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:37",
                    "end": "00:21:41",
                    "text": "it possible to cure more diseases, then"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:39",
                    "end": "00:21:44",
                    "text": "it is right that we expect that it will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:41",
                    "end": "00:21:47",
                    "text": "improve human health. As we use AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:44",
                    "end": "00:21:49",
                    "text": "already every day to find faster"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:47",
                    "end": "00:21:52",
                    "text": "answers,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:49",
                    "end": "00:21:54",
                    "text": "it gives us the opportunity to ask more"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:52",
                    "end": "00:21:57",
                    "text": "questions."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:54",
                    "end": "00:22:01",
                    "text": "The fundamental fuel of human capability"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:57",
                    "end": "00:22:05",
                    "text": "has always been the same, curiosity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:01",
                    "end": "00:22:08",
                    "text": "We need to look at AI as the next great"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:05",
                    "end": "00:22:12",
                    "text": "generator for human curiosity and we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:08",
                    "end": "00:22:14",
                    "text": "need to take some inspiration because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:12",
                    "end": "00:22:16",
                    "text": "we all know the world could use a little"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:14",
                    "end": "00:22:18",
                    "text": "more inspiration."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:16",
                    "end": "00:22:20",
                    "text": "I often think about the following."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:18",
                    "end": "00:22:23",
                    "text": "Before the invention of the washing"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:20",
                    "end": "00:22:27",
                    "text": "machine, it took someone almost always a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:23",
                    "end": "00:22:30",
                    "text": "woman between 6 and 8 hours to wash a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:27",
                    "end": "00:22:32",
                    "text": "load of laundry. But as the washing"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:30",
                    "end": "00:22:34",
                    "text": "machine improved, that was compressed to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:32",
                    "end": "00:22:36",
                    "text": "30 minutes. But do you know what"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:34",
                    "end": "00:22:39",
                    "text": "happened? One thing happened more than"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:36",
                    "end": "00:22:42",
                    "text": "anything else. Everyone wanted to wear"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:39",
                    "end": "00:22:44",
                    "text": "cleaner clothes."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:42",
                    "end": "00:22:47",
                    "text": "Everyone expected to wear cleaner"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:44",
                    "end": "00:22:50",
                    "text": "clothes. People did their laundry a lot"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:47",
                    "end": "00:22:52",
                    "text": "more often. They had better clothes and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:50",
                    "end": "00:22:55",
                    "text": "they had more time and they put that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:52",
                    "end": "00:22:57",
                    "text": "time to work to do more with their"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:55",
                    "end": "00:22:60",
                    "text": "lives."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:57",
                    "end": "00:23:02",
                    "text": "Ultimately that is the question for us"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:60",
                    "end": "00:23:04",
                    "text": "and this too is not just a question for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:02",
                    "end": "00:23:07",
                    "text": "tech companies. This is about what tech"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:04",
                    "end": "00:23:10",
                    "text": "will do to people is people will use"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:07",
                    "end": "00:23:13",
                    "text": "technology to do for people and it's not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:10",
                    "end": "00:23:15",
                    "text": "people who create products. It's every"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:13",
                    "end": "00:23:18",
                    "text": "government. It's every company. It's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:15",
                    "end": "00:23:21",
                    "text": "every nonprofit. It's every employer"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:18",
                    "end": "00:23:24",
                    "text": "because we all have the opportunity to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:21",
                    "end": "00:23:26",
                    "text": "work with our people to manage through"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:24",
                    "end": "00:23:29",
                    "text": "the change that is coming"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:26",
                    "end": "00:23:34",
                    "text": "to show people how with the right AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:29",
                    "end": "00:23:37",
                    "text": "skills, they can create jobs and careers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:34",
                    "end": "00:23:40",
                    "text": "that will be brighter for their future."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:37",
                    "end": "00:23:42",
                    "text": "That will not be easy. But if that's not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:40",
                    "end": "00:23:44",
                    "text": "our goal, then we're missing the big"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:42",
                    "end": "00:23:47",
                    "text": "picture."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:44",
                    "end": "00:23:49",
                    "text": "As we come away from this AI summit, it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:47",
                    "end": "00:23:52",
                    "text": "reflects so much progress. But I would"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:49",
                    "end": "00:23:54",
                    "text": "say one thing as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:52",
                    "end": "00:23:58",
                    "text": "Each of these AI summits is a proud"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:54",
                    "end": "00:24:01",
                    "text": "moment for a great nation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:58",
                    "end": "00:24:03",
                    "text": "But we have an opportunity. Rather than"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:01",
                    "end": "00:24:05",
                    "text": "have summits that are islands that are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:03",
                    "end": "00:24:08",
                    "text": "disconnected from the summits before or"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:05",
                    "end": "00:24:10",
                    "text": "that follow, we need to build bridges."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:08",
                    "end": "00:24:13",
                    "text": "We need to be build bridges between"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:10",
                    "end": "00:24:15",
                    "text": "these summits. We need to define clear"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:13",
                    "end": "00:24:18",
                    "text": "goals. We need to have common"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:15",
                    "end": "00:24:21",
                    "text": "measurement systems. And every year we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:18",
                    "end": "00:24:25",
                    "text": "need to ask the same question. Did we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:21",
                    "end": "00:24:29",
                    "text": "make 12 months of progress in the year"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:25",
                    "end": "00:24:31",
                    "text": "that just preceded our meeting? How can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:29",
                    "end": "00:24:34",
                    "text": "we build on that progress to do more and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:31",
                    "end": "00:24:36",
                    "text": "move faster in the year ahead?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:34",
                    "end": "00:24:38",
                    "text": "For those of us who come to these"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:36",
                    "end": "00:24:42",
                    "text": "meetings in different countries every"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:38",
                    "end": "00:24:45",
                    "text": "year, I hope we will take that away."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:42",
                    "end": "00:24:48",
                    "text": "Let's aim higher, not just for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:45",
                    "end": "00:24:52",
                    "text": "technology, but for what technology can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:48",
                    "end": "00:24:54",
                    "text": "do for people. Let's be clear in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:52",
                    "end": "00:24:59",
                    "text": "defining what we want to accomplish."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:54",
                    "end": "00:25:03",
                    "text": "Let's put in place the ability for us to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:59",
                    "end": "00:25:05",
                    "text": "measure our progress and all hold"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:03",
                    "end": "00:25:10",
                    "text": "ourselves accountable as a global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:05",
                    "end": "00:25:12",
                    "text": "community. If we can do these things,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:10",
                    "end": "00:25:15",
                    "text": "then we can use these summits and we can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:12",
                    "end": "00:25:18",
                    "text": "use this next generation of technology"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:15",
                    "end": "00:25:21",
                    "text": "to build a better world. I know that's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:18",
                    "end": "00:25:23",
                    "text": "what we within these walls want us"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:21",
                    "end": "00:25:26",
                    "text": "ourselves to do. But even more than"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:23",
                    "end": "00:25:30",
                    "text": "that, I know that the people outside"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:26",
                    "end": "00:25:33",
                    "text": "these walls are hoping and expecting us"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:30",
                    "end": "00:25:35",
                    "text": "to do just that. Thank you very much."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:33",
                    "end": "00:25:38",
                    "text": "Important day when the world comes"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:35",
                    "end": "00:25:41",
                    "text": "together under one roof as we have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:38",
                    "end": "00:25:44",
                    "text": "today. It gives us an opportunity"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:41",
                    "end": "00:25:47",
                    "text": "together to ask important questions,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:44",
                    "end": "00:25:50",
                    "text": "even hard questions, and think about how"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:47",
                    "end": "00:25:53",
                    "text": "we want to answer them. As we think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:50",
                    "end": "00:25:55",
                    "text": "about this summit, the first AI summit"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:53",
                    "end": "00:25:58",
                    "text": "in the global south, it's only right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:55",
                    "end": "00:25:60",
                    "text": "that we start, I believe, by focusing on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:58",
                    "end": "00:26:03",
                    "text": "AI and what it means for the global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:60",
                    "end": "00:26:06",
                    "text": "south. In some ways, I think the best"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:03",
                    "end": "00:26:08",
                    "text": "way to start thinking about AI is to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:06",
                    "end": "00:26:12",
                    "text": "look more broadly and think about the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:08",
                    "end": "00:26:14",
                    "text": "state of the world in which we live."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:12",
                    "end": "00:26:17",
                    "text": "We live in a tumultuous time and in a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:14",
                    "end": "00:26:21",
                    "text": "fragmented world. But I think in so many"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:17",
                    "end": "00:26:24",
                    "text": "ways the deepest and most enduring"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:21",
                    "end": "00:26:27",
                    "text": "divide has been the economic divide"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:24",
                    "end": "00:26:29",
                    "text": "between the global north and south. And"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:27",
                    "end": "00:26:32",
                    "text": "what I believe we need to recognize is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:29",
                    "end": "00:26:37",
                    "text": "that this economic divide"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:32",
                    "end": "00:26:40",
                    "text": "is a result more than anything else of a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:37",
                    "end": "00:26:43",
                    "text": "technology divide. the technology divide"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:40",
                    "end": "00:26:45",
                    "text": "created by unequal access to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:43",
                    "end": "00:26:47",
                    "text": "electricity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:45",
                    "end": "00:26:49",
                    "text": "Electricity became one of humanity's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:47",
                    "end": "00:26:52",
                    "text": "most important generalpurpose"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:49",
                    "end": "00:26:55",
                    "text": "technologies, meaning it spread across"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:52",
                    "end": "00:26:58",
                    "text": "economies. It was applied in every"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:55",
                    "end": "00:27:02",
                    "text": "industry. It boosted productivity. Where"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:58",
                    "end": "00:27:05",
                    "text": "electricity went, economic development"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:02",
                    "end": "00:27:07",
                    "text": "and prosperity followed."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:05",
                    "end": "00:27:11",
                    "text": "But as we all know, electricity did not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:07",
                    "end": "00:27:15",
                    "text": "spread everywhere at the same pace. It"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:11",
                    "end": "00:27:17",
                    "text": "was literally 144 years ago that the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:15",
                    "end": "00:27:20",
                    "text": "first electrical power plant started"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:17",
                    "end": "00:27:23",
                    "text": "operating in lower Manhattan. And yet we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:20",
                    "end": "00:27:27",
                    "text": "come together today. And we still live"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:23",
                    "end": "00:27:30",
                    "text": "in a world where 700 million people lack"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:27",
                    "end": "00:27:34",
                    "text": "access to electricity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:30",
                    "end": "00:27:37",
                    "text": "Now comes AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:34",
                    "end": "00:27:39",
                    "text": "AI perhaps more than any other"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:37",
                    "end": "00:27:42",
                    "text": "technology this century will play a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:39",
                    "end": "00:27:46",
                    "text": "bigger role either in closing this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:42",
                    "end": "00:27:49",
                    "text": "economic divide or in exacerbating it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:46",
                    "end": "00:27:52",
                    "text": "and making it even wider."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:49",
                    "end": "00:27:53",
                    "text": "That is perhaps the single most"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:52",
                    "end": "00:27:56",
                    "text": "important question for us I would"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:53",
                    "end": "00:27:59",
                    "text": "suggest today as we think about the role"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:56",
                    "end": "00:28:04",
                    "text": "of AI in the global south. How can we do"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:59",
                    "end": "00:28:07",
                    "text": "better? Because we need to do better."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:04",
                    "end": "00:28:09",
                    "text": "What will it take? I think it's going to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:07",
                    "end": "00:28:12",
                    "text": "take a few things that will require that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:09",
                    "end": "00:28:15",
                    "text": "we all come together and work together."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:12",
                    "end": "00:28:17",
                    "text": "First, the obvious. We need to bring"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:15",
                    "end": "00:28:20",
                    "text": "infrastructure to the global south. That"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:17",
                    "end": "00:28:23",
                    "text": "means data centers and commute compute."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:20",
                    "end": "00:28:27",
                    "text": "It also means more connectivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:23",
                    "end": "00:28:29",
                    "text": "It me it means more electricity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:27",
                    "end": "00:28:32",
                    "text": "That is going to take not only the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:29",
                    "end": "00:28:34",
                    "text": "world's best technology, it's going to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:32",
                    "end": "00:28:36",
                    "text": "require an enormous amount of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:34",
                    "end": "00:28:38",
                    "text": "investment."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:36",
                    "end": "00:28:40",
                    "text": "That's why we at Microsoft announced"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:38",
                    "end": "00:28:42",
                    "text": "yesterday morning that we're on pace to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:40",
                    "end": "00:28:46",
                    "text": "spend $50 billion by the end of this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:42",
                    "end": "00:28:49",
                    "text": "decade to bring AI to the global south."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:46",
                    "end": "00:28:52",
                    "text": "And of all the countries in which we are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:49",
                    "end": "00:28:56",
                    "text": "investing, India, not surprisingly, is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:52",
                    "end": "00:28:56",
                    "text": "one of the largest."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:56",
                    "end": "00:29:00",
                    "text": "The"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:57",
                    "end": "00:29:02",
                    "text": "But we'll need to harness private"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:00",
                    "end": "00:29:05",
                    "text": "capital, investments from tech"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:02",
                    "end": "00:29:07",
                    "text": "companies, other sources of private"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:05",
                    "end": "00:29:09",
                    "text": "capital, government funding. We'll need"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:07",
                    "end": "00:29:11",
                    "text": "governments and others to generate"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:09",
                    "end": "00:29:14",
                    "text": "demand for the use of AI in the global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:11",
                    "end": "00:29:16",
                    "text": "south because that is the only way to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:14",
                    "end": "00:29:20",
                    "text": "get the wheels of the market spinning"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:16",
                    "end": "00:29:22",
                    "text": "and to do what we need to do together."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:20",
                    "end": "00:29:23",
                    "text": "That's the first thing we need to do"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:22",
                    "end": "00:29:25",
                    "text": "together."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:23",
                    "end": "00:29:27",
                    "text": "There's a second thing we need to do."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:25",
                    "end": "00:29:29",
                    "text": "We've talked about it here already today"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:27",
                    "end": "00:29:32",
                    "text": "and it is so clear when you study the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:29",
                    "end": "00:29:35",
                    "text": "history of technology. Infrastructure is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:32",
                    "end": "00:29:36",
                    "text": "not only hardware. It's not only wires"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:35",
                    "end": "00:29:40",
                    "text": "and grids."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:36",
                    "end": "00:29:42",
                    "text": "It's skilling for people. Because the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:40",
                    "end": "00:29:44",
                    "text": "key to enabling a country and a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:42",
                    "end": "00:29:47",
                    "text": "population to use a general purpose"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:44",
                    "end": "00:29:50",
                    "text": "technology at scale is to give people"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:47",
                    "end": "00:29:52",
                    "text": "across the country access to the skills"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:50",
                    "end": "00:29:55",
                    "text": "they need to put it to work. And that's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:52",
                    "end": "00:29:57",
                    "text": "why it's such good news that we seem so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:55",
                    "end": "00:29:59",
                    "text": "many governments supported by the United"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:57",
                    "end": "00:30:04",
                    "text": "Nations and supported by private"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:59",
                    "end": "00:30:05",
                    "text": "companies investing in more skilling."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:04",
                    "end": "00:30:08",
                    "text": "It's actually something that should"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:05",
                    "end": "00:30:11",
                    "text": "speak to all of us. Certainly, as a tech"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:08",
                    "end": "00:30:13",
                    "text": "company, we're committed. We've launched"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:11",
                    "end": "00:30:15",
                    "text": "through Microsoft Elevate new"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:13",
                    "end": "00:30:17",
                    "text": "initiatives, including one we're"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:15",
                    "end": "00:30:20",
                    "text": "announcing this week, Microsoft Elevate"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:17",
                    "end": "00:30:24",
                    "text": "for Educators to equip teachers with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:20",
                    "end": "00:30:27",
                    "text": "access to help their students learn how"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:24",
                    "end": "00:30:30",
                    "text": "to use AI. But the truth, it doesn't"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:27",
                    "end": "00:30:33",
                    "text": "matter where you work, you have a role"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:30",
                    "end": "00:30:37",
                    "text": "to play. Because the lesson of digital"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:33",
                    "end": "00:30:39",
                    "text": "technology, I was an example of this was"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:37",
                    "end": "00:30:43",
                    "text": "that it took employers to open their"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:39",
                    "end": "00:30:46",
                    "text": "doors to computing. It takes employers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:43",
                    "end": "00:30:48",
                    "text": "today to open their doors to new AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:46",
                    "end": "00:30:52",
                    "text": "tools. It will take employers to invest"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:48",
                    "end": "00:30:55",
                    "text": "in the skilling of their employees."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:52",
                    "end": "00:30:57",
                    "text": "It's not just for the next generation,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:55",
                    "end": "00:30:59",
                    "text": "it's for every generation that this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:57",
                    "end": "00:31:01",
                    "text": "fully matters."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:59",
                    "end": "00:31:04",
                    "text": "Then there's a third challenge for the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:01",
                    "end": "00:31:06",
                    "text": "global south. We need to make AI work"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:04",
                    "end": "00:31:08",
                    "text": "effectively for the global south and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:06",
                    "end": "00:31:12",
                    "text": "that requires some special initiatives"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:08",
                    "end": "00:31:15",
                    "text": "at least two. First, we need to make AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:12",
                    "end": "00:31:18",
                    "text": "as effective in every language as it is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:15",
                    "end": "00:31:21",
                    "text": "in English. And today it is not."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:18",
                    "end": "00:31:23",
                    "text": "Performance tests show that's the case."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:21",
                    "end": "00:31:24",
                    "text": "That's why one of the good things to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:23",
                    "end": "00:31:27",
                    "text": "come out of this week is new"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:24",
                    "end": "00:31:30",
                    "text": "announcements to invest upstream in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:27",
                    "end": "00:31:32",
                    "text": "better data in other languages to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:30",
                    "end": "00:31:35",
                    "text": "provide better tools and measurement"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:32",
                    "end": "00:31:38",
                    "text": "systems for AI that is built in other"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:35",
                    "end": "00:31:42",
                    "text": "languages to build out data providence"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:38",
                    "end": "00:31:45",
                    "text": "with a view to linguistic diversity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:42",
                    "end": "00:31:48",
                    "text": "Diversity that we need to advance around"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:45",
                    "end": "00:31:50",
                    "text": "the world. And we need to use AI in the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:48",
                    "end": "00:31:53",
                    "text": "global south to solve the problems that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:50",
                    "end": "00:31:55",
                    "text": "matter to the global south. Oftentimes,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:53",
                    "end": "00:31:58",
                    "text": "as we're doing here in India, that's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:55",
                    "end": "00:32:00",
                    "text": "about improvements in agriculture or as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:58",
                    "end": "00:32:03",
                    "text": "a number of partners including Microsoft"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:00",
                    "end": "00:32:05",
                    "text": "are doing this week, launching a new"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:03",
                    "end": "00:32:09",
                    "text": "initiative to address food security"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:05",
                    "end": "00:32:12",
                    "text": "across Africa. These are just two of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:09",
                    "end": "00:32:15",
                    "text": "myriad of opportunities we have to put"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:12",
                    "end": "00:32:18",
                    "text": "AI to work in ways that will bring"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:15",
                    "end": "00:32:20",
                    "text": "faster benefits to countries in most of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:18",
                    "end": "00:32:23",
                    "text": "the world."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:20",
                    "end": "00:32:25",
                    "text": "If we do those three things well, build"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:23",
                    "end": "00:32:27",
                    "text": "infrastructure, invest in skilling,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:25",
                    "end": "00:32:30",
                    "text": "address real world problems, then I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:27",
                    "end": "00:32:32",
                    "text": "think it may create the foundation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:30",
                    "end": "00:32:35",
                    "text": "to think more and do more about the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:32",
                    "end": "00:32:38",
                    "text": "question I am willing to bet will be a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:35",
                    "end": "00:32:42",
                    "text": "bigger part of the conversation in the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:38",
                    "end": "00:32:46",
                    "text": "next few AI summits. What will AI mean"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:42",
                    "end": "00:32:49",
                    "text": "for the future of work and jobs?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:46",
                    "end": "00:32:52",
                    "text": "Within these walls, I think we're mostly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:49",
                    "end": "00:32:55",
                    "text": "enthusiastic about the future of AI. But"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:52",
                    "end": "00:32:57",
                    "text": "outside these walls, I think we need to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:55",
                    "end": "00:33:00",
                    "text": "recognize that increasingly around the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:57",
                    "end": "00:33:02",
                    "text": "world and especially in some countries,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:00",
                    "end": "00:33:04",
                    "text": "many parents are asking a common"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:02",
                    "end": "00:33:08",
                    "text": "question."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:04",
                    "end": "00:33:11",
                    "text": "What will AI mean for my kids? What will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:08",
                    "end": "00:33:15",
                    "text": "AI mean for my family? What will AI mean"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:11",
                    "end": "00:33:17",
                    "text": "for our future? I get it that some"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:15",
                    "end": "00:33:22",
                    "text": "people are excited and they'll do well,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:17",
                    "end": "00:33:26",
                    "text": "but what about us? Us meaning most of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:22",
                    "end": "00:33:28",
                    "text": "the people who live on this planet. I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:26",
                    "end": "00:33:30",
                    "text": "think we have something to prove. I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:28",
                    "end": "00:33:32",
                    "text": "think we have something to prove not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:30",
                    "end": "00:33:35",
                    "text": "only to communities and countries and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:32",
                    "end": "00:33:38",
                    "text": "our customers, but to ourselves that we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:35",
                    "end": "00:33:41",
                    "text": "can not only embrace but pursue a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:38",
                    "end": "00:33:43",
                    "text": "brighter future for people. It's great"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:41",
                    "end": "00:33:45",
                    "text": "to come to conferences like this and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:43",
                    "end": "00:33:47",
                    "text": "hear people talk about all of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:45",
                    "end": "00:33:50",
                    "text": "advances in technology, but let's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:47",
                    "end": "00:33:55",
                    "text": "remember one other thing as well. Human"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:50",
                    "end": "00:33:56",
                    "text": "capability is neither fixed nor finite."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:55",
                    "end": "00:33:59",
                    "text": "It's great to think about what it would"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:56",
                    "end": "00:34:00",
                    "text": "mean to have computers in a data center"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:59",
                    "end": "00:34:04",
                    "text": "that would be like a country of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:00",
                    "end": "00:34:06",
                    "text": "geniuses. But let's also recognize this."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:04",
                    "end": "00:34:10",
                    "text": "Compared to the people who lived in the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:06",
                    "end": "00:34:13",
                    "text": "bronze age, all of you, all of us are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:10",
                    "end": "00:34:16",
                    "text": "already geniuses."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:13",
                    "end": "00:34:20",
                    "text": "Whenever technology advances, it creates"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:16",
                    "end": "00:34:22",
                    "text": "a new platform, a new foundation that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:20",
                    "end": "00:34:27",
                    "text": "enables people to stand taller and reach"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:22",
                    "end": "00:34:31",
                    "text": "higher if and only if we're committed to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:27",
                    "end": "00:34:34",
                    "text": "using that technology. Well, as AI makes"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:31",
                    "end": "00:34:36",
                    "text": "it possible to cure more diseases, then"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:34",
                    "end": "00:34:39",
                    "text": "it is right that we expect that it will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:36",
                    "end": "00:34:42",
                    "text": "improve human health. As we use AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:39",
                    "end": "00:34:43",
                    "text": "already every day to find faster"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:42",
                    "end": "00:34:47",
                    "text": "answers,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:43",
                    "end": "00:34:48",
                    "text": "it gives us the opportunity to ask more"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:47",
                    "end": "00:34:52",
                    "text": "questions."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:48",
                    "end": "00:34:56",
                    "text": "The fundamental fuel of human capability"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:52",
                    "end": "00:34:59",
                    "text": "has always been the same, curiosity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:56",
                    "end": "00:35:03",
                    "text": "We need to look at AI as the next great"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:59",
                    "end": "00:35:07",
                    "text": "generator for human curiosity and we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:03",
                    "end": "00:35:09",
                    "text": "need to take some inspiration because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:07",
                    "end": "00:35:11",
                    "text": "we all know the world could use a little"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:09",
                    "end": "00:35:12",
                    "text": "more inspiration."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:11",
                    "end": "00:35:15",
                    "text": "I often think about the following."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:12",
                    "end": "00:35:18",
                    "text": "Before the invention of the washing"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:15",
                    "end": "00:35:22",
                    "text": "machine, it took someone almost always a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:18",
                    "end": "00:35:25",
                    "text": "woman between 6 and 8 hours to wash a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:22",
                    "end": "00:35:26",
                    "text": "load of laundry. But as the washing"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:25",
                    "end": "00:35:28",
                    "text": "machine improved, that was compressed to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:26",
                    "end": "00:35:30",
                    "text": "30 minutes. But do you know what"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:28",
                    "end": "00:35:34",
                    "text": "happened? One thing happened more than"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:30",
                    "end": "00:35:36",
                    "text": "anything else. Everyone wanted to wear"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:34",
                    "end": "00:35:38",
                    "text": "cleaner clothes."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:36",
                    "end": "00:35:41",
                    "text": "Everyone expected to wear cleaner"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:38",
                    "end": "00:35:44",
                    "text": "clothes. People did their laundry a lot"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:41",
                    "end": "00:35:46",
                    "text": "more often. They had better clothes and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:44",
                    "end": "00:35:50",
                    "text": "they had more time and they put that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:46",
                    "end": "00:35:51",
                    "text": "time to work to do more with their"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:50",
                    "end": "00:35:54",
                    "text": "lives."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:51",
                    "end": "00:35:56",
                    "text": "Ultimately that is the question for us"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:54",
                    "end": "00:35:58",
                    "text": "and this too is not just a question for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:56",
                    "end": "00:36:02",
                    "text": "tech companies. This is what tech will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:58",
                    "end": "00:36:05",
                    "text": "do to people is people will use"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:02",
                    "end": "00:36:07",
                    "text": "technology to do for people and it's not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:05",
                    "end": "00:36:09",
                    "text": "people who create products. It's every"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:07",
                    "end": "00:36:13",
                    "text": "government. It's every company. It's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:09",
                    "end": "00:36:15",
                    "text": "every nonprofit. It's every employer"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:13",
                    "end": "00:36:18",
                    "text": "because we all have the opportunity to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:15",
                    "end": "00:36:21",
                    "text": "work with our people to manage through"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:18",
                    "end": "00:36:23",
                    "text": "the change that is coming"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:21",
                    "end": "00:36:29",
                    "text": "to show people how with the right AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:23",
                    "end": "00:36:32",
                    "text": "skills, they can create jobs and careers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:29",
                    "end": "00:36:34",
                    "text": "that will be brighter for their future."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:32",
                    "end": "00:36:37",
                    "text": "That will not be easy. But if that's not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:34",
                    "end": "00:36:38",
                    "text": "our goal, then we're missing the big"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:37",
                    "end": "00:36:41",
                    "text": "picture."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:38",
                    "end": "00:36:43",
                    "text": "As we come away from this AI summit, it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:41",
                    "end": "00:36:46",
                    "text": "reflects so much progress. But I would"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:43",
                    "end": "00:36:49",
                    "text": "say one thing as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:46",
                    "end": "00:36:52",
                    "text": "Each of these AI summits is a proud"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:49",
                    "end": "00:36:55",
                    "text": "moment for a great nation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:52",
                    "end": "00:36:57",
                    "text": "But we have an opportunity rather than"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:55",
                    "end": "00:36:59",
                    "text": "have summits that are islands that are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:57",
                    "end": "00:37:02",
                    "text": "disconnected from the summits before or"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:59",
                    "end": "00:37:05",
                    "text": "that follow. We need to build bridges."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:02",
                    "end": "00:37:07",
                    "text": "We need to be build bridges between"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:05",
                    "end": "00:37:10",
                    "text": "these summits. We need to define clear"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:07",
                    "end": "00:37:12",
                    "text": "goals. We need to have common"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:10",
                    "end": "00:37:16",
                    "text": "measurement systems. And every year we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:12",
                    "end": "00:37:20",
                    "text": "need to ask the same question. Did we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:16",
                    "end": "00:37:23",
                    "text": "make 12 months of progress in the year"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:20",
                    "end": "00:37:26",
                    "text": "that just preceded our meeting? How can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:23",
                    "end": "00:37:29",
                    "text": "we build on that progress to do more and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:26",
                    "end": "00:37:31",
                    "text": "move faster in the year ahead?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:29",
                    "end": "00:37:33",
                    "text": "For those of us who come to these"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:31",
                    "end": "00:37:36",
                    "text": "meetings in different countries every"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:33",
                    "end": "00:37:40",
                    "text": "year, I hope we will take that away."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:36",
                    "end": "00:37:43",
                    "text": "Let's aim higher, not just for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:40",
                    "end": "00:37:46",
                    "text": "technology, but for what technology can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:43",
                    "end": "00:37:49",
                    "text": "do for people. Let's be clear in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:46",
                    "end": "00:37:53",
                    "text": "defining what we want to accomplish."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:49",
                    "end": "00:37:57",
                    "text": "Let's put in place the ability for us to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:53",
                    "end": "00:37:59",
                    "text": "measure our progress and all hold"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:57",
                    "end": "00:38:04",
                    "text": "ourselves accountable as a global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:59",
                    "end": "00:38:06",
                    "text": "community. If we can do these things,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:04",
                    "end": "00:38:09",
                    "text": "then we can use these summits and we can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:06",
                    "end": "00:38:12",
                    "text": "use this next generation of technology"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:09",
                    "end": "00:38:15",
                    "text": "to build a better world. I know that's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:12",
                    "end": "00:38:18",
                    "text": "what we within these walls want us"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:15",
                    "end": "00:38:20",
                    "text": "ourselves to do. But even more than"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:18",
                    "end": "00:38:25",
                    "text": "that, I know that the people outside"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:20",
                    "end": "00:38:28",
                    "text": "these walls are hoping and expecting us"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:25",
                    "end": "00:38:30",
                    "text": "to do just that. Thank you very much."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:28",
                    "end": "00:38:33",
                    "text": "Important day when the world comes"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:30",
                    "end": "00:38:36",
                    "text": "together under one roof as we have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:33",
                    "end": "00:38:38",
                    "text": "today. It gives us an opportunity"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:36",
                    "end": "00:38:42",
                    "text": "together to ask important questions,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:38",
                    "end": "00:38:44",
                    "text": "even hard questions, and think about how"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:42",
                    "end": "00:38:47",
                    "text": "we want to answer them. As we think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:44",
                    "end": "00:38:49",
                    "text": "about this summit, the first AI summit"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:47",
                    "end": "00:38:52",
                    "text": "in the global south, it's only right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:49",
                    "end": "00:38:54",
                    "text": "that we start, I believe, by focusing on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:52",
                    "end": "00:38:57",
                    "text": "AI and what it means for the global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:54",
                    "end": "00:39:00",
                    "text": "south. In some ways, I think the best"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:57",
                    "end": "00:39:02",
                    "text": "way to start thinking about AI is to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:00",
                    "end": "00:39:06",
                    "text": "look more broadly and think about the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:02",
                    "end": "00:39:09",
                    "text": "state of the world in which we live. We"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:06",
                    "end": "00:39:12",
                    "text": "live in a tumultuous time and in a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:09",
                    "end": "00:39:15",
                    "text": "fragmented world. But I think in so many"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:12",
                    "end": "00:39:19",
                    "text": "ways the deepest and most enduring"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:15",
                    "end": "00:39:22",
                    "text": "divide has been the economic divide"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:19",
                    "end": "00:39:24",
                    "text": "between the global north and south. And"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:22",
                    "end": "00:39:27",
                    "text": "what I believe we need to recognize is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:24",
                    "end": "00:39:31",
                    "text": "that this economic divide"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:27",
                    "end": "00:39:35",
                    "text": "is a result more than anything else of a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:31",
                    "end": "00:39:37",
                    "text": "technology divide. the technology divide"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:35",
                    "end": "00:39:39",
                    "text": "created by unequal access to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:37",
                    "end": "00:39:42",
                    "text": "electricity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:39",
                    "end": "00:39:43",
                    "text": "Electricity became one of humanity's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:42",
                    "end": "00:39:46",
                    "text": "most important generalpurpose"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:43",
                    "end": "00:39:49",
                    "text": "technologies, meaning it spread across"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:46",
                    "end": "00:39:53",
                    "text": "economies. It was applied in every"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:49",
                    "end": "00:39:56",
                    "text": "industry. It boosted productivity. Where"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:53",
                    "end": "00:39:59",
                    "text": "electricity went, economic development"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:56",
                    "end": "00:40:02",
                    "text": "and prosperity followed."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:59",
                    "end": "00:40:05",
                    "text": "But as we all know, electricity did not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:02",
                    "end": "00:40:09",
                    "text": "spread everywhere at the same pace. It"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:05",
                    "end": "00:40:11",
                    "text": "was literally 144 years ago that the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:09",
                    "end": "00:40:14",
                    "text": "first electrical power plant started"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:11",
                    "end": "00:40:17",
                    "text": "operating in lower Manhattan. And yet we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:14",
                    "end": "00:40:22",
                    "text": "come together today and we still live in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:17",
                    "end": "00:40:25",
                    "text": "a world where 700 million people lack"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:22",
                    "end": "00:40:28",
                    "text": "access to electricity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:25",
                    "end": "00:40:31",
                    "text": "Now comes AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:28",
                    "end": "00:40:34",
                    "text": "AI perhaps more than any other"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:31",
                    "end": "00:40:37",
                    "text": "technology this century will play a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:34",
                    "end": "00:40:37",
                    "text": "bigger role either"
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6996d6cf565455da1fc75ef3",
    "projectId": "298585",
    "projectName": "Yann LeCun Keynote: Deep Dive into Advanced Machine Intelligence",
    "projectDescription": "Analyzing Yann LeCun's Keynote: Challenges and Opportunities in AI",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T09:24:29.269Z",
    "projectUpdatedDate": "2026-02-19T09:24:29.269Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-19T10:54:52.179Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "yann LeCunn trans.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6996e60a565455da1fc7608c",
    "projectId": "754528",
    "projectName": "Sam altmann",
    "projectDescription": "Sam altmann",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T10:29:25.656Z",
    "projectUpdatedDate": "2026-02-19T10:29:25.656Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-19T10:30:22.539Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Transcript Sam Altmann.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6996ede6565455da1fc7612d",
    "projectId": "520137",
    "projectName": "Fireside Chat Nandan nilekani",
    "projectDescription": "Fireside Chat Nandan nilekani",
     "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T11:03:00.166Z",
    "projectUpdatedDate": "2026-02-19T11:03:00.166Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-19T11:03:42.114Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Fireside Chat Nandan Nilekani.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
   {
    "_id": "6996bc20565455da1fc75d38",
    "projectId": "926926",
    "projectName": "Keynote Session: Vishal Sikka, Chief Executive Officer, Vianai ",
    "projectDescription": "Keynote Session: Vishal Sikka, Chief Executive Officer, Vianai | Ai Summit India 2026",
     "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T07:30:39.662Z",
    "projectUpdatedDate": "2026-02-19T07:30:39.662Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-19T08:18:15.040Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "Vishal_Sikka__founder_and_CEO_of_Vianai_address_at_India_AI_Impact_summit__AI_Growth__Job___Delhi(MP3_160K)[1].mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771486195925.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:29",
                    "text": " from a long time that I have spent in AI. My first point is that what we see today in the world of AI, people who know what they are doing with AI are astonishingly effective with AI. Recently, a friend of mine, he and I were students together at Stanford. He has a large service that he runs, open public service."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:29",
                    "end": "00:00:57",
                    "text": " That service was built by 15 people, very world-class engineers, over nine months. Recently, he rebuilt that service entirely by himself in 14 days using one of the generative AI coding tools. So if you are counting, that is about a more than 250 times improvement in productivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:57",
                    "end": "00:01:25",
                    "text": " Now, he's a genius and not everyone gets a 250 times productivity gain. But you will see that people who understand how to use AI are astonishingly effective with it. And I had a similar experience recently with a customer of mine who is a distributor of home goods. And one of their main suppliers shut down their factories in one of their countries."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:26",
                    "end": "00:01:54",
                    "text": " They did analysis using our product, all kinds of simulation scenarios, and over a few days, they reached the decision that they need to exit that country entirely. I asked him, you know, such a monumental decision to get out of an entire country, how long would that have taken you before? And he said, easily it would have taken a year, and it would have involved heavy duty consultants and things like that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:56",
                    "end": "00:02:24",
                    "text": " So we now have instant access to knowledge in any language, a condensation of things that we can present in any way. It is an incredible power. And yes, it is deeply disruptive to the ways that we have worked before, the way that we have done things in the past. But at the same time, and even more importantly, we can do unprecedented things with this, things that we could never do before."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:26",
                    "end": "00:02:54",
                    "text": " So this astonishing effectiveness, and Yoshua referred to this as a jagged frontier, it is not uniform. Not everyone sees this. So that's my second point. Being effective with AI requires not only a knowledge of AI itself, but understanding its limitations and how to overcome those limitations. There is a huge gap between LLMs."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:54",
                    "end": "00:03:25",
                    "text": " and the business users inside enterprises especially, and how to bring value to those users. Overcoming that gap is where a lot of value creating opportunity is. Bridging that gap requires delivering correct systems, trusted, verifiable, reliable systems that deliver value to people. My own company works in this area, a layer that sits above the language models and delivers value."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:25",
                    "end": "00:03:55",
                    "text": " to business users while ensuring correctness and things like that. When we overcome that gap, we can deliver massive value. Mukesh Bhai talked about it just now. We can transform every existing system, legacy systems, enormous complexities inside enterprises can be removed. Industries can be transformed. We can give end users wings. We can amplify them to deliver things that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:55",
                    "end": "00:04:26",
                    "text": " were not possible to do before that, or in the best case, it required professionals to do this. Doing that also requires not just overcoming the limitations of AI, but also imagination to see what is not there, to see what is possible. India has all of this in great abundance. The Honorable Prime Minister has called for a billion entrepreneurs, people who can overcome these and deliver value using AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:26",
                    "end": "00:05:02",
                    "text": " And I think this is exactly what the world needs and what India has the potential for. My third point is that we not only have to master today's AI, but we have to leapfrog it. AI today has enormous limitations. I have worked in AI for the last 38 years. One of our scriptures is Bhaj Govindam. It was written by the Shankaracharya. And it has a beautiful line."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:02",
                    "end": "00:05:36",
                    "text": " What it means is that when you are faced with a life or death situation, the knowledge of a book does not help you. Knowledge without wisdom does not save us. That wisdom comes from living, from doing, from being in the world. AI today has plenty of limitations. Joshua mentioned hallucinations. That's one of the main issues."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:36",
                    "end": "00:06:07",
                    "text": " blocking the use of AI in enterprises. But beyond hallucinations, understanding the world, understanding physical activities, the physical movement, this is one of the next frontiers. Safety, safety of AI, and the Honorable Prime Minister talked about this today, is an existential issue. Swarms of agents can be made to do completely reckless things, and we don't yet have ways to understand or deal with this."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:07",
                    "end": "00:06:39",
                    "text": " And Joshua Demis also talked about this. We have to solve this issue. We have to deliver AI that is safe. We have done this with nuclear power for the last 80 plus years. We can and we must do this with AI. Energy is another one of these issues. Where I live in California on Highway 101, just north of San Thomas, I drive by there every time I go to see my dad. There is a massive data center that is coming up."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:39",
                    "end": "00:07:12",
                    "text": " It's 720 megawatts. This idea that I write a prompt, and these gazillion GPUs blast into existence to produce a response, and then I make a tiny change to that prompt, and then I do that again, it just seems like a completely absurd idea, especially to someone who has been around AI for such a long time. I have thanks to the minister, Ministry of IT, Honorable Ashwini Vashnavji."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:12",
                    "end": "00:07:43",
                    "text": " I have a colleague who has been accompanying me throughout this conference and he told me that yesterday he walked 32,000 steps. And I asked him, what did you eat? And he said, I ate two burgers. Shubham, here, two burgers. You know, we normally eat 2000 calories in a day. That's about a hundred watt light bulb, like less than one of these light bulbs. And out of that, our brain, our nervous system is maybe 15, 20 watts. That's like..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:43",
                    "end": "00:08:15",
                    "text": " When your laptop is in sleep mode, it takes more power than that. So, there are many zeros still to be removed from these models and the models themselves have to be removed. So, I think that there is a tremendous opportunity here. India is a country of the human potential. We have plenty of times before delivered the ability."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:15",
                    "end": "00:08:43",
                    "text": " to, you know, billion plus Indians. Mukesh Bhai talked about Jio and Sunil talked about Airtel and how now we have billion plus Indians who have data and connectivity. When I was young, one of my earliest childhood memories is of worry in my parents' faces around food. They used to tell stories of how there was a shortage of food."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:43",
                    "end": "00:09:15",
                    "text": " and then the green revolution happened, and India is now one of the largest exporters of food in one generation. So I think that when you look at the time of intelligence, it is not only an opportunity to learn about this technology, to learn to master this technology, to understand its limitations, but to leapfrog that and to build the next generation of it. And as this summit so vividly demonstrates,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:16",
                    "end": "00:09:42",
                    "text": " We can be on our way to a human revolution powered by AI, by good AI, by purposeful AI, where every one of us, a billion entrepreneurs is not just making a living, but is making a life. Not some artificial life or some artificial general life, but our own life and the life of others. And that would be so much fun to do. Thank you so much."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:43",
                    "end": "00:10:13",
                    "text": " Thank you so much for that. Wow, wonderful introduction and what an amazing event. I want to share three points from a long time that I have spent in AI. My first point is that what we see today in the world of AI, people who know what they are doing with AI are astonishingly effective with AI. Recently, a friend of mine, he and I were students together at Stanford."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:13",
                    "end": "00:10:43",
                    "text": " He has a large service that he runs, open public service. That service was built by 15 people, very world-class engineers over nine months. Recently, he rebuilt that service entirely by himself in 14 days using one of the generative AI coding tools. So if you're counting, that is about a..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:43",
                    "end": "00:10:60",
                    "text": " more than 250 times improvement in productivity. Now he's a genius and not everyone gets a 250 times productivity gain, but you will see that people who understand how to use AI are astonishingly effective with it."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6996bee5565455da1fc75d6a",
    "projectId": "366316",
    "projectName": "Fire Side Chat AI-impact-summit-2026",
    "projectDescription": "Fire Side Chat",
     "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T07:42:24.828Z",
    "projectUpdatedDate": "2026-02-19T07:42:24.828Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-19T07:43:13.507Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Fire Side Chat.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
   {
    "_id": "6996cfe1565455da1fc75e9d",
    "projectId": "588795",
    "projectName": "Keynote Session: Mukesh Dhirubhai Ambani, Chairman and Managing Director, Reliance",
    "projectDescription": "Keynote Session: Mukesh Dhirubhai Ambani, Chairman and Managing Director, Reliance | AI india Summit - 2026",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T08:54:56.444Z",
    "projectUpdatedDate": "2026-02-19T08:54:56.444Z",
    "projectStatus": "Active",
    "companyId": "697624ed9093c218a6382118",
    "projectCreatedBy": "697624ed9093c218a6382119",
    "projectUpdatedBy": "697624ed9093c218a6382119",
    "projectLastAccessDate": "2026-02-19T08:56:09.989Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO_LINK",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames"
        },
        {
            "col1": "P1",
            "col2": "https://www.youtube.com/watch?v=uGxTnJCg-C4&t=845s",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEOLINK",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:06",
                    "text": "valable principles. First, AI for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:04",
                    "end": "00:00:09",
                    "text": "India's deep tech and advanced"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:06",
                    "end": "00:00:12",
                    "text": "manufacturing leadership reaching not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:09",
                    "end": "00:00:14",
                    "text": "just large enterprises but agriculture,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:12",
                    "end": "00:00:17",
                    "text": "small businesses and the informal"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:14",
                    "end": "00:00:20",
                    "text": "sector. Geo intelligence will not simply"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:17",
                    "end": "00:00:23",
                    "text": "be a search or an ask tool. It'll"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:20",
                    "end": "00:00:25",
                    "text": "primarily be a resource for multiplying"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:23",
                    "end": "00:00:29",
                    "text": "productivity and efficiency."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:25",
                    "end": "00:00:33",
                    "text": "Secondly, worldleading multilingual AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:29",
                    "end": "00:00:35",
                    "text": "capability across all Indian languages."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:33",
                    "end": "00:00:38",
                    "text": "When farmers and articians speak to AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:35",
                    "end": "00:00:41",
                    "text": "in their own words and student learn in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:38",
                    "end": "00:00:45",
                    "text": "their own mother tongue, this is not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:41",
                    "end": "00:00:45",
                    "text": "convenience. This is inclusion."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:58",
                    "end": "00:01:03",
                    "text": "Third,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:00",
                    "end": "00:01:07",
                    "text": "responsibility, security, data residency"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:03",
                    "end": "00:01:11",
                    "text": "and trust as go's core guarantees not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:07",
                    "end": "00:01:14",
                    "text": "afterthought. Fourth, we will provide we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:11",
                    "end": "00:01:18",
                    "text": "will prove that AI does not take away"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:14",
                    "end": "00:01:21",
                    "text": "jobs. Rather, it will create new"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:18",
                    "end": "00:01:26",
                    "text": "highskll work opportunities."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:21",
                    "end": "00:01:29",
                    "text": "And fifth, the AI story has shifted from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:26",
                    "end": "00:01:33",
                    "text": "who has the best model to who can build"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:29",
                    "end": "00:01:37",
                    "text": "the strongest ecosystem for speed and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:33",
                    "end": "00:01:40",
                    "text": "scale of usage. Therefore, we will build"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:37",
                    "end": "00:01:44",
                    "text": "deep partnership ecosystem with Indian"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:40",
                    "end": "00:01:47",
                    "text": "enterprises, startups, IIT, IISC and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:44",
                    "end": "00:01:49",
                    "text": "research institutions. We will work"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:47",
                    "end": "00:01:53",
                    "text": "shoulderto-shoulder with India's leading"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:49",
                    "end": "00:01:55",
                    "text": "industrial groups to embed AI across"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:53",
                    "end": "00:01:58",
                    "text": "manufacturing, logistics, energy,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:55",
                    "end": "00:02:00",
                    "text": "finance, retail, agriculture, and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:58",
                    "end": "00:02:02",
                    "text": "healthcare. We will empower startups"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:00",
                    "end": "00:02:05",
                    "text": "with affordable compute and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:02",
                    "end": "00:02:08",
                    "text": "co-development platforms. We will aspire"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:05",
                    "end": "00:02:10",
                    "text": "to produce global breakthroughs in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:08",
                    "end": "00:02:13",
                    "text": "compute architecture, foundation models"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:10",
                    "end": "00:02:16",
                    "text": "and energy efficiency designed in India,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:13",
                    "end": "00:02:19",
                    "text": "rooted in our values, powered by our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:16",
                    "end": "00:02:21",
                    "text": "talent and scaled for humanity. And we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:19",
                    "end": "00:02:24",
                    "text": "will partner with the very best tech"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:21",
                    "end": "00:02:28",
                    "text": "companies in the world, not as importers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:24",
                    "end": "00:02:31",
                    "text": "of intelligence, but as co-architects of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:28",
                    "end": "00:02:34",
                    "text": "a new AI century."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:31",
                    "end": "00:02:36",
                    "text": "Dear friends, I believe that social"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:34",
                    "end": "00:02:40",
                    "text": "relevance, not momentary craze, should"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:36",
                    "end": "00:02:43",
                    "text": "drive AI growth in India. Jio has"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:40",
                    "end": "00:02:45",
                    "text": "already started AI applications for the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:43",
                    "end": "00:02:49",
                    "text": "most pressing challenges in inclusive"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:45",
                    "end": "00:02:52",
                    "text": "development. In education, we have Gio"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:49",
                    "end": "00:02:56",
                    "text": "Shikshak, an adaptive AI teaching"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:52",
                    "end": "00:02:58",
                    "text": "assistant in 22 languages. When 250"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:56",
                    "end": "00:03:01",
                    "text": "million school children and 50 million"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:58",
                    "end": "00:03:04",
                    "text": "students in higher education are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:01",
                    "end": "00:03:08",
                    "text": "empowered by AI teachers, no power on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:04",
                    "end": "00:03:13",
                    "text": "earth can match India's talent wealth."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:08",
                    "end": "00:03:15",
                    "text": "In healthcare, Gio arrogia AI delivering"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:13",
                    "end": "00:03:18",
                    "text": "first medical guidance in under five"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:15",
                    "end": "00:03:22",
                    "text": "minutes in local languages on any phone."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:18",
                    "end": "00:03:25",
                    "text": "in agriculture. Geocishi converting"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:22",
                    "end": "00:03:28",
                    "text": "satellite imagery and precision weather"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:25",
                    "end": "00:03:31",
                    "text": "into simple voice first advice to 140"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:28",
                    "end": "00:03:35",
                    "text": "million farmers to help improve their"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:31",
                    "end": "00:03:39",
                    "text": "income in everyday life. Gio Bharat IQ a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:35",
                    "end": "00:03:42",
                    "text": "voice first AI companion helping Indians"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:39",
                    "end": "00:03:46",
                    "text": "learn earn and access government"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:42",
                    "end": "00:03:49",
                    "text": "services at bhat scale from variables to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:46",
                    "end": "00:03:51",
                    "text": "fully connected homes"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:49",
                    "end": "00:03:54",
                    "text": "geo frames"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:51",
                    "end": "00:03:58",
                    "text": "an AI glass device and next generation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:54",
                    "end": "00:04:02",
                    "text": "AI devices will make intelligence truly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:58",
                    "end": "00:04:05",
                    "text": "ambient as effortless and as natural as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:02",
                    "end": "00:04:08",
                    "text": "human convers conversations through Geio"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:05",
                    "end": "00:04:11",
                    "text": "Hot Star. AI will multiply Indian"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:08",
                    "end": "00:04:14",
                    "text": "creativity with multilingual"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:11",
                    "end": "00:04:17",
                    "text": "storytelling. We will popularize India's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:14",
                    "end": "00:04:20",
                    "text": "rich cultural heritage with futuristic"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:17",
                    "end": "00:04:24",
                    "text": "technology, enhancing India's soft power"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:20",
                    "end": "00:04:28",
                    "text": "globally. Friends, this inaugural global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:24",
                    "end": "00:04:32",
                    "text": "AI impact summit in India has received a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:28",
                    "end": "00:04:35",
                    "text": "massive response. What does that show?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:32",
                    "end": "00:04:38",
                    "text": "It shows that AI is now becoming a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:35",
                    "end": "00:04:41",
                    "text": "people's movement worldwide. The success"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:38",
                    "end": "00:04:45",
                    "text": "of this movement hinges critically on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:41",
                    "end": "00:04:48",
                    "text": "global cooperation and not polarization."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:45",
                    "end": "00:04:51",
                    "text": "Be it chips or rare earths, AI works its"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:48",
                    "end": "00:04:53",
                    "text": "magic through sharing not holding"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:51",
                    "end": "00:04:56",
                    "text": "through collaborations not conflicts."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:53",
                    "end": "00:04:58",
                    "text": "The unique strength of India is that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:56",
                    "end": "00:05:01",
                    "text": "India serves as the vital bridge"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:58",
                    "end": "00:05:05",
                    "text": "connecting the global south and the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:01",
                    "end": "00:05:08",
                    "text": "global north. After all, south or north,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:05",
                    "end": "00:05:12",
                    "text": "east or west, all of us have only one"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:08",
                    "end": "00:05:15",
                    "text": "earth, one family and one future. Today"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:12",
                    "end": "00:05:18",
                    "text": "at this summit, let us all pledge to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:15",
                    "end": "00:05:21",
                    "text": "transform this noble aspiration into"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:18",
                    "end": "00:05:25",
                    "text": "reality using the most powerful gift of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:21",
                    "end": "00:05:27",
                    "text": "the human mind, AI. Let us combine"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:25",
                    "end": "00:05:30",
                    "text": "intelligence with empathy and let us"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:27",
                    "end": "00:05:34",
                    "text": "build a better future for all. Thank"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:30",
                    "end": "00:05:34",
                    "text": "you, Jin."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:37",
                    "end": "00:05:42",
                    "text": ">> Thank you so much."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:40",
                    "end": "00:05:44",
                    "text": "By 2047,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:42",
                    "end": "00:05:47",
                    "text": "the glorious centenary of our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:44",
                    "end": "00:05:48",
                    "text": "independence."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:47",
                    "end": "00:05:51",
                    "text": "We are deeply honored that our most"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:48",
                    "end": "00:05:54",
                    "text": "respected Prime Minister Sri Narendra"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:51",
                    "end": "00:05:58",
                    "text": "Modi is the guide, philosopher and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:54",
                    "end": "00:06:00",
                    "text": "leader of this summit."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:58",
                    "end": "00:06:04",
                    "text": "Honorable Minister Shivashishna and his"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:00",
                    "end": "00:06:07",
                    "text": "wonderful team deserve full praise for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:04",
                    "end": "00:06:12",
                    "text": "organizing this summit on a grand scale"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:07",
                    "end": "00:06:12",
                    "text": "befitting India's ambition"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:18",
                    "end": "00:06:26",
                    "text": "Holi G's vision of AI powered Vix Bharat"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:22",
                    "end": "00:06:27",
                    "text": "is also a template for a Vix global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:26",
                    "end": "00:06:31",
                    "text": "mouth."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:27",
                    "end": "00:06:35",
                    "text": "If wisely used, I believe AI can usher"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:31",
                    "end": "00:06:39",
                    "text": "in an era of super abundance."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:35",
                    "end": "00:06:42",
                    "text": "A world without poverty and a future of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:39",
                    "end": "00:06:46",
                    "text": "prosperity for all the 8 billion people"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:42",
                    "end": "00:06:50",
                    "text": "on our beautiful planet is now within"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:46",
                    "end": "00:06:52",
                    "text": "sight, within reach. Friends,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:50",
                    "end": "00:06:55",
                    "text": "artificial intelligence is not just"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:52",
                    "end": "00:06:58",
                    "text": "another technology."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:55",
                    "end": "00:07:01",
                    "text": "For the first time, humans are creating"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:58",
                    "end": "00:07:05",
                    "text": "human-like systems that can learn,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:01",
                    "end": "00:07:07",
                    "text": "speak, analyze, move, and produce"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:05",
                    "end": "00:07:12",
                    "text": "autonomously."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:07",
                    "end": "00:07:15",
                    "text": "AI is the mantra that powers every"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:12",
                    "end": "00:07:20",
                    "text": "yantra or every machine and system to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:15",
                    "end": "00:07:25",
                    "text": "work faster, better, and smarter."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:20",
                    "end": "00:07:27",
                    "text": "I see AI as a modernday akshaiatra"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:25",
                    "end": "00:07:31",
                    "text": "the legendary vessel in Madharat that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:27",
                    "end": "00:07:35",
                    "text": "provided endless nourishment to all."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:31",
                    "end": "00:07:38",
                    "text": "Likewise AI offers limitless"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:35",
                    "end": "00:07:40",
                    "text": "augmentation in knowledge, efficiency"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:38",
                    "end": "00:07:44",
                    "text": "and productivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:40",
                    "end": "00:07:44",
                    "text": "We are only"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:44",
                    "end": "00:07:52",
                    "text": "at the dawn of this era. The best of AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:49",
                    "end": "00:07:53",
                    "text": "is yet to come."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:52",
                    "end": "00:07:57",
                    "text": "Distinguished participants in this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:53",
                    "end": "00:07:60",
                    "text": "summit, today the world is def is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:57",
                    "end": "00:08:02",
                    "text": "debating a profound question."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:60",
                    "end": "00:08:05",
                    "text": "Will AI concentrate power in the hands"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:02",
                    "end": "00:08:09",
                    "text": "of a few or will it democratize"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:05",
                    "end": "00:08:14",
                    "text": "opportunity for all? Do we act as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:09",
                    "end": "00:08:18",
                    "text": "isolated nations or as a united global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:14",
                    "end": "00:08:22",
                    "text": "family? Our polarized world stands at a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:18",
                    "end": "00:08:25",
                    "text": "pork in one path has led to a situation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:22",
                    "end": "00:08:28",
                    "text": "where AI is scarce and expensive,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:25",
                    "end": "00:08:31",
                    "text": "compute is concentrated, data is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:28",
                    "end": "00:08:35",
                    "text": "controlled and capability is locked"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:31",
                    "end": "00:08:38",
                    "text": "behind barriers of capital and geography"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:35",
                    "end": "00:08:41",
                    "text": "in the global north. In this dismal"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:38",
                    "end": "00:08:44",
                    "text": "scenario, inequality widens between"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:41",
                    "end": "00:08:47",
                    "text": "nations within societies and across"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:44",
                    "end": "00:08:51",
                    "text": "generations. But there is another path."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:47",
                    "end": "00:08:54",
                    "text": "A future where AI is available,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:51",
                    "end": "00:08:57",
                    "text": "affordable and beneficial to all. As our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:54",
                    "end": "00:09:01",
                    "text": "prime minister said, India believes in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:57",
                    "end": "00:09:04",
                    "text": "this second future. Dear friends,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:01",
                    "end": "00:09:07",
                    "text": "from the podium of this summit today, I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:04",
                    "end": "00:09:11",
                    "text": "want to make a bold prediction. India"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:07",
                    "end": "00:09:13",
                    "text": "will emerge as one of the greatest AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:11",
                    "end": "00:09:16",
                    "text": "powers in in the world in the 21st"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:13",
                    "end": "00:09:16",
                    "text": "century."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:19",
                    "end": "00:09:25",
                    "text": "My confidence is validated by an"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:22",
                    "end": "00:09:28",
                    "text": "undeniable truth. In the coming decades,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:25",
                    "end": "00:09:31",
                    "text": "no country in the world can match"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:28",
                    "end": "00:09:34",
                    "text": "India's strength in demography,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:31",
                    "end": "00:09:38",
                    "text": "democracy, development, digital"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:34",
                    "end": "00:09:41",
                    "text": "infrastructure, data generation, AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:38",
                    "end": "00:09:43",
                    "text": "harvest. Let me begin as was said many"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:41",
                    "end": "00:09:47",
                    "text": "times this morning, what India has"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:43",
                    "end": "00:09:50",
                    "text": "achieved in the past 10 years. First,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:47",
                    "end": "00:09:53",
                    "text": "India is the world's largest mobile data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:50",
                    "end": "00:09:56",
                    "text": "consumer. Nearly 1 billion internet"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:53",
                    "end": "00:09:59",
                    "text": "users data costs among the lowest"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:56",
                    "end": "00:10:01",
                    "text": "globally and in terms of quality there"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:59",
                    "end": "00:10:06",
                    "text": "is no difference between Delhi and the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:01",
                    "end": "00:10:10",
                    "text": "remotest Indian village. Second Aadhaar"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:06",
                    "end": "00:10:13",
                    "text": "1.4 billion digital ids third UPI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:10",
                    "end": "00:10:16",
                    "text": "processes over 12 billion transactions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:13",
                    "end": "00:10:19",
                    "text": "monthly. Fourth, India ranks among the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:16",
                    "end": "00:10:23",
                    "text": "top three startup ecosystems with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:19",
                    "end": "00:10:26",
                    "text": "100,000 startups and 100 plus unicorns."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:23",
                    "end": "00:10:29",
                    "text": "Fifth, India's secure and inclusive"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:26",
                    "end": "00:10:32",
                    "text": "digital public infrastructure stack is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:29",
                    "end": "00:10:36",
                    "text": "now being adopted by countries around"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:32",
                    "end": "00:10:39",
                    "text": "the globe. Friends, in all humility, I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:36",
                    "end": "00:10:43",
                    "text": "wish to state that Jio with over 500"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:39",
                    "end": "00:10:45",
                    "text": "million loyal subscribers was privileged"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:43",
                    "end": "00:10:49",
                    "text": "to play a leading role in this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:45",
                    "end": "00:10:51",
                    "text": "transformation across broadband, 4G, 5G,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:49",
                    "end": "00:10:53",
                    "text": "and home connectivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:51",
                    "end": "00:10:57",
                    "text": "With equal humility, I would like to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:53",
                    "end": "00:10:59",
                    "text": "announce that Jio will play an even"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:57",
                    "end": "00:11:02",
                    "text": "bigger role in India's AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:59",
                    "end": "00:11:05",
                    "text": "transformation. Today on behalf of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:02",
                    "end": "00:11:07",
                    "text": "Reliance Group and Geo Intelligence I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:05",
                    "end": "00:11:11",
                    "text": "want to make three announcements."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:07",
                    "end": "00:11:14",
                    "text": "Announcement one, Jio connected India to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:11",
                    "end": "00:11:19",
                    "text": "the internet era. Jio will now connect"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:14",
                    "end": "00:11:19",
                    "text": "India to the intelligence era."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:20",
                    "end": "00:11:26",
                    "text": "We will deliver intelligence to every"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:23",
                    "end": "00:11:29",
                    "text": "citizen, every sector of the economy and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:26",
                    "end": "00:11:32",
                    "text": "every facet of social development and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:29",
                    "end": "00:11:36",
                    "text": "every service of government. Jio will do"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:32",
                    "end": "00:11:39",
                    "text": "so with the same reliability, quality,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:36",
                    "end": "00:11:41",
                    "text": "scale and extreme affordability that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:39",
                    "end": "00:11:43",
                    "text": "transformed connectivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:41",
                    "end": "00:11:46",
                    "text": "India cannot afford to rent"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:43",
                    "end": "00:11:50",
                    "text": "intelligence. Therefore, we will reduce"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:46",
                    "end": "00:11:55",
                    "text": "the cost of intelligence as dramatically"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:50",
                    "end": "00:11:55",
                    "text": "as we did the cost of data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:57",
                    "end": "00:12:06",
                    "text": "Announcement two, JIO together with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:01",
                    "end": "00:12:11",
                    "text": "Reliance will invest 10 lakh crores over"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:06",
                    "end": "00:12:11",
                    "text": "the next 7 years starting this year."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:13",
                    "end": "00:12:18",
                    "text": "This is not speculative investment. It"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:16",
                    "end": "00:12:22",
                    "text": "is not for chasing valuation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:18",
                    "end": "00:12:25",
                    "text": "This is patient, disciplined nation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:22",
                    "end": "00:12:28",
                    "text": "building capital designed to create"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:25",
                    "end": "00:12:32",
                    "text": "durable economic value and strategic"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:28",
                    "end": "00:12:34",
                    "text": "resilience for decades to come."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:32",
                    "end": "00:12:38",
                    "text": "Distinguished participants, the biggest"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:34",
                    "end": "00:12:41",
                    "text": "constraint in AI today is not talent or"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:38",
                    "end": "00:12:45",
                    "text": "imagination. It is scarcity and high"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:41",
                    "end": "00:12:48",
                    "text": "cost of compute. Therefore, here is my"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:45",
                    "end": "00:12:51",
                    "text": "third announcement. Geo intelligence"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:48",
                    "end": "00:12:54",
                    "text": "will build India's sovereign compute"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:51",
                    "end": "00:12:58",
                    "text": "infrastructure through three bold"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:54",
                    "end": "00:13:02",
                    "text": "initiatives. One, gigawatt scale data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:58",
                    "end": "00:13:05",
                    "text": "centers. We already started construction"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:02",
                    "end": "00:13:08",
                    "text": "on multi- gigawatt AI ready data centers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:05",
                    "end": "00:13:13",
                    "text": "at Jamnagar. Over 120 megawatts will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:08",
                    "end": "00:13:16",
                    "text": "come online in the second half of 2026"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:13",
                    "end": "00:13:18",
                    "text": "this year and a clear path to gigawatt"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:16",
                    "end": "00:13:21",
                    "text": "scale compute for training and large"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:18",
                    "end": "00:13:22",
                    "text": "scale inference."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:21",
                    "end": "00:13:26",
                    "text": "Two,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:22",
                    "end": "00:13:28",
                    "text": "our green energy advantage. We have an"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:26",
                    "end": "00:13:33",
                    "text": "in-house energy advantage with up to 10"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:28",
                    "end": "00:13:35",
                    "text": "gawatt of ready green power surplus"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:33",
                    "end": "00:13:37",
                    "text": "anchored by solar in both Kach and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:35",
                    "end": "00:13:41",
                    "text": "Andhra Pradesh."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:37",
                    "end": "00:13:44",
                    "text": "Three, a nationwide edge compute. An"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:41",
                    "end": "00:13:46",
                    "text": "edge compute layer deeply integrated"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:44",
                    "end": "00:13:50",
                    "text": "with Jio's network will make"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:46",
                    "end": "00:13:53",
                    "text": "intelligence responsive, low latency and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:50",
                    "end": "00:13:57",
                    "text": "affordable close to where Indians live,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:53",
                    "end": "00:13:59",
                    "text": "learn and work. From kirana stores to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:57",
                    "end": "00:14:03",
                    "text": "clinics, from classrooms to farms,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:59",
                    "end": "00:14:07",
                    "text": "intelligence will live at the edge. Our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:03",
                    "end": "00:14:10",
                    "text": "resolve is clear. Make intelligence as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:07",
                    "end": "00:14:13",
                    "text": "ubiquitous as connectivity. When compute"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:10",
                    "end": "00:14:16",
                    "text": "becomes infrastructure,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:13",
                    "end": "00:14:19",
                    "text": "innovation will become inevitable."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:16",
                    "end": "00:14:23",
                    "text": "Friends, geo intelligence is guided by"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:19",
                    "end": "00:14:27",
                    "text": "five non-negotiable principles. First,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:23",
                    "end": "00:14:29",
                    "text": "AI for India's deep tech and advanced"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:27",
                    "end": "00:14:32",
                    "text": "manufacturing leadership. Reaching not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:29",
                    "end": "00:14:34",
                    "text": "just large enterprises but agriculture,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:32",
                    "end": "00:14:37",
                    "text": "small businesses and the informal"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:34",
                    "end": "00:14:40",
                    "text": "sector. Geo intelligence will not simply"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:37",
                    "end": "00:14:43",
                    "text": "be a search or an ask tool. It will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:40",
                    "end": "00:14:47",
                    "text": "primarily be a resource for multiplying"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:43",
                    "end": "00:14:50",
                    "text": "productivity and efficiency. Second,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:47",
                    "end": "00:14:53",
                    "text": "worldleading multilingual AI capability"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:50",
                    "end": "00:14:56",
                    "text": "across all Indian languages. When"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:53",
                    "end": "00:14:58",
                    "text": "farmers and artisians speak to AI in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:56",
                    "end": "00:15:01",
                    "text": "their own words and student learn in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:58",
                    "end": "00:15:04",
                    "text": "their own mother tongue, this is not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:01",
                    "end": "00:15:07",
                    "text": "convenience. This is inclusion."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:04",
                    "end": "00:15:07",
                    "text": "Go ahead."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:18",
                    "end": "00:15:23",
                    "text": "Third,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:20",
                    "end": "00:15:27",
                    "text": "responsibility, security, data residency"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:23",
                    "end": "00:15:31",
                    "text": "and trust as JO's core guarantees, not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:27",
                    "end": "00:15:34",
                    "text": "afterthought. Fourth, we will provide we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:31",
                    "end": "00:15:38",
                    "text": "will prove that AI does not take away"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:34",
                    "end": "00:15:41",
                    "text": "jobs. Rather, it will create new"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:38",
                    "end": "00:15:46",
                    "text": "highskilled work opportunities."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:41",
                    "end": "00:15:49",
                    "text": "And fifth, the AI story has shifted from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:46",
                    "end": "00:15:53",
                    "text": "who has the best model to who can build"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:49",
                    "end": "00:15:56",
                    "text": "the strongest ecosystem for speed and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:53",
                    "end": "00:15:60",
                    "text": "scale of usage. Therefore,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:56",
                    "end": "00:16:04",
                    "text": "we will build deep partnership ecosystem"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:60",
                    "end": "00:16:07",
                    "text": "with Indian enterprises, startups, IIT,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:04",
                    "end": "00:16:09",
                    "text": "IISC and research institutions. We will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:07",
                    "end": "00:16:12",
                    "text": "work shoulderto-shoulder with India's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:09",
                    "end": "00:16:15",
                    "text": "leading industrial groups to embed AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:12",
                    "end": "00:16:18",
                    "text": "across manufacturing, logistics, energy,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:15",
                    "end": "00:16:20",
                    "text": "finance, retail, agriculture, and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:18",
                    "end": "00:16:22",
                    "text": "healthcare. We will empower startups"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:20",
                    "end": "00:16:25",
                    "text": "with affordable compute and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:22",
                    "end": "00:16:28",
                    "text": "co-development platforms. We will aspire"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:25",
                    "end": "00:16:31",
                    "text": "to produce global breakthroughs in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:28",
                    "end": "00:16:33",
                    "text": "compute architecture, foundation models"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:31",
                    "end": "00:16:36",
                    "text": "and energy efficiency"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:33",
                    "end": "00:16:38",
                    "text": "designed in India, rooted in our values,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:36",
                    "end": "00:16:41",
                    "text": "powered by talent and scaled for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:38",
                    "end": "00:16:43",
                    "text": "humanity. And we will partner with the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:41",
                    "end": "00:16:47",
                    "text": "very best tech companies in the world,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:43",
                    "end": "00:16:51",
                    "text": "not as importers of intelligence, but as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:47",
                    "end": "00:16:54",
                    "text": "co-architects of a new AI century."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:51",
                    "end": "00:16:57",
                    "text": "Dear friends, I believe that social"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:54",
                    "end": "00:17:02",
                    "text": "relevance, not momentary craze, should"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:57",
                    "end": "00:17:05",
                    "text": "drive AI growth in India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:02",
                    "end": "00:17:07",
                    "text": "Jio has already started AI applications"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:05",
                    "end": "00:17:11",
                    "text": "for the most pressing challenges in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:07",
                    "end": "00:17:14",
                    "text": "inclusive development. In education, we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:11",
                    "end": "00:17:17",
                    "text": "have Gio Shikshak, an adaptive AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:14",
                    "end": "00:17:20",
                    "text": "teaching assistant in 22 languages. When"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:17",
                    "end": "00:17:23",
                    "text": "250 million school children and 50"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:20",
                    "end": "00:17:26",
                    "text": "million students in higher education are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:23",
                    "end": "00:17:30",
                    "text": "empowered by AI teachers, no power on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:26",
                    "end": "00:17:34",
                    "text": "earth can match India's talent wealth."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:30",
                    "end": "00:17:37",
                    "text": "In healthcare, Gio arrogia AI delivering"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:34",
                    "end": "00:17:40",
                    "text": "first medical guidance in under five"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:37",
                    "end": "00:17:43",
                    "text": "minutes in local languages on any phone."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:40",
                    "end": "00:17:46",
                    "text": "In agriculture, Georishi converting"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:43",
                    "end": "00:17:49",
                    "text": "satellite imagery and precision weather"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:46",
                    "end": "00:17:52",
                    "text": "into simple voice first advice to 140"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:49",
                    "end": "00:17:56",
                    "text": "million farmers to help improve their"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:52",
                    "end": "00:18:00",
                    "text": "income in everyday life. Gio Bharat IQ a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:56",
                    "end": "00:18:03",
                    "text": "voice first AI companion helping Indians"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:00",
                    "end": "00:18:07",
                    "text": "learn earn and access government"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:03",
                    "end": "00:18:10",
                    "text": "services at bhat scale from variables to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:07",
                    "end": "00:18:12",
                    "text": "fully connected homes"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:10",
                    "end": "00:18:16",
                    "text": "geo frames"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:12",
                    "end": "00:18:20",
                    "text": "an AI glass device and next generation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:16",
                    "end": "00:18:24",
                    "text": "AI devices will make intelligence truly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:20",
                    "end": "00:18:28",
                    "text": "ambient as effortless and as natural as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:24",
                    "end": "00:18:31",
                    "text": "human conversations through geo hot star"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:28",
                    "end": "00:18:34",
                    "text": "AI will multiply Indian creativity with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:31",
                    "end": "00:18:36",
                    "text": "multilingual storytelling we will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:34",
                    "end": "00:18:39",
                    "text": "popularize India's rich cultural"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:36",
                    "end": "00:18:42",
                    "text": "heritage with futuristic technology"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:39",
                    "end": "00:18:46",
                    "text": "enhancing India's soft power globally"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:42",
                    "end": "00:18:49",
                    "text": "friends this inaugural global AI impact"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:46",
                    "end": "00:18:54",
                    "text": "summit in India has received a massive"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:49",
                    "end": "00:18:56",
                    "text": "response what does that show it shows"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:54",
                    "end": "00:18:59",
                    "text": "that AI is now becoming a people's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:56",
                    "end": "00:19:02",
                    "text": "movement worldwide. The success of this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:59",
                    "end": "00:19:06",
                    "text": "movement hinges critically on global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:02",
                    "end": "00:19:09",
                    "text": "cooperation and not polarization."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:06",
                    "end": "00:19:12",
                    "text": "Be it chips or rare earths, AI works its"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:09",
                    "end": "00:19:14",
                    "text": "magic through sharing not holding"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:12",
                    "end": "00:19:17",
                    "text": "through collaborations not conflicts."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:14",
                    "end": "00:19:19",
                    "text": "The unique strength of India is that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:17",
                    "end": "00:19:22",
                    "text": "India serves as the vital bridge"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:19",
                    "end": "00:19:25",
                    "text": "connecting the global south and the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:22",
                    "end": "00:19:29",
                    "text": "global north. After all, south or north,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:25",
                    "end": "00:19:32",
                    "text": "east or west, all of us have only one"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:29",
                    "end": "00:19:36",
                    "text": "earth, one family and one future. Today"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:32",
                    "end": "00:19:39",
                    "text": "at this summit, let us all pledge to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:36",
                    "end": "00:19:42",
                    "text": "transform this noble aspiration into"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:39",
                    "end": "00:19:46",
                    "text": "reality using the most powerful gift of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:42",
                    "end": "00:19:48",
                    "text": "the human mind. AI. Let us combine"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:46",
                    "end": "00:19:51",
                    "text": "intelligence with empathy and let us"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:48",
                    "end": "00:19:54",
                    "text": "build a better future for all. Thank"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:51",
                    "end": "00:19:56",
                    "text": "you, Jim."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:54",
                    "end": "00:19:59",
                    "text": "The glorious centenary of our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:56",
                    "end": "00:20:01",
                    "text": "independence."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:59",
                    "end": "00:20:03",
                    "text": "We are deeply honored that our most"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:01",
                    "end": "00:20:07",
                    "text": "respected prime minister Sri Narendra"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:03",
                    "end": "00:20:10",
                    "text": "Modi is the guide, philosopher and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:07",
                    "end": "00:20:13",
                    "text": "leader of this summit."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:10",
                    "end": "00:20:16",
                    "text": "Honorable Minister Shivishna and his"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:13",
                    "end": "00:20:20",
                    "text": "wonderful team deserve full praise for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:16",
                    "end": "00:20:25",
                    "text": "organizing this summit on a grand scale"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:20",
                    "end": "00:20:25",
                    "text": "befitting India's ambition."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:26",
                    "end": "00:20:34",
                    "text": "Oi G's vision of AI powered Vix Bharat"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:31",
                    "end": "00:20:36",
                    "text": "is also a template for a Vix global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:34",
                    "end": "00:20:40",
                    "text": "south."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:36",
                    "end": "00:20:43",
                    "text": "If wisely used, I believe AI can usher"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:40",
                    "end": "00:20:47",
                    "text": "in an era of super abundance."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:43",
                    "end": "00:20:50",
                    "text": "A world without poverty and a future of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:47",
                    "end": "00:20:54",
                    "text": "prosperity for all the 8 billion people"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:50",
                    "end": "00:20:58",
                    "text": "on our beautiful planet is now within"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:54",
                    "end": "00:21:00",
                    "text": "sight, within reach. Friends,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:58",
                    "end": "00:21:03",
                    "text": "artificial intelligence is not just"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:00",
                    "end": "00:21:06",
                    "text": "another technology."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:03",
                    "end": "00:21:09",
                    "text": "For the first time, humans are creating"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:06",
                    "end": "00:21:13",
                    "text": "human-like systems that can learn,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:09",
                    "end": "00:21:15",
                    "text": "speak, analyze, move, and produce"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:13",
                    "end": "00:21:20",
                    "text": "autonomously."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:15",
                    "end": "00:21:24",
                    "text": "AI is the mantra that powers every"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:20",
                    "end": "00:21:28",
                    "text": "yantra or every machine and system to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:24",
                    "end": "00:21:33",
                    "text": "work faster, better, and smarter."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:28",
                    "end": "00:21:36",
                    "text": "I see AI as a modernday akshai patra the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:33",
                    "end": "00:21:40",
                    "text": "legendary vessel in Madharat that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:36",
                    "end": "00:21:44",
                    "text": "provided endless nourishment to all."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:40",
                    "end": "00:21:46",
                    "text": "Likewise AI offers limitless"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:44",
                    "end": "00:21:48",
                    "text": "augmentation in knowledge, efficiency"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:46",
                    "end": "00:21:53",
                    "text": "and productivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:48",
                    "end": "00:21:57",
                    "text": "We are only at the dawn of this era. The"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:53",
                    "end": "00:21:59",
                    "text": "best of AI is yet to come."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:57",
                    "end": "00:22:03",
                    "text": "Distinguished participants in this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:59",
                    "end": "00:22:06",
                    "text": "summit today the world is defound is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:03",
                    "end": "00:22:09",
                    "text": "debating a profound question."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:06",
                    "end": "00:22:12",
                    "text": "Will AI concentrate"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:09",
                    "end": "00:22:16",
                    "text": "power in the hands of a few or will it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:12",
                    "end": "00:22:21",
                    "text": "democratize opportunity for all? Do we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:16",
                    "end": "00:22:24",
                    "text": "act as isolated nations or as a united"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:21",
                    "end": "00:22:27",
                    "text": "global family? Our polarized world"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:24",
                    "end": "00:22:31",
                    "text": "stands at a pork."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:27",
                    "end": "00:22:34",
                    "text": "One path has led to a situation where AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:31",
                    "end": "00:22:36",
                    "text": "is scarce and expensive, compute is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:34",
                    "end": "00:22:40",
                    "text": "concentrated, data is controlled and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:36",
                    "end": "00:22:44",
                    "text": "capability is locked behind barriers of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:40",
                    "end": "00:22:47",
                    "text": "capital and geography in the global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:44",
                    "end": "00:22:50",
                    "text": "north. In this dismal scenario,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:47",
                    "end": "00:22:53",
                    "text": "inequality widens between nations within"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:50",
                    "end": "00:22:58",
                    "text": "societies and across generations. But"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:53",
                    "end": "00:23:00",
                    "text": "there is another path, a future where AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:58",
                    "end": "00:23:04",
                    "text": "is available, affordable and beneficial"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:00",
                    "end": "00:23:07",
                    "text": "to all. As our prime minister said,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:04",
                    "end": "00:23:10",
                    "text": "India believes in this second future."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:07",
                    "end": "00:23:13",
                    "text": "Dear friends, from the podium of this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:10",
                    "end": "00:23:19",
                    "text": "summit today, I want to make a bold"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:13",
                    "end": "00:23:22",
                    "text": "prediction. India will emerge as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:19",
                    "end": "00:23:26",
                    "text": "one of the greatest AI powers in the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:22",
                    "end": "00:23:26",
                    "text": "world in the 21st century."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:29",
                    "end": "00:23:35",
                    "text": "My confidence is validated by an"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:32",
                    "end": "00:23:37",
                    "text": "undeniable truth. In the coming decades,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:35",
                    "end": "00:23:41",
                    "text": "no country in the world can match"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:37",
                    "end": "00:23:43",
                    "text": "India's strength in demography,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:41",
                    "end": "00:23:47",
                    "text": "democracy, development, digital"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:43",
                    "end": "00:23:50",
                    "text": "infrastructure, data generation, AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:47",
                    "end": "00:23:52",
                    "text": "harvest. Let me begin as was said many"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:50",
                    "end": "00:23:55",
                    "text": "times this morning, what India has"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:52",
                    "end": "00:23:59",
                    "text": "achieved in the past 10 years. First,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:55",
                    "end": "00:24:01",
                    "text": "India is the world's largest mobile data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:59",
                    "end": "00:24:04",
                    "text": "consumer. Nearly 1 billion internet"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:01",
                    "end": "00:24:07",
                    "text": "users data costs among the lowest"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:04",
                    "end": "00:24:10",
                    "text": "globally and in terms of quality there"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:07",
                    "end": "00:24:14",
                    "text": "is no difference between Delhi and the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:10",
                    "end": "00:24:18",
                    "text": "remotest Indian village. Second Aadhaar"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:14",
                    "end": "00:24:21",
                    "text": "1.4 billion digital IDs. Third UPI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:18",
                    "end": "00:24:24",
                    "text": "processes over 12 billion transactions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:21",
                    "end": "00:24:27",
                    "text": "monthly. Fourth, India ranks among the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:24",
                    "end": "00:24:31",
                    "text": "top three startup ecosystems with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:27",
                    "end": "00:24:34",
                    "text": "100,000 startups and 100 plus unicorns."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:31",
                    "end": "00:24:37",
                    "text": "Fifth, India's secure and inclusive"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:34",
                    "end": "00:24:40",
                    "text": "digital public infrastructure stack is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:37",
                    "end": "00:24:44",
                    "text": "now being adopted by countries around"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:40",
                    "end": "00:24:47",
                    "text": "the globe. Friends, in all humility, I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:44",
                    "end": "00:24:50",
                    "text": "wish to state that Jio with over 500"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:47",
                    "end": "00:24:53",
                    "text": "million loyal subscribers was privileged"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:50",
                    "end": "00:24:56",
                    "text": "to play a leading role in this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:53",
                    "end": "00:24:59",
                    "text": "transformation across broadband, 4G, 5G,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:56",
                    "end": "00:25:01",
                    "text": "and home connectivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:59",
                    "end": "00:25:05",
                    "text": "With equal humility, I would like to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:01",
                    "end": "00:25:07",
                    "text": "announce that Jio will play an even"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:05",
                    "end": "00:25:10",
                    "text": "bigger role in India's AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:07",
                    "end": "00:25:13",
                    "text": "transformation. Today on behalf of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:10",
                    "end": "00:25:15",
                    "text": "Reliance Group and Geo Intelligence I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:13",
                    "end": "00:25:19",
                    "text": "want to make three announcements."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:15",
                    "end": "00:25:22",
                    "text": "Announcement one, Jio connected India to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:19",
                    "end": "00:25:27",
                    "text": "the internet era. Jio will now connect"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:22",
                    "end": "00:25:27",
                    "text": "India to the intelligence era."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:28",
                    "end": "00:25:34",
                    "text": "We will deliver intelligence to every"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:31",
                    "end": "00:25:37",
                    "text": "citizen, every sector of the economy and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:34",
                    "end": "00:25:40",
                    "text": "every facet of social development and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:37",
                    "end": "00:25:44",
                    "text": "every service of government. Jio will do"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:40",
                    "end": "00:25:47",
                    "text": "so with the same reli"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:44",
                    "end": "00:25:50",
                    "text": "distinguished guests, my fellow Indians,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:47",
                    "end": "00:25:53",
                    "text": "namaste."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:50",
                    "end": "00:25:58",
                    "text": "The global AI impact summit is a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:53",
                    "end": "00:26:03",
                    "text": "defining moment in India's tech history."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:58",
                    "end": "00:26:07",
                    "text": "A moment when India pledges to make AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:03",
                    "end": "00:26:11",
                    "text": "one of the driving forces to realize its"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:07",
                    "end": "00:26:15",
                    "text": "dream of a Vixs Bharat, the dream of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:11",
                    "end": "00:26:16",
                    "text": "becoming a fully developed nation by"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:15",
                    "end": "00:26:18",
                    "text": "2047,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:16",
                    "end": "00:26:21",
                    "text": "the glorious centenary of our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:18",
                    "end": "00:26:23",
                    "text": "independence."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:21",
                    "end": "00:26:26",
                    "text": "We are deeply honored that our most"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:23",
                    "end": "00:26:29",
                    "text": "respected prime minister Shri Narendra"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:26",
                    "end": "00:26:32",
                    "text": "Modi is the guide, philosopher and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:29",
                    "end": "00:26:35",
                    "text": "leader of this summit."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:32",
                    "end": "00:26:38",
                    "text": "Honorable Minister Shriishna and his"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:35",
                    "end": "00:26:42",
                    "text": "wonderful team deserve full praise for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:38",
                    "end": "00:26:46",
                    "text": "organizing this summit on a grand scale"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:42",
                    "end": "00:26:46",
                    "text": "befitting India's ambition."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:48",
                    "end": "00:26:56",
                    "text": "Muli's vision of AI powered Vix Bharat"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:53",
                    "end": "00:26:58",
                    "text": "is also a template for a Vix global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:56",
                    "end": "00:27:02",
                    "text": "south."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:58",
                    "end": "00:27:06",
                    "text": "If wisely used, I believe AI can usher"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:02",
                    "end": "00:27:09",
                    "text": "in an era of super abundance."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:06",
                    "end": "00:27:13",
                    "text": "A world without poverty and a future of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:09",
                    "end": "00:27:16",
                    "text": "prosperity for all the 8 billion people"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:13",
                    "end": "00:27:20",
                    "text": "on our beautiful planet is now within"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:16",
                    "end": "00:27:23",
                    "text": "sight within reach. Friends,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:20",
                    "end": "00:27:25",
                    "text": "artificial intelligence is not just"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:23",
                    "end": "00:27:29",
                    "text": "another technology."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:25",
                    "end": "00:27:32",
                    "text": "For the first time, humans are creating"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:29",
                    "end": "00:27:36",
                    "text": "human-like systems that can learn,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:32",
                    "end": "00:27:38",
                    "text": "speak, analyze, move, and produce"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:36",
                    "end": "00:27:42",
                    "text": "autonomously."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:38",
                    "end": "00:27:46",
                    "text": "AI is the mantra that powers every"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:42",
                    "end": "00:27:50",
                    "text": "yantra or every machine and system to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:46",
                    "end": "00:27:55",
                    "text": "work faster, better, and smarter."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:50",
                    "end": "00:27:58",
                    "text": "I see AI as a modernday akshaiatra"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:55",
                    "end": "00:28:02",
                    "text": "the legendary vessel in Madharat that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:58",
                    "end": "00:28:06",
                    "text": "provided endless nourishment to all."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:02",
                    "end": "00:28:08",
                    "text": "Likewise AI offers limitless"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:06",
                    "end": "00:28:11",
                    "text": "augmentation in knowledge, efficiency"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:08",
                    "end": "00:28:15",
                    "text": "and productivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:11",
                    "end": "00:28:20",
                    "text": "We are only at the dawn of this era. The"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:15",
                    "end": "00:28:23",
                    "text": "best of AI is yet to come. Distinguished"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:20",
                    "end": "00:28:26",
                    "text": "participants in this summit today the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:23",
                    "end": "00:28:28",
                    "text": "world is defound is debating a profound"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:26",
                    "end": "00:28:31",
                    "text": "question."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:28",
                    "end": "00:28:34",
                    "text": "Will AI concentrate power in the hands"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:31",
                    "end": "00:28:38",
                    "text": "of a few or will it democratize"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:34",
                    "end": "00:28:42",
                    "text": "opportunity for all? Do we act as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:38",
                    "end": "00:28:46",
                    "text": "isolated nations or as a united global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:42",
                    "end": "00:28:49",
                    "text": "family? Our polarized world stands at a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:46",
                    "end": "00:28:52",
                    "text": "pork."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:49",
                    "end": "00:28:55",
                    "text": "One path has led to a situation where AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:52",
                    "end": "00:28:57",
                    "text": "is scarce and expensive,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:55",
                    "end": "00:29:00",
                    "text": "compute is concentrated, data is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:57",
                    "end": "00:29:04",
                    "text": "controlled, and capability is locked"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:00",
                    "end": "00:29:07",
                    "text": "behind barriers of capital and geography"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:04",
                    "end": "00:29:11",
                    "text": "in the global north. In this dismal"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:07",
                    "end": "00:29:13",
                    "text": "scenario, inequality widens between"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:11",
                    "end": "00:29:17",
                    "text": "nations within societies and across"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:13",
                    "end": "00:29:21",
                    "text": "generations. But there is another path."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:17",
                    "end": "00:29:24",
                    "text": "A future where AI is available,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:21",
                    "end": "00:29:27",
                    "text": "affordable and beneficial to all. As our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:24",
                    "end": "00:29:30",
                    "text": "prime minister said, India believes in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:27",
                    "end": "00:29:31",
                    "text": "this second future."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:30",
                    "end": "00:29:34",
                    "text": "Dear friends,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:31",
                    "end": "00:29:37",
                    "text": "from the podium of this summit today, I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:34",
                    "end": "00:29:41",
                    "text": "want to make a bold prediction. India"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:37",
                    "end": "00:29:47",
                    "text": "will emerge as one of the greatest AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:41",
                    "end": "00:29:47",
                    "text": "powers in the world in the 21st century."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:49",
                    "end": "00:29:56",
                    "text": "My confidence is validated by an"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:53",
                    "end": "00:29:58",
                    "text": "undeniable truth. In the coming decades,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:56",
                    "end": "00:30:02",
                    "text": "no country in the world can match"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:58",
                    "end": "00:30:05",
                    "text": "India's strength in demography,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:02",
                    "end": "00:30:08",
                    "text": "democracy, development, digital"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:05",
                    "end": "00:30:12",
                    "text": "infrastructure, data generation, AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:08",
                    "end": "00:30:14",
                    "text": "harvest. Let me begin as was said many"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:12",
                    "end": "00:30:17",
                    "text": "times this morning, what India has"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:14",
                    "end": "00:30:20",
                    "text": "achieved in the past 10 years. First,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:17",
                    "end": "00:30:23",
                    "text": "India is the world's largest mobile data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:20",
                    "end": "00:30:26",
                    "text": "consumer. Nearly 1 billion internet"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:23",
                    "end": "00:30:29",
                    "text": "users data costs among the lowest"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:26",
                    "end": "00:30:32",
                    "text": "globally and in terms of quality there"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:29",
                    "end": "00:30:36",
                    "text": "is no difference between Delhi and the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:32",
                    "end": "00:30:41",
                    "text": "remotest Indian village. Second Aadhaar"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:36",
                    "end": "00:30:44",
                    "text": "1.4 billion digital IDs. Third UPI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:41",
                    "end": "00:30:47",
                    "text": "processes over 12 billion transactions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:44",
                    "end": "00:30:50",
                    "text": "monthly. Fourth, India ranks among the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:47",
                    "end": "00:30:53",
                    "text": "top three startup ecosystems with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:50",
                    "end": "00:30:57",
                    "text": "100,000 startups and 100 plus unicorns."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:53",
                    "end": "00:30:60",
                    "text": "Fifth, India's secure and inclusive"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:57",
                    "end": "00:31:02",
                    "text": "digital public infrastructure stack is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:60",
                    "end": "00:31:06",
                    "text": "now being adopted by countries around"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:02",
                    "end": "00:31:10",
                    "text": "the globe. Friends, in all humility, I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:06",
                    "end": "00:31:15",
                    "text": "wish to state that Jio with over 500"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:10",
                    "end": "00:31:18",
                    "text": "million loyal subscribers was"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:15",
                    "end": "00:31:20",
                    "text": "privileged to play a leading role in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:18",
                    "end": "00:31:24",
                    "text": "this transformation across broadband,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:20",
                    "end": "00:31:26",
                    "text": "4G, 5G and home connectivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:24",
                    "end": "00:31:29",
                    "text": "With equal humility, I would like to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:26",
                    "end": "00:31:32",
                    "text": "announce that Jio will play an even"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:29",
                    "end": "00:31:34",
                    "text": "bigger role in India's AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:32",
                    "end": "00:31:37",
                    "text": "transformation. Today, on behalf of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:34",
                    "end": "00:31:39",
                    "text": "Reliance Group and Geo Intelligence, I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:37",
                    "end": "00:31:43",
                    "text": "want to make three announcements."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:39",
                    "end": "00:31:46",
                    "text": "Announcement one, Jio connected India to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:43",
                    "end": "00:31:51",
                    "text": "the internet era. Jio will now connect"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:46",
                    "end": "00:31:51",
                    "text": "India to the intelligence era."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:52",
                    "end": "00:31:57",
                    "text": "We will deliver intelligence to every"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:54",
                    "end": "00:32:00",
                    "text": "citizen, every sector of the economy and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:57",
                    "end": "00:32:03",
                    "text": "every facet of social development and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:00",
                    "end": "00:32:06",
                    "text": "every service of government. Jio will do"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:03",
                    "end": "00:32:09",
                    "text": "so with the same reliability, quality,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:06",
                    "end": "00:32:12",
                    "text": "scale and extreme affordability that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:09",
                    "end": "00:32:14",
                    "text": "transformed connectivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:12",
                    "end": "00:32:17",
                    "text": "India cannot afford to rent"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:14",
                    "end": "00:32:21",
                    "text": "intelligence. Therefore, we will reduce"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:17",
                    "end": "00:32:26",
                    "text": "the cost of intelligence as dramatically"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:21",
                    "end": "00:32:26",
                    "text": "as we did the cost of data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:28",
                    "end": "00:32:36",
                    "text": "Announcement two, Jio together with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:32",
                    "end": "00:32:42",
                    "text": "Reliance will invest 10 lakh crores over"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:36",
                    "end": "00:32:42",
                    "text": "the next 7 years starting this year."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:43",
                    "end": "00:32:49",
                    "text": "This is not speculative investment. It"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:46",
                    "end": "00:32:52",
                    "text": "is not for chasing valuation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:49",
                    "end": "00:32:56",
                    "text": "This is patient, disciplined nation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:52",
                    "end": "00:32:59",
                    "text": "building capital designed to create"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:56",
                    "end": "00:33:02",
                    "text": "durable economic value and strategic"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:59",
                    "end": "00:33:05",
                    "text": "resilience for decades to come."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:02",
                    "end": "00:33:09",
                    "text": "Distinguished participants, the biggest"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:05",
                    "end": "00:33:12",
                    "text": "constraint in AI today is not talent or"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:09",
                    "end": "00:33:16",
                    "text": "imagination. It is scarcity and high"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:12",
                    "end": "00:33:19",
                    "text": "cost of compute. Therefore, here is my"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:16",
                    "end": "00:33:21",
                    "text": "third announcement. Geo intelligence"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:19",
                    "end": "00:33:24",
                    "text": "will build India's sovereign compute"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:21",
                    "end": "00:33:28",
                    "text": "infrastructure through three bold"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:24",
                    "end": "00:33:32",
                    "text": "initiatives. One, gigawatt scale data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:28",
                    "end": "00:33:35",
                    "text": "centers. We already started construction"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:32",
                    "end": "00:33:38",
                    "text": "on multi- gigawatt AI ready data centers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:35",
                    "end": "00:33:43",
                    "text": "at Jamnagar. Over 120 megawatts will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:38",
                    "end": "00:33:46",
                    "text": "come online in the second half of 2026"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:43",
                    "end": "00:33:49",
                    "text": "this year and a clear path to gigawatt"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:46",
                    "end": "00:33:51",
                    "text": "scale compute for training and large"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:49",
                    "end": "00:33:53",
                    "text": "scale inference."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:51",
                    "end": "00:33:56",
                    "text": "Two,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:53",
                    "end": "00:33:59",
                    "text": "our green energy advantage. We have an"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:56",
                    "end": "00:34:02",
                    "text": "in-house energy advantage with up to 10"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:59",
                    "end": "00:34:06",
                    "text": "gawatt of ready green power"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:02",
                    "end": "00:34:08",
                    "text": "surplus anchored by solar in both Kach"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:06",
                    "end": "00:34:12",
                    "text": "and Andhra Pradesh."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:08",
                    "end": "00:34:15",
                    "text": "Three, a nationwide edge compute. An"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:12",
                    "end": "00:34:17",
                    "text": "edge compute layer deeply integrated"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:15",
                    "end": "00:34:20",
                    "text": "with Jio's network will make"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:17",
                    "end": "00:34:23",
                    "text": "intelligence responsive, low latency and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:20",
                    "end": "00:34:28",
                    "text": "affordable close to where Indians live,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:23",
                    "end": "00:34:30",
                    "text": "learn and work. From kirana stores to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:28",
                    "end": "00:34:34",
                    "text": "clinics, from classrooms to farms,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:30",
                    "end": "00:34:38",
                    "text": "intelligence will live at the edge. Our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:34",
                    "end": "00:34:41",
                    "text": "resolve is clear. Make intelligence as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:38",
                    "end": "00:34:44",
                    "text": "ubiquitous as connectivity. When compute"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:41",
                    "end": "00:34:46",
                    "text": "becomes infrastructure, innovation will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:44",
                    "end": "00:34:50",
                    "text": "become inevitable."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:46",
                    "end": "00:34:53",
                    "text": "Friends, geo intelligence is guided by"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:50",
                    "end": "00:34:57",
                    "text": "five non-negotiable principles. First,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:53",
                    "end": "00:34:59",
                    "text": "AI for India's deep tech and advanced"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:57",
                    "end": "00:35:02",
                    "text": "manufacturing leadership reaching not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:59",
                    "end": "00:35:04",
                    "text": "just large enterprises but agriculture,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:02",
                    "end": "00:35:07",
                    "text": "small businesses and the informal"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:04",
                    "end": "00:35:10",
                    "text": "sector. Geo intelligence will not simply"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:07",
                    "end": "00:35:13",
                    "text": "be a search or an ask tool. It will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:10",
                    "end": "00:35:16",
                    "text": "primarily be a resource for multiplying"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:13",
                    "end": "00:35:20",
                    "text": "productivity and efficiency."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:16",
                    "end": "00:35:23",
                    "text": "Secondly, worldleading multilingual AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:20",
                    "end": "00:35:26",
                    "text": "capability across all Indian languages."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:23",
                    "end": "00:35:29",
                    "text": "When farmers and artisians speak to AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:26",
                    "end": "00:35:32",
                    "text": "in their own words and student learn in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:29",
                    "end": "00:35:36",
                    "text": "their own mother tongue, this is not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:32",
                    "end": "00:35:36",
                    "text": "convenience. This is inclusion."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:49",
                    "end": "00:35:53",
                    "text": "Third,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:50",
                    "end": "00:35:57",
                    "text": "responsibility, security, data residency"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:53",
                    "end": "00:36:01",
                    "text": "and trust as go core guarantees not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:57",
                    "end": "00:36:05",
                    "text": "afterthought. Fourth, we will provide we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:01",
                    "end": "00:36:08",
                    "text": "will prove that AI does not take away"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:05",
                    "end": "00:36:11",
                    "text": "jobs. Rather, it will create new"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:08",
                    "end": "00:36:16",
                    "text": "highskll work opportunities."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:11",
                    "end": "00:36:20",
                    "text": "And fifth, the AI story has shifted from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:16",
                    "end": "00:36:23",
                    "text": "who has the best model to who can build"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:20",
                    "end": "00:36:27",
                    "text": "the strongest ecosystem for speed and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:23",
                    "end": "00:36:30",
                    "text": "scale of usage. Therefore, we will build"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:27",
                    "end": "00:36:35",
                    "text": "deep partnership ecosystem with Indian"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:30",
                    "end": "00:36:37",
                    "text": "enterprises, startups, IIT, IISC and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:35",
                    "end": "00:36:40",
                    "text": "research institutions. We will work"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:37",
                    "end": "00:36:43",
                    "text": "shoulderto-shoulder with India's leading"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:40",
                    "end": "00:36:46",
                    "text": "industrial groups to embed AI across"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:43",
                    "end": "00:36:48",
                    "text": "manufacturing, logistics, energy,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:46",
                    "end": "00:36:51",
                    "text": "finance, retail, agriculture, and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:48",
                    "end": "00:36:53",
                    "text": "healthcare. We will empower startups"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:51",
                    "end": "00:36:56",
                    "text": "with affordable compute and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:53",
                    "end": "00:36:58",
                    "text": "co-development platforms. We will aspire"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:56",
                    "end": "00:37:01",
                    "text": "to produce global breakthroughs in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:58",
                    "end": "00:37:04",
                    "text": "compute architecture, foundation models"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:01",
                    "end": "00:37:06",
                    "text": "and energy efficiency designed in India,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:04",
                    "end": "00:37:09",
                    "text": "rooted in our values, powered by our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:06",
                    "end": "00:37:12",
                    "text": "talent and scaled for humanity. And we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:09",
                    "end": "00:37:15",
                    "text": "will partner with the very best tech"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:12",
                    "end": "00:37:18",
                    "text": "companies in the world, not as importers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:15",
                    "end": "00:37:21",
                    "text": "of intelligence, but as co-architects of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:18",
                    "end": "00:37:24",
                    "text": "a new AI century."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:21",
                    "end": "00:37:27",
                    "text": "Dear friends, I believe that social"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:24",
                    "end": "00:37:30",
                    "text": "relevance, not momentary craze should"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:27",
                    "end": "00:37:34",
                    "text": "drive AI growth in India. Jio has"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:30",
                    "end": "00:37:36",
                    "text": "already started AI applications for the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:34",
                    "end": "00:37:40",
                    "text": "most pressing challenges in inclusive"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:36",
                    "end": "00:37:43",
                    "text": "development. In education, we have Gio"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:40",
                    "end": "00:37:46",
                    "text": "Shikshak, an adaptive AI teaching"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:43",
                    "end": "00:37:49",
                    "text": "assistant in 22 languages. When 250"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:46",
                    "end": "00:37:51",
                    "text": "million school children and 50 million"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:49",
                    "end": "00:37:55",
                    "text": "students in higher education are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:51",
                    "end": "00:37:59",
                    "text": "empowered by AI teachers, no power on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:55",
                    "end": "00:38:03",
                    "text": "earth can match India's talent wealth."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:59",
                    "end": "00:38:06",
                    "text": "In healthcare, Gio arrogia AI delivering"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:03",
                    "end": "00:38:09",
                    "text": "first medical guidance in under five"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:06",
                    "end": "00:38:12",
                    "text": "minutes in local languages on any phone."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:09",
                    "end": "00:38:15",
                    "text": "in agriculture, geoci converting"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:12",
                    "end": "00:38:19",
                    "text": "satellite imagery and precision weather"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:15",
                    "end": "00:38:22",
                    "text": "into simple voice first advice to 140"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:19",
                    "end": "00:38:26",
                    "text": "million farmers to help improve their"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:22",
                    "end": "00:38:30",
                    "text": "income in everyday life. Gio Bharat IQ a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:26",
                    "end": "00:38:32",
                    "text": "voice first AI companion helping Indians"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:30",
                    "end": "00:38:37",
                    "text": "learn earn and access government"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:32",
                    "end": "00:38:39",
                    "text": "services at bhat scale from variables to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:37",
                    "end": "00:38:42",
                    "text": "fully connected homes"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:39",
                    "end": "00:38:45",
                    "text": "geo frames"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:42",
                    "end": "00:38:49",
                    "text": "an AI glass device and next generation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:45",
                    "end": "00:38:53",
                    "text": "AI devices will make intelligence truly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:49",
                    "end": "00:38:56",
                    "text": "ambient as effortless and as natural as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:53",
                    "end": "00:39:00",
                    "text": "human conversations through Geio Hot"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:56",
                    "end": "00:39:03",
                    "text": "Star. AI will multiply Indian creativity"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:00",
                    "end": "00:39:05",
                    "text": "with multilingual storytelling. We will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:03",
                    "end": "00:39:09",
                    "text": "popularize India's rich cultural"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:05",
                    "end": "00:39:12",
                    "text": "heritage with futuristic technology,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:09",
                    "end": "00:39:16",
                    "text": "enhancing India's soft power globally."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:12",
                    "end": "00:39:19",
                    "text": "Friends, this inaugural global AI impact"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:16",
                    "end": "00:39:23",
                    "text": "summit in India has received a massive"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:19",
                    "end": "00:39:26",
                    "text": "response. What does that show? It shows"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:23",
                    "end": "00:39:29",
                    "text": "that AI is now becoming a people's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:26",
                    "end": "00:39:33",
                    "text": "movement worldwide. The success of this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:29",
                    "end": "00:39:36",
                    "text": "movement hinges critically on global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:33",
                    "end": "00:39:39",
                    "text": "cooperation and not polarization. Be it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:36",
                    "end": "00:39:42",
                    "text": "chips or rare earths, AI works its magic"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:39",
                    "end": "00:39:45",
                    "text": "through sharing, not holding, through"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:42",
                    "end": "00:39:48",
                    "text": "collaborations not conflicts. The unique"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:45",
                    "end": "00:39:50",
                    "text": "strength of India is that India serves"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:48",
                    "end": "00:39:53",
                    "text": "as the vital bridge connecting the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:50",
                    "end": "00:39:57",
                    "text": "global south and the global north. After"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:53",
                    "end": "00:40:00",
                    "text": "all, south or north, east or west, all"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:57",
                    "end": "00:40:04",
                    "text": "of us have only one earth, one family,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:00",
                    "end": "00:40:07",
                    "text": "and one future. Today at this summit,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:04",
                    "end": "00:40:10",
                    "text": "let us all pledge to transform this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:07",
                    "end": "00:40:13",
                    "text": "noble aspiration into reality using the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:10",
                    "end": "00:40:16",
                    "text": "most powerful gift of the human mind,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:13",
                    "end": "00:40:20",
                    "text": "AI. Let us combine intelligence with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:16",
                    "end": "00:40:24",
                    "text": "empathy and let us build a better future"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:20",
                    "end": "00:40:24",
                    "text": "for all. Thank you. Jim,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:24",
                    "end": "00:40:31",
                    "text": "distinguished guests, my fellow Indians,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:28",
                    "end": "00:40:34",
                    "text": "namaste."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:31",
                    "end": "00:40:38",
                    "text": "The global AI impact summit is a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:34",
                    "end": "00:40:44",
                    "text": "defining moment in India's tech history."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:38",
                    "end": "00:40:47",
                    "text": "A moment when India pledges to make AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:44",
                    "end": "00:40:53",
                    "text": "one of the driving forces to realize its"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:47",
                    "end": "00:40:56",
                    "text": "dream of a Vix bharat. The dream of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:53",
                    "end": "00:40:57",
                    "text": "becoming a fully developed nation by"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:56",
                    "end": "00:40:60",
                    "text": "2047,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:57",
                    "end": "00:41:02",
                    "text": "the glorious centenary of our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:60",
                    "end": "00:41:04",
                    "text": "independence."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:02",
                    "end": "00:41:06",
                    "text": "We are deeply honored that our most"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:04",
                    "end": "00:41:10",
                    "text": "respected prime minister Shri Narendra"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:06",
                    "end": "00:41:13",
                    "text": "Modi is the guide, philosopher and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:10",
                    "end": "00:41:15",
                    "text": "leader of this summit."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:13",
                    "end": "00:41:19",
                    "text": "Honorable Minister Shriishna and his"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:15",
                    "end": "00:41:22",
                    "text": "wonderful team deserve full praise for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:19",
                    "end": "00:41:27",
                    "text": "organizing this summit on a grand scale"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:22",
                    "end": "00:41:27",
                    "text": "befitting India's ambition."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:29",
                    "end": "00:41:37",
                    "text": "Mhi G's vision of AI powered Vix Bharat"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:33",
                    "end": "00:41:39",
                    "text": "is also a template for a Vix global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:37",
                    "end": "00:41:43",
                    "text": "south."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:39",
                    "end": "00:41:46",
                    "text": "If wisely used, I believe AI can usher"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:43",
                    "end": "00:41:50",
                    "text": "in an era of super abundance."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:46",
                    "end": "00:41:53",
                    "text": "A world without poverty and a future of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:50",
                    "end": "00:41:57",
                    "text": "prosperity for all the 8 billion people"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:53",
                    "end": "00:42:01",
                    "text": "on our beautiful planet is now within"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:57",
                    "end": "00:42:04",
                    "text": "sight within reach. Friends,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:01",
                    "end": "00:42:06",
                    "text": "artificial intelligence is not just"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:04",
                    "end": "00:42:09",
                    "text": "another technology."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:06",
                    "end": "00:42:13",
                    "text": "For the first time, humans are creating"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:09",
                    "end": "00:42:17",
                    "text": "human-like systems that can learn,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:13",
                    "end": "00:42:19",
                    "text": "speak, analyze, move, and produce"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:17",
                    "end": "00:42:23",
                    "text": "autonomously."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:19",
                    "end": "00:42:27",
                    "text": "AI is the mantra that powers every"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:23",
                    "end": "00:42:31",
                    "text": "yantra or every machine and system to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:27",
                    "end": "00:42:36",
                    "text": "work faster, better, and smarter."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:31",
                    "end": "00:42:39",
                    "text": "I see AI as a modernday akshaiatra,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:36",
                    "end": "00:42:43",
                    "text": "the legendary vessel in Madharat that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:39",
                    "end": "00:42:46",
                    "text": "provided endless nourishment to all."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:43",
                    "end": "00:42:49",
                    "text": "Likewise, AI offers limitless"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:46",
                    "end": "00:42:51",
                    "text": "augmentation in knowledge, efficiency,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:49",
                    "end": "00:42:56",
                    "text": "and productivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:51",
                    "end": "00:43:01",
                    "text": "We are only at the dawn of this era. The"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:56",
                    "end": "00:43:04",
                    "text": "best of AI is yet to come. Distinguished"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:01",
                    "end": "00:43:07",
                    "text": "participants in this summit today the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:04",
                    "end": "00:43:09",
                    "text": "world is defound is debating a profound"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:07",
                    "end": "00:43:12",
                    "text": "question."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:09",
                    "end": "00:43:15",
                    "text": "Will AI concentrate power in the hands"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:12",
                    "end": "00:43:19",
                    "text": "of a few or will it democratize"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:15",
                    "end": "00:43:23",
                    "text": "opportunity for all? Do we act as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:19",
                    "end": "00:43:27",
                    "text": "isolated nations or as a united global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:23",
                    "end": "00:43:30",
                    "text": "family? Our polarized world stands at a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:27",
                    "end": "00:43:33",
                    "text": "fork."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:30",
                    "end": "00:43:36",
                    "text": "One path has led to a situation where AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:33",
                    "end": "00:43:39",
                    "text": "is scarce and expensive, compute is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:36",
                    "end": "00:43:42",
                    "text": "concentrated, data is controlled and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:39",
                    "end": "00:43:46",
                    "text": "capability is locked behind barriers of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:42",
                    "end": "00:43:49",
                    "text": "capital and geography in the global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:46",
                    "end": "00:43:52",
                    "text": "north. In this dismal scenario,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:49",
                    "end": "00:43:56",
                    "text": "inequality widens between nations within"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:52",
                    "end": "00:44:00",
                    "text": "societies and across generations. But"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:56",
                    "end": "00:44:03",
                    "text": "there is another path. A future where AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:00",
                    "end": "00:44:07",
                    "text": "is available, affordable and beneficial"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:03",
                    "end": "00:44:10",
                    "text": "to all. As our prime minister said,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:07",
                    "end": "00:44:12",
                    "text": "India believes in this second future."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:10",
                    "end": "00:44:15",
                    "text": "Dear friends,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:12",
                    "end": "00:44:18",
                    "text": "from the podium of this summit today, I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:15",
                    "end": "00:44:22",
                    "text": "want to make a bold prediction. India"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:18",
                    "end": "00:44:27",
                    "text": "will emerge as one of the greatest AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:22",
                    "end": "00:44:27",
                    "text": "powers in the world in the 21st century."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:30",
                    "end": "00:44:37",
                    "text": "My confidence is validated by an"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:34",
                    "end": "00:44:39",
                    "text": "undeniable truth. In the coming decades,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:37",
                    "end": "00:44:43",
                    "text": "no country in the world can match"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:39",
                    "end": "00:44:46",
                    "text": "India's strength in demography,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:43",
                    "end": "00:44:49",
                    "text": "democracy, development, digital"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:46",
                    "end": "00:44:53",
                    "text": "infrastructure, data generation, AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:49",
                    "end": "00:44:55",
                    "text": "harvest. Let me begin as was said many"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:53",
                    "end": "00:44:58",
                    "text": "times this morning, what India has"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:55",
                    "end": "00:45:01",
                    "text": "achieved in the past 10 years. First,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:58",
                    "end": "00:45:04",
                    "text": "India is the world's largest mobile data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:01",
                    "end": "00:45:07",
                    "text": "consumer. Nearly 1 billion internet"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:04",
                    "end": "00:45:10",
                    "text": "users, data costs among the lowest"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:07",
                    "end": "00:45:13",
                    "text": "globally and in terms of quality there"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:10",
                    "end": "00:45:17",
                    "text": "is no difference between Delhi and the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:13",
                    "end": "00:45:22",
                    "text": "remotest Indian village. Second, Aadhaar"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:17",
                    "end": "00:45:24",
                    "text": "1.4 billion digital IDs. Third, UPI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:22",
                    "end": "00:45:28",
                    "text": "processes over 12 billion transactions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:24",
                    "end": "00:45:30",
                    "text": "monthly. Fourth, India ranks among the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:28",
                    "end": "00:45:34",
                    "text": "top three startup ecosystems with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:30",
                    "end": "00:45:38",
                    "text": "100,000 startups and 100 plus unicorns."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:34",
                    "end": "00:45:40",
                    "text": "Fifth, India's secure and inclusive"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:38",
                    "end": "00:45:43",
                    "text": "digital public infrastructure stack is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:40",
                    "end": "00:45:47",
                    "text": "now being adopted by countries around"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:43",
                    "end": "00:45:51",
                    "text": "the globe. Friends, in all humility, I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:47",
                    "end": "00:45:54",
                    "text": "wish to state that Jio with over 500"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:51",
                    "end": "00:45:57",
                    "text": "million loyal subscribers was privileged"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:54",
                    "end": "00:46:00",
                    "text": "to play a leading role in this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:57",
                    "end": "00:46:02",
                    "text": "transformation across broadband, 4G, 5G,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:00",
                    "end": "00:46:05",
                    "text": "and home connectivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:02",
                    "end": "00:46:08",
                    "text": "With equal humility, I would like to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:05",
                    "end": "00:46:11",
                    "text": "announce that Jio will play an even"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:08",
                    "end": "00:46:14",
                    "text": "bigger role in India's AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:11",
                    "end": "00:46:17",
                    "text": "transformation. Today on behalf of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:14",
                    "end": "00:46:19",
                    "text": "Reliance Group and Geo Intelligence I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:17",
                    "end": "00:46:23",
                    "text": "want to make three announcements."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:19",
                    "end": "00:46:26",
                    "text": "Announcement one, Jio connected India to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:23",
                    "end": "00:46:31",
                    "text": "the internet era. Jio will now connect"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:26",
                    "end": "00:46:31",
                    "text": "India to the intelligence era."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:32",
                    "end": "00:46:38",
                    "text": "We will deliver intelligence to every"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:35",
                    "end": "00:46:41",
                    "text": "citizen, every sector of the economy and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:38",
                    "end": "00:46:44",
                    "text": "every facet of social development and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:41",
                    "end": "00:46:47",
                    "text": "every service of government. Jio will do"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:44",
                    "end": "00:46:50",
                    "text": "so with the same reliability, quality,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:47",
                    "end": "00:46:52",
                    "text": "scale and extreme affordability that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:50",
                    "end": "00:46:54",
                    "text": "transformed connectivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:52",
                    "end": "00:46:58",
                    "text": "India cannot afford to rent"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:54",
                    "end": "00:47:01",
                    "text": "intelligence. Therefore, we will reduce"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:58",
                    "end": "00:47:07",
                    "text": "the cost of intelligence as dramatically"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:01",
                    "end": "00:47:07",
                    "text": "as we did the cost of data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:09",
                    "end": "00:47:17",
                    "text": "Announcement two, Jio together with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:12",
                    "end": "00:47:23",
                    "text": "Reliance will invest 10 lakh crores over"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:17",
                    "end": "00:47:23",
                    "text": "the next seven years starting this year."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:24",
                    "end": "00:47:30",
                    "text": "This is not speculative investment. It"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:27",
                    "end": "00:47:33",
                    "text": "is not purchasing valuation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:30",
                    "end": "00:47:36",
                    "text": "This is patient, disciplined,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:33",
                    "end": "00:47:39",
                    "text": "nationbuilding capital designed to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:36",
                    "end": "00:47:41",
                    "text": "create durable economic value and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:39",
                    "end": "00:47:43",
                    "text": "strategic resilience for decades to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:41",
                    "end": "00:47:46",
                    "text": "come."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:43",
                    "end": "00:47:50",
                    "text": "Distinguished participants, the biggest"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:46",
                    "end": "00:47:53",
                    "text": "constraint in AI today is not talent or"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:50",
                    "end": "00:47:56",
                    "text": "imagination. It is scarcity and high"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:53",
                    "end": "00:47:60",
                    "text": "cost of compute. Therefore, here is my"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:56",
                    "end": "00:48:02",
                    "text": "third announcement. Geo intelligence"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:60",
                    "end": "00:48:05",
                    "text": "will build India's sovereign compute"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:02",
                    "end": "00:48:09",
                    "text": "infrastructure through three bold"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:05",
                    "end": "00:48:12",
                    "text": "initiatives. One, gigawatt scale data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:09",
                    "end": "00:48:16",
                    "text": "centers. We already started construction"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:12",
                    "end": "00:48:19",
                    "text": "on multi- gigawatt AI ready data centers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:16",
                    "end": "00:48:24",
                    "text": "at Jamnagar. Over 120 megawatts will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:19",
                    "end": "00:48:27",
                    "text": "come online in the second half of 2026"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:24",
                    "end": "00:48:30",
                    "text": "this year and a clear path to gigawatt"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:27",
                    "end": "00:48:32",
                    "text": "scale compute for training and large"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:30",
                    "end": "00:48:34",
                    "text": "scale inference."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:32",
                    "end": "00:48:37",
                    "text": "Two,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:34",
                    "end": "00:48:40",
                    "text": "our green energy advantage. We have an"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:37",
                    "end": "00:48:44",
                    "text": "in-house energy advantage with up to 10"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:40",
                    "end": "00:48:47",
                    "text": "gawatt of ready green power surplus"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:44",
                    "end": "00:48:49",
                    "text": "anchored by solar in both Kach and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:47",
                    "end": "00:48:53",
                    "text": "Andhra Pradesh."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:49",
                    "end": "00:48:56",
                    "text": "Three, a nationwide edge compute. An"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:53",
                    "end": "00:48:58",
                    "text": "edge compute layer deeply integrated"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:56",
                    "end": "00:49:01",
                    "text": "with Jio's network will make"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:58",
                    "end": "00:49:04",
                    "text": "intelligence responsive, low latency and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:01",
                    "end": "00:49:08",
                    "text": "affordable close to where Indians live,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:04",
                    "end": "00:49:11",
                    "text": "learn and work. From kirana stores to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:08",
                    "end": "00:49:14",
                    "text": "clinics, from classrooms to farms,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:11",
                    "end": "00:49:18",
                    "text": "intelligence will live at the edge. Our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:14",
                    "end": "00:49:22",
                    "text": "resolve is clear. Make intelligence as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:18",
                    "end": "00:49:25",
                    "text": "ubiquitous as connectivity. When compute"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:22",
                    "end": "00:49:27",
                    "text": "becomes infrastructure, innovation will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:25",
                    "end": "00:49:31",
                    "text": "become inevitable."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:27",
                    "end": "00:49:34",
                    "text": "Friends, geo intelligence is guided by"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:31",
                    "end": "00:49:38",
                    "text": "five non-negotiable principles. First,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:34",
                    "end": "00:49:40",
                    "text": "AI for India's deep tech and advanced"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:38",
                    "end": "00:49:43",
                    "text": "manufacturing leadership reaching not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:40",
                    "end": "00:49:45",
                    "text": "just large enterprises but agriculture,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:43",
                    "end": "00:49:48",
                    "text": "small businesses and the informal"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:45",
                    "end": "00:49:51",
                    "text": "sector. Geo intelligence will not simply"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:48",
                    "end": "00:49:54",
                    "text": "be a search or an ask tool. It will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:51",
                    "end": "00:49:58",
                    "text": "primarily be a resource for multiplying"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:54",
                    "end": "00:50:02",
                    "text": "productivity and efficiency. Second,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:58",
                    "end": "00:50:04",
                    "text": "worldleading multilingual AI capability"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:02",
                    "end": "00:50:07",
                    "text": "across all Indian languages. When"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:04",
                    "end": "00:50:10",
                    "text": "farmers and articians speak to AI in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:07",
                    "end": "00:50:12",
                    "text": "their own words and student learn in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:10",
                    "end": "00:50:17",
                    "text": "their own mother tongue, this is not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:12",
                    "end": "00:50:17",
                    "text": "convenience. This is inclusion."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:29",
                    "end": "00:50:34",
                    "text": "Third,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:31",
                    "end": "00:50:38",
                    "text": "responsibility, security, data residency"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:34",
                    "end": "00:50:42",
                    "text": "and trust as GEO's core guarantees, not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:38",
                    "end": "00:50:45",
                    "text": "afterthought. Fourth, we will provide we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:42",
                    "end": "00:50:49",
                    "text": "will prove that AI does not take away"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:45",
                    "end": "00:50:52",
                    "text": "jobs. Rather, it will create new"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:49",
                    "end": "00:50:57",
                    "text": "highskll work opportunities."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:52",
                    "end": "00:51:00",
                    "text": "And fifth, the AI story has shifted from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:57",
                    "end": "00:51:04",
                    "text": "who has the best model to who can build"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:00",
                    "end": "00:51:08",
                    "text": "the strongest ecosystem for speed and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:04",
                    "end": "00:51:12",
                    "text": "scale of usage. Therefore, we will build"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:08",
                    "end": "00:51:16",
                    "text": "deep partnership ecosystem with Indian"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:12",
                    "end": "00:51:18",
                    "text": "enterprises, startups, IIT, IISC and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:16",
                    "end": "00:51:21",
                    "text": "research institutions. We will work"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:18",
                    "end": "00:51:24",
                    "text": "shoulderto-shoulder with India's leading"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:21",
                    "end": "00:51:26",
                    "text": "industrial groups to embed AI across"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:24",
                    "end": "00:51:29",
                    "text": "manufacturing, logistics, energy,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:26",
                    "end": "00:51:31",
                    "text": "finance, retail, agriculture, and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:29",
                    "end": "00:51:33",
                    "text": "healthcare. We will empower startups"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:31",
                    "end": "00:51:36",
                    "text": "with affordable compute and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:33",
                    "end": "00:51:39",
                    "text": "co-development platforms. We will aspire"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:36",
                    "end": "00:51:42",
                    "text": "to produce global breakthroughs in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:39",
                    "end": "00:51:45",
                    "text": "compute architecture, foundation models"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:42",
                    "end": "00:51:47",
                    "text": "and energy efficiency designed in India,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:45",
                    "end": "00:51:50",
                    "text": "rooted in our values, powered by our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:47",
                    "end": "00:51:53",
                    "text": "talent and scaled for humanity. And we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:50",
                    "end": "00:51:55",
                    "text": "will partner with the very best tech"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:53",
                    "end": "00:51:59",
                    "text": "companies in the world, not as importers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:55",
                    "end": "00:52:02",
                    "text": "of intelligence, but as co-architects of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:59",
                    "end": "00:52:05",
                    "text": "a new AI century."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:02",
                    "end": "00:52:08",
                    "text": "Dear friends, I believe that social"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:05",
                    "end": "00:52:11",
                    "text": "relevance, not momentary craze should"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:08",
                    "end": "00:52:14",
                    "text": "drive AI growth in India. Jio has"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:11",
                    "end": "00:52:17",
                    "text": "already started AI applications for the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:14",
                    "end": "00:52:21",
                    "text": "most pressing challenges in inclusive"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:17",
                    "end": "00:52:24",
                    "text": "development. In education, we have Gio"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:21",
                    "end": "00:52:27",
                    "text": "Shikshak, an adaptive AI teaching"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:24",
                    "end": "00:52:30",
                    "text": "assistant in 22 languages. When 250"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:27",
                    "end": "00:52:32",
                    "text": "million school children and 50 million"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:30",
                    "end": "00:52:36",
                    "text": "students in higher education are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:32",
                    "end": "00:52:39",
                    "text": "empowered by AI teachers, no power on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:36",
                    "end": "00:52:44",
                    "text": "earth can match India's talent wealth."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:39",
                    "end": "00:52:47",
                    "text": "In healthcare, Geo Aarogia AI delivering"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:44",
                    "end": "00:52:50",
                    "text": "first medical guidance in under five"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:47",
                    "end": "00:52:53",
                    "text": "minutes in local languages on any phone."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:50",
                    "end": "00:52:56",
                    "text": "in agriculture, geoci converting"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:53",
                    "end": "00:52:59",
                    "text": "satellite imagery and precision weather"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:56",
                    "end": "00:53:03",
                    "text": "into simple voice first advice to 140"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:59",
                    "end": "00:53:07",
                    "text": "million farmers to help improve their"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:03",
                    "end": "00:53:10",
                    "text": "income in everyday life. Gio Bharat IQ a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:07",
                    "end": "00:53:13",
                    "text": "voice first AI companion helping Indians"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:10",
                    "end": "00:53:17",
                    "text": "learn earn and access government"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:13",
                    "end": "00:53:20",
                    "text": "services at bhat scale from variables to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:17",
                    "end": "00:53:23",
                    "text": "fully connected homes"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:20",
                    "end": "00:53:26",
                    "text": "geo frames"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:23",
                    "end": "00:53:30",
                    "text": "an AI glass device and next generation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:26",
                    "end": "00:53:34",
                    "text": "AI devices will make intelligence truly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:30",
                    "end": "00:53:36",
                    "text": "ambient as effortless and as natural as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:34",
                    "end": "00:53:40",
                    "text": "human conversation. ations. Through Geo"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:36",
                    "end": "00:53:42",
                    "text": "Hot Star, AI will multiply Indian"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:40",
                    "end": "00:53:45",
                    "text": "creativity with multilingual"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:42",
                    "end": "00:53:48",
                    "text": "storytelling. We will popularize India's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:45",
                    "end": "00:53:51",
                    "text": "rich cultural heritage with futuristic"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:48",
                    "end": "00:53:56",
                    "text": "technology, enhancing India's soft power"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:51",
                    "end": "00:53:59",
                    "text": "globally. Friends, this inaugural global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:56",
                    "end": "00:54:03",
                    "text": "AI impact summit in India has received a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:59",
                    "end": "00:54:06",
                    "text": "massive response. What does that show?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:03",
                    "end": "00:54:09",
                    "text": "It shows that AI is now becoming a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:06",
                    "end": "00:54:13",
                    "text": "people's movement worldwide. The success"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:09",
                    "end": "00:54:16",
                    "text": "of this movement hinges critically on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:13",
                    "end": "00:54:19",
                    "text": "global cooperation and not polarization."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:16",
                    "end": "00:54:22",
                    "text": "Be it chips or rare earths, AI works its"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:19",
                    "end": "00:54:25",
                    "text": "magic through sharing not holding"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:22",
                    "end": "00:54:27",
                    "text": "through collaborations not conflicts."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:25",
                    "end": "00:54:30",
                    "text": "The unique strength of India is that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:27",
                    "end": "00:54:32",
                    "text": "India serves as the vital bridge"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:30",
                    "end": "00:54:36",
                    "text": "connecting the global south and the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:32",
                    "end": "00:54:39",
                    "text": "global north. After all, south or north,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:36",
                    "end": "00:54:43",
                    "text": "east or west, all of us have only one"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:39",
                    "end": "00:54:47",
                    "text": "earth, one family, and one future. Today"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:43",
                    "end": "00:54:49",
                    "text": "at this summit, let us all pledge to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:47",
                    "end": "00:54:53",
                    "text": "transform this noble aspiration into"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:49",
                    "end": "00:54:56",
                    "text": "reality using the most powerful gift of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:53",
                    "end": "00:54:59",
                    "text": "the human mind, AI. Let us combine"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:56",
                    "end": "00:55:02",
                    "text": "intelligence with empathy and let us"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:59",
                    "end": "00:55:05",
                    "text": "build a better future for all. Thank"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:02",
                    "end": "00:55:05",
                    "text": "you, Jahin."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:05",
                    "end": "00:55:12",
                    "text": "Distinguished guests, my fellow Indians,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:09",
                    "end": "00:55:15",
                    "text": "namaste."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:12",
                    "end": "00:55:19",
                    "text": "The global AI impact summit is a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:15",
                    "end": "00:55:25",
                    "text": "defining moment in India's tech history."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:19",
                    "end": "00:55:28",
                    "text": "A moment when India pledges to make AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:25",
                    "end": "00:55:33",
                    "text": "one of the driving forces to realize its"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:28",
                    "end": "00:55:36",
                    "text": "dream of a Vixs Bharat, the dream of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:33",
                    "end": "00:55:38",
                    "text": "becoming a fully developed nation by"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:36",
                    "end": "00:55:40",
                    "text": "2047,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:38",
                    "end": "00:55:43",
                    "text": "the glorious centenary of our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:40",
                    "end": "00:55:45",
                    "text": "independence."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:43",
                    "end": "00:55:47",
                    "text": "We are deeply honored that our most"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:45",
                    "end": "00:55:50",
                    "text": "respected prime minister Shri Narendra"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:47",
                    "end": "00:55:54",
                    "text": "Modi is the guide, philosopher and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:50",
                    "end": "00:55:56",
                    "text": "leader of this summit."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:54",
                    "end": "00:55:60",
                    "text": "Honorable Minister Shriishnau and his"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:56",
                    "end": "00:56:03",
                    "text": "wonderful team deserve full praise for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:60",
                    "end": "00:56:08",
                    "text": "organizing this summit on a grand scale"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:03",
                    "end": "00:56:08",
                    "text": "befitting India's ambition."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:10",
                    "end": "00:56:18",
                    "text": "Mi G's vision of AI powered Vixed Bharat"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:14",
                    "end": "00:56:20",
                    "text": "is also a template for a Vix global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:18",
                    "end": "00:56:24",
                    "text": "south."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:20",
                    "end": "00:56:27",
                    "text": "If wisely used, I believe AI can usher"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:24",
                    "end": "00:56:31",
                    "text": "in an era of super abundance."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:27",
                    "end": "00:56:34",
                    "text": "A world without poverty and a future of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:31",
                    "end": "00:56:38",
                    "text": "prosperity for all the 8 billion people"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:34",
                    "end": "00:56:42",
                    "text": "on our beautiful planet is now within"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:38",
                    "end": "00:56:44",
                    "text": "sight within reach. Friends,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:42",
                    "end": "00:56:47",
                    "text": "artificial intelligence is not just"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:44",
                    "end": "00:56:50",
                    "text": "another technology."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:47",
                    "end": "00:56:54",
                    "text": "For the first time, humans are creating"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:50",
                    "end": "00:56:58",
                    "text": "human-like systems that can learn,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:54",
                    "end": "00:56:60",
                    "text": "speak, analyze, move, and produce"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:58",
                    "end": "00:57:04",
                    "text": "autonomously."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:60",
                    "end": "00:57:08",
                    "text": "AI is the mantra that powers every"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:04",
                    "end": "00:57:12",
                    "text": "yantra or every machine and system to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:08",
                    "end": "00:57:17",
                    "text": "work faster, better, and smarter."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:12",
                    "end": "00:57:20",
                    "text": "IC AI as a modernday akshaiatra"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:17",
                    "end": "00:57:24",
                    "text": "the legendary vessel in Madharat that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:20",
                    "end": "00:57:27",
                    "text": "provided endless nourishment to all."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:24",
                    "end": "00:57:30",
                    "text": "Likewise AI offers limitless"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:27",
                    "end": "00:57:32",
                    "text": "augmentation in knowledge, efficiency"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:30",
                    "end": "00:57:37",
                    "text": "and productivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:32",
                    "end": "00:57:42",
                    "text": "We are only at the dawn of this era. The"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:37",
                    "end": "00:57:45",
                    "text": "best of AI is yet to come. Distinguished"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:42",
                    "end": "00:57:48",
                    "text": "participants in this summit today the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:45",
                    "end": "00:57:50",
                    "text": "world is defound is debating a profound"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:48",
                    "end": "00:57:53",
                    "text": "question."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:50",
                    "end": "00:57:56",
                    "text": "Will AI concentrate power in the hands"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:53",
                    "end": "00:57:60",
                    "text": "of a few or will it democratize"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:56",
                    "end": "00:58:04",
                    "text": "opportunity for all? Do we act as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:60",
                    "end": "00:58:08",
                    "text": "isolated nations or as a united global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:04",
                    "end": "00:58:10",
                    "text": "family? Our polarized world stands at a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:08",
                    "end": "00:58:14",
                    "text": "fork."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:10",
                    "end": "00:58:17",
                    "text": "One path has led to a situation where AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:14",
                    "end": "00:58:20",
                    "text": "is scarce and expensive, compute is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:17",
                    "end": "00:58:23",
                    "text": "concentrated, data is controlled and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:20",
                    "end": "00:58:27",
                    "text": "capability is locked behind barriers of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:23",
                    "end": "00:58:30",
                    "text": "capital and geography in the global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:27",
                    "end": "00:58:33",
                    "text": "north. In this dismal scenario,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:30",
                    "end": "00:58:37",
                    "text": "inequality widens between nations within"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:33",
                    "end": "00:58:41",
                    "text": "societies and across generations. But"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:37",
                    "end": "00:58:44",
                    "text": "there is another path, a future where AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:41",
                    "end": "00:58:48",
                    "text": "is available, affordable and beneficial"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:44",
                    "end": "00:58:51",
                    "text": "to all. As our prime minister said,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:48",
                    "end": "00:58:53",
                    "text": "India believes in this second future."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:51",
                    "end": "00:58:56",
                    "text": "Dear friends,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:53",
                    "end": "00:58:59",
                    "text": "from the podium of this summit today, I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:56",
                    "end": "00:59:03",
                    "text": "want to make a bold prediction. India"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:59",
                    "end": "00:59:08",
                    "text": "will emerge as one of the greatest AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:03",
                    "end": "00:59:08",
                    "text": "powers in the world in the 21st century."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:11",
                    "end": "00:59:18",
                    "text": "My confidence is validated by an"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:15",
                    "end": "00:59:20",
                    "text": "undeniable truth. In the coming decades,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:18",
                    "end": "00:59:24",
                    "text": "no country in the world can match"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:20",
                    "end": "00:59:26",
                    "text": "India's strength in demography,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:24",
                    "end": "00:59:30",
                    "text": "democracy, development, digital"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:26",
                    "end": "00:59:33",
                    "text": "infrastructure, data generation, AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:30",
                    "end": "00:59:36",
                    "text": "harvest. Let me begin as was said many"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:33",
                    "end": "00:59:39",
                    "text": "times this morning, what India has"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:36",
                    "end": "00:59:42",
                    "text": "achieved in the past 10 years. First,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:39",
                    "end": "00:59:45",
                    "text": "India is the world's largest mobile data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:42",
                    "end": "00:59:48",
                    "text": "consumer. Nearly 1 billion internet"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:45",
                    "end": "00:59:51",
                    "text": "users, data costs among the lowest"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:48",
                    "end": "00:59:54",
                    "text": "globally and in terms of quality there"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:51",
                    "end": "00:59:58",
                    "text": "is no difference between Delhi and the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:54",
                    "end": "01:00:02",
                    "text": "remotest Indian village. Second, Aadhaar"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:58",
                    "end": "01:00:05",
                    "text": "1.4 billion digital ids. Third, UPI"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:02",
                    "end": "01:00:08",
                    "text": "processes over 12 billion transactions"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:05",
                    "end": "01:00:11",
                    "text": "monthly. Fourth, India ranks among the"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:08",
                    "end": "01:00:15",
                    "text": "top three startup ecosystems with"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:11",
                    "end": "01:00:19",
                    "text": "100,000 startups and 100 plus unicorns."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:15",
                    "end": "01:00:21",
                    "text": "Fifth, India's secure and inclusive"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:19",
                    "end": "01:00:24",
                    "text": "digital public infrastructure stack is"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:21",
                    "end": "01:00:28",
                    "text": "now being adopted by countries around"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:24",
                    "end": "01:00:32",
                    "text": "the globe. Friends, in all humility, I"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:28",
                    "end": "01:00:35",
                    "text": "wish to state that Jio with over 500"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:32",
                    "end": "01:00:38",
                    "text": "million loyal subscribers was privileged"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:35",
                    "end": "01:00:41",
                    "text": "to play a leading role in this"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:38",
                    "end": "01:00:43",
                    "text": "transformation across broadband, 4G, 5G,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:41",
                    "end": "01:00:46",
                    "text": "and home connectivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:43",
                    "end": "01:00:49",
                    "text": "With equal humility, I would like to"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:46",
                    "end": "01:00:52",
                    "text": "announce that Jio will play an even"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:49",
                    "end": "01:00:55",
                    "text": "bigger role in India's AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:52",
                    "end": "01:00:57",
                    "text": "transformation. Today on behalf of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:55",
                    "end": "01:00:60",
                    "text": "Reliance Group and Geo Intelligence I"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:57",
                    "end": "01:01:03",
                    "text": "want to make three announcements."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:60",
                    "end": "01:01:07",
                    "text": "Announcement one, Jio connected India to"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:01:03",
                    "end": "01:01:12",
                    "text": "the internet era. Jio will now connect"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:01:07",
                    "end": "01:01:12",
                    "text": "India to the intelligence era."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:01:13",
                    "end": "01:01:19",
                    "text": "We will deliver intelligence to every"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:01:15",
                    "end": "01:01:21",
                    "text": "citizen, every sector of the economy and"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:01:19",
                    "end": "01:01:25",
                    "text": "every facet of social development and"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:01:21",
                    "end": "01:01:28",
                    "text": "every service of government. Jio will do"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:01:25",
                    "end": "01:01:31",
                    "text": "so with the same reliability, quality,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:01:28",
                    "end": "01:01:33",
                    "text": "scale and extreme affordability that"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:01:31",
                    "end": "01:01:35",
                    "text": "transformed connectivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:01:33",
                    "end": "01:01:39",
                    "text": "India cannot afford to rent"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:01:35",
                    "end": "01:01:42",
                    "text": "intelligence. Therefore, we will reduce"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:01:39",
                    "end": "01:01:45",
                    "text": "the cost of intelligence as dramatically"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:01:42",
                    "end": "01:01:50",
                    "text": "as we"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:01:45",
                    "end": "01:01:50",
                    "text": "did the cost of data."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:01:52",
                    "end": "01:01:59",
                    "text": "Announcement two, Jio together with"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:01:55",
                    "end": "01:02:04",
                    "text": "Reliance will invest 10 lakh crores over"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:01:59",
                    "end": "01:02:06",
                    "text": "the next seven years starting this year."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:02:04",
                    "end": "01:02:09",
                    "text": "No,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:02:06",
                    "end": "01:02:12",
                    "text": "this is not speculative investment. It"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:02:09",
                    "end": "01:02:15",
                    "text": "is not purchasing valuation. This is"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:02:12",
                    "end": "01:02:18",
                    "text": "patient, disciplined nation building"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:02:15",
                    "end": "01:02:21",
                    "text": "capital designed to create durable"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:02:18",
                    "end": "01:02:24",
                    "text": "economic value and strategic resilience"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:02:21",
                    "end": "01:02:27",
                    "text": "for decades to come."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:02:24",
                    "end": "01:02:31",
                    "text": "Distinguished participants, the biggest"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:02:27",
                    "end": "01:02:34",
                    "text": "constraint in AI today is not talent or"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:02:31",
                    "end": "01:02:37",
                    "text": "imagination. It is scarcity and high"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:02:34",
                    "end": "01:02:41",
                    "text": "cost of compute. Therefore, here is my"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:02:37",
                    "end": "01:02:43",
                    "text": "third announcement. Geo intelligence"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:02:41",
                    "end": "01:02:46",
                    "text": "will build India's sovereign compute"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:02:43",
                    "end": "01:02:50",
                    "text": "infrastructure through three bold"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:02:46",
                    "end": "01:02:53",
                    "text": "initiatives. One, gigawatt scale data"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:02:50",
                    "end": "01:02:56",
                    "text": "centers. We already started construction"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:02:53",
                    "end": "01:03:00",
                    "text": "on multi- gawatt AI ready data centers"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:02:56",
                    "end": "01:03:03",
                    "text": "at Jamnagar. Over 120 megawatts will"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:03:00",
                    "end": "01:03:07",
                    "text": "come online in"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:03:03",
                    "end": "01:03:10",
                    "text": "the second half of 2026 this year and a"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:03:07",
                    "end": "01:03:13",
                    "text": "clear path to gigawatt scale compute for"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:03:10",
                    "end": "01:03:15",
                    "text": "training and large scale inference."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:03:13",
                    "end": "01:03:18",
                    "text": "Two,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:03:15",
                    "end": "01:03:21",
                    "text": "our green energy advantage. We have an"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:03:18",
                    "end": "01:03:25",
                    "text": "in-house energy advantage with up to 10"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:03:21",
                    "end": "01:03:27",
                    "text": "gawatt of ready green power surplus"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:03:25",
                    "end": "01:03:29",
                    "text": "anchored by solar in both Kach and"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:03:27",
                    "end": "01:03:33",
                    "text": "Andhra Pradesh."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:03:29",
                    "end": "01:03:36",
                    "text": "Three, a nationwide edge compute. An"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:03:33",
                    "end": "01:03:39",
                    "text": "edge compute layer deeply integrated"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:03:36",
                    "end": "01:03:42",
                    "text": "with Jio's network will make"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:03:39",
                    "end": "01:03:45",
                    "text": "intelligence responsive, low latency and"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:03:42",
                    "end": "01:03:49",
                    "text": "affordable close to where Indians live,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:03:45",
                    "end": "01:03:52",
                    "text": "learn and work. From kirana stores to"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:03:49",
                    "end": "01:03:56",
                    "text": "clinics, from classrooms to farms,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:03:52",
                    "end": "01:03:60",
                    "text": "intelligence will live at the edge. Our"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:03:56",
                    "end": "01:04:03",
                    "text": "resolve is clear. Make intelligence as"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:03:60",
                    "end": "01:04:06",
                    "text": "ubiquitous as connectivity. When compute"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:04:03",
                    "end": "01:04:08",
                    "text": "becomes infrastructure, innovation will"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:04:06",
                    "end": "01:04:12",
                    "text": "become inevitable."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:04:08",
                    "end": "01:04:15",
                    "text": "Friends, geo intelligence is guided by"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:04:12",
                    "end": "01:04:19",
                    "text": "five non-negotiable principles. First,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:04:15",
                    "end": "01:04:21",
                    "text": "AI for India's deep tech and advanced"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:04:19",
                    "end": "01:04:24",
                    "text": "manufacturing leadership reaching not"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:04:21",
                    "end": "01:04:26",
                    "text": "just large enterprises but agriculture,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:04:24",
                    "end": "01:04:29",
                    "text": "small businesses and the informal"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:04:26",
                    "end": "01:04:32",
                    "text": "sector. Geo intelligence will not simply"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:04:29",
                    "end": "01:04:35",
                    "text": "be a search or an ask tool. It will"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:04:32",
                    "end": "01:04:39",
                    "text": "primarily be a resource for multiplying"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:04:35",
                    "end": "01:04:43",
                    "text": "productivity and efficiency. Second,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:04:39",
                    "end": "01:04:45",
                    "text": "worldleading multilingual AI capability"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:04:43",
                    "end": "01:04:48",
                    "text": "across all Indian languages. When"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:04:45",
                    "end": "01:04:51",
                    "text": "farmers and articians speak to AI in"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:04:48",
                    "end": "01:04:53",
                    "text": "their own words and student learn in"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:04:51",
                    "end": "01:04:57",
                    "text": "their own mother tongue, this is not"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:04:53",
                    "end": "01:05:00",
                    "text": "convenience. This is inclusion."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:04:57",
                    "end": "01:05:00",
                    "text": "Go ahead."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:05:11",
                    "end": "01:05:15",
                    "text": "Third,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:05:12",
                    "end": "01:05:19",
                    "text": "responsibility, security, data residency"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:05:15",
                    "end": "01:05:23",
                    "text": "and trust as JO's core guarantees not"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:05:19",
                    "end": "01:05:26",
                    "text": "afterthought. Fourth, we will provide we"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:05:23",
                    "end": "01:05:32",
                    "text": "will prove that AI does not take away"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:05:26",
                    "end": "01:05:33",
                    "text": "jobs rather it will create new highskll"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:05:32",
                    "end": "01:05:38",
                    "text": "work opportunities."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:05:33",
                    "end": "01:05:41",
                    "text": "And fifth, the AI story has shifted from"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:05:38",
                    "end": "01:05:45",
                    "text": "who has the best model to who can build"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:05:41",
                    "end": "01:05:49",
                    "text": "the strongest ecosystem for speed and"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:05:45",
                    "end": "01:05:52",
                    "text": "scale of usage. Therefore, we will build"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:05:49",
                    "end": "01:05:57",
                    "text": "deep partnership ecosystem with Indian"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:05:52",
                    "end": "01:05:59",
                    "text": "enterprises, startups, IIT, IISC and"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:05:57",
                    "end": "01:06:02",
                    "text": "research institutions. We will work"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:05:59",
                    "end": "01:06:05",
                    "text": "shoulderto-shoulder with India's leading"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:06:02",
                    "end": "01:06:08",
                    "text": "industrial groups to embed AI across"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:06:05",
                    "end": "01:06:10",
                    "text": "manufacturing, logistics, energy,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:06:08",
                    "end": "01:06:13",
                    "text": "finance, retail, agriculture, and"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:06:10",
                    "end": "01:06:14",
                    "text": "healthcare. We will empower startups"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:06:13",
                    "end": "01:06:18",
                    "text": "with affordable compute and"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:06:14",
                    "end": "01:06:20",
                    "text": "co-development platforms. We will aspire"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:06:18",
                    "end": "01:06:23",
                    "text": "to produce global breakthroughs in"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:06:20",
                    "end": "01:06:26",
                    "text": "compute architecture, foundation models"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:06:23",
                    "end": "01:06:28",
                    "text": "and energy efficiency designed in India,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:06:26",
                    "end": "01:06:31",
                    "text": "rooted in our values, powered by our"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:06:28",
                    "end": "01:06:34",
                    "text": "talent and scaled for humanity. And we"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:06:31",
                    "end": "01:06:36",
                    "text": "will partner with the very best tech"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:06:34",
                    "end": "01:06:40",
                    "text": "companies in the world. Not as importers"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:06:36",
                    "end": "01:06:44",
                    "text": "of intelligence but as co-architects"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:06:40",
                    "end": "01:06:47",
                    "text": "of a new AI century."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:06:44",
                    "end": "01:06:49",
                    "text": "Dear friends, I believe that social"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:06:47",
                    "end": "01:06:53",
                    "text": "relevance not momentary craze should"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:06:49",
                    "end": "01:06:56",
                    "text": "drive AI growth in India. Jio has"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:06:53",
                    "end": "01:06:58",
                    "text": "already started AI applications for the"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:06:56",
                    "end": "01:07:02",
                    "text": "most pressing challenges in inclusive"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:06:58",
                    "end": "01:07:05",
                    "text": "development. In education, we have Gio"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:07:02",
                    "end": "01:07:08",
                    "text": "Shikshak, an adaptive AI teaching"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:07:05",
                    "end": "01:07:11",
                    "text": "assistant in 22 languages. When 250"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:07:08",
                    "end": "01:07:13",
                    "text": "million school children and 50 million"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:07:11",
                    "end": "01:07:16",
                    "text": "students in higher education are"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:07:13",
                    "end": "01:07:20",
                    "text": "empowered by AI teachers,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:07:16",
                    "end": "01:07:23",
                    "text": "no power on earth can match India's"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:07:20",
                    "end": "01:07:27",
                    "text": "talent wealth. In healthcare, Gio"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:07:23",
                    "end": "01:07:29",
                    "text": "arrogia AI delivering first medical"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:07:27",
                    "end": "01:07:33",
                    "text": "guidance in under five minutes in local"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:07:29",
                    "end": "01:07:36",
                    "text": "languages on any phone. in agriculture,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:07:33",
                    "end": "01:07:39",
                    "text": "geoci converting satellite imagery and"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:07:36",
                    "end": "01:07:42",
                    "text": "precision weather into simple voice"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:07:39",
                    "end": "01:07:45",
                    "text": "first advice to 140 million farmers to"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:07:42",
                    "end": "01:07:49",
                    "text": "help improve their income in everyday"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:07:45",
                    "end": "01:07:53",
                    "text": "life. Gio Bharat IQ a voice first AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:07:49",
                    "end": "01:07:57",
                    "text": "companion helping Indians learn earn and"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:07:53",
                    "end": "01:08:01",
                    "text": "access government services at bhat scale"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:07:57",
                    "end": "01:08:04",
                    "text": "from variables to fully connected homes"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:08:01",
                    "end": "01:08:07",
                    "text": "geo frames"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:08:04",
                    "end": "01:08:10",
                    "text": "an AI glass device and next generation"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:08:07",
                    "end": "01:08:15",
                    "text": "AI devices will make intelligence truly"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:08:10",
                    "end": "01:08:17",
                    "text": "ambient as effortless and as natural as"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:08:15",
                    "end": "01:08:21",
                    "text": "human conversation. ations. Through geo"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:08:17",
                    "end": "01:08:23",
                    "text": "hot star AI will multiply Indian"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:08:21",
                    "end": "01:08:26",
                    "text": "creativity with multilingual"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:08:23",
                    "end": "01:08:29",
                    "text": "storytelling. We will popularize India's"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:08:26",
                    "end": "01:08:32",
                    "text": "rich cultural heritage with futuristic"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:08:29",
                    "end": "01:08:37",
                    "text": "technology enhancing India's soft power"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:08:32",
                    "end": "01:08:40",
                    "text": "globally. Friends, this inaugural global"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:08:37",
                    "end": "01:08:44",
                    "text": "AI impact summit in India has received a"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:08:40",
                    "end": "01:08:47",
                    "text": "massive response. What does that show?"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:08:44",
                    "end": "01:08:50",
                    "text": "It shows that AI is now becoming a"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:08:47",
                    "end": "01:08:53",
                    "text": "people's movement worldwide. The success"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:08:50",
                    "end": "01:08:57",
                    "text": "of this movement hinges critically on"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:08:53",
                    "end": "01:09:00",
                    "text": "global cooperation and not polarization."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:08:57",
                    "end": "01:09:03",
                    "text": "Be it chips or rare earths, AI works its"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:09:00",
                    "end": "01:09:06",
                    "text": "magic through sharing not holding"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:09:03",
                    "end": "01:09:08",
                    "text": "through collaborations not conflicts."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:09:06",
                    "end": "01:09:11",
                    "text": "The unique strength of India is that"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:09:08",
                    "end": "01:09:13",
                    "text": "India serves as the vital bridge"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:09:11",
                    "end": "01:09:17",
                    "text": "connecting the global south and the"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:09:13",
                    "end": "01:09:20",
                    "text": "global north. After all, south or north,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:09:17",
                    "end": "01:09:24",
                    "text": "east or west, all of us have only one"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:09:20",
                    "end": "01:09:27",
                    "text": "earth, one family and one future. Today"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:09:24",
                    "end": "01:09:30",
                    "text": "at this summit, let us all pledge to"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:09:27",
                    "end": "01:09:34",
                    "text": "transform this noble aspiration into"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:09:30",
                    "end": "01:09:37",
                    "text": "reality using the most powerful gift of"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:09:34",
                    "end": "01:09:40",
                    "text": "the human mind, AI. Let us combine"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:09:37",
                    "end": "01:09:42",
                    "text": "intelligence with empathy and let us"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:09:40",
                    "end": "01:09:46",
                    "text": "build a better future for all. Thank"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:09:42",
                    "end": "01:09:46",
                    "text": "you. Jim,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:09:46",
                    "end": "01:09:53",
                    "text": "distinguished guests, my fellow Indians,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:09:50",
                    "end": "01:09:56",
                    "text": "namaste."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:09:53",
                    "end": "01:10:00",
                    "text": "The global AI impact summit is a"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:09:56",
                    "end": "01:10:06",
                    "text": "defining moment in India's tech history."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:10:00",
                    "end": "01:10:09",
                    "text": "A moment when India pledges to make AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:10:06",
                    "end": "01:10:14",
                    "text": "one of the driving forces to realize its"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:10:09",
                    "end": "01:10:17",
                    "text": "dream of a Vixs Bharat, the dream of"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:10:14",
                    "end": "01:10:19",
                    "text": "becoming a fully developed nation by"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:10:17",
                    "end": "01:10:21",
                    "text": "2047,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:10:19",
                    "end": "01:10:24",
                    "text": "the glorious centenary of our"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:10:21",
                    "end": "01:10:26",
                    "text": "independence."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:10:24",
                    "end": "01:10:28",
                    "text": "We are deeply honored that our most"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:10:26",
                    "end": "01:10:31",
                    "text": "respected prime minister Shri Narendra"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:10:28",
                    "end": "01:10:35",
                    "text": "Modi is the guide, philosopher and"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:10:31",
                    "end": "01:10:37",
                    "text": "leader of this summit."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:10:35",
                    "end": "01:10:40",
                    "text": "Honorable Minister Shriishnau and his"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:10:37",
                    "end": "01:10:44",
                    "text": "wonderful team deserve full praise for"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:10:40",
                    "end": "01:10:49",
                    "text": "organizing this summit on a grand scale"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:10:44",
                    "end": "01:10:49",
                    "text": "befitting India's ambition."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:10:51",
                    "end": "01:10:59",
                    "text": "Miji's vision of AI powered vixed bharat"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:10:55",
                    "end": "01:11:01",
                    "text": "is also a template for a vixith global"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:10:59",
                    "end": "01:11:04",
                    "text": "south."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:11:01",
                    "end": "01:11:08",
                    "text": "If wisely used, I believe AI can usher"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:11:04",
                    "end": "01:11:12",
                    "text": "in an era of super abundance."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:11:08",
                    "end": "01:11:15",
                    "text": "A world without poverty and a future of"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:11:12",
                    "end": "01:11:19",
                    "text": "prosperity for all the 8 billion people"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:11:15",
                    "end": "01:11:23",
                    "text": "on our beautiful planet is now within"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:11:19",
                    "end": "01:11:25",
                    "text": "sight within reach. Friends,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:11:23",
                    "end": "01:11:28",
                    "text": "artificial intelligence is not just"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:11:25",
                    "end": "01:11:31",
                    "text": "another technology."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:11:28",
                    "end": "01:11:34",
                    "text": "For the first time, humans are creating"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:11:31",
                    "end": "01:11:39",
                    "text": "human-like systems that can learn,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:11:34",
                    "end": "01:11:40",
                    "text": "speak, analyze, move, and produce"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:11:39",
                    "end": "01:11:45",
                    "text": "autonomously."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:11:40",
                    "end": "01:11:48",
                    "text": "AI is the mantra that powers every"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:11:45",
                    "end": "01:11:53",
                    "text": "yantra or every machine and system to"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:11:48",
                    "end": "01:11:58",
                    "text": "work faster, better, and smarter."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:11:53",
                    "end": "01:12:00",
                    "text": "IC AI as a modernday akshaiatra"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:11:58",
                    "end": "01:12:04",
                    "text": "the legendary vessel in Madharat that"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:12:00",
                    "end": "01:12:08",
                    "text": "provided endless nourishment to all."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:12:04",
                    "end": "01:12:11",
                    "text": "Likewise AI offers limitless"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:12:08",
                    "end": "01:12:13",
                    "text": "augmentation in knowledge, efficiency"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:12:11",
                    "end": "01:12:18",
                    "text": "and productivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:12:13",
                    "end": "01:12:22",
                    "text": "We are only at the dawn of this era. The"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:12:18",
                    "end": "01:12:26",
                    "text": "best of AI is yet to come. Distinguished"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:12:22",
                    "end": "01:12:29",
                    "text": "participants in this summit today the"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:12:26",
                    "end": "01:12:30",
                    "text": "world is defound is debating a profound"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:12:29",
                    "end": "01:12:33",
                    "text": "question."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:12:30",
                    "end": "01:12:36",
                    "text": "Will AI concentrate power in the hands"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:12:33",
                    "end": "01:12:40",
                    "text": "of a few or will it democratize"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:12:36",
                    "end": "01:12:45",
                    "text": "opportunity for all? Do we act as"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:12:40",
                    "end": "01:12:49",
                    "text": "isolated nations or as a united global"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:12:45",
                    "end": "01:12:51",
                    "text": "family? Our polarized world stands at a"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:12:49",
                    "end": "01:12:55",
                    "text": "fork."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:12:51",
                    "end": "01:12:58",
                    "text": "One path has led to a situation where AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:12:55",
                    "end": "01:13:01",
                    "text": "is scarce and expensive, compute is"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:12:58",
                    "end": "01:13:04",
                    "text": "concentrated, data is controlled and"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:13:01",
                    "end": "01:13:08",
                    "text": "capability is locked behind barriers of"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:13:04",
                    "end": "01:13:11",
                    "text": "capital and geography in the global"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:13:08",
                    "end": "01:13:14",
                    "text": "north. In this dismal scenario,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:13:11",
                    "end": "01:13:18",
                    "text": "inequality widens between nations within"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:13:14",
                    "end": "01:13:22",
                    "text": "societies and across generations. But"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:13:18",
                    "end": "01:13:25",
                    "text": "there is another path. A future where AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:13:22",
                    "end": "01:13:29",
                    "text": "is available, affordable and beneficial"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:13:25",
                    "end": "01:13:32",
                    "text": "to all. As our prime minister said,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:13:29",
                    "end": "01:13:34",
                    "text": "India believes in this second future."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:13:32",
                    "end": "01:13:37",
                    "text": "Dear friends,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:13:34",
                    "end": "01:13:40",
                    "text": "from the podium of this summit today, I"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:13:37",
                    "end": "01:13:44",
                    "text": "want to make a bold prediction. India"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:13:40",
                    "end": "01:13:49",
                    "text": "will emerge as one of the greatest AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:13:44",
                    "end": "01:13:49",
                    "text": "powers in the world in the 21st century."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:13:52",
                    "end": "01:13:59",
                    "text": "My confidence is validated by an"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:13:56",
                    "end": "01:14:01",
                    "text": "undeniable truth. In the coming decades,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:13:59",
                    "end": "01:14:05",
                    "text": "no country in the world can match"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:14:01",
                    "end": "01:14:07",
                    "text": "India's strength in demography,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:14:05",
                    "end": "01:14:11",
                    "text": "democracy, development, digital"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:14:07",
                    "end": "01:14:14",
                    "text": "infrastructure, data generation, AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:14:11",
                    "end": "01:14:16",
                    "text": "harvest. Let me begin as was said many"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:14:14",
                    "end": "01:14:20",
                    "text": "times this morning, what India has"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:14:16",
                    "end": "01:14:23",
                    "text": "achieved in the past 10 years. First,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:14:20",
                    "end": "01:14:26",
                    "text": "India is the world's largest mobile data"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:14:23",
                    "end": "01:14:29",
                    "text": "consumer. Nearly 1 billion internet"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:14:26",
                    "end": "01:14:32",
                    "text": "users, data costs among the lowest"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:14:29",
                    "end": "01:14:35",
                    "text": "globally and in terms of quality there"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:14:32",
                    "end": "01:14:39",
                    "text": "is no difference between Delhi and the"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:14:35",
                    "end": "01:14:43",
                    "text": "remotest Indian village. Second, Aadhaar"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:14:39",
                    "end": "01:14:46",
                    "text": "1.4 billion digital IDs. Third, UPI"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:14:43",
                    "end": "01:14:49",
                    "text": "processes over 12 billion transactions"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:14:46",
                    "end": "01:14:52",
                    "text": "monthly. Fourth, India ranks among the"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:14:49",
                    "end": "01:14:56",
                    "text": "top three startup ecosystems with"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:14:52",
                    "end": "01:14:59",
                    "text": "100,000 startups and 100 plus unicorns."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:14:56",
                    "end": "01:15:02",
                    "text": "Fifth, India's secure and inclusive"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:14:59",
                    "end": "01:15:05",
                    "text": "digital public infrastructure stack is"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:15:02",
                    "end": "01:15:09",
                    "text": "now being adopted by countries around"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:15:05",
                    "end": "01:15:13",
                    "text": "the globe. Friends, in all humility, I"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:15:09",
                    "end": "01:15:16",
                    "text": "wish to state that Jio with over 500"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:15:13",
                    "end": "01:15:18",
                    "text": "million loyal subscribers was privileged"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:15:16",
                    "end": "01:15:22",
                    "text": "to play a leading role in this"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:15:18",
                    "end": "01:15:24",
                    "text": "transformation across broadband, 4G, 5G,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:15:22",
                    "end": "01:15:27",
                    "text": "and home connectivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:15:24",
                    "end": "01:15:30",
                    "text": "With equal humility, I would like to"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:15:27",
                    "end": "01:15:33",
                    "text": "announce that Jio will play an even"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:15:30",
                    "end": "01:15:36",
                    "text": "bigger role in India's AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:15:33",
                    "end": "01:15:38",
                    "text": "transformation. Today on behalf of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:15:36",
                    "end": "01:15:41",
                    "text": "Reliance Group and Geo Intelligence I"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:15:38",
                    "end": "01:15:44",
                    "text": "want to make three announcements."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:15:41",
                    "end": "01:15:48",
                    "text": "Announcement one, Jio connected India to"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:15:44",
                    "end": "01:15:53",
                    "text": "the internet era. Jio will now connect"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:15:48",
                    "end": "01:15:53",
                    "text": "India to the intelligence era."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:15:54",
                    "end": "01:15:60",
                    "text": "We will deliver intelligence to every"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:15:56",
                    "end": "01:16:02",
                    "text": "citizen, every sector of the economy and"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:15:60",
                    "end": "01:16:05",
                    "text": "every facet of social development and"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:16:02",
                    "end": "01:16:09",
                    "text": "every service of government. Jio will do"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:16:05",
                    "end": "01:16:12",
                    "text": "so with the same reliability, quality,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:16:09",
                    "end": "01:16:14",
                    "text": "scale and extreme affordability that"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:16:12",
                    "end": "01:16:16",
                    "text": "transformed connectivity."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:16:14",
                    "end": "01:16:19",
                    "text": "India cannot afford to rent"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:16:16",
                    "end": "01:16:23",
                    "text": "intelligence. Therefore, we will reduce"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:16:19",
                    "end": "01:16:28",
                    "text": "the cost of intelligence as dramatically"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:16:23",
                    "end": "01:16:28",
                    "text": "as we did the cost of data."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:16:31",
                    "end": "01:16:39",
                    "text": "Announcement two, Jio together with"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:16:34",
                    "end": "01:16:44",
                    "text": "Reliance will invest 10 lakh crores over"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:16:39",
                    "end": "01:16:44",
                    "text": "the next 7 years starting this year."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:16:46",
                    "end": "01:16:51",
                    "text": "This is not speculative investment. It"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:16:49",
                    "end": "01:16:54",
                    "text": "is not purchasing valuation."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:16:51",
                    "end": "01:16:57",
                    "text": "This is patient, disciplined,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:16:54",
                    "end": "01:17:01",
                    "text": "nationbuilding capital designed to"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:16:57",
                    "end": "01:17:03",
                    "text": "create durable economic value and"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:17:01",
                    "end": "01:17:05",
                    "text": "strategic resilience for decades to"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:17:03",
                    "end": "01:17:07",
                    "text": "come."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:17:05",
                    "end": "01:17:11",
                    "text": "Distinguished participants, the biggest"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:17:07",
                    "end": "01:17:14",
                    "text": "constraint in AI today is not talent or"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:17:11",
                    "end": "01:17:18",
                    "text": "imagination. It is scarcity and high"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:17:14",
                    "end": "01:17:22",
                    "text": "cost of compute. Therefore, here is my"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:17:18",
                    "end": "01:17:24",
                    "text": "third announcement. Geo intelligence"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:17:22",
                    "end": "01:17:27",
                    "text": "will build India's sovereign compute"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:17:24",
                    "end": "01:17:31",
                    "text": "infrastructure through three bold"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:17:27",
                    "end": "01:17:34",
                    "text": "initiatives. One, gigawatt scale data"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:17:31",
                    "end": "01:17:37",
                    "text": "centers. We already started construction"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:17:34",
                    "end": "01:17:41",
                    "text": "on multi- gigawatt AI ready data centers"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:17:37",
                    "end": "01:17:46",
                    "text": "at Jamnagar. Over 120 megawatts will"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:17:41",
                    "end": "01:17:49",
                    "text": "come online in the second half of 2026"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:17:46",
                    "end": "01:17:51",
                    "text": "this year and a clear path to gigawatt"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:17:49",
                    "end": "01:17:54",
                    "text": "scale compute for training and large"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:17:51",
                    "end": "01:17:55",
                    "text": "scale inference."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:17:54",
                    "end": "01:17:59",
                    "text": "Two,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:17:55",
                    "end": "01:18:01",
                    "text": "our green energy advantage. We have an"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:17:59",
                    "end": "01:18:05",
                    "text": "in-house energy advantage with up to 10"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:18:01",
                    "end": "01:18:08",
                    "text": "gawatt of ready green power"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:18:05",
                    "end": "01:18:10",
                    "text": "surplus anchored by solar in both Kach"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:18:08",
                    "end": "01:18:14",
                    "text": "and Andhra Pradesh."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:18:10",
                    "end": "01:18:17",
                    "text": "Three, a nationwide edge compute. An"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:18:14",
                    "end": "01:18:20",
                    "text": "edge compute layer deeply integrated"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:18:17",
                    "end": "01:18:23",
                    "text": "with Jio's network will make"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:18:20",
                    "end": "01:18:26",
                    "text": "intelligence responsive, low latency and"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:18:23",
                    "end": "01:18:30",
                    "text": "affordable close to where Indians live,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:18:26",
                    "end": "01:18:33",
                    "text": "learn and work. From kirana stores to"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:18:30",
                    "end": "01:18:36",
                    "text": "clinics, from classrooms to farms,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:18:33",
                    "end": "01:18:40",
                    "text": "intelligence will live at the edge. Our"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:18:36",
                    "end": "01:18:44",
                    "text": "resolve is clear. Make intelligence as"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:18:40",
                    "end": "01:18:47",
                    "text": "ubiquitous as connectivity. When compute"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:18:44",
                    "end": "01:18:49",
                    "text": "becomes infrastructure, innovation will"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:18:47",
                    "end": "01:18:52",
                    "text": "become inevitable."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:18:49",
                    "end": "01:18:56",
                    "text": "Friends, geo intelligence is guided by"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:18:52",
                    "end": "01:18:59",
                    "text": "five non-negotiable principles. First,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:18:56",
                    "end": "01:19:02",
                    "text": "AI for India's deep tech and advanced"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:18:59",
                    "end": "01:19:05",
                    "text": "manufacturing leadership reaching not"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:19:02",
                    "end": "01:19:06",
                    "text": "just large enterprises but agriculture,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:19:05",
                    "end": "01:19:10",
                    "text": "small businesses and the informal"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:19:06",
                    "end": "01:19:13",
                    "text": "sector. Geo intelligence will not simply"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:19:10",
                    "end": "01:19:16",
                    "text": "be a search or an ask tool. It will"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:19:13",
                    "end": "01:19:18",
                    "text": "primarily be a resource for multiplying"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:19:16",
                    "end": "01:19:22",
                    "text": "productivity and efficiency."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:19:18",
                    "end": "01:19:26",
                    "text": "Second, worldleading multilingual AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:19:22",
                    "end": "01:19:28",
                    "text": "capability across all Indian languages."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:19:26",
                    "end": "01:19:32",
                    "text": "When farmers and articians speak to AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:19:28",
                    "end": "01:19:34",
                    "text": "in their own words and student learn in"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:19:32",
                    "end": "01:19:38",
                    "text": "their own mother tongue, this is not"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:19:34",
                    "end": "01:19:38",
                    "text": "convenience. This is inclusion."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:19:51",
                    "end": "01:19:56",
                    "text": "Third,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:19:53",
                    "end": "01:19:60",
                    "text": "responsibility, security, data residency"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:19:56",
                    "end": "01:20:04",
                    "text": "and trust as GO's core guarantees not"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:19:60",
                    "end": "01:20:08",
                    "text": "afterthought. Fourth, we will provide we"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:20:04",
                    "end": "01:20:11",
                    "text": "will prove that AI does not take away"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:20:08",
                    "end": "01:20:14",
                    "text": "jobs. Rather, it will create new"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:20:11",
                    "end": "01:20:19",
                    "text": "highskll work opportunities."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:20:14",
                    "end": "01:20:22",
                    "text": "And fifth, the AI story has shifted from"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:20:19",
                    "end": "01:20:26",
                    "text": "who has the best model to who can build"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:20:22",
                    "end": "01:20:30",
                    "text": "the strongest ecosystem for speed and"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:20:26",
                    "end": "01:20:33",
                    "text": "scale of usage. Therefore, we will build"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:20:30",
                    "end": "01:20:37",
                    "text": "deep partnership ecosystem with Indian"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:20:33",
                    "end": "01:20:40",
                    "text": "enterprises, startups, IIT, IISC and"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:20:37",
                    "end": "01:20:42",
                    "text": "research institutions. We will work"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:20:40",
                    "end": "01:20:46",
                    "text": "shoulderto-shoulder with India's leading"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:20:42",
                    "end": "01:20:48",
                    "text": "industrial groups to embed AI across"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:20:46",
                    "end": "01:20:51",
                    "text": "manufacturing, logistics, energy,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:20:48",
                    "end": "01:20:53",
                    "text": "finance, retail, agriculture, and"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:20:51",
                    "end": "01:20:55",
                    "text": "healthcare. We will empower startups"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:20:53",
                    "end": "01:20:58",
                    "text": "with affordable compute and"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:20:55",
                    "end": "01:21:01",
                    "text": "co-development platforms. We will aspire"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:20:58",
                    "end": "01:21:03",
                    "text": "to produce global breakthroughs in"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:21:01",
                    "end": "01:21:06",
                    "text": "compute architecture, foundation models"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:21:03",
                    "end": "01:21:09",
                    "text": "and energy efficiency designed in India,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:21:06",
                    "end": "01:21:12",
                    "text": "rooted in our values, powered by our"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:21:09",
                    "end": "01:21:14",
                    "text": "talent and scaled for humanity. And we"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:21:12",
                    "end": "01:21:17",
                    "text": "will partner with the very best tech"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:21:14",
                    "end": "01:21:21",
                    "text": "companies in the world, not as importers"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:21:17",
                    "end": "01:21:24",
                    "text": "of intelligence, but as co-architects of"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:21:21",
                    "end": "01:21:27",
                    "text": "a new AI century."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:21:24",
                    "end": "01:21:29",
                    "text": "Dear friends, I believe that social"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:21:27",
                    "end": "01:21:33",
                    "text": "relevance, not momentary craze should"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:21:29",
                    "end": "01:21:36",
                    "text": "drive AI growth in India. Jio has"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:21:33",
                    "end": "01:21:38",
                    "text": "already started AI applications for the"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:21:36",
                    "end": "01:21:42",
                    "text": "most pressing challenges in inclusive"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:21:38",
                    "end": "01:21:45",
                    "text": "development. In education, we have Gio"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:21:42",
                    "end": "01:21:49",
                    "text": "Shikshak, an adaptive AI teaching"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:21:45",
                    "end": "01:21:52",
                    "text": "assistant in 22 languages. When 250"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:21:49",
                    "end": "01:21:54",
                    "text": "million school children and 50 million"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:21:52",
                    "end": "01:21:58",
                    "text": "students in higher education are"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:21:54",
                    "end": "01:22:01",
                    "text": "empowered by AI teachers, no power on"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:21:58",
                    "end": "01:22:06",
                    "text": "earth can match India's talent wealth."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:22:01",
                    "end": "01:22:08",
                    "text": "In healthcare, Gio arrogia AI delivering"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:22:06",
                    "end": "01:22:12",
                    "text": "first medical guidance in under five"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:22:08",
                    "end": "01:22:15",
                    "text": "minutes in local languages on any phone."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:22:12",
                    "end": "01:22:18",
                    "text": "in agriculture. Geocishi converting"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:22:15",
                    "end": "01:22:21",
                    "text": "satellite imagery and precision weather"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:22:18",
                    "end": "01:22:24",
                    "text": "into simple voice first advice to 140"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:22:21",
                    "end": "01:22:29",
                    "text": "million farmers to help improve their"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:22:24",
                    "end": "01:22:32",
                    "text": "income in everyday life. Gio Bharat IQ a"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:22:29",
                    "end": "01:22:35",
                    "text": "voice first AI companion helping Indians"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:22:32",
                    "end": "01:22:39",
                    "text": "learn earn and access government"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:22:35",
                    "end": "01:22:42",
                    "text": "services at bhat scale from variables to"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:22:39",
                    "end": "01:22:44",
                    "text": "fully connected homes"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:22:42",
                    "end": "01:22:48",
                    "text": "geo frames"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:22:44",
                    "end": "01:22:51",
                    "text": "an AI glass device and next generation"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:22:48",
                    "end": "01:22:56",
                    "text": "AI devices will make intelligence truly"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:22:51",
                    "end": "01:22:58",
                    "text": "ambient as effortless and as natural as"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:22:56",
                    "end": "01:23:01",
                    "text": "human convers conversations through geo"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:22:58",
                    "end": "01:23:04",
                    "text": "hot star AI will multiply Indian"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:23:01",
                    "end": "01:23:07",
                    "text": "creativity with multilingual"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:23:04",
                    "end": "01:23:10",
                    "text": "storytelling. We will popularize India's"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:23:07",
                    "end": "01:23:13",
                    "text": "rich cultural heritage with futuristic"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:23:10",
                    "end": "01:23:18",
                    "text": "technology enhancing India's soft power"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:23:13",
                    "end": "01:23:21",
                    "text": "globally. Friends, this inaugural global"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:23:18",
                    "end": "01:23:25",
                    "text": "AI impact summit in India has received a"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:23:21",
                    "end": "01:23:28",
                    "text": "massive response. What does that show?"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:23:25",
                    "end": "01:23:31",
                    "text": "It shows that AI is now becoming a"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:23:28",
                    "end": "01:23:34",
                    "text": "people's movement worldwide. The success"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:23:31",
                    "end": "01:23:38",
                    "text": "of this movement hinges critically on"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:23:34",
                    "end": "01:23:41",
                    "text": "global cooperation and not polarization."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:23:38",
                    "end": "01:23:44",
                    "text": "Be it chips or rare earths, AI works its"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:23:41",
                    "end": "01:23:46",
                    "text": "magic through sharing not holding"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:23:44",
                    "end": "01:23:49",
                    "text": "through collaborations not conflicts."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:23:46",
                    "end": "01:23:52",
                    "text": "The unique strength of India is that"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:23:49",
                    "end": "01:23:54",
                    "text": "India serves as the vital bridge"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:23:52",
                    "end": "01:23:58",
                    "text": "connecting the global south and the"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:23:54",
                    "end": "01:24:01",
                    "text": "global north. After all, south or north,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:23:58",
                    "end": "01:24:05",
                    "text": "east or west, all of us have only one"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:24:01",
                    "end": "01:24:08",
                    "text": "earth, one family, and one future. Today"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:24:05",
                    "end": "01:24:11",
                    "text": "at this summit, let us all pledge to"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:24:08",
                    "end": "01:24:14",
                    "text": "transform this noble aspiration into"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:24:11",
                    "end": "01:24:18",
                    "text": "reality using the most powerful gift of"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:24:14",
                    "end": "01:24:20",
                    "text": "the human mind, AI. Let us combine"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:24:18",
                    "end": "01:24:23",
                    "text": "intelligence with empathy and let us"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:24:20",
                    "end": "01:24:27",
                    "text": "build a better future for all. Thank"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:24:23",
                    "end": "01:24:27",
                    "text": "you, Jin."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:24:27",
                    "end": "01:24:34",
                    "text": "Distinguished guests, my fellow Indians,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:24:31",
                    "end": "01:24:37",
                    "text": "namaste."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:24:34",
                    "end": "01:24:41",
                    "text": "The global AI impact summit is a"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:24:37",
                    "end": "01:24:47",
                    "text": "defining moment in India's tech history."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:24:41",
                    "end": "01:24:50",
                    "text": "A moment when India pledges to make AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:24:47",
                    "end": "01:24:55",
                    "text": "one of the driving forces to realize its"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:24:50",
                    "end": "01:24:58",
                    "text": "dream of a Vixs Bharat, the dream of"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:24:55",
                    "end": "01:24:60",
                    "text": "becoming a fully developed nation by"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:24:58",
                    "end": "01:25:02",
                    "text": "2047,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:24:60",
                    "end": "01:25:04",
                    "text": "the glorious centenary of our"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:25:02",
                    "end": "01:25:06",
                    "text": "independence."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:25:04",
                    "end": "01:25:09",
                    "text": "We are deeply honored that our most"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:25:06",
                    "end": "01:25:09",
                    "text": "respected prime minister"
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
}, {
    "_id": "6996afb1565455da1fc75c32",
    "projectId": "267910",
    "projectName": "Inaugural Session AI-impact-summit-2026- Spanish ",
    "projectDescription": "Inaugural Session - Spanish",
    "projectCreatedDate": "2026-02-19T06:37:34.181Z",
    "projectUpdatedDate": "2026-02-19T06:37:34.181Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-19T06:38:43.406Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Inaugral Session.pdf",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Spanish"
},
    {
    "_id": "6996af4f565455da1fc75c18",
    "projectId": "996134",
    "projectName": "Inaugural Session -Hindi",
    "projectDescription": "Inaugural Session -Hindi",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T06:35:56.245Z",
    "projectUpdatedDate": "2026-02-19T06:35:56.245Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-19T06:37:02.084Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Inaugral Session.pdf",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Hindi"
},
    {
    "_id": "6996aaef565455da1fc75bda",
    "projectId": "207027",
    "projectName": "Inaugural Session | AI-impact-summit-2026",
    "projectDescription": "Inaugural Session | AI-impact-summit-2026",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-19T06:17:16.320Z",
    "projectUpdatedDate": "2026-02-19T06:17:16.320Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-19T06:17:59.506Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Inaugral Session.pdf",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6995b917565455da1fc759b7",
    "projectId": "851116",
    "projectName": "India_s Intelligence Infrastructure for Sovereign AI Impact Summit India 2026 -Japanese",
    "projectDescription": "India_s Intelligence Infrastructure for Sovereign AI Impact Summit India 2026 -Japanese",
     "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T13:05:22.161Z",
    "projectUpdatedDate": "2026-02-18T13:05:22.161Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-18T13:17:23.814Z",
    "projectDeleted": false,
    "projectDataLength": 3,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "India_s Intelligence Infrastructure for Sovereign AI.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "India_s Intelligence Infrastructure for Sovereign AI.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Japanese"
},
{
    "_id": "6995b8be565455da1fc759a5",
    "projectId": "167916",
    "projectName": "India_s Intelligence Infrastructure for Sovereign AI Impact Summit India 2026",
    "projectDescription": "India_s Intelligence Infrastructure for Sovereign AI Impact Summit India 2026",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T13:03:55.929Z",
    "projectUpdatedDate": "2026-02-18T13:03:55.929Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-18T13:17:59.248Z",
    "projectDeleted": false,
    "projectDataLength": 3,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "India_s Intelligence Infrastructure for Sovereign AI.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "India_s Intelligence Infrastructure for Sovereign AI.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
  
{
    "_id": "6995b971565455da1fc759c7",
    "projectId": "342713",
    "projectName": "India_s Intelligence Infrastructure for Sovereign AI Impact Summit India 2026 - Hindi",
    "projectDescription": "India_s Intelligence Infrastructure for Sovereign AI Impact Summit India 2026 - Hindi",
     "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T13:06:55.353Z",
    "projectUpdatedDate": "2026-02-18T13:06:55.353Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-18T13:16:44.696Z",
    "projectDeleted": false,
    "projectDataLength": 3,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "India_s Intelligence Infrastructure for Sovereign AI.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "India_s Intelligence Infrastructure for Sovereign AI.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Hindi"
},
   {
    "_id": "6995b518565455da1fc758f1",
    "projectId": "309147",
    "projectName": "Empowering Courts with AI Tools Insights AI Impact Summit India 2026 - Japanese",
    "projectDescription": "Empowering Courts with AI Tools Insights AI Impact Summit India 2026",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T12:48:22.100Z",
    "projectUpdatedDate": "2026-02-18T12:48:22.100Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-18T12:49:21.810Z",
    "projectDeleted": false,
    "projectDataLength": 3,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "New Text Document (3).txt",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "New Text Document (3).txt",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Japanese"
}, 
    {
    "_id": "6995b467565455da1fc758bd",
    "projectId": "780686",
    "projectName": "Empowering Courts with AI Tools Insights AI Impact Summit India 2026 - Hindi",
    "projectDescription": "Empowering Courts with AI Tools Insights AI Impact Summit India 2026",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T12:45:24.757Z",
    "projectUpdatedDate": "2026-02-18T12:45:24.757Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-18T12:46:22.780Z",
    "projectDeleted": false,
    "projectDataLength": 3,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "New Text Document (3).txt",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "New Text Document (3).txt",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Hindi"
},
{
    "_id": "6995b325565455da1fc75871",
    "projectId": "174143",
    "projectName": "Empowering Courts with AI Tools Insights Impact AI Impact Summit India 2026 - Hindi",
    "projectDescription": "Empowering Courts with AI Tools Insights Impact AI Impact Summit India 2026",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T12:40:03.493Z",
    "projectUpdatedDate": "2026-02-18T12:40:03.493Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-18T12:43:06.582Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "Empowering Courts with AI_ Tools_ Insights _ Impact(MP3_128K).mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771418362085.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:28",
                    "text": " G and human rights related to AI. Today we had a different scenario. We had our assistant director general opening the session and then launching one of the most important briefs that we have so far. He's on his way, stuck at another meeting as all of us are. So we're gonna start with a panel discussion to start and then we will switch and then go to the launch afterwards. So we're gonna start with our esteemed panelists today. We have..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:28",
                    "end": "00:00:58",
                    "text": " We have three distinguished panelists that are joining us and we're going to start with Ms. Jalak and I'll introduce each of our speakers as we go ahead also to save time. So Ms. Jalaka Khar, Executive Director, Center for Communication Governance at NLU Delhi. Jalak, India's judiciary faces both first huge overloads in terms of case loads and at the same time immense technological opportunities and potential. What are the key ethical,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:58",
                    "end": "00:01:16",
                    "text": " or procedural, let's say, safeguards that you think India's courts should prioritize when engaging with AI systems, especially given the diversity and scale of the justice ecosystem? Thanks for that question, and thanks for organizing this and having me here today."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:17",
                    "end": "00:01:47",
                    "text": " If you look at AI integration into the judiciary, you can broadly say there are two areas it can be integrated. One is in sort of helping streamline administrative load of courts, whether that's scheduling of cases. Or it can be used in more substantive matters like research, translation of documents, or then even decision making. And we've seen a spectrum of applications being developed that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:47",
                    "end": "00:02:14",
                    "text": " that help across the cross-section of these tasks. Some have higher risk, such as decision making. Some have sort of downstream impacts on significant decision making, like if you're translating documents and if there are errors over there. And then, of course, it may seem that administrative decision making being taken over by AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:14",
                    "end": "00:02:40",
                    "text": " may not impact justice, but actually even use of AI and scheduling can impact justice, because AI deciding which case gets listed first, which case is extraneous, can sort of really impact someone's timelines for justice delivery. So I think we need to, of course, categorize AI use in judiciary across various risks."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:40",
                    "end": "00:03:03",
                    "text": " categories and then for each category we need to think about what is the level of assessment that needs to be undertaken. There are various risk assessment auditing mechanisms that are being established in the AI domain and there needs to be thought about how those can be translated and used in the judicial context. So I think."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:03",
                    "end": "00:03:26",
                    "text": " That sort of one piece is auditing and assessments for questions of bias, for questions of transparency, for questions of explainability. The second piece that I think we need to be thinking about is, judiciary sits typically in any country on a large trove of data. I know within the Indian judicial ecosystem,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:26",
                    "end": "00:03:53",
                    "text": " as part of the ecommittee process. The Supreme Court was really thinking about how do you unlock the value of that data and use it to streamline court processes? How can you give it to stakeholders in the ecosystem, whether it's startups, other public interest stakeholders, researchers, to sort of go through that data and figure out, okay, you know, these are the lessons and this is how we can streamline processes to move through caseload quickly."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:53",
                    "end": "00:04:18",
                    "text": " But that also means that you need to have very robust data management practices, data protection principles being applied, because this can, in many instances, be sensitive and personal data that is getting shared, mixed in with non-personal data. So I think that's the second piece the judiciary needs to think about. I think the third piece, and of course there's so many things, but these are my three, is."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:18",
                    "end": "00:04:47",
                    "text": " Judiciary doesn't have the expertise to build these things in-house. So they're going to rely on third-party vendors. So to engender public trust, I think public procurement has to happen in an open and transparent way. And there needs to be clear guidelines that are set up for what is the assessment of these systems on a regular basis by judiciary external auditors to engender public trust."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:47",
                    "end": "00:05:16",
                    "text": " in the use of AI in courts. Because the last thing you want to do is erode public trust in the court system through the integration of AI. So you have to carry the public along with you. Thank you. Just a quick follow up. If you can briefly tell us, how can India balance between this AI efficiency, AI-driven efficiency and constitutional protections, especially in light of this multilingual, multi-constitutional context?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:16",
                    "end": "00:05:45",
                    "text": " I think that's a great question and something we really need to think deeply about because I think there's immense pressure to adopt technology because there's so much case load and it's burgeoning. But I think we have to do it in a cautious way where efficiency doesn't trump due process, rule of law, rights of individuals. When you think about multilingual systems, the thing to think about is that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:45",
                    "end": "00:06:07",
                    "text": " A lot of these systems are based on things that are built for English context. So how are you going to translate it into use in context where you have low resource languages? And then there aren't necessarily enough data sets for low resource languages, and especially data sets that are primed."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:07",
                    "end": "00:06:32",
                    "text": " for use in more legal context. So I think that is definitely going to be a journey. Because if we want to ensure, as you're saying, the protection of constitutional rights of people and the true delivery of justice, you will need to balance that. I think one piece that we need to really think about is capacity building of judges. And I know that's something UNESCO has focused on heavily over the last, you know,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:32",
                    "end": "00:06:59",
                    "text": " several years, in fact, the last close to a decade under its AI and rule of law program at NLU Delhi at CCG. We were part of the first MOOC that was designed for judges. And I think you train thousands of judges around the world. We are on the technical committee for the second MOOC that you all are designing for judges with Oxford University. And we've also done trainings for South Asian judges."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:59",
                    "end": "00:07:28",
                    "text": " at NLU Delhi on AI and rule of law. I think bringing awareness to judges on what, that AI is not a neutral technology, that it's a socio-technical technology, and that it's not gonna make neutral decisions. If it's trained on biased data sets, it's going to give you biased output. And building that awareness of where AI is useful, but what its limitations and challenges can be is extremely important."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:28",
                    "end": "00:07:54",
                    "text": " a conference organized by the Supreme Court of India with the Supreme Court of Singapore. And they brought together a lot of Supreme Court judges, a lot of senior high court judges. This was a few years back. And what was very telling to me during that conference is these are extremely distinguished jurists, the top legal minds of our country, right? People who've been in this space for 30, 40, 50 sort of years."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:55",
                    "end": "00:08:11",
                    "text": " They had very, very fundamental questions on AI and they were really grappling with, okay, where is it appropriate to use these systems and where is it not? And I think we need to bridge that divide where people working on AI and technology need to sort of sit in the room."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:11",
                    "end": "00:08:26",
                    "text": " With a public interest lens, people working on it with a public interest lens need to sit in the room and really break down what are the risks we're seeing, what's appropriate, so that they don't need to reinvent the wheel. And I think UNESCO's playing a great role in bridging that divide."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:26",
                    "end": "00:08:53",
                    "text": " Thank you. Very interesting points here. I would like to go now, because you mentioned a very important component, which is the capacity building. But before we go to our next speaker, I'd like to introduce Professor Sri Krishna Deva Rao, Vice Chancellor for Naltzar University of Law. And my question to you is exactly what Jalak was saying. From an academic perspective, how should legal education for lawyers and for judges..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:53",
                    "end": "00:09:18",
                    "text": " find some sort of adaptation to prepare judges for this AI-driven justice environment. What skills, what ethical, perhaps, reflections should be also included? Just now, Jalak mentioned about, I'll start with the multilingualism, what she said, and then I'll come back to how the legal academia has to respond to, is a major..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:18",
                    "end": "00:09:42",
                    "text": " new age thinking as a relearning the law. But multilingual is an important which is India is a land of several languages and several dialects and of course the Anglo-Saxon legal system was able to reach out only to the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:42",
                    "end": "00:10:12",
                    "text": " communities, those who are privileged. And this is a challenge before the legal education. I think that may take another 10 or 20 years. How do we really the massive justice gap can be at least temporarily answered once the legal education moves into more and more into the language teaching in the mother tongues and other. That's where the AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:12",
                    "end": "00:10:42",
                    "text": " can really help, but that requires not just use of the AI translation into the several languages, but we require several experts in the legal academia, those who can understand, journalists, and several things. But the legal academia, I would look at when national law universities were started in India in 1980s, we used to say what we need is a professionally competent,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:42",
                    "end": "00:11:08",
                    "text": " socially relevant and technically sound that's what the slogan which started is aim and objective now we will say in addition to this we need technologically literate technologically equipped lawyers entire legal fraternity as a law teacher always we look at how the new curriculum has to be developed but when the curriculum is developed"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:08",
                    "end": "00:11:31",
                    "text": " We need to have the entire participation of the legal fraternity, the law firms, IT companies, all have to come together and to see how they can create a more challenging, more engaging curriculum and the pedagogy. For instance,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:31",
                    "end": "00:11:56",
                    "text": " We teach what we call in the legal academia, bread and butter courses, constitutional law, corporate law, criminal law, and several other various areas, property and several other. But now we need to think about introducing maybe second year onwards a specialized honors in the law technology with focus on the artificial intelligence."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:56",
                    "end": "00:12:26",
                    "text": " As someone who has been associated with the legal aid movement in India over three decades and the paralegal movement in India, I would say that it requires several levels of first, legal literacy. How the AI can help in the massive legal literacy which is required in several of the languages and using a new pedagogical method. Second is community dispute resolution."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:26",
                    "end": "00:12:50",
                    "text": " how the online dispute resolution with the participation of both lawyers, judges and the legal academia must help. In fact, generally in the legal education sometimes we drew parallels to the medical education. Because like a medical education, a medical college in India will not be given a permission unless you have a teaching hospital."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:50",
                    "end": "00:13:20",
                    "text": " Why don't we have a teaching course? AI, law and technology provides that an opportunity. For a consumer dispute, I wanted to settle because the public have a confidence in legal academia rather than any of the legal fraternity. Simple family disputes, I wanted to resolve. Land disputes. Because that's a major challenge of massive backlogs. We can start where legal academia takes the major responsibility and duty, not only creating a new curriculum."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:20",
                    "end": "00:13:43",
                    "text": " and pedagogical tools and method, where 40% in the classroom, 60% in the technological companies. That's a clinical legal education, clinical pedagogy can be used. Second, we need to train, orient the legal fraternity, starting from the law students first, and then the lawyers and judges and others. So I'll stop here."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:43",
                    "end": "00:13:55",
                    "text": " and I think I have taken sufficient time. Not at all. Thank you very much for your very relevant contribution because you're right, everything that we do has to have this multi-stakeholder approach."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:55",
                    "end": "00:14:17",
                    "text": " It cannot be done by only one side only, not only judicial institutes, but also across the board. Allow me to welcome our Assistant Director General, Dr. Taufit Jalassi, for joining us. We're gonna come to you. We have pushed the launch, actually, to the end of the session instead of the beginning. We have one more speaker here, Mr. Argiya."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:17",
                    "end": "00:14:44",
                    "text": " Bhattacharya, and I'm really sorry about my pronunciation, if I mispronounce any of the names, co-founder and CTO of Adalat AI. So, Argiya, as CTO of Adalat AI, you are powering courtrooms across India with tools like real-time transcription and case management. What real-world applications are you seeing that reduce barriers, especially for marginalized or rural communities, while aligning with principles of fairness?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:44",
                    "end": "00:15:09",
                    "text": " transparency and non-discrimination. Yeah, firstly thank you so much for having me here. I think there's two tracks to how AI is being used to sort of improve access to justice. The first track is more direct, which is directly with the communities and I think over there a big part of the problem in the global south is people have no idea of..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:09",
                    "end": "00:15:36",
                    "text": " getting information about their cases. They have no idea what's going on with their cases. It's extremely difficult to navigate archaic government websites to find out, where is my case? What's the judgment? What do I have to do next? There are complex jargons that lawyers love to use, like interlocutory order and res judicata, which common citizens just don't understand, right? And so I think one big use case of AI is to be able to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:36",
                    "end": "00:16:02",
                    "text": " make sure that communities are able to access this information easily. And to that extent, at the Dalit AI, we recently launched a WhatsApp chat bot where you can have a conversation in any language that you speak, not just English, and you can give your name, maybe your PIN code, and find out if you have a case, what's the next action item to take. And so that's more of the direct sort of access to justice solution."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:02",
                    "end": "00:16:26",
                    "text": " The second track is more around the indirect access to justice sort of solution, which is to improve capacity of courts. Now, when we talk about improving capacity of courts, just to paint a picture, because now, along with my co-founder, who's a lawyer, I've been down deep in the trenches. You know, courts in India are severely broken. There's paper everywhere."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:26",
                    "end": "00:16:37",
                    "text": " one of the use cases is that you know judges write by hand so if we were all in a courtroom and you know I asked as a lawyer I asked a witness where were you on the night of 21st April"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:37",
                    "end": "00:16:57",
                    "text": " the judge would stop us there, write this by hand, and then the next thing would be dictated. Now, as you can imagine, this becomes a serious bottleneck to court proceedings moving fast. And to that extent, we've built AI transcription solutions so that you don't have to write things by hand. You can just go about your witness deposition, and then you get a transcript of the same."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:57",
                    "end": "00:17:24",
                    "text": " By virtue of the process itself, everything is verified on spot. What we are seeing is that solutions like this, which are simple, that don't go into sort of being an AI judge or suggesting how to make a judgment, are able to improve productivity at a daily and weekly level by two to three X, which at scale could reduce case resolution timelines by 30 to 50%, depending on case stage and type."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:24",
                    "end": "00:17:50",
                    "text": " I think the big point that I'm trying to drive home is that even apart from the frontier of technology, even apart from just being able to predict cases or being able to sort of summarize or translate cases, there are so many other very basic tasks that AI could help with. Transcription is something that has been there for about 20 years now, speech to text, that is, right? And so there are a lot of..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:50",
                    "end": "00:18:18",
                    "text": " technologies that are safe to use today, and I think we should really work towards making sure that our judiciary has access to those, and that's kind of what Adalat AI works on. We are in 20% of India's courts today, and that's kind of where we've observed this statistic. Now, coming to this last part of your question, which is, how do you make this fair, trustworthy, and make sure that this doesn't discriminate? I think that's the hard part about AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:18",
                    "end": "00:18:40",
                    "text": " When it comes to being fair, I think given the current state of technology, fairness means making sure that you're only grounded to information that's available and not being suggestive. Not giving any sort of judicial advice. Let's say if you're building a WhatsApp chat bot for communities to find out information about their cases."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:40",
                    "end": "00:18:54",
                    "text": " AI systems cannot give advice today. That's not safe at all. Those things have not been audited yet. We have no benchmarks for doing that. And so I think that's more around the faithfulness angle. In terms of trust, I think what needs to happen is..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:54",
                    "end": "00:19:19",
                    "text": " AI systems need to acknowledge when they don't know something. That's something that the entire AI space is working towards, making sure that we can get our models to say no more often. And I think that's what it means to be trustworthy right now. And in terms of making sure that they don't discriminate, I think the biggest thing that AI models need to do is make sure they account for the linguistic diversity. And that's the..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:19",
                    "end": "00:19:42",
                    "text": " the largest problem that we have in the country, and a lot of other players in the ecosystem are working to make sure that every language and every dialect is covered, because just giving information in English is not good enough. So I think these are some of the challenges towards fairness, trust, and making sure AI systems don't discriminate, and possible solutions as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:42",
                    "end": "00:20:10",
                    "text": " Thank you. Thank you very much. It's interesting to hear the different perspectives, but it's also interesting to know that while AI is global, we always have to look at the context. Implementation in all countries and regions should not be the same. India has this diversity and multilingualism that perhaps is not there in many countries. So my next question is for you, Dr. Jalasi. We've heard from different perspectives, from UNESCO's perspective."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:10",
                    "end": "00:20:43",
                    "text": " what would you say would be the urgent next steps for member states to actually more, how do you say, institutionalize this AI training for justice across the board? But also, more importantly, how this could translate into real judicial independence and accountability in practice. Thank you, Mali. Good afternoon, everybody, and sorry for arriving late. I was speaking at another session. So obviously, I'm trying to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:43",
                    "end": "00:21:09",
                    "text": " to gather what was said from the previous speakers. Well, as you know, UNESCO, which this year celebrated its 80th anniversary, had as a constitutional mission to build peace in the minds of men and women. Justice is a foundation for peace, as is education, culture, science, communication and information. So for us..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:10",
                    "end": "00:21:36",
                    "text": " Today, when we look at the result of a survey we conducted in 2023, it revealed that 90% of judiciary professionals use AI without any prior training nor guidelines. This is frightening. 90% of them use AI, but they never had any guidelines nor any training. And we know that some of these AI systems..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:36",
                    "end": "00:22:06",
                    "text": " unfortunately use data sets that carry some biases, whether it is about gender, ethnicity, language, geography. So what I'm saying is that these AI system could, and in some cases that you have documented, did misguide, misinform the judiciary. So needless to say that the outcome would be rather negative. And in this case, AI will not be a force for good, but a force for risk and even for harm."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:06",
                    "end": "00:22:21",
                    "text": " So UNESCO developed training, including massive open online courses on AI and the rule of law. And we have trained to date more than 10,000 judges, prosecutors in a number of countries around the world."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:21",
                    "end": "00:22:50",
                    "text": " for the proper use of AI. So yes, of course, AI can help on a number of dimensions. It can reduce the time to process judicial cases. It can reduce the cost of it. It could help with maybe past cases. As you know, in AI, there is one field called case-based reasoning. It could bring to the fore some past cases and how they were dealt with. However, the integrity of justice cannot be automated."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:51",
                    "end": "00:23:18",
                    "text": " and that we should keep in mind. So AI and AI systems in courts and tribunals as a help, as a support system to the judiciary professionals, but not as a replacement for them. And this has been a guiding line in the work of UNESCO, which has been anchored, as you know, Mali, in the respect of human rights, in respect of human dignity, in ethics, in equity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:18",
                    "end": "00:23:48",
                    "text": " So we have to keep these values and these principles as guiding our work as far as artificial intelligence and the judiciary. Finally, I would say, when I mentioned these human rights, human dignity and the like, you may know that for now 13 years, UNESCO has trained more than 36,000 judges and prosecutors in 160 countries."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:48",
                    "end": "00:24:16",
                    "text": " on freedom of expression, safety of journalists, because quite often when there is a legal case against a media professional, the judge and the prosecutor have to process that. So we want to bring the awareness or the knowledge up to date of the judiciary regarding international standards on freedom of expression and on safety of journalists and on media, because you believe that freedom of expression"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:17",
                    "end": "00:24:38",
                    "text": " is a human right that enables other human rights and should be taken as a foundation for what follows as far as our work on AI and the judiciary. Thank you, ADG. I have a follow-up question. So as we said, international standards are non-negotiable, human rights are non-negotiable, but there are sometimes..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:38",
                    "end": "00:25:08",
                    "text": " There are certain contexts, as we said, the multilingualism, the diversity in India and others, that requires some adaptation of the tools that UNESCO and other organizations offer. So how do you see what would UNESCO be able to do with India's leadership in this AI adoption? How can we support these regional adaptations? You are absolutely right. It's not one size fits all. UNESCO has to work on a global scale because UNESCO has 194 member states. It's the world."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:08",
                    "end": "00:25:36",
                    "text": " So of course what we provide in terms of guidelines, in terms of framework, in terms of the UNESCO recommendation on the ethics of AI which was adopted in 2021. That's two solid years before the advent of generative AI and the advent of CHED GPT. The work on this recommendation started in 2018 and it was approved in 2021 by all"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:36",
                    "end": "00:26:06",
                    "text": " 194 member states calling for an ethical responsible use of AI. So what I'm saying is our work is global like the guidelines for the use of AI systems in tribunals and courts. These are global guidelines but of course they need to be contextualized. So of course taking into account by the judiciary in different countries, regions around the world, contextualize them taking into account the specificities of that country, cultural,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:06",
                    "end": "00:26:30",
                    "text": " legal, socioeconomic specificities and that's why I'm saying it's not one size fits all. But at least national judiciary systems and courts and tribunals have a good blueprint on which to build. Take the UNESCO work instead of starting from scratch and then of course adapted, not adopted as is, but adapted to the local or the national context."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:30",
                    "end": "00:26:54",
                    "text": " Well, since we're talking about guidelines and tools, I think this is a moment for us to launch here, in your presence, our new policy brief related exactly to the same topic. So Mr. Jalassi, I invite you to the podium, perhaps, and you could maybe launch, share with us the latest policy brief, and then we can take some questions from the audience. Or if you can speak here as well. Do we have the documents?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:55",
                    "end": "00:27:23",
                    "text": " I don't have a copy of the document, but since you don't want AI, it's a virtual document. It exists, but for now, it is in a digital forum, not in a paper forum. So we are delighted having authored, developed this policy brief. We believe, again, that it is useful resource, and I say it's a resource that, of course, member states and the judiciary may want to use or not."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:23",
                    "end": "00:27:50",
                    "text": " This is the work of UNESCO. I mean, part of our work is advocacy, but advocacy based on evidence-based work. And we believe that advocacy could lead to national policies or to regulation or to implementation. And we do help as well with that. It's not only we do research, we publish, we do advocacy, but we help member states at their request develop national policies and implement them."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:50",
                    "end": "00:28:15",
                    "text": " In addition to our normative work, when I mentioned the UNESCO recommendation on the ethics of AI, it's not a convention because otherwise a convention would be legally binding. We don't make it legally binding, but a recommendation to say this is a set of principles that was developed by more than 100 international experts, then it was endorsed or approved."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:15",
                    "end": "00:28:42",
                    "text": " by all our 194 member states. We believe it has a value for each country around the world, but it's up to the national authorities, the professional communities to decide to what extent they want to adapt and use these guidelines or this policy brief which Mali mentioned. You are looking for it. Thank you. I actually have it in a QR code for people to scan. Yes, please. And in the meantime, I can..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:42",
                    "end": "00:29:19",
                    "text": " coordinate the Q's and A's, perhaps, the more I stand while I do that. We do have some time for Q's and A's, and I think it is as important for you to be able to approach the speakers and have questions than the speakers just having a one-sided discussion. So, yes, we can start taking questions. I don't know if we have a microphone, if someone can help us give to the speakers. We have the first, the audience, we have one question there. Ashita, can you help us with the, please, the questions? The microphone. I will. Hello."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:19",
                    "end": "00:29:42",
                    "text": " Yeah, so my question is generally to the panel here. What is the, so basically we were hearing, I heard all the panelists and everybody gave their views on the reliance of AI in making decisions and judgments. Do you think going forward and in future, what will be the liability of AI? How can we make AI liable, you know, in..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:42",
                    "end": "00:30:12",
                    "text": " making judgments as well. Because as Sir said rightly, a lot of judges are relying on AI and making decisions. And so just wanted to ask you what will be, there'll be a code set up for making AI liable or not. I think Sir just referred to the fact that, you know, a very high number of judges are using AI without"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:12",
                    "end": "00:30:41",
                    "text": " any kind of guidance being given to them in a national context. So I think before we get to liability, we perhaps need to ensure that guidelines are developed domestically around AIUs by judges, what it's appropriate to use it for, what it's not appropriate to use it for, where you are using it, what kind of safeguards, process checks need to be in place around its use. When courts are acquiring"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:41",
                    "end": "00:31:07",
                    "text": " AI technology, whether it's for translation, transcription, administrative management, research purposes, what kind of vetting, auditing, assessments of the system is being done at the point of acquisition of this technology, but also post-deployment assessments and audits that are being conducted."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:07",
                    "end": "00:31:37",
                    "text": " how open and transparent is public procurement, and then what is the kind of transparency that is remaining. So first you have to put into place these safeguards, right? I think liability then becomes the next layer. I think at this point, adoption of AI and courts is being used as a tool for judges, rather than direct impact, where it's directly impacting citizen rights. I mean, there have been instances of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:37",
                    "end": "00:32:02",
                    "text": " Compass and Heart and other technologies which were deployed in jurisdictions like the US, Brazil, the UK, which didn't, you know, had their own sort of challenges. I think people are more cognizant of it now. So I think decision making, I would say, is largely still vested in judges, so I'm not sure where that liability comes in, but I think..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:02",
                    "end": "00:32:11",
                    "text": " overarchingly governments are increasingly thinking about liability of AI systems more generally, right? Not specific to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:11",
                    "end": "00:32:40",
                    "text": " AI use in judiciary, but I think more generally. And I think that is still an evolving conversation. There's a lot of thinking going into, okay, from strict liability frameworks that have been developed, taught liability frameworks that have been developed across the last hundreds of years, what learnings can be drawn in the context of AI liability. There's been a lot of conversations"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:40",
                    "end": "00:33:10",
                    "text": " around, okay, is there something to learn from intermediary liability principles that have been imposed on internet platforms, social media platforms, the due diligence obligations that are imposed on them. So what learnings can we take from all these spaces and then translate them appropriately into the context of AI. So I think there's still, I don't think there's quite consensus on what liability looks like, but there's definitely ongoing conversations on this issue."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:10",
                    "end": "00:33:27",
                    "text": " supplement Jalak yesterday I think some of you have read Supreme Court Chief Justice of India mentioned about some of these and what we could see is that it is the lawyers those who use and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:27",
                    "end": "00:33:57",
                    "text": " through their arguments, they mention with the use of the artificial intelligence. And some of the, even in Kerala, we can see Bangalore, there's an income tax tribunal, entire, when the judgment came, they realized that the lawyer who mentioned about this case in the income tax tribunal, the case does not exist. The principle does not exist. So, it depends more on what we are mentioning about UNESCO,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:57",
                    "end": "00:34:04",
                    "text": " the Toolkit will help and also individual even the Supreme Court"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:04",
                    "end": "00:34:34",
                    "text": " and other high court, the Department of Justice has to come out with the comprehensive guidelines. But till that time, it's an individual judge he has to use. Of course, we can see that some instances are coming up where the judges are using the AI for the research and the judgment. But all of us, what we have seen in the panel discussion, we must remember that the legal representation will remain. Lawyer cannot be replaced by AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:34",
                    "end": "00:35:07",
                    "text": " Teacher cannot be replaced by AI, and judges cannot be replaced by AI. Thank you. Questions, and I'm glad we have the time. Okay, let's take, let's see, if there's a lady, let's do, no? Not interested? Okay, we'll take second question and third question here. Okay, we'll take her first and then we'll come to you, sorry, thank you. Yes, so we are talking a lot about AI for law, but what about law for AI?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:07",
                    "end": "00:35:35",
                    "text": " So when we are talking about we are having a lot of frameworks in education, in communication, in jurisdiction and everything. But what about a common framework or guideline or a policy to use AI or to use the data of AI that has been produced or developed or a common understanding to the audience that what is law for AI can be."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:35",
                    "end": "00:36:08",
                    "text": " What are the benchmark for using the AI or any kind of AI tool? What is right or what is wrong? That is what I wanted to know from any of you, if you would like to. Thank you. Maybe we can take two more questions and then there are similarities. And if you also want to address it to a particular speaker, please do so, unless you want to address it to the entire panel. My question is for Argya. You mentioned two specific streams where AI is sort of broadly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:08",
                    "end": "00:36:38",
                    "text": " being looked at as a potential use case. Do you also see a potential need or opportunity for point solutions, for example, QC of critical documents, et cetera, which have a significant bearing on how an individual case moves forward, the delay, the outcome as well? Could be a POCSO-related charge sheet, could be a domestic abuse-related charge sheet, coming from a very narrow perspective, but QC-related opportunities as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:38",
                    "end": "00:36:51",
                    "text": " Thank you. That's a great question. I think that's the territory where I advise AI practitioners and companies building to stay away from."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:51",
                    "end": "00:37:19",
                    "text": " And just to tie that up with the previous point of training judges in the context of India, we have a thing called the Adalat AI Academy, which, in fact, the curriculum of which we've been building in partnership with UNESCO. As part of that, we've done a lot of trainings with judges in India. A lot of them don't know how to update their browsers. They don't know how to update their Chrome browsers yet. And so the first thing is for them to understand that if they use AI outputs,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:19",
                    "end": "00:37:38",
                    "text": " they're liable today, right? And to that extent, giving these tools in these settings to people who don't have the knowledge yet can be extremely dangerous. So I can tell you a few examples where AI could help even with sensitive matters. One is perhaps transcription, because again, it's..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:38",
                    "end": "00:38:05",
                    "text": " very sort of embedded technology right now. You're just saying and it's writing what it is. The second is you have these long documents that you have to always go through. And this is a common problem for lawyers and judges. AI can be used to ask questions and not get answers, but the AI then scrolling and taking you to the right part of the document, right? And often the problem becomes that you just have to scroll through thousands and thousands of pages. And so if you think about it, a small trick,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:05",
                    "end": "00:38:27",
                    "text": " makes AI a lot more auditable in this case as opposed to you asking a question and then it generating a huge answer where you could have fake cases being generated as well, right? And so it's really AI plus how you design the user interface that's going to do it. But generally I would avoid using generative AI in sensitive cases."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:29",
                    "end": "00:38:57",
                    "text": " So my question is from a team representative of NLU Delhi and Adalat AI. Considering the UNESCO mandate on linguistic rights, what are the challenges that the AI can cater in aspect of English only mandate through Article 348 of High Court and Supreme Court? And how multi-linguistic criteria can be integrated in a judicial system of India?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:57",
                    "end": "00:39:35",
                    "text": " My question is actually twofold. The first part would be, a lot was said on guidelines on use of AI, but how do we ascertain guidelines which are indicative in any way, in all ways actually, are being used or are being applied to in the right way in all the existing matters? Because they are indicative for us. So for us,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:35",
                    "end": "00:40:02",
                    "text": " How do we ensure that the steps that are being taught or that are being told is being adhered to or not? Because there is no way to reverse engineer this and know if a particular case is affected by those guidelines not being in place or there in place or not. The second part to that would be, how would I know if my case, because for example, as Sir rightly gave that example as a case which didn't exist."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:02",
                    "end": "00:40:30",
                    "text": " was cited and an order was passed on that. So how would I know that my case is not being affected by a bias or by some case which was not in existence, but the opposite side has argued that case. Thank you. So we have two questions, one also two-fold. Would you like to start first? You were already starting to respond to the question. I can start with the question on the multilingual aspect."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:30",
                    "end": "00:40:47",
                    "text": " It's a very hard problem. It's important that we acknowledge that we have many dialects. The way that people speak, let's say, English in Kerala is very different from the way that people speak English in Delhi, let alone the fact that we have thousands of dialects."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:47",
                    "end": "00:41:16",
                    "text": " There is a certain way in which you can build translation systems that can actually address this. I can maybe give you an example of some of the efforts that are going on by the government, by Supreme Court, and also by Adalat AI, which is to build these systems using expert preferences. What I mean by that is, you know, let's say there is a legal document that needs to be translated to, let's say, a local language to make sure that it's accessible. What we are doing is we actually give"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:16",
                    "end": "00:41:44",
                    "text": " copies from humans and a lot of different AIs to lawyers. And then we ask them to choose which one they like the most, right? And an AI solution is only allowed once it reaches a point where 80% of the responses are from AI and they're being preferred by humans. So that's one of the things that we've done in Karnataka and some of the translation systems in Karnataka are run by Adalat AI and we've seen that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:44",
                    "end": "00:42:14",
                    "text": " You know, humans actually prefer AI output 80 to 90% of the time, and these are legal professionals. These are not just any annotators. So that's one very important thing, which is to make sure that the expert human is in the loop, right? And I think the second challenge overall is it's going to take time. It's important to understand that AI systems are not just going to do it overnight."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:14",
                    "end": "00:42:35",
                    "text": " And so we still have to build capacity for translators in the state, which is quite lagging right now. Yeah? Would anyone like to respond to the other question that we had? Can you maybe quickly sum it up in a couple of seconds?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:35",
                    "end": "00:42:57",
                    "text": " guidelines being indicative. So how do we establish the fact that these are being, the steps that are taught or that are told are being applied in that order. For example, if you skip one step, you may not go in the right order to the next step."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:57",
                    "end": "00:43:21",
                    "text": " And the other part was that how do I know that my case is not being affected by AI or the bias involved in that because anyway, it was engineering. Thanks for summing that up. I think within the judicial system, there will have to be, you know, like you have office for data protection officers, right? The office of data protection is being set up in many courts. Similarly,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:21",
                    "end": "00:43:49",
                    "text": " either within that office or a separate office depending on different contexts, different judiciaries may choose to do it differently, you will have to have an arm that is set up within the court system whose function is to oversee, enforce, and look out for patterns and challenges that may be emerging at any given point in time."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:49",
                    "end": "00:44:12",
                    "text": " In terms of how do you know that your case is not being affected, I think lawyers and participants in the judicial system have to be vigilant and point to challenges that emerge, surface challenges that emerge, so that patterns of harm get visibleized within the judicial system and they can be flagged."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:12",
                    "end": "00:44:31",
                    "text": " where there are harmful instances or challenges arising, there needs to be a feedback loop into the judiciary, a grievance redressal mechanism, sort of where it's flagged that this AI use is being problematic. But a lot of AI use and its harmful impact may be invisibilized, where you do not realize that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:31",
                    "end": "00:44:45",
                    "text": " This decision is being made based on AI systems, which is why you need third party audits, assessments, and other practices, which may not pick it up at an individual level."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:45",
                    "end": "00:45:14",
                    "text": " at a more sort of meta level, it will be able to identify that these are the challenges in the deployment of this system, and these changes and tweaks need to be made to the system to ensure that it is non-discriminatory, that it is maintaining privacy, that it's not implicating freedom of speech and expression norms. But I'll defer to Dr. Krishnadevaraya over here, since he has decades of experience."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:14",
                    "end": "00:45:36",
                    "text": " in these kind of questions. Just 90s, we were looking at in the legal academy, particularly in India, court managers and case management. And the Supreme Court has come out with each district they have appointed a court manager who is a management graduate. They did not work. But now we need"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:36",
                    "end": "00:46:01",
                    "text": " with the AI we need a technologist at the each not one person a group of persons at the district course data protection data analysis helping the judge in the research summary translation so many things but as we all have agreed that AI will not substitute for a judge or a teacher or a lawyer so but some of these safeguards"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:01",
                    "end": "00:46:24",
                    "text": " It has to be built up. I think that's where it's a concerted effort has to be made. Everybody and technology, particularly the IT companies. In fact, when we say that I am passionate that even the law students must be trained to develop a software development. Engineering is developing maybe."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:24",
                    "end": "00:46:52",
                    "text": " If the engineer may not understand the technicalities of a law or a court and other, so we need to think about now how software development of, whether this online dispute resolution, judicial research, preventing the bias, discrimination, preservation of constitutional values, equality, all this, only a law student. But that requires a major challenge of rethinking. Thank you. OK."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:52",
                    "end": "00:47:23",
                    "text": " Mr. ATG, would you like to add something? Just some final words. Maybe before the final words, I see some people here wanting to ask questions. Look, if you promise, your question will be straight to the point, fast. We'll take one lady, one here, and one here. He asked first, but you're welcome to ask later. Two questions? Okay, so can we take one here and one from there, and then you can talk to the speakers. Hi, my name is Abhishek."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:23",
                    "end": "00:47:51",
                    "text": " and I'm a podcast host doing podcasts with judges of Supreme Court and high courts mostly. So I have met so many judges in my time and 99% are reluctant to do work with AI and I do not blame them at all because of the facts which you have just mentioned that AI is still hallucinating and all."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:51",
                    "end": "00:48:25",
                    "text": " So how do we make them comfortable? How do we make sure that their judgments, in case they are relying on AI or some lawyer is relying on AI, then they are not misleading. Hi, panelists. Thank you for such an insightful talk. My question to you is how do you protect the creativity? So we see in India a lot of judgments and everything is based on emotions. How do you protect creativity? That is one question."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:25",
                    "end": "00:48:55",
                    "text": " Secondly, the older generation is still on, you know, they are more towards documents and commentaries that are available. However, the new generation that is coming, they are more tech-based. And for them, for me, experience matters over tech. And how do you create a distinction? Because it will also affect the employment in the legal fraternity. How do you protect that? We go there?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:57",
                    "end": "00:49:29",
                    "text": " Can you turn it on? There we go. Okay, last question. I'm Hofmeister with the German Ministry for Digitalization. And towards the end, one provocative question. Don't you think that maybe the use of AI for judges by judges should become mandatory in the future? The legal system is made on trust and on acceptance. And if we see AI being more accurate in the future, the judgments being more precise in the future, and comparing one judgment which is extremely precise with the help of AI, and one court deciding"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:29",
                    "end": "00:49:59",
                    "text": " not so precise, the trust of society goes down. So I just provocatively ask, should we not make use of AI by judges mandatory in the near future? Thank you. So we have three questions, less than five minutes, and Mr. Julasi would like to close and I would like to tell you something. So I don't know if each of you would like to answer very briefly, straight to the point, each of these three questions. I think it requires a judicial education where I think some of the questions were raised."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:59",
                    "end": "00:50:15",
                    "text": " I think very challenging how to simple English writing and then how AI can supplement and maybe a judge may mention that he used the AI so that safety,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:15",
                    "end": "00:50:46",
                    "text": " And using that preserving the essence of the judgment, how it is to be right. I know it's difficult to write in a simple language within a small paragraph, maybe five pages of judgment. It's difficult. Some of them, they used to say they refer to the Privy Council, some of the outstanding judgments. Yeah, to answer your question more directly, AI is actually mandated in a lot of courts in India. So for example, there's a state called Kerala and Adalat AI is actually mandatory."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:46",
                    "end": "00:51:12",
                    "text": " in every courtroom in Kerala, you cannot record witness depositions by hand, you need to do it using a Dalit AI transcription, right? And to that extent, before any such mandate kicks in for other use cases like judgment writing, I think what's important is to do a serious economic impact evaluation. It's important to see that the impact is measurable, and it's also important to see that across the board,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:12",
                    "end": "00:51:34",
                    "text": " Generally, you see better and fairer justice, and only then can AI be mandated. As long as that infrastructure can be developed, I don't see a harm in mandating AI. Thank you. Have we answered all the questions? We had the old generation versus young generation. Anyone has a message to say? I guess calculators came in. We didn't get bad at math. And so we'll have to figure out very similar."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:35",
                    "end": "00:52:00",
                    "text": " Yeah, maybe we did, but we're still moving on in terms of doing more innovation. We're still building more, right? And so we'll have to figure out something similar. I think the educational curriculum will have to change to account for AI as the new calculator as well. But all generations must learn AI. Thank you. Dr. Jalasi would like to say a few words. Thank you for coming, but I just want to say if you can scan this for the publication if you haven't already."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:01",
                    "end": "00:52:30",
                    "text": " Okay, done. One second. And then there's the other one where you have all the resources about all the guidelines that UNESCO has and can offer. So one, two, here. And Dr. Jilas, I'll give you the floor. You can scan this code for all the different material. Thank you, Mali. The mission of UNESCO has five Cs in it. I think our meeting here first is to convene, to converse, to comprehend, to commit."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:30",
                    "end": "00:52:58",
                    "text": " to collaborate. And there is a sixth C, which is capacity building. And I think what you mentioned here, our work on AI and the rule of law, our guidelines for the use of AI systems in courts and tribunals, the AI essentials for judges, which we officially launched here, these are just examples. These are three key capacity building means. We have to ensure that together,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:58",
                    "end": "00:53:23",
                    "text": " AI remains aligned with the human rights, with the rule of law, because as I said earlier, the integrity of justice must not be automated, and you have to ensure that AI strengthens, not sidelines, judicial independence, fairness, due diligence, and the integrity of the judicial process."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:23",
                    "end": "00:53:49",
                    "text": " On the last exchange between some questions and the panelists, when we said, you know, what can we do with AI? I think generally speaking, we cannot stop technological advances. We cannot. But we need to provide the framework. We need to provide some sort of an overarching regulation. It's not a regulation that hinders creativity and innovation because the lady had a question about that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:49",
                    "end": "00:54:12",
                    "text": " It's not either or. We need to have some overarching guidelines, principles for using AI responsibly, ethically, and make it a force for good. Thank you very much. Thank you all for joining. The message was clear. AI can change maybe the way justice operates, but not what it stands for. Thank you very much. Have a wonderful summit."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
    {
    "_id": "6995aba9565455da1fc75772",
    "projectId": "260279",
    "projectName": "Scaling Intelligence NVIDIA NeMo and the Future of National AI - AI Impact Summit 2026 - English",
    "projectDescription": "Scaling Intelligence NVIDIA NeMo and the Future of National AI - AI Impact Summit",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T12:08:08.407Z",
    "projectUpdatedDate": "2026-02-18T12:08:08.407Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T12:24:41.003Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "Scaling Intelligence_ NVIDIA NeMo and the Future of National AI _ AI Impact Summit 2026(MP3_128K) (1).mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771416478498.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:01",
                    "end": "00:01:36",
                    "text": " Hello? Morning. Sorry, I was running a little bit late. I need to put my... Can I use my laptop? Oh, okay. So, first, let me introduce myself. I'm Bernard Nguyen. I'm a director of engineering at NVIDIA, and I'm responsible for... I oversee the development of Nemo framework. Today, I want to speak with you about advancing LLM training at scale. I'll be talking about..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:36",
                    "end": "00:02:09",
                    "text": " the process of taking a model from pre-training through fine-tuning through post-training, and some of the libraries and open source AI that NVIDIA offers to help facilitate this. So I'll just begin. Maybe you've seen this slide. It's a good level-setting context for this presentation. So in the beginning of LLM model training,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:09",
                    "end": "00:02:38",
                    "text": " A lot of the focus was on scaling pre-training models. Essentially, we saw scaling laws where the larger you make the model, the more intelligent it would become. Now, if you follow this to the logical conclusion, if I want my model to be extremely intelligent, then it needs to keep getting bigger and bigger and bigger. But that's not particularly practical because at some point you run out of resources and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:38",
                    "end": "00:03:08",
                    "text": " You wanna deploy these models on edge devices and you cannot fit these super large models on such environments. So there was an advent of post-training scaling. This is where you can take a pre-trained large model and adapt it to a specific domain. You can even shrink down that model and make it fit. But the model is still trained to be more of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:08",
                    "end": "00:03:36",
                    "text": " without reasoning and thinking. So many of you are probably familiar with, you know, chat GPT or other such AI applications where now the models can actually think. So it takes time. You can ask it to think harder about a question or a prompt, and it'll give a more intelligent answer. So how do you enable a model to become thinking? And that's..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:36",
                    "end": "00:04:05",
                    "text": " what's happening in this post-training to test time scaling long thinking. So we'll spend some time talking about this particular latter half. Before we do that, I'd like to take some time to talk about just open source AI and NVIDIA's stance towards it. So open source software has been around for a long time, I mean since the days of Linux, but AI has not."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:05",
                    "end": "00:04:34",
                    "text": " been open source that long. For example, a lot of the models and applications that we use, you know, a chat GPT is not open source. So, recently, AI has become more and more open source. What you're seeing is an explosion of more models being developed and then released out in the open. Now,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:34",
                    "end": "00:04:58",
                    "text": " There are different definitions of open source when it comes to models. For example, there are models where you get access to the weights. So you download the weights, you can take the model, deploy the model architecture and the weights, and you can run it. But if I want to..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:58",
                    "end": "00:05:27",
                    "text": " Say, fine tune that model, or if I want to retrain that model, if I want to know what was the recipe to train that model, that's not always made available with the model weights. So when I am talking about open source with respect to NVIDIA's stance towards open source AI, NVIDIA is wanting to provide all of the things required to train the model. So not just the model and the weights, also the data sets, the recipes."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:27",
                    "end": "00:06:04",
                    "text": " the libraries that were used to train those models. I might be missing anything, but we try to provide everything that you need to learn and reproduce the models that come out of NVIDIA. Sorry, I'm just gonna turn off my phone. So there's a number of areas that we open source software from health AI to physics to agentic AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:04",
                    "end": "00:06:36",
                    "text": " We'll spend most of our time talking today about Nemotron. Also, Cosmos for physical AI, Groot Robotics, and AV. Jensen spent quite a bit of time talking about Alpamio at the CES in January. And you can see here, as far as GitHub repositories, NVIDIA is leading by far as the total number of repositories. So, just to reinforce, yeah, data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:36",
                    "end": "00:07:11",
                    "text": " models, libraries, training scripts, which are the recipes in which to take these libraries and train the models, and blueprints in which you can take the models and piece together them into solutions. So, Nemotron, this is a fine balance that we're trying to make with Nemotron, where we try to make it efficient, intelligent, and open. And I'll show some..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:11",
                    "end": "00:07:42",
                    "text": " leaderboard charts later to demonstrate that. So for example, Nemotron 3, Nano, was a model that we released most recently. There is Super and Ultra coming out soon. Nano 3, in its weight class, so this is a 30 billion parameter model. The closest models that are in this class would be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:42",
                    "end": "00:08:13",
                    "text": " GPTOS S20B and QWEN 330B. You can see that NVIDIA's Nematron 3 is at least or more intelligent than these. Now where NVIDIA really spends time is on efficiency. So when we're talking about efficiency, we're talking about tokens and the number of tokens that you can get out of the model with the least amount of compute. Or put another way,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:13",
                    "end": "00:08:44",
                    "text": " which models given a certain compute can give you the most tokens output. Here you can see that Nemotron 3.0 provides 350, 60 tokens per second, which is much higher than other models. And so what this means is you can essentially ask more prompts to this model in a certain period of time for less cost. As I mentioned, Nano 3.0 is already available."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:44",
                    "end": "00:09:14",
                    "text": " super and ultra are coming soon, super very soon, and then ultra is a short follow. And so these are MOE architectures, so even though the total number of parameters is much larger, like 30B, 100, 500, how we achieve the efficiency is that at inference time, only a small portion of the total parameters are actually active."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:14",
                    "end": "00:09:47",
                    "text": " So A3B means only three billion parameters are active out of the 30B, and this is leveraging the mixture of experts architecture. Same with 100B, only 10B are active, 550B. So 10% essentially of the parameters are active during inference time. I'm gonna transition into agentic AI and how agentic AI is..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:47",
                    "end": "00:10:18",
                    "text": " post-training and reasoning is required for agentic AI. So agentic AI is essentially models where an AI agent can act on behalf to accomplish a set of tasks or goals. This example is where a user is interacting with an agent via speech. So I'm talking to this agent. This agent is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:18",
                    "end": "00:10:52",
                    "text": " then guiding certain tasks and decisions to different LMs in the back, so language models, vision language models, reasoning models, accessing data, accomplishing a job, and then routing it back from speech back to text. Now, or sorry, text to speech. Nemotron, we have different models for reasoning, vision, information retrieval, content safety, and speech."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:52",
                    "end": "00:11:24",
                    "text": " So what I'm trying to share here is that there is a broad portfolio of nematron models that you can use for your agentic AI. So you can take a lot of these models and piece them into the agentic AI architecture. The data, just a little bit more detail on that. We provide, yeah, essentially all of these are different data sets that are available on open source via GitHub."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:29",
                    "end": "00:11:58",
                    "text": " I'm gonna, I wanna get through some other slides. These are some examples of blueprints. This one is essentially user interacting with an agent to do report generation. Give me a report on, I think this is probably functionality that folks might be used to already if you use deep thinking on any of the most recent AI chats applications."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:58",
                    "end": "00:12:26",
                    "text": " ask it to do deep research for you and it'll come back with a full report in a couple minutes of a subject that might have taken hours to Google and compile on your own. And this is a blueprint for how to implement this yourself. Same, there's searching through hours of video and then summarizing. This is another agentic AI use case and a blueprint for it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:27",
                    "end": "00:12:55",
                    "text": " So let's dive deeper into AI agents specifically. So we have the starting with a prompt. So what do I want my agent to do? For example, plan me an itinerary and keep it within a certain budget over a number of days. And so the agent will do a cycle of perception, understanding my prompt, reasoning through it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:55",
                    "end": "00:13:22",
                    "text": " and then taking a course of action, the agent may use tools such as search to go search the web, calculators to calculate things, can even run code through Python interpreters, essentially any tool that can be made available through a protocol. There's protocols such as MCP where you can enable any kind of web service to be accessible from an agent."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:23",
                    "end": "00:13:54",
                    "text": " Agents can also even directly use your computer. If you give it access, it can access local file system. It can search files and your information for you and use that as part of the planning context. Memory is an important aspect of agentic AI because you might have many agents acting in concert together. The agents, a more brute force way is the agent passes all the data from one to another."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:54",
                    "end": "00:14:25",
                    "text": " over the network, that would be very expensive and slow, but if you can run, say, deploy all of your agents in something like a rack scale, Grace Blackwell rack, for example, all the GPUs can share the memory together and so the agents, you could have a whole agent ecosystem running in a GPU rack and sharing the context over high speed NVLink. Agents can also talk to other agents."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:25",
                    "end": "00:14:57",
                    "text": " and you can have specialized agents, you can have like a travel booking service agent speaking with an agent that is really good at planning rental cars. I mean, I'm giving kind of simple examples, but you can imagine there'll be specialized agents. Speaking of specialized agents, so let's see. We were talking about larger models and now we're talking about smaller models and this is where"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:57",
                    "end": "00:15:25",
                    "text": " you can create more specialized small models, and these small models can be used to power these specialized agents. These small models can be much more domain-specific, they can be adapted, and they're very cost-efficient, and the data IP control, this is assuming that..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:25",
                    "end": "00:15:58",
                    "text": " Yeah, you're running your agent in a data center either that you trust that you're operating or with a cloud service provider that is also trusted That would be part of the sovereign AI theme also that word that we've been discussing at this conference okay, and These are just examples of different specialized agents. All right, let's get more into post training so Post training is essentially taking an existing model"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:58",
                    "end": "00:16:31",
                    "text": " that you've either pre-trained from scratch yourself or say downloaded from open source and giving it additional capabilities, aligning it. You can make the model more helpful and say less harmful. A little bit of the recent evolution of post-training. So in the beginning there was"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:31",
                    "end": "00:17:02",
                    "text": " essentially just taking a model and being able to predict the next token. This is just text completion. Then with ChatGPT, I think we all saw that become a first in where you can have a conversation with a model. And this was essentially taking that model and teaching it how to have a multi-turn conversation. Then the next wave was, okay, now..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:02",
                    "end": "00:17:31",
                    "text": " I can do the multi-turn conversation, but please think before you respond. And so thinking and reasoning was added into these language models. Now, where we are today with agentic AI is the reasoning can lead to multiple steps. So there can be a multiple course of action and the agents can use tools, which we discussed previously."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:31",
                    "end": "00:17:59",
                    "text": " where we are headed, if you ask me, in the next year or so, we want to get to a longer horizon. So what does that mean? Let me go to the next slide and we can talk about that more. So today, this is actually a little bit out of date. GPT, we're at GPT 5.2 now, as of early February, and..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:60",
                    "end": "00:18:32",
                    "text": " I believe Claude's Sonnet 4.6 was just released yesterday. Yeah, okay, 4.6, yes. So the AI field is moving fast. That said, the point of this slide is basically trying to explain that these are models where, so context, or horizon. So this is a 30 minute horizon."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:32",
                    "end": "00:19:01",
                    "text": " You know, about one, two years ago, a model, if you gave it a task that would take a human to accomplish in 30 minutes, a model could do. Now we're at a point where models can accomplish a task that would take multiple hours for a human to do and have a 50% chance of getting it right. So that's fine. I mean, when..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:01",
                    "end": "00:19:31",
                    "text": " when we're talking about long horizon, we're talking about a task that would take, you know, you or I days or weeks or months, you know, the longer and longer, that's long horizon. And the reason why this is, it's difficult to do long horizon is because if you think of a task, say it has a hundred steps, say in any given step, there's a 99% success rate for the model in that step."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:31",
                    "end": "00:19:60",
                    "text": " So you have 100 steps, 99 to 100. Eventually, that's, I don't know, I'm not doing the math, but it's probably 20, 30% at the end of that. So at the end of those 100 steps, there's probably a 20% chance that it succeeded, which is a pretty low rate of success. So we need to get the models to where they can succeed close to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:60",
                    "end": "00:20:32",
                    "text": " you know, multiple nines of success rate in any given step. And the models need to be also smart enough to detect if they've chosen a step, gone kind of off course, and then, you know, re-correct and try again. So there's, you know, there's work to do as part of long horizon where we want to get to. Let's dive more into"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:32",
                    "end": "00:21:03",
                    "text": " what's going on in post-training reasoning. Here is a thinking trace. So, you know, solve a math equation, and this is the model thinking over here in this right-hand side box, and then it gives a final answer, aiming for the answer to be both high quality as well as helpful, safe, and respectful. So, originally post-training was using supervised"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:03",
                    "end": "00:21:32",
                    "text": " training, fine-tuning a model. So this is an example of, say, an SFT data set, which is given a particular prompt, here's the answer, here's another prompt, here's the answer. So you could train, you could basically force the model to know what's the right answer, given a prompt, by giving it examples of prompts and right and wrong answers. But that's not really thinking, that's just"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:32",
                    "end": "00:22:04",
                    "text": " imitation. So RL is where now you actually train the model to try a number of different reasoning generations, a number of different steps, and you reward the model when it chooses a particular generation that is more correct than the others. So in this simple example, rollout is essentially"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:04",
                    "end": "00:22:34",
                    "text": " opportunities for the model to respond to the question and You know the first one is Looking more correct. So we gave it a reward of one hundred percent good job The response to you is not very helpful at all So we gave it zero reward and this one created a ticket which is just a really bad I did not ask this thing to create a ticket. I was just asking to give information. So the reward is just"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:34",
                    "end": "00:23:05",
                    "text": " We completely mark we don't want this response at all. So that's learning through exploration. I'll pass this one. So how do you actually determine the reward? So we have to create a reward model. And this is a simple example of training a reward model where given a prompt and a response candidate, the reward model outputs a particular reward."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:05",
                    "end": "00:23:32",
                    "text": " In this case, we're showing an example where a reward is not just one number, but a vector saying how helpful, correct, coherent, complex, verbose this particular prompt was. There are two main categories of reward models today. The first category of verifiable rewards, these are..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:32",
                    "end": "00:23:60",
                    "text": " Fairly intuitive these are rule-based award rewards where they can be verified, you know completely Whether it's true or false. So for example math You can verify whether the response is correct for code You can execute the code run unit test to verify if the code is correct Instruction following you can verify if it follow the instructions. So these are high, you know, very easy to verify"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:60",
                    "end": "00:24:27",
                    "text": " and very easy to assign the reward based on how verifiable it is. There's also a class of model-based rewards where it's less verifiable, it's maybe more objective, and so this is where you can have, say, models reasoning as a judge. You can use an LLM to actually judge how..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:27",
                    "end": "00:24:57",
                    "text": " correct a particular response was. Okay, so this is going through, so today, there's many different algorithms used in RL. One of the most common ones today is this algorithm called GRPO. It's essentially a group-based policy algorithm. So here policy is the model that we are,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:57",
                    "end": "00:25:27",
                    "text": " Trying to train and giving it we want to give this model thinking ability so We give it a prompt and so one of the examples earlier, you know or here it is, you know solve this math problem and Given that prompt we give the model Say 16 tries generation and 1 through 16 now out of these 16 tries then we pass it through"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:27",
                    "end": "00:25:55",
                    "text": " separate reward model which we just discussed to score the reward for each of those 16 tries and here you know only only one out of the 16 was correct and and then so then we we take the average of the group and we update the model with with with that average and nudge the model towards the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:55",
                    "end": "00:26:25",
                    "text": " the more correct generation. This reference model is essentially just making sure that the original model doesn't stray too far from a particular RL loop. And so this is an iterative process. The key takeaway here is that you can run this GRPO loop many, many, many times with actually not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:25",
                    "end": "00:26:49",
                    "text": " You don't need that many prompts. You don't need that much data. As long as you have a decent reward model, or multiple reward models, the model will get smarter. It will keep getting smarter. There will be a point of saturation where there's diminishing gains. But this is, if you remember that previous slide at the very beginning, the..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:49",
                    "end": "00:27:22",
                    "text": " the post-training scaling law and test time loop scaling law, this is where that comes into play, where the scaling law becomes more about the number of iterations that one does to make the model more intelligent, not how much bigger the model gets, but just how much more we refine the model. And so this is essentially benchmarks. We can see quantitatively,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:22",
                    "end": "00:27:52",
                    "text": " after you can see that the model got much smarter, for example, in this math rewards upper left, after just a hundred different loops of that algorithm of the GRPO. It still kept getting smarter until about 800, 600, 800, then it leveled out. Here in one other thing you can see here, generation length. This is how..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:52",
                    "end": "00:28:20",
                    "text": " how much more is the model thinking in its generations? How much more thinking tokens is it using? So you can see after more and more loops, it also grows in its thinking output, which is something actually one needs to balance. You don't want the model to be thinking so long, then it becomes kind of like, okay, it's thinking too long."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:20",
                    "end": "00:28:53",
                    "text": " I don't want it to think that long. I want it to be smart, but not think too long, because I still need to get my job done. This is an example of what I was discussing. At a certain point, token budget, then the intelligence here, the accuracy of how accurate is it in beating those benchmarks, it does plateau."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:54",
                    "end": "00:29:28",
                    "text": " I don't want the model, I can limit the number of tokens that the model spends thinking, and I'll have reached the highest level of intelligence that it can get, and that's okay. So here's some tools that NVIDIA offers Nemo RL and Nemo Gym to accomplish what we were just discussing. This is basically a high level picture of GRPO that we were just discussing."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:28",
                    "end": "00:29:60",
                    "text": " Nemo RL being the library that we can use for generating. We'll go into more of that. And then Nemo Gym offers these different verifiers and environments as part of that RL process. Verifiers, we kind of talked about this before. This is where you can have rule-based, LM as a judge. These are essentially different."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:60",
                    "end": "00:30:30",
                    "text": " different ways of providing a reward score for a given model's prompt generation. Okay. So, RL ingredients. Getting back to what you can expect from NVIDIA to provide, NVIDIA is providing Nemotron models. You can also, of course, pre-train your own model."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:30",
                    "end": "00:31:04",
                    "text": " or take any other model from open source and start with that. We are also providing different data sets. We discussed open environments. So NemoGym, we have a number of different environments that we provide for, say, math reasoning, science reasoning, coding, instruction following, engineering, search chat, agentic tool use, and long context. So this is an example of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:04",
                    "end": "00:31:34",
                    "text": " how much reasoning improved the model's intelligence. So, Nemotron V2 was released earlier in 2025, and it was getting a score of 15 in the artificial analysis intelligence index for small models. You can see that Nemotron version three and the difference between V2 and V3,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:34",
                    "end": "00:32:09",
                    "text": " Not that much architectural difference, but more so we added this GRPO post-training loop to it. It improved its intelligence by 67%. This is an example where in one of those GRPO loops, multiple environments are meaning essentially multiple reward models. In one given loop,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:09",
                    "end": "00:32:43",
                    "text": " we would have the model try to provide multiple generations for prompt and see how it's scored against multiple different environments. All right, we're gonna get into NEMO framework a bit here. So this is a bit of an overwhelming slide, but if you go to this GitHub URL address, you'll be able to see all of these different"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:43",
                    "end": "00:33:11",
                    "text": " NVIDIA NEMO repositories, and these are all available open source. And what we are providing is end-to-end what one might need to train their own model and pre-train, post-train, and then even deploy for inference. So we have these data generation and pre-processing libraries, pre-training libraries, Megatron Bridge, AutoModel, post-training, RL and Jim."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:11",
                    "end": "00:33:45",
                    "text": " and then various inference-related libraries. All of these are developed out in the open, and also if you find them useful, you're also welcome to submit your own pull requests and help us improve these repositories. We welcome open source contributions anytime. All right. Just some shout-out to Megatron Bridge. This is a project that provides optimized recipes."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:45",
                    "end": "00:34:14",
                    "text": " When I say optimized, meaning these are recipes for training models that are optimized for NVIDIA specific hardware. An example of this is we have recipes in Megatron Bridge for many, many models. You can pre-train DeepSeag V3 or Quen 3 if you want in Megatron Bridge. And so this is different hardware from H100,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:14",
                    "end": "00:34:43",
                    "text": " to Grace Blackwell, the most recent Grace Blackwell being GB300. So you can see for what we're showing here is this is different data types, BF16, MXFP8. If you're wondering where NVFP4 is, those numbers are, we're optimizing those recipes as we speak. So I didn't have them available for this slide, but we will have them soon."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:43",
                    "end": "00:35:14",
                    "text": " But essentially you can imagine NVFP4 will be even higher number of teraflops. If you're not sure what teraflops for your GPU, just think of it as getting the most efficiency out of the GPU. So you can see FP8 versus BF16. Essentially MXFPA is using half the number of bits and it is, as you can expect,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:14",
                    "end": "00:35:39",
                    "text": " more efficient than BF16 and if MVFP4 will be even more efficient but there's always a trade-off between low precision and accuracy and then what this different graph is showing is within a given data type for H100 you're getting essentially you know three over here like"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:39",
                    "end": "00:36:08",
                    "text": " 3x boost bf16 going from h100 to grace blackwell In fp8 you're actually getting four times boost going from h100 to gb nvfp4 I think that's going to look even higher more like 5x plus boost Which is meaningful in a sense that okay if you're if you're training the model and it takes five days and you just switch to a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:08",
                    "end": "00:36:41",
                    "text": " better hardware, it can bring it down to one day, just by using a higher class GPU. Okay, let's talk a little bit about how Nemo RL, this GitHub project, fits into the training that we were just discussing. So, this is kind of like a high level set of features that Nemo RL, the project, provides. We talked mainly about GRPO, the algorithm. There's..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:41",
                    "end": "00:37:07",
                    "text": " and many different RL algorithms, and many of these are supported in NEMO RL, and there's new algorithms coming out all the time. Actually, this is missing PPO, but PPO, DPO, so SFTU is probably the main algorithm two years ago. PPO, DPO is the main thing one year ago, and so now..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:07",
                    "end": "00:37:37",
                    "text": " Now we're into the GRPO, GSPO, DAPO regime. And so you can expect the algorithms in the RL space are evolving very quickly. What is considered state of the art yesterday becomes quickly outdated thereafter. So one of the challenges is we want to also make sure that the most latest meaningful algorithms are supported in RL."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:37",
                    "end": "00:38:06",
                    "text": " and available for use. Training backends, what does this mean? So you can basically choose a path where in auto model we have a backend where it is essentially compatible with pure native PyTorch and using native PyTorch parallelisms and the advantage here is you get access to a large number of models. There's essentially"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:06",
                    "end": "00:38:37",
                    "text": " Any model on HogniFace you can use on Nemo RL. You can post-train it on RL using the AutoModel backend. Megatron is a more optimized backend. The numbers that I showed you earlier were using the Megatron backend. The only thing here is the Megatron backend just has a subset, a smaller coverage of models. And generation backends. So this is where we are in that GRPO algorithm."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:37",
                    "end": "00:39:07",
                    "text": " prompting the model that we are post-training for outputs for generations, and here we support different inference backends. So VLM, Megatron inference, SGLang is actually another one that's upcoming, and in the end we use Ray for orchestrating. On the performance side, there are a number of different models, model architectures supported. We have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:07",
                    "end": "00:39:37",
                    "text": " large MOE with all kinds of parallelisms. We support FP8 end-to-end with accuracy matching and as that previous slide showed, you do get a 20% speed up of using FP8 over BF16. So just showing that you can use FP8 end-to-end from pre-training through post-training with RL. And NVFP4 is coming soon."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:37",
                    "end": "00:40:08",
                    "text": " Async RL, I won't spend too much time talking about that, but this is essentially a feature in Nemo RL where you can run, you can do something called off-policy post-training. You can basically scale out with more infrastructure using Async RL, and so you do get, yeah, I would say more optimal end-to-end performance."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:08",
                    "end": "00:40:46",
                    "text": " in the RL loop. Okay. All right, so this is motivating why one could use a Nemo Gym project. So when one is training, doing post-training, there needs to be something that's orchestrating the different tools that we were discussing, the tool calling that a model can use."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:46",
                    "end": "00:41:16",
                    "text": " different, all the different environments, those can become different servers and resources that need to be managed as part of the RL loop. And yeah, being able to have an orchestrator that manages the data and the rollout efficiently across the policy model and the different environments. So our answer to that is Nemo Jim."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:16",
                    "end": "00:41:46",
                    "text": " NemoGym is this orchestration layer that you can use with NemoRL. NemoGym is compatible with other RL frameworks, but it's highly optimized for NemoRL. And it's been battle-tested, meaning that NemoRL, NemoGym were used to post-train Nemotron 3. And we are using it currently for the super and the ultra variants."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:50",
                    "end": "00:42:17",
                    "text": " So where does Nemo Gym come into play? So here's an example of preparing a training environment. You start with your benchmark model and what are the capabilities that we want to add to this benchmark model. And we can use existing environments that Nemo Gym provides."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:17",
                    "end": "00:42:47",
                    "text": " You can also develop your own new environments and add those into Nemo Gym and that prepares the gym environment. Now the gym environment is ready and we can begin the model training. You start training your model and verify the train model against benchmarks and validate capacity capability improvement. I am reaching the end of time here. Okay."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:49",
                    "end": "00:43:17",
                    "text": " This is a more pictorial or graphical view of what's happening between Nemo So RL is a training framework on the left hand side and gym is the gym environment on the right hand side So what we have going on here is the policy model itself is hosted and trained inside the RL framework When we do a given GRPO loop, we'll send the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:17",
                    "end": "00:43:46",
                    "text": " We'll send the the the prompts to the agent coordinator in emojim That will call the model or models to do the generations, you know that 1 through 16 attempts it'll pass those generations to the the verifiers or the environments and get reward scores for each of those generations and you'll take"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:46",
                    "end": "00:44:17",
                    "text": " the output from those and bring it back into the training loop for RL and update the weights for the original model. And so this given loop is GRPO brought down from conceptual level into how it's actually the pieces that are involved in the RL and NVIDIA Nemo RL and Gym projects. Again, you can extend Nemo Gym. You can bring your own"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:17",
                    "end": "00:44:53",
                    "text": " your own verifiers, you can bring your own training data and different HTTP endpoints to act as resources, resource servers. These are the ones that are provided out of the box by NemoJim. We talked about them a little bit already. Okay, so I'll just end with a couple of slides about how NemoTron 3, how we went from the pre-train to post-train."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:53",
                    "end": "00:45:23",
                    "text": " and delivered the final model that is now available on Hugging Face, given the context of everything we just discussed this past hour or so. So we started with pre-trained data. There's 15 categories of pre-trained data. We'll get into these in different slides. So we did the pre-train, extended the context. This is where we made it extendable to a one million token context."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:23",
                    "end": "00:45:56",
                    "text": " So now you have your pre-trained base model. That pre-trained base model, then we ran SFT on that pre-trained base model. We gave it reasoning control. And I'll talk about that in the next slide. And now you have your fine-tuned checkpoint. That fine-tuned checkpoint, then we ran it through an RL, this is a GRPO loop using RL with verifiable reward models. So RLVR."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:56",
                    "end": "00:46:25",
                    "text": " via GRPO, now you have your RLVR checkpoint. Then we ran it through RLHF using reward model, the help steer data set, this is what I mentioned earlier where it has those different five categories of safety, helpfulness, accuracy, and so this is another loop. And then we ran it through RLVR again."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:25",
                    "end": "00:46:53",
                    "text": " via GRPO. We have the final BF16 checkpoint. We ran push training quantization to shrink the models down from BF16 to FP8 data types. So it's essentially half the size now. Half the size, but still just as smart. And so now you have your final FP8 checkpoint. Going into the pre-training a little bit. Data designer."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:53",
                    "end": "00:47:24",
                    "text": " is a project that you can use for synthetically generating data. When you are, yeah, to help add more data, you can use Data Designer. Okay, base model. This is the architecture. It's a hybrid Mamba 2 MOE architecture. So the unique thing here is back originally, all the LMs were attention, attention, attention, attention, very expensive."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:24",
                    "end": "00:47:55",
                    "text": " So now we're using Mamba2 with MOEs, and there's only very few attention blocks in the overall architecture, which is how we've enabled both long context and efficiency. This is the SFT loop. The main thing here is you're able to turn off on and off thinking. We trained it with some small percent of the samples have this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:55",
                    "end": "00:48:31",
                    "text": " thinking, end of thinking token. So the model became aware of what it means to turn on and off thinking. And finally, the RLVR loops. And here's a slide just, you know, talking about Neymar RL. Again, just that one URL I showed earlier. You can get to all of these things. So, yeah, you can take a picture or just go back to, where is it? This one."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:31",
                    "end": "00:48:40",
                    "text": " Just go here. It's all there. Thank you very much for your time today. Hope this was somewhat helpful. Appreciate it."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
    {
    "_id": "6995b110565455da1fc757f4",
    "projectId": "822060",
    "projectName": "Scaling Intelligence NVIDIA NeMo and the Future of National AI - AI Impact Summit 2026 - Spanish",
    "projectDescription": "Scaling Intelligence_ NVIDIA NeMo and the Future of National AI _ AI Impact Summit 2026",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T12:31:10.565Z",
    "projectUpdatedDate": "2026-02-18T12:31:10.565Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T12:32:19.869Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Scaling Intelligence_ NVIDIA NeMo and the Future of National AI _ AI Impact Summit.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Spanish"
},
    {
    "_id": "6995b174565455da1fc75810",
    "projectId": "295220",
    "projectName": "Scaling Intelligence NVIDIA NeMo and the Future of National AI - AI Impact Summit 2026 - Hindi",
    "projectDescription": "Scaling Intelligence NVIDIA NeMo and the Future of National AI - AI Impact Summit 2026",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T12:32:50.303Z",
    "projectUpdatedDate": "2026-02-18T12:32:50.303Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T12:33:49.114Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Scaling Intelligence_ NVIDIA NeMo and the Future of National AI _ AI Impact Summit.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Hindi"
},

    {
    "_id": "69959088565455da1fc75374",
    "projectId": "473895",
    "projectName": "Leading Through AI Transitions Technology Energy and Security - English",
    "projectDescription": "Leading Through AI Transitions Technology Energy and Security",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T10:12:24.145Z",
    "projectUpdatedDate": "2026-02-18T10:12:24.145Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T10:17:10.766Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "Leading Through AI Transitions_ Technology_ Energy_ and Security(MP3_128K).mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771409536091.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:31",
                    "text": " We're driving the system bottom up. So essentially, it is from school to space. We have labs in about 10,000 schools. We're adding another 50,000. Fascinating elements of stories which are happening there. We're getting ready for you, and as Sir mentioned, the workforce of tomorrow. These are the startups of today and tomorrow. At the second level, we have one of the highest number of incubators sitting with us, the Atal Incubation Mission itself. In fact, two weeks ago, we crossed 100."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:31",
                    "end": "00:00:59",
                    "text": " And this is where we are picking up these startups, we are picking up this AI, we are picking up new technologies, nurturing them and trying to see how they are bringing about a real grassroots impact. You know what's interesting? Of those 100 incubators, do I have time? Okay. Of those 100 incubators, currently over 20 of them are in small towns and cities and they are not looking to become a financial unicorn."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:60",
                    "end": "00:01:30",
                    "text": " These are my social unicorns. Each one of them is touching issues which are touching a billion lives, not just for India, but everywhere else. And then we have the high-end ones, which are at the space, et cetera, which is what we work with directly. So it is actually creating this entire ecosystem which goes from within the government, for example, creating an entity in the Ministry of Defense, which is how can we bring transformation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:30",
                    "end": "00:01:58",
                    "text": " And how can we bring technology and innovation into the government? That's a big thing. And government everywhere will be a big adopter. So how do you help the government pick up and bring that in? And we are now taking it across a number of other ministries. And similarly, we're working with the state governments to bring in similar activities. So with this entire effort is actually joining the dots and creating that web of innovation and creating that ground up impact."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:58",
                    "end": "00:02:23",
                    "text": " which you started to beginning to see now. Thank you. Well, happy 10th birthday. And thank you for those very concrete examples. And Sarit, we were talking before about the idea that regulation should be or can be an enabler of innovation. Tell us a little bit about the Israeli approach to productive regulation to enable critical AI innovations."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:25",
                    "end": "00:02:55",
                    "text": " So first of all, thank you. Thank you for inviting me to this panel with this distinguished and honorable guests and thank you all for being here. So a few words about Israel. You might know it from other things, but I'll tell you something. A very small country and a very small market. Okay, we're 10 million people. Third of only New Delhi, I think, right? So we are a small market and we are also a country with almost"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:55",
                    "end": "00:03:21",
                    "text": " we don't have almost any natural resources. And that leads us to be very tech dependent. Our tech sector is really crucial to our economy. And when I say that we are a small market, if Israel will set its own specific regulation, I assume that companies will not develop their products and services according to our regulation. When a company..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:21",
                    "end": "00:03:50",
                    "text": " develop their products or services, they think about bigger markets, right? The EU, 400 million people, the US, Asia, not a market of 10 million people. And we have to be conscious about that when we set regulation on AI. But in the same time, we cannot also sit on the fence and let products and services come in and that might harm safety or other public interest."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:50",
                    "end": "00:04:20",
                    "text": " So the approach Israel has taken towards the AI regulation is, first of all, we'll do it for the time being sector-specific. We're not going to think about it as a horizontal regulation that covers all artificial intelligence wherever it is, but rather each sector, if it's finance, health, education, or the public sector, the government, will think about the opportunities of AI in this sector, whether the current regulation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:20",
                    "end": "00:04:50",
                    "text": " blocks it, or where the regulation doesn't give enough protection. If we think about finance, for example, or law enforcement, when we do the regulation to these sectors, we have something in mind. If it's finance, we want to think about social inclusion. We don't want people to be deprived of credit because it's crucial for their life. And if it's law enforcement, we don't want people to risk their freedom with no justified reason. So I think the idea in Israel is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:50",
                    "end": "00:05:19",
                    "text": " For the time being, we give regulators to control their areas in many other fields, whether it's privacy, consumer protection, so they can do it also in the field of AI. And for the time being, it's working. I'm not sure. There's also disadvantage to this approach. There can be a regulatory arbitrage when one regulator decides to take a very strict approach and one regulator decides on a very light approach."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:19",
                    "end": "00:05:45",
                    "text": " But recently in Israel we've established an AI Directorate. It was supposed to harmonize the activity of each sector. So that's one thing we're doing. The second thing that we're doing in terms of regulation, so I'm a lawyer and I perceive like when I, in my government, when I go to parliament, I think they see me with a stop sign on my head or something like that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:45",
                    "end": "00:06:14",
                    "text": " But it's not like my former boss used to say that artificial intelligence requires human trust. People have to trust it. If they'll trust it, they'll adopt it. If they'll adopt it, it will lead to wide implementation and so on. And when you think about it, when you're driving a car or when you put your money in the bank, you have trust in this institution. You think some, I hope so, that someone somewhere"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:14",
                    "end": "00:06:43",
                    "text": " has made enough rules that your money is safe or that the car that you're driving is safe. So regulation, in a sense, is also a way to gain the public's trust. Now, I think regulation should be measured proportionate, risk-based, but it shouldn't be perceived as a contradiction, innovation and regulation. Regulation is much of a player"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:43",
                    "end": "00:07:13",
                    "text": " in achieving innovation as much as a public investment by the country. So this is like in a nutshell our approach. We can speak afterwards on other things. Thank you for that very clear view and we'll definitely come back to the trust component that you mentioned. Nick, we've been hearing a lot about the role of government, but I'd also like to hear your perspective on the role of firms and firm level governance."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:13",
                    "end": "00:07:40",
                    "text": " in this story because I think it's certainly part of the picture. How do you see firms engaging governance to really galvanize critical innovations? Yeah, thank you Tanya and thank you everyone for being here. It's a great privilege. Tanya, when you started you laid out kind of three big goals. We want kind of resilient climate enhancing infrastructure, energy infrastructure. We want a really dynamic innovative economy that creates opportunities and solves problems for us."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:40",
                    "end": "00:07:51",
                    "text": " And we want really clear, safe, secure, resilient, self-determining, stable societies with a good, you know, defence wrapper as part of all of that. And the problem is that triangle."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:51",
                    "end": "00:08:21",
                    "text": " is a little bit unstable, right? When it comes to new technologies, often the Jevons Paradox means the more efficient we get at delivering a service, the more emissions we produce because demand goes up. States excited about new technologies like AGI can start to worry about their geopolitical kind of relationships and get into a race mentality in technologies. We've seen that before. And when it comes to the climate resilient infrastructure versus security, often cost and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:21",
                    "end": "00:08:49",
                    "text": " alliances are relied upon to get you know materials flowing and so you know the point that my esteemed colleague from Israel has made is it's really important to think about what are the rules that then stop that from fracturing so that we can rely on having all three good things rather than just allowing us to fall into one corner of the triangle but it's completely right because once you have no matter what the rules and regime is and speaking as a lawyer"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:49",
                    "end": "00:09:18",
                    "text": " A regulation that's written down is worth absolutely nothing to industry. We know from, particularly from studies in areas like work health and safety, unless you have very clear signals of intent and enforcement of those rules, people will just push the boundaries of it. And Australia, like Israel, and also like India with the governance guidelines released last year, we're taking a bit of a light touch approach to this. We're saying that what should happen is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:18",
                    "end": "00:09:45",
                    "text": " Each sector should be looking at technology neutral rules as much as possible and what we're seeing in Australia is we're trying to really give life to the point that you made which is regulation and governance is an enabler of innovation and so it's in the interest of every business, particularly businesses with a bit of public profile, to invest in actually doing the right thing from the beginning."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:45",
                    "end": "00:10:14",
                    "text": " The one problem I see with governance, one of the key ones is organizations as they experiment with artificial intelligence, they sometimes get caught in some of those triangular instabilities in terms of starting to race against another competitor. And if you race and you think if you're a chief executive or you're a senior executive in a company, if someone says I sat next to a tech"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:14",
                    "end": "00:10:33",
                    "text": " a tick person on an aeroplane or a bus and I think we should invest in AI. And that's the reason you invest in AI in your company. You are much more likely to fall foul of the regulation and the governance issues than if you say what are the problems we're trying to solve and then what are the processes we need to put in place."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:33",
                    "end": "00:11:03",
                    "text": " to build that trust with our clients, customers and stakeholders. And so for me, this big picture and this small picture of decision making that all of us make within the organizations and even in our families, it's a really important link that we have to be consistent in and policy really has to support the incentives for us to think like that throughout the stack. Thank you. And I really want to continue on this theme of trust, which I think is already emerging as very critical to this conversation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:03",
                    "end": "00:11:31",
                    "text": " One thing that it's worth acknowledging is that when we think about transitions or disruptions, they come with trade-offs. So if we think about defense and security, for example, the need for defense innovation comes with higher public spending. I know that India in the budget will be increasing its public spending on defense by about more than 15%. India is, of course, also a country of huge scale. And so I'm curious about..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:31",
                    "end": "00:12:01",
                    "text": " how public trust is created in these new resilience building measures across the population. And I'd like to also just start this question by acknowledging that, you know, this event in itself is highly inclusive. People have come from all parts to be here. It's also very evident still the G20 signs from 2023 that are very evident across the country that that was an event that was really across states."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:01",
                    "end": "00:12:31",
                    "text": " and involve many, many people. So Mr. Sringla, perhaps you can tell us about how you build public trust when it comes to new resilience building measures. Tanya, you're right. I mean, I think the word trust is very important here. And when you talk of summit like this, you know, our whole, I think, purpose of being able to host a summit of this nature is to make it as inclusive as possible."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:31",
                    "end": "00:12:59",
                    "text": " to reach out to as many people in our own country as possible. So if you notice, it's not just those of us who are involved with AI, but there's a lot of people from the public who are here at the conference venue, at the exhibition. Yesterday, I was at a restaurant and the waiter said, I'd like to visit the exhibition. I was a bit surprised because I said, hello, isn't it for professionals? He said, no, my children are very keen. Then I realized that there's a public interface here."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:59",
                    "end": "00:13:26",
                    "text": " Which is great. I think this is the way it should be. And G20 was similar. G20 was a rare moment when we assumed the presidency. And so I think the Prime Minister decided that instead of making it an event that was just based in the capital, that we would broad base it and take it to every city, every state of our country. And as a result, I think people across the country, whether it's in..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:26",
                    "end": "00:13:54",
                    "text": " you know, in Gujarat or in Nagaland or in Jammu and Kashmir or Kerala were able to feel that they were a part of that exercise. So inclusivity is critical. I saw that the CEO of Anthropic yesterday spoke about, you know, this opportunity of the AI summit to democratize the AI. He said that India has the opportunity to democratize the AI to make it more effective. And I think this is what it is all about. I mean, our effort is to democratize it not only in our own country."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:54",
                    "end": "00:14:20",
                    "text": " but also across the global south. Your point is important that as you invest in AI and as you invest in systems that would be important as you move ahead into the future, the costs are more. Of course, public trust is important here. The public needs to be part of that exercise."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:20",
                    "end": "00:14:46",
                    "text": " If you look at it in one way, I think the public understands that there is a need for an investment in your public security. In other words, that you face external challenges, you face internal threats. How do you make sure that you are able to provide a secure environment for your people? How does the same technology work to your benefit, whether it's the use of drones in agriculture and farming, for field mapping?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:46",
                    "end": "00:15:16",
                    "text": " or for land zoning, there's always property disputes in India because of the fact that demarcation of land holdings is not clear. We're now using drones to map it out and AI to make sure that this particular mapping is accurate. So use of AI, I think, is critical in building the public trust. Many of us here would remember that during COVID, we had a software called COWIN that ensured that two billion vaccinations were administered to our people."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:16",
                    "end": "00:15:44",
                    "text": " We did it across the country in a very, very smooth manner which meant that you could really take it up to scale and at the same time you had systems that enabled everybody to be identified in a database. But imagine if AI were to be used in that, how much simpler that exercise would become. So I think again the public is an important part of that. They have to understand that whatever you do is for their benefit."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:44",
                    "end": "00:16:13",
                    "text": " I personally don't believe that any money should be spent without public trust and confidence. I mean, it's very important, you know, I have to leave a little early because I have to go for a parliamentary committee that scrutinizes the budget that is now present before us. And I think it's very important to be there because I want to make sure that, you know, I am present to see that whatever is proposed is something which is, which is, you know, what we can live with. It's for the public interest. I mean, everything is for the public interest, but you have to make sure that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:13",
                    "end": "00:16:42",
                    "text": " every small part of that proposal lives up to that commitment. And so I think when you talk about EI, when you talk about futuristic systems, you're investing in the future. And I think for India, it's very important that we invest in the future. We must leapfrog technologies. We cannot afford to be moving in the normal way. We have already found that in many senses, certain countries have done that and have got ahead in a way that is not always"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:42",
                    "end": "00:17:11",
                    "text": " helpful or constructive. We want to be constructive players. We believe in human-centric globalization. How do you use technology for the benefit of people across the world and in our own country? And we want to be that player, that constructive player in the global stage that can use technology, use development to help our own people and help others. Thank you. And Mr. Bagla, in creating a culture, let's say, of innovation and entrepreneurship,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:11",
                    "end": "00:17:42",
                    "text": " across demographics in the country, how do you feel that that contributes to this development of social and economic resilience? So the first point, we go back to inclusion and empowerment. The moment you have that, you have a greater sense of trust which comes in. But let's look at trust per se. And you're thinking of the government's role in creating that trust. Won't you think, so one is policy reflects the mood of society, right? So you need to read the mood."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:42",
                    "end": "00:18:09",
                    "text": " and then create a policy around it to create that trust. Won't trust be a responsibility of every stakeholder in the system, which is every citizen, every private entity, every government entity, the government as a whole and as an individual? So it's a very collective element which happens. And now think about the ability of creating trust. You know, in the Indian context, you will notice that whenever there is a challenge between innovation and regulation,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:10",
                    "end": "00:18:42",
                    "text": " I think the sense always is that we tend to tilt towards innovation. I mean, we are more innovation happy and as you're saying, we need to create the new ideas and leapfrog into new situations. But at the same time, look at the role of a policymaker. Is it to enable or is it to regulate? That's a very difficult role in the current context as we are going forward. Also because as we discussed,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:42",
                    "end": "00:19:13",
                    "text": " disruption is happening every moment. When the core is changing, let's say you're entering into a world where everything is possible, but nothing is certain. Now, to create a structure around that scenario with creative destruction at its peak, I think one of the most challenging jobs in today's world is that of a policymaker. And the element of creating trust,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:13",
                    "end": "00:19:43",
                    "text": " also comes in with a lot of other elements which are changing by the day. So from that perspective, what we are now trying to work is that the moment you have inclusion and empowerment, if every individual is empowered to see what they want is right or wrong for them, and have the ability to be included in the process, then they are a stakeholder in the process. And the moment you are a stakeholder in the process,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:43",
                    "end": "00:20:13",
                    "text": " You are a part of the trust-building effort. And I think that is where when you're looking at a bottom-up approach and a top-down, and they all meet. So we moved away from an essentially top-down approach. It's all driven bottom-up. And that is what creates the trust. That is what creates the flexibility. That is what reflects the mood of the society. And that's what enables a collective sense of trust to get evolved. That's what we are doing."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:14",
                    "end": "00:20:40",
                    "text": " Thank you, and I want to pick up on what you mentioned around the role of the the policymaker and the idea that trust takes coordination it takes new kinds of capacity building as well and I think what often we're seeing across the three domains we're talking about critical technologies Energy innovations defense and security innovations is that often they don't neatly fit the structures of government"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:40",
                    "end": "00:21:11",
                    "text": " And I think that this can be very, very challenging for those who are trying to think through the policy and regulatory landscape. So, Sarit, I'd like to ask you about the experience in Israel of inter-agency cooperation, of how you see ministerial collaboration happening, and perhaps the cross-sectoral level as well. And if I can add, it would be interesting to hear about the role of the new artificial intelligence directorate playing that coordinating function. So, thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:12",
                    "end": "00:21:42",
                    "text": " Well, I think I think Israel will not be special in that sense that we work also in silos and AI being the new and I'll be a bit cynical now, but like the new shining toy everybody also wants to have like a stake in it and In these circumstances, I won't lie. It's hard to cooperate. Everyone wants to control like AI and and also in Israel we we experience lately like an political we're not like"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:42",
                    "end": "00:22:09",
                    "text": " in political instability, which means that we have like elections quite often, which makes it hard to think of long-term policies. So in that sense, we are really looking forward to this new directorate of AI, which is led by professionals who will have the opportunity to think about it long-term, how Israel wants to lead in AI in five, 10, 15 years."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:09",
                    "end": "00:22:37",
                    "text": " I don't know if we went through 2047, but at least 10 years ahead. So I think that's one way to break the silo, and also this directorate will be located at the prime minister office, and I think that's also to give a sense how important, like the importance that Israel seen it. Now, until now, there was, I would say, organized chaos in Israel for what concerns the AI, but I think it's..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:37",
                    "end": "00:23:06",
                    "text": " It was also a good thing, right, because it gave some kind of, like people on the ground, companies, they worked on AI, they tried to promote AI, and it gave us, like as government, like a period of time when we can see where the market is heading, what are the needs of Israeli companies in order to thrive in AI. So I think the directorate comes in a good momentum, not too early and not too late."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:06",
                    "end": "00:23:36",
                    "text": " And Israel will be to understand where Israel has advantages, what are our disadvantages, and to allocate our resources. And resources are not just money. It's time, right? It's time of policymakers who have a lot on their mind, and you have to catch their attention for a certain moment. And think about regulators in the transportation or on the education. Their main role"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:36",
                    "end": "00:24:03",
                    "text": " is not to encourage AI. If you're the Minister of Education, your main role is that the education system will function. You as a Directorate will have to give them resources and the motivation to invest time in AI. If you're a financial regulator, it's the same. You think about your consumers or financial stability. You don't think necessarily about why AI is important. And I think the Directorate role will be as well to..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:03",
                    "end": "00:24:30",
                    "text": " give them resources to catch their attention and to push them in the direction to invest in AI and to make AI part of their daily thinking. Because I think all sectors will be impacted. There is no way around it. And another thing that I want to mention regarding the directorate is I think in Israel at least, the government"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:30",
                    "end": "00:24:56",
                    "text": " is you spoke about public spending. So our public spending in defense was always high and in the last three years it's skyrocketed. So we need to think about how we make our government more efficient and AI is a part of it. I'll give you just an example and it also has to do with trust, right? When you think about how you can use AI,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:56",
                    "end": "00:25:24",
                    "text": " We see all these sometimes like doomsday example, right, like incarceration. But why do we need to start there? We can start like in so many simple things that can improve our governance. I'll give you an example of history in Israel. In Israel, there is a long wait for public housing. We don't have a lot of public housing and the demands exceed what we have to offer. Sometimes people rate..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:24",
                    "end": "00:25:53",
                    "text": " six, seven months to get an answer whether they're eligible to, and why. Because the people who ask for public housing, they, when I was a law student, I volunteered in a center like that. They come with a bunch of paper. They don't get into a form in the computer and put like very organized, like this is my salary, the number of kids I have. They come with a lot of paper. And until, I don't know, six months ago, seven clerks took this, all of these papers."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:53",
                    "end": "00:26:22",
                    "text": " and made a chart of whether they're eligible or not and brought that to a committee. It took about six, seven hours to sum up one request. They brought AI to do this summary. There is a human in the loop that is looking that there were no written mistakes and that it shortened the time of the committee to receive about 70%. I assume it will be better because the system will be better and the human in the loop will have less work."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:22",
                    "end": "00:26:51",
                    "text": " Now, it's not a risky use. All the AI is doing is making information more organized. It helps with public trust because people know that this AI works for them. In that sense, I think the government is leading by example, that we adopt AI, we adopt it in a responsible manner. I think that will also help."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:51",
                    "end": "00:27:22",
                    "text": " also in gaining trust, but also to show the companies or regulators, the government is doing it, the government is using it, that means we are in a state where we can go to the next level with AI. Thank you. Mr. Schringler, I know you have to go soon, so I'll come to you for one additional comment. I want to draw on your experience as Foreign Secretary. I think a lot of the challenges that we're talking about cross borders, we're thinking about energy defense."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:22",
                    "end": "00:27:54",
                    "text": " critical technologies, these are national stories, but they're very much also regional. And so I'm curious to hear from you about the role of international cooperation in galvanizing the changes that you'd like to see, and in particular to think about the role of regional alliances in this picture. Well, you know, you have a global governance structure. You've got the United Nations, you've got the Bretton Woods institutions, you've got the WTO."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:54",
                    "end": "00:28:20",
                    "text": " But none of them are geared up for technologies that represent the future. So whether it's AI, or whether it is use of, say, critical minerals, or whether it is any futuristic technology, how do you, as Deepakji said, how do you regulate that space? How do you ensure that there is some level of international governance where every country has a certain responsibility?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:20",
                    "end": "00:28:50",
                    "text": " So how do you use AI responsibly? I mean, I think I fully agree about the disruptions that AI could cause. I mean, the disruptions could be in many senses. I mean, there's already concern in our country about the disruptions in terms of employment or unemployment that AI could cause because it would make a whole lot of people redundant. On the other hand, AI could also be used in a manner that could be socially, I would say, of concern. For example, to the social media."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:50",
                    "end": "00:29:20",
                    "text": " fake messages, fake posts, using AI to morph and change pictures and so on and so forth. This is something that could be beyond just individuals, but it could affect societies. It could lead to incidents. It could lead to unrest. So how do you deal with that and how do you deal with it at a global scale? Take a simple thing like cybercrime. Cybercrime today is a global syndicated operation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:20",
                    "end": "00:29:49",
                    "text": " Your something taken from your account illegally through the means of technology is going into 60 accounts and then you know 600 accounts and then going into Into cryptocurrency and you can't trace it. So how do you regulate crypto for example? So I think you need another global governance Let's say framework that can help countries across the board provide some regulation I don't know in what sense I think the AI summit is a good place for a discussion"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:49",
                    "end": "00:30:15",
                    "text": " of all of those concerns. And I think some work is being done by the United Nations as well. But I think it needs more than just the UN because this is technology which is cutting edge, which is not just there but might come up in future. You need people who have a good idea and some of our friends here are, I think, well-placed to provide us with that. And Deepak quite correctly pointed out that I'm probably the only exception in this room."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:15",
                    "end": "00:30:45",
                    "text": " My only familiarity with AI is that my son deals with AI. But having said that, I think there's a lot more that we can do in terms of bringing some sort of control over this entire framework that you have, and not just nationally, but internationally. So I'll stop there and I'll also beg your indulgence to- Sorry to come in the middle of the session. As Sri Harsh Vardhan Singla has to leave, so I request Saroj Bharatwaj from our team to give a memento to him."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:04",
                    "end": "00:31:32",
                    "text": " Thank you very much. Thank you sir. Thank you. And Nick, I know that you have both very extensive experience when it comes to international cooperation but also a particular view on the role of perhaps the global competition narrative that's emerging. Tell us about your thoughts on this."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:32",
                    "end": "00:31:58",
                    "text": " The point that was made at the beginning of the session was regulation as an enabler. And certainly international coordination is also an incredible enabler of the safe and responsible use of AI systems for many reasons. But one of them is because it doesn't just lead to products and services that are more likely to be trusted, but they're actually more trustworthy."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:58",
                    "end": "00:32:27",
                    "text": " because you can move from country to country, service to service, and feel, okay, there is a standard that I'm aware of that this is based upon. At the organizational level, the kind of predominant standard in the AI space is 42,001, ISO 42,001, which is a management standard. So it's a standard that attaches to the use of the organization, the processes around by which you know whether what you have."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:27",
                    "end": "00:32:54",
                    "text": " works and is safe. And in my mind here, on both the international level and all the way down to us using things individually, there are three big questions that I think we have to be really clear on. And this includes in work in the military domain that I've done with the Global Commission. And the first one is that it's the right tool for the job. So within an organization or across countries,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:54",
                    "end": "00:33:21",
                    "text": " this is actually going to do what it says, and it is suited to that task. And just for example, I don't use Claude or GPT-5 for most of my summarization work. I have GPT-OSS running on my hard drive, and it's a much lighter use of the thing. I don't need an internet connection, and I know how much energy it burns, which is directly from my battery. And choosing the right tool for a job, whether it's robotic process automation or others, is an absolute key."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:21",
                    "end": "00:33:45",
                    "text": " But the second one is the team, the group of people that are using it together. And I think this is the most under-investigated thing when it comes to generative AI. We all use it individually, but at an organizational level, it's actually how teams use it together to get an outcome in a ministry or your example of lawyers helping someone make an application. The whole process has to be adjusted and people have to know what the..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:45",
                    "end": "00:34:14",
                    "text": " system is doing. That is also true at the international level. So when groups of countries come together and increasingly track two and track 1.5 dialogues where these are the key people in labs and tech firms talking to each other about geopolitically sensitive questions, there has to be enough transparency about what is happening in these systems. That's probably the sticking point at the moment in terms of international competition."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:14",
                    "end": "00:34:42",
                    "text": " At the very bottom level, you need to understand the trade-offs. And I think that's where sessions like this really start to think, well what trade-offs are the most important for us to manage? And in the military realm, one of the most important ones is human agencies and control over the deadliest of systems. And that's the one where a lot of focus has been at the UN level recently on military systems. Thank you, Nick."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:42",
                    "end": "00:35:10",
                    "text": " I want to give both of our additional speakers a minute just to talk about their country's perceptions of the role of international cooperation before I hand over to Shyam Krishnakumar for some closing remarks. So as I mentioned before, Israel is a small market. We want to make our own unique regulations where we are very much looking towards the world to set some standard that we can adopt because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:10",
                    "end": "00:35:40",
                    "text": " we understand the limits of our regulations. So we are looking with anticipation for these standards to sink in. I do want to say also that sometimes, I'm not a policy maker, I'm a legal advisor to policy makers. And when I have to come to parliament or when I have to come to meet with the minister and I have to come with a recommendation, I look at international papers. And there are so many of them. And they're the OECD and then the UN and the UNESCO and..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:40",
                    "end": "00:36:07",
                    "text": " and they're all very valuable, but I think we need to start narrowing it down and think what is the unique advantage of each of these papers and what we can take from them and think how we, I think we are in a stage of limited resources also for these organizations and how we make this work more collaborative and how we can take the most out of it because we, and I'm sure that Israel is not unique in that sense."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:07",
                    "end": "00:36:34",
                    "text": " We're very much looking to see these works and to learn from them and to implement them when we do our AI regulations. So that's one thing. I also want to say, even though it's not, but I'm Israeli. Just because of the discussion that was here earlier, there is also so much that you can expect from regulation. It was said here, for example, this information and fake news. Regulation will not..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:34",
                    "end": "00:37:02",
                    "text": " encourage critical thinking among the kids, right? Regulation will not explain my father why not everything in child GPT is true and he will call me and to say, you know, child GPT told me that and that and I have to say to him that it's wrong. So and regulation will not make any caution like all cautionary measures that has to do with AI. So I think you were talking about trade-off and everything but I think also we need to think about what we do except regulation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:02",
                    "end": "00:37:36",
                    "text": " if you want to have trust in AI. And so I'll sum up with that, and thank you so much again for having me. You know, you would have seen where global cooperation is concerned or partnership. I think India's position is very clear. Look at our ethos. You know, in Sanskrit, and the prime minister often uses it, is Vasubhadev Katam. Vasubhadev Katam, right? The world is one family for us. We are very clear. It's nothing but that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:36",
                    "end": "00:38:07",
                    "text": " And for us, family is number one, like for all of us. And within that family, we expect everyone to cooperate. And we want to do it. So for us, cooperation is a part of our DNA. The second interesting thing is that for the first time in human history, until now, if you look at our entire history, it was defined by geographical borders. And then came nation-states, etc. For the first time, we are now coming to a position"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:07",
                    "end": "00:38:31",
                    "text": " where there is a new agent which doesn't understand physical boundaries or borders. That is a completely different challenge. And for us, that challenge fits in very well with us because we in any case look at it as a good big partnership family. Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:31",
                    "end": "00:38:44",
                    "text": " We are fortunate to have partnered on this event with two organizations, with the Pranava Institute and with RKA India, and I'd now like to turn to Shyam Krishnakumar for a short closing remark."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:44",
                    "end": "00:39:32",
                    "text": " As we approach the closing of the session, we would like to thank Shri Deepak Baglaji for his presence and insights shared. As he's heading towards another session soon, we would like to present his memento. I invite Ms. Saroj Bharadwaj, Ms. Shefali Shyam and Pooja Banerjee from METI to present this token of appreciation to him. Before handing over to Shyam for closing remarks,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:32",
                    "end": "00:39:48",
                    "text": " You're welcome to join us tomorrow afternoon at 3.30, where we will continue the conversation on tech diplomacy at the British Pavilion. Thank you. And over to Shyam. Thank you, Tanya. And there's two minutes to go on the clock. I'm Shyam Krishnagama from the Pranava Institute. It was an honor."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
   {
    "_id": "69959206565455da1fc75396",
    "projectId": "745731",
    "projectName": "Leading Through AI Transitions Technology Energy and Security - Hindi",
    "projectDescription": "Leading Through AI Transitions Technology Energy and Security - Hindi",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T10:18:45.314Z",
    "projectUpdatedDate": "2026-02-18T10:18:45.314Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T10:19:35.817Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Leading Through AI Transitions_ Technology_ Energy_ and Security.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Hindi"
},
   {
    "_id": "699592bd565455da1fc753a6",
    "projectId": "866583",
    "projectName": "Leading Through AI Transitions Technology Energy and Security - Spanish",
    "projectDescription": "Leading Through AI Transitions Technology Energy and Security - Spanish",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T10:21:47.539Z",
    "projectUpdatedDate": "2026-02-18T10:21:47.539Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T10:22:46.965Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Leading Through AI Transitions_ Technology_ Energy_ and Security.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Spanish"
}, {
    "_id": "6995975a565455da1fc75422",
    "projectId": "980366",
    "projectName": "AI for Energy Digital Twins and India's Energy Stack - English",
    "projectDescription": "AI for Energy Digital Twins and India's Energy Stack - English",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T10:41:29.560Z",
    "projectUpdatedDate": "2026-02-18T10:41:29.560Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T10:45:46.223Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "AI for Energy_ Digital Twins and India_s Energy Stack(MP3_128K).mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771411280194.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:11:05",
                    "end": "00:11:33",
                    "text": " Thank you for joining us today and good evening ladies and gentlemen. On behalf of the International Solar Alliance and our partners, it is my privilege to welcome you to this session spotlighting the global mission on AI for energy. Today's session brings together global leadership to discuss what could become one of the most consequential missions of this decade. To formally open the session, I invite Mr. Karan Mangothra"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:33",
                    "end": "00:12:09",
                    "text": " chief of strategy and new initiatives International Solar Alliance to deliver the welcome address. Good evening everyone. You know will AI overwhelm our grids or will it optimize them? Will solar remain underutilized or will it be intelligently orchestrated across rooftop, substations and markets? Will this transition be centralized and exclusive or decentralized and citizen driven? These were the questions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:09",
                    "end": "00:12:38",
                    "text": " of why we began this journey. A few years ago, the conversation was simple. Deploy more solar, and we did. The world added record levels of solar capacity, roughly around 400 gigawatt in a single year. Costs fell, ambitions rose, and the solar revolution took hold. But as solar scaled, systems grew more complex. Rooftop panels multiplied."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:38",
                    "end": "00:13:08",
                    "text": " distribution networks became bidirectional. Electric vehicles and storage entered the grid. Now at the same time, AI surged. Powerful, energy intensive, yet transformative. Now what we realized was something very fundamental, that AI needs energy, but energy systems also need AI. And if we shape this convergence well, the future is extraordinarily promising. Now imagine grids."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:08",
                    "end": "00:13:32",
                    "text": " that anticipate stress before it happens, distribution networks that self-optimize, and rooftop solar aggregated intelligently, peer-to-peer markets empowering citizens, and cleaner power driving smarter economies. Now, it's not just about technology for its own sake. It's about reliability, affordability, and resilience for people."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:32",
                    "end": "00:13:58",
                    "text": " At the International Solar Alliance, a global south anchored intergovernmental body, we are committed to ensure that this transformation is inclusive. AI-enabled energy systems are deployed not only in advanced markets, but where the demand is rising the fastest and the impacts matter most. Through practical, use case-driven collaboration, we are moving from ideas to implementation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:58",
                    "end": "00:14:27",
                    "text": " Now, on behalf of the ISA, I would like to sincerely thank the Ministry of New and Renewable Energy, Ministry of Power, Rural Electrification Corporation, and FSR Global for their leadership and partnership on this. I would also like to acknowledge the ISA Multi-Donor Trust Fund partners, GAP, and our knowledge partners, Systemic, for helping us shape this important initiative. Most importantly, I would like to thank and welcome all our distinguished panelists and participants."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:27",
                    "end": "00:14:59",
                    "text": " for being a part of this journey. So we began all of this with solar deployment, but today we stand at the intersection of solar and AI. And the future, if we design it well, is intelligent, inclusive, and deeply citizen-centric. So welcome you all and look forward to an engaging session. Thank you, Mr. Mangotra, for setting the tone and reminding us why this movement demands collective leadership. As we move from vision to architecture, this next segment of our session highlights a key enabler."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:59",
                    "end": "00:15:35",
                    "text": " the India Energy Stack. Let's begin by watching a short explainer video."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:35",
                    "end": "00:16:05",
                    "text": " solar panels on roofs, wind farms, electric vehicles, batteries, smart meters and smart appliances. So, as India grows, the challenge is not generating power. It's helping these parts coordinate in real time. This gets complex. Let's break it down. Meet Rani. She runs a small catering business from her home kitchen. Her family has rooftop solar and an EV scooter."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:05",
                    "end": "00:16:33",
                    "text": " Rani wants four simple things. Power she can afford, power she can rely on, bills she can understand and options that let her benefit from the energy she produces. Today, this experience is fragmented for everyone. Joining a new energy program means paperwork, manual checks and long waits. Data is locked in different systems."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:33",
                    "end": "00:17:01",
                    "text": " services struggle to connect to the grid. This is the coordination gap that we need to solve for. To bridge this gap, India is building India Energy Stack. IES is Digital Public Infrastructure for Electricity. Think of it as a network of digital highways that allow all utilities, service providers, consumers and even devices to communicate and work with each other securely."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:01",
                    "end": "00:17:32",
                    "text": " IES is built using foundational building blocks that can be used to build energy services, identity and addressability, registries and directories, interaction protocols, energy credentials and policy as code. What does this change? For utilities and system operators it enables faster problem solving, better planning and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:32",
                    "end": "00:18:02",
                    "text": " Lower costs because everyone can finally talk to each other. For regulators, stronger oversight with clear data, transparent reporting, better decision making. For innovators, a bigger opportunity. Startups can build new energy services without starting from scratch every time. And for Rani, real empowerment. She's no longer just paying for power."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:02",
                    "end": "00:18:29",
                    "text": " she's earning from it. She gets paid fairly for her extra solar energy, charges her EV smoothly, understands her bills and gets help quickly when she needs it. IES opens the door to this kind of energy agency and livelihoods at scale. Imagine women led groups in small villages and local entrepreneurs across the country."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:29",
                    "end": "00:18:59",
                    "text": " delivering trusted energy services to their communities. IES is not just a piece of technology. It is the foundation for coordination, trust and reliability on a national scale. It creates a cleaner, more reliable power system, where electricity becomes not just a cost, but an avenue for livelihoods and empowerment."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:59",
                    "end": "00:19:50",
                    "text": " One digital grid. India Energy Stack. An initiative by the Ministry of Power. Built in India for the world. To help deepen our understanding, I now invite Ms. Shweta Ravi Kumar, Executive Director, FSR Global, to please make her presentation. Good afternoon, everyone. Thank you so much for this opportunity. I'm Shweta. I head FSR Global. And I'm going to unpack some elements of the video that you just saw. So it's an absolute privilege to be opening"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:50",
                    "end": "00:20:19",
                    "text": " on the India Energy Stack at a global forum at this session. So let's just rewind back. India in 2014 was at the same crossroads trying to transform the banking sector. And today the energy sector is at the same crossroads. What does this mean? Back in 2014 as well, India had banks. It had phones. It had a lot of money that was transacting. But it did not have a common digital rail to connect them all."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:19",
                    "end": "00:20:45",
                    "text": " UPI happened. It was not a new bank that the government came up with. It was not a new app either. It was a simple protocol that unleashed the power of a two trillion digital payment economy today competing with the likes of the visas and the masters across the world. So this is the unlock that a simple digital public infrastructure approach can enable at the world level. So coming to the power sector."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:45",
                    "end": "00:21:13",
                    "text": " We currently have all of these different moving parts, be it a rooftop solar owner in Jaipur, or an EV startup or a CPO in Bangalore, or an energy regulator who wants to make sure it's a consumer centric plants are being made in Chennai, or a cooperative in rural Maharashtra. All of them use different systems and interact with a different set of stakeholders. We are now enabling through IES, the common digital rails for them to communicate with each other in a seamless way."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:14",
                    "end": "00:21:39",
                    "text": " So the sector is now ready to connect. So what does this mean? For all of us who are in the sector, we know that there is a lot of deployment in terms of SCADA systems, smart meters. India's planning to connect 300 million households through smart meters. But all of these different systems have its own unique language that they interact with. Interoperability does exist."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:39",
                    "end": "00:22:06",
                    "text": " but it is not in a seamless way that we would be able to extract at the scale that we're talking in the case of India. So here, how can we move from siloed to more shared information? How can you have interoperability at the heart of design so that you make that seamless interactions possible? And create an ecosystem that allows for innovation to scale up all of these leapfrogs that we are envisioning for the sector."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:07",
                    "end": "00:22:34",
                    "text": " So this is where IES comes in as a trust bridge. Let's look at some of the examples that are more closer to all of us in our day-to-day lives. For instance, how if every meter, a solar panel, or an EV charger had a verified digital identity, much like how all of us as citizens of this country have Aadhaar, which was a DPI. Again, if an app can discover, transact, or settle energy services,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:34",
                    "end": "00:23:03",
                    "text": " across utilities, say we move from one city to another, that would be very similar to the UPI that we have today, that we also generously use on a day-to-day basis. I don't think any of us carry cash anymore in our wallets. If there are anyone, you can raise your hand. For regulators, how can they write and make evidence-based decision-making more easily, much like how GST unified the country through a more centralized and federated way of going about this?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:03",
                    "end": "00:23:28",
                    "text": " Likewise, if you want to have prosumers, like we saw in the case of Rani in the video before, wants to sell her surplus power and exchange it with another customer from within her city or beyond in another state, then you have other DPI's like ONDC, which will facilitate for this. So, India Energy Stack is not either other UPI, ONDC or GST. It's all of the above."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:28",
                    "end": "00:23:52",
                    "text": " So it's nothing but a set of protocols and specifications that enable for this interaction to happen in a trustworthy manner. What it essentially does is it gives us a common language, a taxonomy, a data model that machines and humans alike can understand, and to help us scale solutions from pilot to actual implementation across the country. So how is this being done?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:52",
                    "end": "00:24:08",
                    "text": " When we say a digital public infrastructure, it's not a monolithic solution that is being developed to be deployed across the ecosystem. It's built on foundational building blocks. So think of this a bit like Lego bricks that you're trying to put together to make whatever."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:08",
                    "end": "00:24:36",
                    "text": " that is that you're building as that particular stakeholder. So as of now, it's work in progress for us as the team. It's about six months old that this concept is in various forms of iteration. But we have five building blocks here. Identity and addressing so that we recognize each other, be it a system or a person. Registries and directories, which is like a lookup table for information to exchange across multiple systems. Interaction protocols. The power system already has"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:36",
                    "end": "00:25:06",
                    "text": " a lot of protocols that's being used. But if you want to bring in new services, how do you sort of make the old meet the new? And how do you sort of bind them together? You have energy credentials, as the video was also mentioning. How can I make, like something like EKYC, how do I make my onboarding as a consumer into the utility ecosystem more seamless, be it a commercial consumer or a residential consumer? And what if we can codify all of the policy that we all engage with on a daily basis?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:06",
                    "end": "00:25:30",
                    "text": " It could be for a utility to understand what tariff applies to which consumer, or it could be for researchers to understand how do you bring in the next kind of policy intervention to work with the policymakers. So if we can codify a lot of the policy, it becomes easy for us to scale and enable analytics over and above that. This is where the AI unlock happens. What IES will do..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:30",
                    "end": "00:25:59",
                    "text": " because of its standard data models and organizing information in a common way, would allow for the data that is needed for AI to be able to become that intelligence layer on top to give any of us whatever questions we may have. It could be a grid operator asking for what should be the reliability or flexibility measures that they need to keep in mind, or it could simply be me as a consumer trying to understand my own behavior, or it could be for policymakers or businesses to know"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:59",
                    "end": "00:26:24",
                    "text": " Where do I park my money? And how do I make those investments? So basically, IES would act as that bedrock over which AI can literally unlock AI agents for a billion. And that is what we intend to demonstrate in two days from now and continue to do so through the many months and years. And not just that, we as a country are now moving away from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:24",
                    "end": "00:26:48",
                    "text": " the conversation on energy access to energy agency. Because we are democratizing power here, that means self-help groups can now invest into decentralized systems and participate as an active participant in the ecosystem. College students can become part of the solution building because we're building or opening up the innovation ecosystem through accelerators."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:48",
                    "end": "00:27:12",
                    "text": " The agrarian society, the farmers, can now participate in offering flexibility and in demand response by simply adjusting some of their consumption patterns. So IES unlocks micro-entrepreneurs and livelihoods at a scale of a billion. And so consumers are now becoming participants, passive becomes active, and access becomes agency."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:12",
                    "end": "00:27:42",
                    "text": " And that's the vision that we have at the India Energy Stack. So with that, this is the idea that we would love to give built in India for the world. Thank you. Thank you, Shweta, for that very insightful overview of how digital public infrastructure can translate ambition into scalable systems. Now moving on to the core conversation of this session, a fireside chat titled From Vision to a Citizen-Centric AI Mission."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:42",
                    "end": "00:28:34",
                    "text": " It is my pleasure to invite our distinguished speakers, Mr. Hemang Jani, Senior Advisor to the Executive Director for India, Bhutan, Bangladesh and Sri Lanka World Bank. Mr. Henry Verdeer, Director General, INRIA Foundation. Please join us on stage, sir. Mr. JVN Subramaniam, Joint Secretary, Ministry of New and Renewable Energy, Government of India. And the Chair for this segment, Mr. Ashish Khanna, Director General, International Solar Alliance. A big round of applause, please. Over to you, sir."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:39",
                    "end": "00:29:12",
                    "text": " Thank you, Aarti. A very warm welcome to all of you and to this distinguished panel. I will begin by saying three broad aspects just to introduce. First, a lot of you might be thinking, why is International Solar Alliance talking about India Energy Stack and AI for Energy? The answer is very simple. The world did 1,000 gigawatt of solar in just the last two years."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:13",
                    "end": "00:29:42",
                    "text": " 45% of that was decentralized renewable energy, which means either on the rooftop or on the pumps. And while India has achieved 150 GW of cumulative, only 15% is currently decentralized. Which means India will see a huge quantum leap in solar rooftops, solar pumps, PM Kusum, PM Sudeghar and a host of decentralized renewable energy."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:42",
                    "end": "00:30:13",
                    "text": " A lot of distribution companies believe it creates more problems for them if you have a lot of solar rooftop, which actually may not be true as we will experience in the session today. And therefore we believe in International Solar Alliance having digitization and AI is absolutely foundational element for doing lot more solar. Second, while a lot of you, if you are in the sector, would have seen the data center demand and therefore"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:13",
                    "end": "00:30:43",
                    "text": " What are the energy requirements for AI? Now that is true for a lot of countries, especially in the West, and it's not a trivial matter, but for developing countries, the real game changer might be AI for energy, which means what are the AI applications that will fundamentally transform the way energy sector is dealt, as you saw in India Energy Stack presentation. Because what you saw is what Amazon did to retail"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:43",
                    "end": "00:31:13",
                    "text": " would happen exactly the same way in energy in five years from now, millions of people buying and selling power. And finally, as intertwining of AI and energy is going to happen, it is very important to have a framework, which is what we are here today for in terms of our overall global AI mission. Which kind of a utility, which kind of an organization should be aspiring for what? And therefore, there should be a framework."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:13",
                    "end": "00:31:43",
                    "text": " for you have to do this in step one, this in step two. So all these three elements is what ISI is there for. And since I have a great set of panel, I'm going to begin. I'll start with you Hemang. Hemang is not only a senior advisor in Indian ED's office, he has worked very closely on capacity building commission, prime minister's office, government of Gujarat, and World Bank as well. So Hemang, you have seen it from an Indian level and the global angle. Tell us a little bit."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:43",
                    "end": "00:32:03",
                    "text": " what do you think needs to be done to accelerate the adoption of AI, especially it has a lot of cross-sectoral elements, both digital and energy coming together, but also the way this interoperability and products of technical assistance and lending as you look at the World Bank. What will be your perspective?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:03",
                    "end": "00:32:28",
                    "text": " Thank you, Ashish. First of all, thank you very much for inviting me and gathering such a stellar set of people. It itself is an honor. But let us also begin by thanking Honorable Prime Minister for bringing the AI Summit here. After France, in maybe three months or six months, we are sitting here with the AI Summit. It's a big thing."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:28",
                    "end": "00:32:54",
                    "text": " when I sit in Washington D.C. and I tell people that I'm going to AI Summit and it's in India, and they all get surprised because nobody had thought of it. So we have to thank him and thanks to REC and ISA because you are sitting here in this hall on the very first day when the honorable Prime Minister is inaugurating the summit. So that's a big thing for this audience, this hall."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:54",
                    "end": "00:33:21",
                    "text": " begin to answer your question by answering Swetha's question, I still query cash, but only outside India. Not in India. So that in a nutshell needs to change for AI, for energy, for the globe. If we can stop carrying cash outside India, we have changed it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:21",
                    "end": "00:33:49",
                    "text": " Let me begin by what we look at it from DC at the bank. We have around 700 million people to be connected by 2030 to electricity. We at the bank have a mission of connecting 300 million only in Africa, the continent Africa. But if we want to do that, as I saw in one of the presentations, India is going to add 300 million new customers with smart meters."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:49",
                    "end": "00:34:17",
                    "text": " Let me put it in a different way. India is going to add 300 million intelligent customers to the grid. So while the rest of the world will add 300 customers, we will add 300 intelligent customers. And that is what AI can do. So unless we look at AI not as a technology, particularly for energy sector, but as a development infrastructure itself, the way we look at the grid, the way we look at these cons, and the way."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:17",
                    "end": "00:34:44",
                    "text": " we look at meters, we have to look at AI. That it is also one of the infrastructure layer that we need to create simultaneously. And that is where I think the game will change. Another aspect to it is what AI cannot do is AI cannot bring the fiscal prudence, AI cannot bring the governance reforms, AI cannot bring all those things. But AI can enable a lot of those things."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:44",
                    "end": "00:35:04",
                    "text": " So if you are worried about your losses and that is what is the biggest contributing factor to your fiscal issues, then AI can immediately come in and help you. So how do we look at AI as a problem solver for the problems which are non-technical in nature? That also is a very important part of it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:04",
                    "end": "00:35:30",
                    "text": " And the third and the important part is what I will borrow from our senior colleague from Jaipur. She said we don't think in terms of pilots. So you can't think of in terms of pilots when you are talking about a technology like AI. You have to think big. You have to think at scale. And that is what IEAS is trying to do. And to end my intervention,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:30",
                    "end": "00:35:54",
                    "text": " What Shweta mentioned about IES, that it is going to become another banking kind of a technology. I think we should not think that small. We should think even bigger than that. We shouldn't just stop being a finnacle. We should become the Google of AI for energy for the world. And that is what AI for energy or AI or IES for energy for the world would mean. So thanks again."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:57",
                    "end": "00:36:26",
                    "text": " Thank you so much, Hemang. Indeed, the global air mission for energy is indeed that level of ambition. Let me go to Henry. Henry, thank you so much for joining. You have not only been a French ambassador for digital affairs, but also DG of a leading research agency. Tell us a little bit. You mentioned very clearly that you're much more of an AI person than energy. When you hear this big push of trying to bring energy and AI together,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:26",
                    "end": "00:36:57",
                    "text": " What's your reaction on what will it take to create this ecosystem of interoperability, stack for payments, and citizen-centric approach? Thank you. Very happy to be in India again, even if I'm not anymore an ambassador. We need to work hard to build proper ecosystems, but we need open standards, ecosystems, and good regulation if you want to."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:57",
                    "end": "00:37:26",
                    "text": " to win this thing. But first, I want to recall, you know this, but I will say it from an AI, from a digital perspective. The energy transition is accelerating everywhere, as you said. Our power systems are becoming more and more complex, with more renewable energy, more decentralization, higher expectation for resilience from citizens, from businesses, from governments. I believe"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:26",
                    "end": "00:37:55",
                    "text": " that artificial intelligence might be the only way to shift to our prioritizing renewable energies. We cannot build a new system in the old organization. So we must use it. But we also must remember that innovation is not always a progress. And that sometimes, especially in the field of digital, it did turn bad. We have bad experience with social network, for example. The AI didn't play a good role."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:55",
                    "end": "00:38:24",
                    "text": " the uncultured AI. We did start with open, and genuine, and decentralized network. But some people took control, putting companies on the top of the infrastructure. So we have to learn some lessons from this. I want to add, before answering briefly to your question, France comes to this conversation with a third objective. We want the digital revolution and the environmental transition to move forward together."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:24",
                    "end": "00:38:50",
                    "text": " And that's very important. So I don't know if you know, but in France, in 2021, we did adopt a law specifically aimed at reducing the environmental footprint of the digital sector. Because, of course, there is a growing issue regarding the costs, the consumption of AI. And we have to invest fiercely to... You did see the..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:50",
                    "end": "00:39:19",
                    "text": " IEA reports noting that AI could represent a meaningful share of global electricity demand soon in 2030. So we should ensure that the AI for energy does not unintentionally create a new stress on power systems. So just to mention this, that's why France helped to launch with India the International Sustainable AI Coalition during the Paris Summit and we expect to go further during this impressive summit."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:20",
                    "end": "00:39:46",
                    "text": " I did mention this because if we come back to your question, I told you open standards, ecosystems, and regulation. Open standards are the heart of the digital revolution itself. Everything is based on TCPIP, HTML, and other open standards. And that's not a coincidence. With open standards, first you share innovation everywhere. Because you don't have to."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:46",
                    "end": "00:40:13",
                    "text": " to spend time and money to make money. You can decrease the return on invest and you can encourage a lot of innovation. And of course, people don't notice usually. You give a strategic possibility to everyone to cooperate and to discover much efficient, much more efficient way. I don't go further because I think that's it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:13",
                    "end": "00:40:43",
                    "text": " And the experience of the Indian digital grid will go further to this. But when you have open standards, you need big, extensive ecosystems with a huge variety of factors. If you pay attention to the history of humankind, you don't have any example of anything important that was not within a very diverse and vibrant ecosystem."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:43",
                    "end": "00:41:12",
                    "text": " with this important concept of co-opetition. For example, in Bangalore, they all want to build the next DPI. So they do cooperate, they do exchange, they do share views, but they also want to be the one. So this is an example of cooperation and competition, a co-opetition. But art, music, science, industry, everything important was born in this kind of context. And if you don't have a strong ecosystem with you,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:12",
                    "end": "00:41:40",
                    "text": " You can be like Nokia, for example, they were the universal leader of phone but they didn't find any idea to fight the smartphone and they basically quite died and now they come back because they didn't have enough energy, ideas, diversity to surround them. And the last point is about regulation and I want to say sovereignty because..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:40",
                    "end": "00:42:08",
                    "text": " For us, people from the history of internet and digital revolution, we feel a bit cheated. We did work hard to propose a world of cooperation, knowledge, citizen empowerment, diversity, decentralization. And because of a failure of regulation, because we did misregulate, it's currently turning into few big tech companies."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:08",
                    "end": "00:42:34",
                    "text": " trying to control all the network, including the submarine cables. Because as you have seen, the last three years, 80% of the new submarine cables were built by the five GAFAM. And that's probably because we did miss one point when social network did appear. We did consider them as a passive infrastructure. And they were media."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:34",
                    "end": "00:43:02",
                    "text": " media with curation and promotion and a very active content curation. And because of this mistake, we have very strong actors that can oppose firmly to Europe, for example, as you can see easily. With AI, it could be worse. It could be worse because if you can take control of an ecosystem through AI, you can take all the added value."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:02",
                    "end": "00:43:30",
                    "text": " They can transform every industry into a kind of Uber driver. So you are still there, but you are just a servant of the system. And the only answer to fight this is to be sure. And I think that here, Europe, India, and even on a certain perspective, China, are on the same line. We need enough open source AI model to be sure that everyone will be able to innovate without asking permission."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:30",
                    "end": "00:43:59",
                    "text": " public, private, small companies, innovators, et cetera. And we need the strong policies to encourage the sectorial models for energy, for lawyers, for farmers, et cetera, to have a huge diversity of solutions. We cannot afford a world with three AI giants controlling every aspect of the economy. So again, open standards that has the bottom of everything, strong and diverse and vibrant ecosystem."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:59",
                    "end": "00:44:29",
                    "text": " serious regulation or we are working for three sociopathic billionaires. Thank you. Thank you. So very clear messages. I like that concept of collaborate and compete at the same time. I think that's very good as part of ecosystem and also the issues of concentration and interoperability. I think those are very clear messages which you will see in a global AI mission. But let me come to Mr. Subramaniam. Mr. Subramaniam, you look after the entire"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:29",
                    "end": "00:44:59",
                    "text": " distributed renewable energy program of India as part of Ministry of New and Renewable Energy. Tell us a little bit of your experience as India is now close to 3 million households, the target of 10 million. How are you seeing digitization and AI in terms of an essential piece? Thank you Ashish. Thanks for having me here. In fact, both of us have been together just a week, 10 days back in Davos and we have seen with our own eyes the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:59",
                    "end": "00:45:14",
                    "text": " scale and the speed at which India is implementing the distributed renewables has been amazing and that is what we have presented to the globe. As you rightly mentioning"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:14",
                    "end": "00:45:42",
                    "text": " the overall picture on renewables. Let me put things in context. As a country, we're close to almost about 500 gigawatts plus now, 520 gigawatts to be precise. Out of which, almost 52 percent comes from the renewable sources or the non-fossil sources to be precise. Out of which, almost about 140 gigawatts come from solar. Within this entire piece,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:43",
                    "end": "00:46:11",
                    "text": " We're talking close to about 35 gigawatts of distributed renewables currently installed in the country. But the even more amazing part of it is that, out of this, what we have done in the last 10 years and what we have done in the last 15 months, the last 15 months has added almost close to 18 gigawatts to the distributed renewable energy space, both under the Pradhan Mantri Suryagar Most Bijli Yojana, which is the rooftop solarization program."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:11",
                    "end": "00:46:34",
                    "text": " one of the world's largest. And the Pradhan Mantri Kusum program, which is again the agriculture pump solarization program, again one of the world's largest. Between both the programs, we spent close to 9 billion USD on rooftop solarization and about $4 billion on the Kusum agriculture solarization."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:34",
                    "end": "00:46:58",
                    "text": " Now, what happened with this kind of a speed that is coming up on the distributed renewables? Of course, distributed renewables is something which we are going up in a big way. But where AI would help and the technologies would help, we have seen that with our own eyes in the Pradhan Mantri Suryahar, where it's not that rooftops have not been installed in India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:58",
                    "end": "00:47:27",
                    "text": " in the last 10 years. They have been. But what has changed significantly? Why is that we could do almost 50% of what we have done in the last 10 years, in the last one year, is because of the technology solutions that we have incorporated. And the technology solutions have benefited the consumer, the people who are on the field, the vendors, the banks, and the discoms alike."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:27",
                    "end": "00:47:57",
                    "text": " when you build a single portal, national portal, have all the stakeholders come onto it, so it adds that much of value and credence. Now having said that, what AI is going to do, in fact, I'm sure what Shweta has presented in the India Energy Stack, it's going to be a game changer, especially for the DRE. One, because it has got all the elements of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:57",
                    "end": "00:48:24",
                    "text": " what an energy stack should do. It is not providing software, it is providing governance. Now, what is that we are looking at from here to scale up DRE? DRE has scaled up and it is scaled up in a big way moving forward also. But where our challenge lies is on the distribution systems, where unidirectional flow of transformers are there."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:24",
                    "end": "00:48:52",
                    "text": " where millions of prosumers are now generating, where demand response becomes important, where the assets that are being developed across the country need to be maintained. Can we build in predictability into this? There are a lot of generators who require weather forecasting, predictive analytics. Can we get this onto the AI tool? Lastly, as a consumer, as a prosumer,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:52",
                    "end": "00:49:19",
                    "text": " is my rooftop system, is my solar pump generating enough when compared to my peers? Can I be able to sell my electricity to the other peers, which the P2P is trying to enable? So these are enabling decisions. So if you ask me what the AI is going to do in the future, it will not just make us act, it will not just make us react, but it will make us act."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:19",
                    "end": "00:49:44",
                    "text": " So that is the story that we are looking forward as far as DRE and the AI convergence is concerned. Yeah. Thank you. Before I hand over back to Aarti, I'm actually going to ask a very quick one-sentence question to all three of you in the same order. There is incredible noise around AI, I feel, and it's very important to lead it to specificity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:44",
                    "end": "00:50:20",
                    "text": " So my one-line question is what would success in next one or two years on AI for energy should mean? Start with you, Hema. See, I think this is a very fast-moving, evolving technology. So you shouldn't look for very early successes upfront. You should look for very long-term building of platforms, long-term building of infrastructure. For example, let me give you one example."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:20",
                    "end": "00:50:48",
                    "text": " International Solar Alliance. Honorable Prime Minister conceived it in 2014. Now it has 42 countries. It will have 80 countries in no time. And that is the digital public infrastructure for deploying AI to the world, right? So we should think in that long term sense. So what is one layer that we can create from India and give to the world? That is how I look at it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:48",
                    "end": "00:51:17",
                    "text": " So we are 125 countries, but we need to go much bigger. Henry, what will success look like in the next one or two years? I think that we should prove that we are integrating as much renewable energies as possible in our grid, thanks to a much better gestion of the grid. Because I think you are the professional, but renewable energies are less predictable, and you need much more."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:17",
                    "end": "00:51:56",
                    "text": " provision, anticipation, control of the network. You cannot just add more solar system. So that's something to AI that we can have a very significant proportion of. Okay, very clear. I think in the next two to three years time, success in AI and the RE convergence would be where the overall cost of power to the consumers go down, our industrial competitiveness goes up."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:56",
                    "end": "00:52:24",
                    "text": " And consumer empowerment becomes a prosumer empowerment. And our grids become ready for energy transition. This is all what I would look like. Wonderful. So with that, thank you very much. Please give a round of applause to this set of panel. We are going to have lots more people joining. Aarti, over to you. Thank you, sir. Thank you, Chair, and thank you, panelists, for that very rich and forward-looking discussion. And I request that you continue to stay on stage with us."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:24",
                    "end": "00:52:54",
                    "text": " We now move to a significant milestone for the evening, the launch of the report that lays the foundation for the global mission on AI for Energy. To mark this launch, I request all our panelists to please join us on the stage. Ms. Aarti Dogra, Chairperson, Discoms and Managing Director, Jaipur Vidyut Vithran Nigam Limited, Mr. Abhishek Ranjan, CEO, BRPL, Mr. Sujit Nair, CEO and Co-Founder, FIDE, and Mr. Jeremy Oppenheim, Managing Partner, Systemic."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:58",
                    "end": "00:53:33",
                    "text": " While the panelists take their seats, we'll begin the segment with a video preview of the report."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:33",
                    "end": "00:54:02",
                    "text": " Across the world, AI for energy efforts remain fragmented. Pilots, proofs of concept, isolated breakthroughs, but no coordinated backbone. Utilities face data silos. Regulators lack modern tools. Capital hesitates. Innovation moves, but not at the speed this decade demands. The latest ISA report makes a bold case. The world does not need more pilots. It needs a global mission for AI for energy."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:02",
                    "end": "00:54:32",
                    "text": " A citizen-centric mission to institutionalize delivery, to bring urgency and focus, to scale what works, to move from experimentation to execution. Because no single country and no single institution can do this alone. Under the leadership of the International Solar Alliance, representing over 120 countries, this mission will turn AI potential into measurable outcomes, faster renewables integration, lower system costs."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:32",
                    "end": "00:55:02",
                    "text": " Greater reliability. Stronger energy security. It will stand on five pillars. Smarter policy. Skilled people. Secure data systems. Innovation at scale. And finance that unlocks deployment. But this mission is not just about technology. It is about people. Citizens who understand their energy use. Optimize their bills. Participate in electricity markets. Become producers, not just consumers."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:02",
                    "end": "00:55:28",
                    "text": " A citizen-centric power system is more efficient, more resilient, and fairer. The coming decade will define the future of global energy. Collaboration today will determine affordability, security, and sustainability for generations. The technology is ready. The opportunity is clear. This is the moment for a global mission for AI for energy. And the time to act is now."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:33",
                    "end": "00:56:08",
                    "text": " I now invite Ms. Arushi Chopra, Senior Director Systemic, to make her presentation on the report, please. Hi, everyone. And we're delighted to be launching this report here today with you all. Solar growth has been outstripping all projections made to date. Wind has been rising rapidly. So as we know, generation is no longer the bottleneck. The grid is. And this grid was built for a linear world."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:08",
                    "end": "00:56:37",
                    "text": " There's concentrated power. There are unidirectional flows. But that world has now changed. And the system that we now live in is dynamic. So upstream, you'll see that generation is now intermittent and decentralized. And downstream, you'll see that demand is being reshaped by heat waves, electric vehicles, data centers. We're living in a new dynamic system now. But the grid is not dynamic. The grid is linear."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:37",
                    "end": "00:57:06",
                    "text": " Therefore, that's the mismatch of the energy transition. The good news is that digitalization can turn the grid from passive infrastructure into something that's truly intelligent and autonomous. This report here really tries to set out what that intelligent energy system looks like and what does it take to build it. We think of it as this digital ladder, if you like. At the very bottom of this digital ladder,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:06",
                    "end": "00:57:32",
                    "text": " are if you like the ears and eyes of the energy system. This includes smart meters, sensors. They really help you build visibility. Then as you move to the blue rung, the second rung if you like, these are almost like the arms and legs of your digital ladder, right? Of the intelligent energy system. This helps you act on the data that has been collected in the bottom most layer. So take the example of, for example, if..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:32",
                    "end": "00:58:15",
                    "text": " it's 6 PM and all electric vehicles have been plugged in at the same time, this will act on that data and smoothen the demand to ensure that the system doesn't fail. Then at the very top is effectively the brain of the energy system. This is not just reacting to the data that you get, but proactively anticipating what's going to happen in the future. So it almost helps you predict and look ahead. So based on, let's say, temperature readings today, it might help you predict."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:15",
                    "end": "00:58:45",
                    "text": " Same rung, right? Not every country is actually operating at the same level. So what we've done is, if we move to the next page, we'll see that there are effectively, if we move to the next page, we'll see that there are effectively four different archetypes of countries that we look at, right? And if you look at certain high-income countries, for example, if you look at the United Kingdom, right? Here, they already have visibility in place. Now it's about optimization."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:45",
                    "end": "00:59:14",
                    "text": " Their challenge is optimization. So take a winter evening in Britain. Demand surges at home as everyone turns on their heating. Now in a traditional system, that would mean that further grid build out is needed, and that would mean more cost. But in an intelligent system, that doesn't happen. The system is optimizing itself. So it's really going to turn on storage. It's going to ensure that demand reshapes itself, and there's forecasting built into the dynamic. So that's what it's going to do in the UK."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:14",
                    "end": "00:59:43",
                    "text": " system-level optimization. But if you look at low-income countries, let's say, I mean, there, it's primarily about visibility, right? Where is it that energy is coming from? In what direction is it flowing? Is it being lost? Are people paying for it? Those are the kind of questions that are being answered, that are being asked in lower-income countries. So we looked at the use case of Ghana, really, right? And in Ghana, what we saw was that putting smart prepaid meters"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:43",
                    "end": "01:00:11",
                    "text": " is what actually reduced theft and billing gaps. And consumption rose by 13%. So that's the power of customized use cases, depending on the country you find yourself in. And if you look across all of these four different archetypes, the technological and investment pathways are different. And that's what this report documents, in addition to, of course, all the use cases. So in conclusion, what I would like to say is the technology is already here. The use cases are already proven."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:11",
                    "end": "01:00:41",
                    "text": " And our report brings up several of these showing the value that is also going to be generated for the system and for the consumer. Now what will determine success is execution. And that's what this report launches. It will be talked about later as well. The global mission to really make this a reality and execute seamlessly across the 125 ISR countries. So the linear grid is what effectively built the last 100 years. But the next 100 will be built on an intelligent grid. Thank you very much."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:45",
                    "end": "01:01:01",
                    "text": " Thank you, Arushi. Please, will you stay on stage with us? And I'm going to invite Karan Mangotra, Ms. Shweta Ravi Kumar, and Ms. Nidhi Sarin from JIAB to please join us on the stage for the formal unveiling of the report. The panelists, may I please request you to please come in front?"
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
    {
    "_id": "699598a1565455da1fc7546a",
    "projectId": "939910",
    "projectName": "AI for Energy Digital Twins and India's Energy Stack - Hindi",
    "projectDescription": "AI for Energy Digital Twins and India's Energy Stack - Hindi",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T10:46:56.041Z",
    "projectUpdatedDate": "2026-02-18T10:46:56.041Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T10:47:52.323Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "AI for Energy_ Digital Twins and India_s Energy Stack.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Hindi"
},
    {
    "_id": "69959918565455da1fc7548c",
    "projectId": "262682",
    "projectName": "AI for Energy Digital Twins and India's Energy Stack - French",
    "projectDescription": "AI for Energy Digital Twins and India's Energy Stack - French",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T10:48:55.194Z",
    "projectUpdatedDate": "2026-02-18T10:48:55.194Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T10:49:48.650Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "AI for Energy_ Digital Twins and India_s Energy Stack.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "French"
},
    {
    "_id": "699572a0565455da1fc751e2",
    "projectId": "255474",
    "projectName": "Israel's AI Model Building a Better Future Through Innovation - English",
    "projectDescription": "Israel's AI Model_ Building a Better Future Through Innovation",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T08:04:47.383Z",
    "projectUpdatedDate": "2026-02-18T08:04:47.383Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T08:10:09.584Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "Israel_s AI Model_ Building a Better Future Through Innovation(MP3_128K).mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771401874919.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:23",
                    "text": " My honor to invite His Excellency, Mr. Reuven Azar, Ambassador of Israel to India, to deliver the opening remarks. Thank you, Arpita. Good morning, everybody. It's really a great pleasure to be here. I want to thank all my colleagues that came from Israel all the way to honor us with their presence. We have here the people that..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:23",
                    "end": "00:00:57",
                    "text": " standing at the forefront of the Israeli effort on AI. I want to thank of course the Embassy of Israel for organizing this, especially Maya and all the team. Thank you very much. India and Israel trust each other because we have built this partnership for the last 30 years and we have together developed the capabilities that allow us to face the challenges of today, the security challenges we have."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:57",
                    "end": "00:01:26",
                    "text": " But we have many other challenges because we are two old civilizations that have established their young states and are lagging behind in building infrastructure and giving all our citizens the opportunity to flourish and to develop what they need in order to be part of modern life."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:26",
                    "end": "00:01:54",
                    "text": " other challenges and the challenge of AI that is going basically to give us the opportunity to harness all the knowledge that humanity has gathered and our challenge is to make it accessible to each and every citizen, each and every entrepreneur. So we want to hear a little bit today from our participants here how they're going to do it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:54",
                    "end": "00:02:24",
                    "text": " because this is something that is new for everybody and we want to make sure that we would be able to build the capability and create the access that we need so we can together continue to flourish and to develop new projects that will make sure that we face also the challenges of the future. Okay, so I'd like to thank you very much for being here."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:24",
                    "end": "00:03:04",
                    "text": " And you will invite the next speaker. That's great. Thank you. Thank you, Ambassador, for your valuable insights. Now, our next speaker will take us into the strategic core of Israel's AI journey. Please join me in welcoming Brigadier General Reserve, Eris Askel, Head of the AI Directorate at the Prime Minister's Office. Thank you, Ambassador, and thank you for the organizers. I'm very glad to be here in this stage."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:04",
                    "end": "00:03:32",
                    "text": " I am Erez Askal, the head of the new AI Directorate under the Prime Minister, before I retired from the military as the head of the digital transformation and AI of the IDF all over the world. And we established our Directorate under the Prime Minister because in Israel we believe that AI is an issue of national security."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:32",
                    "end": "00:04:02",
                    "text": " and it's going to change all our life, of course. So, we want to lead Israel. We have two main goals. First, to lead Israel to be among the top three of the world in this field, and we're going to be there. And the second is to improve the quality of life of every Israeli citizen and our allies like India, of course. How are we going to do it? We're going to use our competitive advantage."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:02",
                    "end": "00:04:32",
                    "text": " I believe that is where we are the best in the world to solve the real-life problems and because that we're going to build three domains. First, as I said, labs of real-life solutions with academia, with the industry, all the guys here. And we're going to find 50 verticals and to solve them. The second is to use our strong in cybersecurity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:32",
                    "end": "00:04:57",
                    "text": " cyber security for AI, safe foundational models, et cetera. And the third is edge solutions. We believe that if we want to implement AI in the edge or in the real life, we must to build something in the edge, not to trust data centers hundreds kilometer from us."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:57",
                    "end": "00:05:26",
                    "text": " We want something local in the surgery room, in the classroom, or in the border or drones. So the idea is to use our industry from the military, from the civil industry, and to build it together. I want to invite you all to be part of that. India, they are a good friend of us, of course, and I believe that we can't do it alone."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:26",
                    "end": "00:05:54",
                    "text": " We need to build it together, all the big countries, Israel and India, of course. Thank you very much. And we're going to show you our products, but just we have a lot of companies and a lot of products just at the beginning, like I said. Thank you. Thank you, General Askel. Now I invite Dr. Victor Alcáceres."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:54",
                    "end": "00:06:27",
                    "text": " Senior Research Scientist and Head of the Center of AI in Agriculture at the Volcani Institute, Ministry of Agriculture. Thank you. Thank you for being here. Thank you, Ambassador, for having us here. It's a great opportunity to give you in a few minutes our vision about using AI in agriculture. I come from the research, the Volcani Center."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:27",
                    "end": "00:06:53",
                    "text": " It's the research arm of the Ministry of Agriculture. We deal with all aspects of agriculture, including the AI and the technologies. So our mission is to feed the world, to feed the world in the future with the exponentially growing population."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:53",
                    "end": "00:07:13",
                    "text": " and at the same time do not exhaust our planet, planet Earth. So we would like to have both sustainability of the agriculture and also produce enough food for everybody."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:13",
                    "end": "00:07:40",
                    "text": " The common and the conventional way to do it is to maintain sustainability, is to reduce the inputs, reduce chemicals, reduce fertilizers, do not pollute. On the other hand, this reduces the yields and feeding the world, it won't be possible. So there is a kind of contradiction of whether we can use"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:40",
                    "end": "00:08:08",
                    "text": " high technology in order to solve this problem. And we think that the sustainability of agriculture does not necessarily go together with reducing the use of technology. The other way around. Technology today can help us solve problems and do things more efficiently than we did once."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:08",
                    "end": "00:08:38",
                    "text": " So we can use tools like AI in order to do precision agriculture, digital agriculture, find where are the specific points where we can use those advanced technologies, the specific points and local areas that we can fertilize, where the need is most prominent, where we have to use herbicides and pesticides."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:38",
                    "end": "00:09:07",
                    "text": " and in a very early stage we can detect it. So AI today gives us the tools to do that. We can do it by having a lot of data acquired in pests, diseases, and decide when is the best time to use those chemicals. Then we can centralize and use"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:07",
                    "end": "00:09:36",
                    "text": " the knowledge, which is today only in the farmer's head and sometimes in the advisor's head, and to make it available to all the farmers all over the country, all over the world. And last but not least is to use the digital twins models. Just to give you an example of what we are trying to do, we use, of course, all the modern and state of the art tools of deep learning and computer vision."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:36",
                    "end": "00:10:02",
                    "text": " just to detect very early weeds in the field, diseases in the leaves or pests, and decide when is the best time to act. We want to use all the LLMs, the large language models, that are specific for agriculture, that do not really exist today. Not only because nobody has trained them, but also that the knowledge..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:02",
                    "end": "00:10:30",
                    "text": " that those models have to be trained from is not available in a written form. Sometimes it's just experience of farmers and we have to collect it and put it in the LLMs. And the digital twins enables us to make simulations, but not just simple simulations, but spatial and simulations with various kind of conditions that will find the optimal"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:30",
                    "end": "00:10:54",
                    "text": " the optimal conditions to grow the crops in each one of the places. Speaking of AI tools, it's not only in the implementation of the trivial or known things, but also in genomics, genomics and phonomics. We can use AI tools to go through the phenotyping of crops."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:54",
                    "end": "00:11:21",
                    "text": " find how the genes are connected to the phenotype, that means how the plants grow, and find the best types and varieties that they are resilient to the specific conditions that the crop is grown. This is specifically important for the current situation where climate change makes the..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:21",
                    "end": "00:11:51",
                    "text": " climatic conditions very different than they were once. So how we do it? In Volcani we have established a center which is a national center in Israel that brings everything together. Brings the biological infrastructure of genomics and all the omics together with the sensing and the AI tools in the algorithm part and also a computational center where all this can be implemented."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:51",
                    "end": "00:12:19",
                    "text": " So research is ongoing in order to find what are the best tools in the AI that bridge all of them together. But of course, this is not enough. In order to get the innovation to impact, we have to go through some barriers that they are really known from other topics as well, but also in agriculture, the legal..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:19",
                    "end": "00:12:53",
                    "text": " societal and the financial barrier have to be overcome in order to have an impact. And as have been also told a lot in this conference before, the trust of the farmers is a difficult part that has to be gained in order to have a great impact. Thank you very much. Thank you, Dr. Victor, for sharing how AI is shaping a more sustainable future. Now moving on, AI is not only transforming industries,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:53",
                    "end": "00:13:47",
                    "text": " it is reshaping how we educate the next generation. And to discuss the same, please welcome Ms. Mirav Zarbib, Deputy Director General for Innovation and Technology at the Ministry of Education."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:47",
                    "end": "00:14:17",
                    "text": " I'm the Deputy Director General for Innovation and Technology, and I'm honored to be here with you today. I'm going to give you only a brief glimpse of what we're doing in the Israeli Ministry of Education in order to implement AI in the education system. I'll start with a vision, then I will present four pillars in the strategic Israeli plan."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:17",
                    "end": "00:14:47",
                    "text": " strategic plan, then I will give an example of how we implement AI into the system. I will elaborate on the 720 personalization project and I will talk about our sandboxes and of course we will talk about collaboration. Okay, so let's start. I will read out loud our vision. Our vision is to promote flourishing at every level."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:47",
                    "end": "00:15:18",
                    "text": " individual, community, and the nation, our education system will provide personalization to drive students and teachers growth across all dimensions of life. And now to our strategic plan. Our strategic plan is based on four main pillars, which are aligned with the Oxford AI Readiness, and of course, to this amazing conference principles."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:18",
                    "end": "00:15:45",
                    "text": " The four pillars are human capital, infrastructure, regulation, and innovation. Okay, so let's dive into the human capital. The human capital pillar is, I would like to present, the competency taxonomy. This is a framework we developed in Israel of how to promote AI competency from the basic level to the topest."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:45",
                    "end": "00:16:14",
                    "text": " level. So as you can see in Israel we have only 2.3 million students, not more than that. We have basic AI competency for all. I will give an example how we gained this goal and the next level is AI in subject areas. We integrated AI into different subject areas. What's unique about these two steps is that this is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:14",
                    "end": "00:16:42",
                    "text": " how students use AI, in the next three layers we have students learning how AI is built. From the basic level, AI and STEM, to the highest level. Our 1,500 students learning as AI scientists how to build the future Israeli LLM for the future of Israel."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:43",
                    "end": "00:17:06",
                    "text": " Of course, with an alignment to what Erez has just presented. I will have the video just behind me. What I would like to say that, as you can see in the picture, we have every year, a year, a month of AI. What's very unique in this month that, in only one month,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:07",
                    "end": "00:17:22",
                    "text": " More than 70% of the students in Israel gain the AI competency. We used it in a very special approach, integrating a top-down and bottom-up approach."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:22",
                    "end": "00:17:51",
                    "text": " and competitions, hackathons. You can see the curious faces and the happy faces of the students. We understood that if we want to implement AI into the curriculum, we should do it in a really different ways. Now to the next pillar, infrastructure. In Israel, from the fourth grade, every student can use a special bot that we developed in the Ministry of Education."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:51",
                    "end": "00:18:20",
                    "text": " A special bot that teaches AI. This is a Q2 bot. Then next to that, 7th to the 12th graders, we have the notebook and Gemini. Every student has an access to these tools. And then we have our amazing 720 project that I'm going to present to you today. So the 720 project is a personalized, we're talking about a personalized system."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:20",
                    "end": "00:18:48",
                    "text": " We are going to talk about four versions of personalized systems. In every system, students, teachers and principals have their own AI agents. The agents are connected, so everything is personalized. We are talking about a unified, holistic system that every school can use. I will talk about the sandboxes later and how we mitigated..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:48",
                    "end": "00:19:34",
                    "text": " the risks okay so let's see it in the video that we have here the vision is clear the technology is in place now let's see how the system works in practice"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:34",
                    "end": "00:20:01",
                    "text": " The student enters a 360-degree learning environment where everything about their school experience lives in one place. Learning, attendance, progress, mentoring, and goals. Together, this creates a complete picture of each student, enabling personalized learning and meaningful growth. The experience begins with a short personal onboarding. Students share their interests, their goals, and how they prefer to learn. Meet Maya."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:01",
                    "end": "00:20:23",
                    "text": " She wants to move to a more advanced class in science. She learns best through visuals. Based on this, the system builds her a learning path, not a fixed curriculum, but one that responds as she progresses. From Maya's point of view, learning feels clear and active. She moves forward, explores, and receives support exactly when she needs it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:23",
                    "end": "00:20:46",
                    "text": " As Maya learns, the system quietly adapts with her, helping surface patterns, strengths, and challenges early, before they become obstacles. Leaders see the full picture. Learning pathways across grades, usage and engagement, teacher development, matriculation predictions based on live data. This is 720 Vision."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:46",
                    "end": "00:21:16",
                    "text": " 360 degrees of the student, and 360 degrees of educator support, all connected in one system. We believe that when learning is personalized, potential is unlocked. When every child feels seen, valued, and empowered, education transforms into a journey of growth and discovery, and together, shaping a better future for thousands of children who deserve nothing less. Okay, as I said, only a glimpse."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:16",
                    "end": "00:21:25",
                    "text": " You didn't see the whole video about how we personalize this to the teachers' experience."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:25",
                    "end": "00:21:54",
                    "text": " this time that the clock is ticking. Now to the final pillar regulation. In order to promote all these tools we had to move from frameworks to national scales and we did it in a very special framework that we built to regulatory sandboxes. We started by setting risk framework with multi-stakeholder committee. The defined risk set new sandboxes"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:54",
                    "end": "00:22:24",
                    "text": " Then we work together, the ministry, industry, local authorities, teachers, and we co-run together experiments. Then we monitor, measure, and learn from real classroom data, set new evidence-based regulation, and scale it up. This is an example of the QBOT, how we set the risk categories from technological problems."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:24",
                    "end": "00:22:49",
                    "text": " a pedagogical, professional identity, the risk level, what is the result of the risk and how we mitigate the risks. And now, finally, I know I have only 10 seconds, the Jerusalem declarations. Two weeks ago, there was a pre-conference to this conference, a conference in Israel, and we're here."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:49",
                    "end": "00:23:20",
                    "text": " as I said before, to strengthen the Israel-India cooperation in AI and STEM education. And in our conference we had the Jerusalem declaration that was adopted by 15 countries at Education 2026 that reflects a shared commitment to integrate AI and education and to scale success through international collaboration. Will India be the 16th? Okay, thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:20",
                    "end": "00:23:56",
                    "text": " Thank you, Ms. Zarif, for this wonderful presentation and explaining it so well. AI is redefining the labor market globally. In Israel, it is seen as a strategic engine for growth. It is my pleasure to introduce Ms. Inbal Mishash, Director General of the Israeli Employment Service, to speak about how AI supports nationwide workforce adaptation and strengthens Israel's innovation-driven economy."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:56",
                    "end": "00:24:36",
                    "text": " Good morning, all. I'm honored to be here in your beautiful country. This is my first time in India, so I'm very excited to be here. Shalom and Namaste. When we speak about AI and the future of work, many countries speak in terms of disruptions. Israeli speaks in terms of design."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:36",
                    "end": "00:25:07",
                    "text": " For us, the response to AI is not a reactive response to technological change. It is a strategic engine which we expand innovation across the entire labor market through to our identity as a startup nation. Israel's strength has always been scale. It has been speed, dynasty, talent, and ecosystem."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:07",
                    "end": "00:25:43",
                    "text": " alignment. Today, the strength is clearly visible in our AI adoption patterns. According to the Innovation Authority, 95% of high-tech employees in Israel use generation and IT tools regularly."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:43",
                    "end": "00:26:10",
                    "text": " compared to only 18% across other sectors. But the story of AI in Israel is not the same for all. Israel is strong in AI, but adoption is not equal across the economy. When we analyzed almost 2 million job seekers, we see a clear and systematic guardians."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:10",
                    "end": "00:26:39",
                    "text": " Core urban areas show high exposure to AI. In the geographic periphery, it drops. In the socioeconomic periphery, it declines further. And in the double periphery, exposure and effective use of language and image model tools can fall below the national baseline."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:39",
                    "end": "00:27:12",
                    "text": " This is not anecdotal. It reflects human capital spread firm capabilities and access to training and employers. It is not technological gap. It is sexual labor market. Just a minute. It's not working. It is not technological gap. It is sexual labor market challenges."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:12",
                    "end": "00:27:40",
                    "text": " And this is exactly where public policy must intervene. Israel's approach is therefore comprehensive. We treat AI not as a sectoral innovation, but as a future of work infrastructure challenges. Our national strategy operates on a multi-level pathway. First,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:40",
                    "end": "00:28:11",
                    "text": " basic AI exposure at scale, safe and productive use of language and image tools for every worker and job seekers. Second, workflow integration, embedding AI into processes, templates, quality checks, knowledge bases, automation with governance and result tracking. Third, advanced tools."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:11",
                    "end": "00:28:38",
                    "text": " Upskilling and re-skilling into data, automation, operation, AI systems, AI security, and product rules that are directly job linked. First, advanced and sectoral models. Domain specific applications in health, finance, industries, and public services."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:38",
                    "end": "00:29:07",
                    "text": " deployed in secure environment. But adoption alone is not enough. The labor market is becoming more dynamic, occupational transition are accelerating, skill scales are shortening. We more and more frame training as ongoing support. Modular learning, micro-courses, regular review,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:07",
                    "end": "00:29:36",
                    "text": " and employer-linked pathways that close a clear loop between skill signals and job outcomes. In this ecosystem, the Israeli public employment service plays a central role. We position ourselves not only as a placement agency, but as a bridge across the changing labor market."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:38",
                    "end": "00:30:07",
                    "text": " We connect four domains at scale. One, job seekers. Two, occupations and skills. Three, employers. First, and training system in Israel. Israel model is built on a simple principle. Innovation must move beyond the core. The fourth of our work"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:07",
                    "end": "00:30:36",
                    "text": " is not something that happens to us. It is something we need to make happen. As a startup nation, we know how to build technologics. Now we are focused on translating technological straight into a broad base capabilities, including in the double periphery. This is how AI becomes not only innovation story,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:36",
                    "end": "00:31:18",
                    "text": " but a national growth strategy. I think that Israel and India both are the future. Thank you all. Thank you, Ms. Bashash. Now, please join me in welcoming Dr. Victor Israel Gozelka, head of the Horizon Line Division, who will share how AI is reshaping scientific research methodologies and what this means for the future of discovery. Hello to everyone. Thank you for having me here."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:18",
                    "end": "00:31:55",
                    "text": " During the past several decades, we see more and more researchers to work just to stay in the same rate of progress in science and stay in the same rate. You can see it in the graph here behind me regarding the Moore law, but it's the same in other fields of science."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:55",
                    "end": "00:32:27",
                    "text": " The definition to this phenomena, the name is the society's productivity paradox. When we're speaking about AI and science, there is three perspectives or three different ways to look at it. One is to understand the impact of AI on society, economy,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:27",
                    "end": "00:32:53",
                    "text": " and the ethics and other things in the life. The second way to understand the relationship between the AI and the science is to discover how to enhancement the artificial intelligence itself. And the third one is what I'm focusing my presentation, it's how..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:53",
                    "end": "00:33:25",
                    "text": " we can use the AI to advance the science. As you can see in the slide behind me, the AI brings a change in the way we are doing the science. The scope of the change is revolutionary. It's the same like the microscope discovery or the telescope or computer and their implication in the science."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:27",
                    "end": "00:33:56",
                    "text": " This is my main slide that explains that AI has changed every aspect or every part of the process of science process from generating the hypothesis through the literature synthesis to doing the experiments itself. The tools for doing all this, the literature is already there, copilot,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:56",
                    "end": "00:34:25",
                    "text": " Prism, AlphaFold, and so on. So you just need to take all these tools and implement it in the AI in science. And we think that the future of science is not just using AI as a tool. It's not tools. It's co-scientists. It's not already there. It's work like."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:25",
                    "end": "00:34:51",
                    "text": " with other partners or other scientists, it's not just another tool like computer or Microsoft telescope. As I mentioned, it's a revolutionary paradigm of science. In the Israel Ministry of Innovation, Science and Technology,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:51",
                    "end": "00:35:19",
                    "text": " We set the goal to advance the scientific paradigm by developing my implementation capabilities, specifically dedicated to science. And that's the way we are working to resolve the productivity paradox that I mentioned in the first slide. And we are looking for the collaboration with India because we believe that together we can go to the future of science."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:19",
                    "end": "00:35:56",
                    "text": " by using those tools and find new discoveries in sea science. Thank you very much. So, part of us are leaving, but please stay and continue enjoying our presentations. Thank you very much. Thank you, Dr. Gusalka. Finally, no AI ecosystem can thrive without strong industrial leadership."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:56",
                    "end": "00:36:27",
                    "text": " Now join me in welcoming Mr. Avner Vilan, Chief Operating Officer of AI21, who will speak about how Israel's AI industry, alongside multinational R&D centers, sets global standards and strengthens collaboration between academia, startups, and government. Thank you. I'm going to steal Inbal's shalom and namaste."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:27",
                    "end": "00:36:58",
                    "text": " Thank you Ambassador Erez and the Israeli Embassy here for organizing this event. I'm very honored to represent the Israeli industry in this panel. How do I turn this? Because you know what they say. They say that it takes a village to raise a child. It takes an ecosystem to raise a global AI leader. In Israel we see ourselves as a global AI leader."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:58",
                    "end": "00:37:27",
                    "text": " We have the government, we have the academia, and we have obviously the industry. The government provides our infrastructure, our roads, our electricity for our village. Academia is our school, is our intellectual basis to build our village, to raise our village. But the industry, the industry is the home, is the heart of where the raising of the ecosystem happens."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:27",
                    "end": "00:37:55",
                    "text": " And Israel is one of the only nations in the world that actually possesses the whole AI stack. And our ability to control all the layers of the stack, which I will go into in a second, is what makes us resilient, strong and innovative in pushing this forward. So let's go into details."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:55",
                    "end": "00:38:23",
                    "text": " On the bottom layer, we have the foundation. So we don't have huge infrastructure like other countries, but we do see the infrastructure layer as a very important basis for everything that we do. We need to make sure that we have enough sovereign compute power for our sustainable self-use as well as to support our research, our scientists, our innovators."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:23",
                    "end": "00:38:52",
                    "text": " that they have all the GPUs that they need to develop their innovations. Next layer is the muscle. It's where we build the hardware of the AI. Hardware chip design. Israel is the home of some of the leaders in AI chip design, some hardware manufacturing, edge AI, companies like Habana Labs and Halo."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:52",
                    "end": "00:39:20",
                    "text": " or companies that are focused on this layer. Next level is the brain, is the software infrastructure for AI. This is companies like ourselves, like AI21. At AI21, we build LLMs. We're the Israeli LLM. We build foundation models. We build agent orchestration platform built for accuracy, for scale."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:20",
                    "end": "00:39:48",
                    "text": " and to make sure that we target the mission critical applications because AI is right now just at the beginning of the adoption curve and one of the key challenges for the entire industry, not just in Israel, would be to create the trustworthiness of the AI solutions. So the AIs could deliver accurate, relevant,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:48",
                    "end": "00:40:13",
                    "text": " and timely results. And all these layers are actually what drive the most important layer, which is the value. And this is what my colleagues have discussed today. It's where the applications and real-world solutions across all of the critical pillars we've discussed today, agriculture, education, employment, science, future of work, so."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:13",
                    "end": "00:40:43",
                    "text": " All of these layers are built on top of the foundations below. And in Israel, we were famous as the startup nation. And we're really good at solving real-world problems. And we have a lot of companies working with the government to solve these problems built on top of this stack. So what does our ecosystem look like? So we have four major players in our ecosystem. On top, we have the regulators."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:43",
                    "end": "00:41:13",
                    "text": " define the direction, they set the policy, they block, sorry, they remove, they don't block, they remove. Regulatory blockages, hopefully. And they are the basis for what we do. On the right-hand side, we have the academia. AI is a knowledge-based industry. We must have very strong academia that will raise our..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:13",
                    "end": "00:41:38",
                    "text": " next generation of researchers, of developers that build everything. On the left hand side is actually the private sector, the operators and the public sector. So the people who are consuming AI inside of Israel. In the past Israel's economy or Israel's startups used to try to go to the US or to Europe."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:38",
                    "end": "00:42:02",
                    "text": " And what we've seen in recent years, which is very interesting, that we are focusing now on local market. Israel is on its way to become a trillion-dollar economy. And we see that our partners, both in the public sector as well as the private sector, are actually the best-designed partners that we could ask for. They are demanding. They are, you know..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:02",
                    "end": "00:42:28",
                    "text": " taking us to the limit, but this is what actually enables us to take and build this technology and then spread it to the world. And at the heart, we have the Israeli industry. So it's a combination of local startups. You see some of the logos here. I didn't have enough room. I would need probably three slides to put the entire industry. And we have the multinationals. We have..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:28",
                    "end": "00:42:58",
                    "text": " huge R&D centers of all the big players, Microsoft, Google, NVIDIA. Just in the last CES in January, Jensen, the CEO of NVIDIA mentioned that in one of the core AI projects of NVIDIA, four out of the six chips were developed in Israel. So they're doubling down and there's a lot of investment going on in multinational R&D centers. And this creates these core"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:58",
                    "end": "00:43:28",
                    "text": " AI technology companies create a spillover effect because our company, we're about 250 people. About 150 of them are scientists. Already five startups have spun out of our company. So the people who work at these core companies go and build the companies that are all over the stack that I mentioned before. So to conclude, our vision that Israel would go from the startup nation to the AI nation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:28",
                    "end": "00:43:58",
                    "text": " In Israel, we are famous in dealing with uncertainty. Oh boy, we've had a lot of uncertainty in the last few years. But uncertainty builds resilience. And our strength is in our ability to work through uncertainty with our resilience and to build sustainable company that would power the international AI infrastructure."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:58",
                    "end": "00:44:27",
                    "text": " We're not after the hype cycle, we're after the impact cycle. And this is what we do. We invite everybody to join our village, not just as visitors or as users of our technology, but to help us build and expand our ecosystem to an international ecosystem because what we build can benefit with cooperation with India, with the rest of the region, with the rest of the world."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:27",
                    "end": "00:45:05",
                    "text": " because everything that we do is becoming global. Thank you very much. Thank you, Mr. Villain. Now I invite Mr. Joel Ben-Ohr, who is Director of Digital Health from the Ministry of Health. Okay, so thank you for having me. My name is Joel Ben-Ohr. I'm from the Ministry of Health, from the Digital Health Division. I will try in the very few minutes I have just to talk about our strategy in the past two years."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:05",
                    "end": "00:45:34",
                    "text": " Every slide here can take one hour, a couple of hours of presentation, but feel free to connect me afterwards if you're interested in one of the subjects more deeply. So let's start. So first, what we're trying to achieve in health, I think we have two main goals here. One is to redefine healthcare services, not just adding AI to current service models. If we want to create impact, we need to really change the way we deliver care."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:34",
                    "end": "00:46:03",
                    "text": " The second, I think, is the growth engine. In Israel, we like to see the healthcare services not just as an expenditure, but as said just before, seeing as an investment in creating a powerhouse for companies to expand globally. So this is a win-win situation between tackling our health challenges and a global expansion to powerhouse of AI companies. So this is the vision. How are we planning to get there? So we have three pillars in our strategies."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:03",
                    "end": "00:46:26",
                    "text": " One is obviously data, regulation and transformation. So we'll talk really briefly about each one of them. Data. So we started in Israel very early to digitalize, about 30 years ago. We have a lot of data. Healthcare is a data-driven industry. But because we started early, this industry is working with silos. There was no regulation at the time we began."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:26",
                    "end": "00:46:48",
                    "text": " And currently in Israel, there's a big project called Health Data Portability Law, working very hard to standardize the data to being much more interoperable and navigating and helping data to flow from to every point of care. And obviously this is the first step for AI revolution, right? If you don't get all the relevant data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:48",
                    "end": "00:47:17",
                    "text": " at the point of care, you cannot do anything. So this is something that it's a big project, it's an infrastructure project. Nobody likes to do infrastructure products. People like to do the value side of things, but you have to start with it. And there's a lot of lessons to be learned how to scale this infrastructure project. That would be great to share if anyone is more interested. Second is AI regulation. So we're seeing a big shift from regulating the products."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:17",
                    "end": "00:47:45",
                    "text": " to risk management that is being done by health organizations. Once we, as a small country, looked at the FDA or the CA, the global regulators of software as medical device, and we said, OK, they approved it. Now we can just implement it in Israel or India or any other places. Today, AI became a platform. There is no regulator in the world who can regulate every AI agent that will be implemented in every health care system. That is not a sustainable approach."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:45",
                    "end": "00:48:03",
                    "text": " Therefore, risk should be managed by the health organization themselves. And this carries a lot of responsibility of the health organization. They're coming back to ask us, well, it's great that you're giving us the responsibility. Now we can run fast. But how are we going to do it? Please advise us. Please guide us."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:03",
                    "end": "00:48:33",
                    "text": " very closely with health organization in Israel to building the AI governance for health organization. How can they manage the risk by themselves? And of course, we have government will be accountable for putting these processes in place in each of the health organization. So this is the second pillar. The third is transformation. We want to see the impact of this data and governance of the data infrastructure and the governance infrastructure."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:33",
                    "end": "00:49:02",
                    "text": " And here, there's a lot of use cases going on in Israel, but I just want to talk about how we separate between two main themes. One is, I call them the obvious but complex use case. If it's obvious and not complex, then it's not interested from the government perspective. There's a lot of obvious use cases that are not complex. And I just wanted to give you two examples. One is, obviously, most discharge summaries going out of the hospitals."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:02",
                    "end": "00:49:21",
                    "text": " in two, three years. We would not think that physicians would write them by themselves. Maybe they will go over them. But this is technology that already exists. But to deploy it in scale in every health organization is very, very complex. There's a lot of regulatory issues, accountability issues."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:21",
                    "end": "00:49:47",
                    "text": " So this is what I call the obvious but complex use cases. We see very high impact. The technology is already there. And now the question is how you can scale them. Maybe another example of these kind of projects. The possibility of a physician to communicate with clinical guidelines."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:47",
                    "end": "00:50:16",
                    "text": " take every place a physician meets a patient, we want them to be able, in the context of these specific patients with the relevant data, to be able to consult, to have a consultancy with an LLM that reads all the academic research that is relevant to this patient. This is also a technology that is already in place in many sectors, but putting it in large scale in every hospital in Israel or India, it's a very complex but obvious use case. This is something that we should do."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:16",
                    "end": "00:50:43",
                    "text": " The moonshot projects are also very interesting. And this is, I think, a different mindset and different approach how can government should tackle them is working on things that are more riskier, but with more impact. And this, maybe I'll give an example. Do we really need a physician to write every prescription?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:43",
                    "end": "00:51:04",
                    "text": " or maybe if someone is a chronic patient and he wants to renew every three months, he renews his prescription in Israel. So maybe you don't need every time a physician to write it. So these are much riskier projects and we're now currently building regulatory sandbox for these new kinds of projects."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:04",
                    "end": "00:51:41",
                    "text": " With this, I just want to finish. So as we all said, we're just in the beginning. Collaboration, sharing, and learning from the mistake of each other is the way to succeed. So happy to be here and thank you very much. Thank you, Mr. Yoel. Now, for the closing remarks, I would like to invite Mr. Fares, Deputy Chief of Mission at the Israeli Embassy in New Delhi. Hello, everyone. I will be very brief. I know that we are exceeding our time now. First, I want to thank"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:41",
                    "end": "00:52:11",
                    "text": " the AI Summit organizers and of course the team that's handling us. Thank you very much for this wonderful event. And of course the embassy's team, Maya, Indira, who worked a lot on this event and the entire team. I want to thank our delegation from Israel, representing Israeli government and private sector, showing the Israeli strength. And of course, if we are talking about AI as the future of humanity, I think that India and Israel creates a very important and very powerful engine."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:11",
                    "end": "00:52:38",
                    "text": " into creating this future. And I am sure that the collaboration between our both nations, our countries, is going to create a wonderful future for humanity. So I want to thank you very much for coming here. Of course, all the government representative, the private sector, AI21, Dream, who are also here around us. And thank you, of course, everybody here, the participants and the media for covering this. And I wish you a very successful summit and very successful summit for India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:38",
                    "end": "00:53:25",
                    "text": " Wonderful, wonderful host. Thank you."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
    {
    "_id": "69957407565455da1fc751fa",
    "projectId": "835448",
    "projectName": "Israel's AI Model Building a Better Future Through Innovation - Hindi",
    "projectDescription": "Israel's AI Model Building a Better Future Through Innovation",

    "projectCreatedDate": "2026-02-18T08:10:46.250Z",
    "projectUpdatedDate": "2026-02-18T08:10:46.250Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T08:11:52.673Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Israel_s AI Model_ Building a Better Future Through Innovation.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Hindi"
},
    {
    "_id": "6995746f565455da1fc7520e",
    "projectId": "751123",
    "projectName": "Israel's AI Model Building a Better Future Through Innovation - Spanish",
    "projectDescription": "Israel's AI Model Building a Better Future Through Innovation",
    "projectCreatedDate": "2026-02-18T08:12:30.121Z",
    "projectUpdatedDate": "2026-02-18T08:12:30.121Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T08:13:27.604Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Israel_s AI Model_ Building a Better Future Through Innovation.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Spanish"
},
    {
    "_id": "6995960e565455da1fc753e8",
    "projectId": "887796",
    "projectName": "Fueling the Revolution Democratizing Compute for AI Startups Economic Growth - Hindi",
    "projectDescription": "Fueling the Revolution Democratizing Compute for AI Startups Economic Growth - Hindi",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T10:35:56.999Z",
    "projectUpdatedDate": "2026-02-18T10:35:56.999Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T10:36:55.196Z",
    "projectDeleted": false,
    "projectDataLength": 3,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "Fueling the Revolution_ Democratizing Compute for AI Startups _ Economic Growth(MP3_128K).docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "Fueling the Revolution_ Democratizing Compute for AI Startups _ Economic Growth(MP3_128K).docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Hindi"
},
{
    "_id": "69959435565455da1fc753b8",
    "projectId": "474663",
    "projectName": "Fueling the Revolution Democratizing Compute for AI Startups _ Economic Growth - English",
    "projectDescription": "Fueling the Revolution Democratizing Compute for AI Startups _ Economic Growth - English",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T10:28:04.861Z",
    "projectUpdatedDate": "2026-02-18T10:28:04.861Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T10:32:14.658Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 2,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "Fueling the Revolution_ Democratizing Compute for AI Startups _ Economic Growth(MP3_128K).mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771410473981.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:27",
                    "text": " what the NVIDIA ecosystem is all about and what you have to offer for Indian startups and founders. Thank you, Ari, for the kind introduction. Hello, everybody. One of the pleasures of this job is that I could sit in company of people like them, particularly Swakhu's finding a company like what you're doing. And I saw the show of hands from many of you. So that's amazing, great opportunity. And what you do is disrupt industries."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:27",
                    "end": "00:00:34",
                    "text": " create new businesses and what we do is to create a platform for you which accelerates your workload."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:34",
                    "end": "00:01:03",
                    "text": " Right, and then we build that platform in a five layer cake so that you can take advantage of the platform with whichever layer which you want to build. And then accelerate leveraging that to workloads which you are working on. And that's our continued effort across the developer ecosystem. The company treats everybody as a developer so that you can take advantage of what we are building to do what you want to do. We become a time machine for achieving your results. We have, at any point of time,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:03",
                    "end": "00:01:18",
                    "text": " large amount of developer tools available across domains, across stacks, which enables to accelerate your workloads. That's a real treasure that the company has beyond the semiconductor layer, which we grew on generation to generation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:18",
                    "end": "00:01:42",
                    "text": " Over the last four, five years, particularly in AI, we have developed that stack to reach across domains from everything you can think about, from autonomous all the way to healthcare and then into manifested and physical AI. To support this enablement for the startup ecosystem particularly, what we're doing is enabling them across their journey. We consider any..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:42",
                    "end": "00:01:60",
                    "text": " company, which is less than 10 years as a startup within the program framework, which we call Inception. Everybody is welcome. It's a free program. And then we enable your journey based on where you are in your current growth of the journey. If you're a founder just starting off."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:60",
                    "end": "00:02:28",
                    "text": " Then we give you enablement tools for building what you want, learning tools for getting you foundationally close onto the topics which you want at various levels of expertise with an IP which we have built called Deep Learning Institute, NVIDIA Deep Learning Institute. Very modularly built so that in a very tight time-constrained environment, you can take your learning journey like a Lego brick and then grow up the teams."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:28",
                    "end": "00:02:54",
                    "text": " learning curve also very faster, right? And then enable you in the domain by putting in efforts from our side with our technical teams and route to you all the benefits that the company can offer across various departments to help and enable what you're wanting to do. And once you reach a place where you would want to get connected with the capital, we connect you to friends like Purvi out here who bring the capital to you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:54",
                    "end": "00:03:11",
                    "text": " We will also look at encouraging the entire ecosystem to network between each other so that they can get to know each other. If it makes sense from a perspective of alignment with what you want to do with them, then probably there is an opportunity for our investment arm called Inventures to look at."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:11",
                    "end": "00:03:38",
                    "text": " look and invest into those companies. And then region to region where you would want to go to market, we enable that go to market with supporting ecosystem enablement for you both on the tech side and in the market reach side. What we noticed over this last two years from advantage of where we could see in a general terms is that the capacity for most of the companies are moved really from POC stage into production."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:38",
                    "end": "00:03:51",
                    "text": " across the main areas. They're finding an accelerated number of companies coming up with real practical problems, which they are ready to now take to the market, not just for India, but for the globe. So that's really very exciting."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:51",
                    "end": "00:04:17",
                    "text": " Thank you. Thank you. Thank you. So actually I'd now like to redirect to Schwark, right? So you've had an interesting journey. You've been on both sides of the table. You've chosen this moment in time to step away from being a venture investor, now becoming a founder. So I'd love to get your perspective on what has changed in the last couple of years and what sort of made you move in the direction of becoming a founder and broadly also comment on the AI ecosystem and infrastructure. Yeah, I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:17",
                    "end": "00:04:42",
                    "text": " One of the things I have the good or bad fortune of doing is actually living through the last two bubbles. One was the internet bubble back in the late 90s and then the financial bubble that happened in the mid 2000s. And one of the interesting things with all these things is that there's a big innovation boom that happens when the bubble happens. But the real opportunities are five years out from..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:42",
                    "end": "00:05:02",
                    "text": " the bubble actually popping. And that basically means that you have a tremendous amount of long-term opportunity that's just starting to build. So I'm not talking about all the companies are going to zero to a hundred in eight months right now, which I think is great. But I think there's a lot more sustainable opportunities that'll come. There are 10 times as much in volume."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:02",
                    "end": "00:05:21",
                    "text": " over the next five or 10 years. So when I have the option of either being a VC or a founder, the founder out seemed very, very exciting, especially at this particular point in time. And the reason why I think this is interesting is bubbles tend to..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:21",
                    "end": "00:05:49",
                    "text": " build a lot of foundation, like NVIDIA is creating a lot of hardware, folks like Coreweave and others are basically creating a lot of hosting and actual access to that hardware. There's also people that are building foundational models that can then be used to build higher level applications. And then there's a lot of plumbing that needs to be built to put all of this stuff together. And I think that's the big opportunity that's not fully solved yet because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:49",
                    "end": "00:06:19",
                    "text": " think of today's AI as using a little piece of functionality on its own. But if you can string a lot of those things together in complex and interesting ways, you can do things that are almost unthinkable today. And to make that happen, you need all of the different building blocks. And so we think that there's a lot of missing pieces of infrastructure for the next generation of AI to come together. And that's why I jumped back in as a founder."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:19",
                    "end": "00:06:40",
                    "text": " Fantastic. And you know, when VCs are starting to become founders, you know, it's a good time to be a SADA founder in India. So let's just slightly, you know, change track now and talk a little bit about capital. So what we want to know is about what's the landscape of financing AI companies in India right now, and specifically about the India-US cross border corridor. So I'll first direct my question to Arun."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:40",
                    "end": "00:07:08",
                    "text": " So, you know, IDTA just released an AI and deep tech investment report with venture intelligence and you have a significant amount of commitments. You've got Nvidia on board, the government's backing you. Would love to sort of get your take, Arun, on what the state of AI financing in India is and what's fundamentally different this time around about financing AI companies versus, you know, financing SaaS or any other tech startups in earlier times. Yeah. Sure. Yeah. Thank you, Hari."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:08",
                    "end": "00:07:37",
                    "text": " So the state of AI financing in India today is that there's a reasonable amount of capital available for early stage, for like seed stage, early stage. And then as you go forward, the amount of capital actually is not adequate. So the adequacy in the beginning and then tapering off. Now that is being corrected and being corrected in a pretty big way. So the government announced the Research Development and Innovation Fund, RDIF."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:37",
                    "end": "00:08:00",
                    "text": " for $100,000 crores, $12.5 billion, which would be a big fill-up to the deep tech investment ecosystem. And deep tech today almost invariably includes AI, a huge overlap between the two. So that's one. Two is that venture capital firms have been interested in deep tech. And many of us got together."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:00",
                    "end": "00:08:28",
                    "text": " and a group of us got together and we announced that we will come together under the umbrella of the India Deep Tech Alliance and together commit significant amounts of money towards deep tech investment in India over the next five years. So the firms have committed two and a half billion dollars and I think it's going to increase and off that at least one billion is going to be very much directed towards AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:28",
                    "end": "00:08:56",
                    "text": " In addition to that, we had Nvidia join us as a founding strategic member. They were the first corporate member to join us. And they've been followed by another five very good companies that have joined us. And there'll be more. And the ones that join us include Applied Materials, Lamb Research, Micron, Qualcomm, LNT, CG Power. And they will work along with the investors to..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:56",
                    "end": "00:09:25",
                    "text": " really develop the ecosystem for Deep Tech Investment India. What does that mean? It means mentoring for founders, for entrepreneurs. It means opening labs so that founders and entrepreneurs can take advantage of these large companies. And many of these large companies themselves are customers, they may be acquirers, so that tremendous amount of synergies that come together from investors and large corporates coming together to work with entrepreneurs."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:25",
                    "end": "00:09:49",
                    "text": " So that's the big story about IDTA and generally the AI financing landscape and how that is going to change with this increased focus and with the support of the RDIF that has been launched by the government. Thank you, Arun. Can I also just ask you, how do you think deep tech investing is different from the rest of it? Yeah, it's a very good question. First of all,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:49",
                    "end": "00:10:17",
                    "text": " You know, in a lot of venture investments, you may be looking at what's the size of the market? How quickly can it be attacked? Can the founders of the hustle do it? But in deep tech, there's one additional really important point, which is how deep is the technology? Is the technology going to be a true moat? Is it going to be defensible? That's going to be a very important aspect for most every deep tech investment because often large amounts of money have to be committed"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:17",
                    "end": "00:10:47",
                    "text": " before you actually see results. It's not like in many other areas, consumer areas, very quickly you can say, I've got so many users, 100,000 users, a million users. That's not the case in deep tech. So this large amount of commitment, whether it is to build a chip and go through a tape out, more than one, or whatever, which means a certain amount of time, time it takes to sign up customers, the amount of capital can be quite a lot before you have any sense of what the result will be, which means that you have to have pretty good confidence"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:47",
                    "end": "00:11:16",
                    "text": " in the defensibility of the technology and the talent of the team to execute on it. I think that would be one major difference between other kinds of investment and deep tech. Second, within deep tech, there are many areas. If you're doing a semiconductor chip, it's got a different cycle. If you're making an application at the top of the stack, it'd be a much shorter cycle. So again, there's a spectrum. But the main thing I would look for is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:16",
                    "end": "00:11:35",
                    "text": " Is that deep and defensible technology? And can this team execute it? And can this team also have the temperament to add other people to the team so that once the product is ready, you're getting ready, they can scale the company to reach global markets? Thanks, Arun."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:35",
                    "end": "00:11:50",
                    "text": " So from deep tech, let's move on to enterprise AI. So Purvi, you know, what is your investment thesis today? And what are you seeing as far as startups are concerned in the India-US quarter? You may want to use Schwach's mic because that's, I think. No, we changed the mic. Okay. I think this is good, right? Yeah, we're good now."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:50",
                    "end": "00:12:11",
                    "text": " No see what has changed and see my thesis like I invest in a lot in infrastructure middleware dev tools and then applications as well horizontal as well as vertical see but the biggest thing and the realization that has happened in the last and investment thesis keeps changing every six months to be very fair because things are moving too fast, but one thing that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:11",
                    "end": "00:12:39",
                    "text": " we have come to realize is that an engineer in India versus an engineer in the US is actually fairly using the same tools. They are using the same amount, they are seeing the same problems that probably somebody else in the West is seeing, and that is why they are able to think through what gaps exist, what are the kind of problems they can solve, and that's the reason we're seeing a lot of DevTools, infrastructure, middleware. We're investors in a lot of companies in that space, and they are doing very well. The other thing between India and US is, see,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:39",
                    "end": "00:12:50",
                    "text": " I think historically, and this is again realization from SAS, that we really started late as an ecosystem, right? We were almost 10, 15 years back, I think."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:50",
                    "end": "00:13:20",
                    "text": " I think Salesforce was the first IPO in the US in the SaaS era. We were fresh works for the first one for us, which happened almost 15, 17 years after. In AI, we're almost starting at the same time. Maybe a couple of years here and there, maybe some of the research work happened a lot faster. But I think the information arbitrage, to say so, is actually reducing. And that's why we're seeing Indian entrepreneurs, Indian founders attacking very high quality problems and building solutions that probably in the last decade we have not."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:20",
                    "end": "00:13:42",
                    "text": " seen and I am personally very excited about the quality of products and I think we have seen that it's starting to happen that the quality of product that are coming out of Indian entrepreneurs is actually very very good. I think the one problem that I see though is I think Indian adoption is still low. Enterprise AI I think you know the flywheel that needs to happen for an ecosystem to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:42",
                    "end": "00:14:11",
                    "text": " to really become big is that there need to be funders, which I think all of us are. There's a lot of capital in the market, if someone wants to build a company. So I don't think there is lack of capital. I think there are exceptional founders coming up, very highly technical, really solving smart problems. I do feel like the domestic market, when it comes to adoption, has to really grow. Because if that doesn't grow, everyone is just going to move out, and they will start selling where people are going to buy. So that's where there's been a lot of urge, and we have an internal team."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:11",
                    "end": "00:14:41",
                    "text": " not just for an AI portfolio, because AI portfolio is obviously having an AI adoption, but all our non-AI companies and consumer tech, fintech, we actually have teams which work with them to make sure that they are focusing on AI and adopting AI internally in their teams. And that's something that we keep pushing other teams to do a lot more as well. So that domestic market has to really grow. No, definitely, I mean, AI is not just AI startup's agenda, but everyone's agenda. No, it definitely makes sense. No, thanks for bringing that perspective. So I think, you know,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:41",
                    "end": "00:15:10",
                    "text": " I think we can probably move on to sort of talking about the next sort of set of questions, which I largely had around what is a better place to start a new company today, right? And Schwark, maybe this is to you, right? So you spent time in India, you worked in the Valley and would love to sort of understand if there's a young founder today looking to start up, should they sort of think about, consider, let's say starting up in a Chennai or a Bangalore, or should they think about how could I move to San Francisco, right? How do you sort of think about that and make that decision?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:12",
                    "end": "00:15:33",
                    "text": " That's an excellent question. And I think the way to think about it, I think Purvi had a good framework for it, which is it's about where the founders or engineering is and where the market is. I think in India, the market is probably less mature. So you probably have less people buying."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:33",
                    "end": "00:16:02",
                    "text": " in and around Silicon Valley, you probably have more people buying. Generally, I'd say across the United States, you probably have a little bit more of a buying mindset right now for different kinds of technologies. So you definitely want to be selling there. In terms of where you can build, I think you can totally build out of Bangalore. There's no reason you can't build out of Bangalore anymore or Chennai in our case. And I think we can build technologies pretty much anywhere in the world as long as you have the right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:02",
                    "end": "00:16:26",
                    "text": " go to market infrastructure in the right markets that have the most demand. The quality of engineering and the engineering talent in India is substantially different. I actually started a company in India in 2000, and we're doing this again now in 2026. In those 25 years or so that's transpired in between, the quality of the talent that's available in India is kind of leapfrogged by..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:26",
                    "end": "00:16:44",
                    "text": " leaps and bounds. So the talents here, the infrastructure is there, like Union team are making all that stuff available to folks here. So there's capital from folks like Purvi and others. So there is differently, all the different ingredients to try and build are available here."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:44",
                    "end": "00:17:10",
                    "text": " And especially if the Indian government actually makes even more of that GPU-type compute available in a subsidized manner to startups, that'll be an even bigger reason to try and build out of India. But I think building out of India just makes a lot more sense as long as you have the right bridge to your go-to-market markets. Great. Yeah, Arun, please, yeah. Yeah, I'd like to just build on that point because I think..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:10",
                    "end": "00:17:38",
                    "text": " India's huge competitive advantage today is going to be the capital efficiency. Because when you have the talent and you have the availability of the capital, the capital can go a long, much further way in India. And I can give you examples, you know, I'm on, invested on the board of Agnikul, which is a rocket company. What they have done, building a rocket in Chennai, it would have cost five to ten times that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:38",
                    "end": "00:18:09",
                    "text": " in the United States. I mean, there's no question about it. We've invested in oncology diagnostic company in Pune. Again, the same thing to do in the U.S. would cost at least five times that. So, you know, I totally agree with your point, and I think this should be really a... The call to action is that we have... It's a moment when India can be very capital-efficient. Companies out of here can beat companies elsewhere using less capital."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:09",
                    "end": "00:18:34",
                    "text": " I'm so thankful that Arun called us out and they all made great points. In fact, I wanted to put these three, four things in perspective out here, building on what Arun said first. So the RDIF is a great initiative. It will help ensure that patient capital is available for you all to build something great. In fact, one of the incremental things which I'm hoping will happen and we are really keen to make."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:34",
                    "end": "00:18:58",
                    "text": " possible from whatever we can do to support that, is to also nurture the researchers, all great research researchers, but particularly in AI, which has enabled all the disruptive breakthroughs which we are seeing and nurture them, give them freedom to build that research which they are making into a global company and take it global, right? Potential to build from India and take it global is day one."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:58",
                    "end": "00:19:27",
                    "text": " for Indian companies now. And like I said, the program we run in India itself has 4,200 plus startups. And some of them are building amazing products at the level of capital efficiency. Atnikul is an inception startup. 4,200 plus companies built by founders like you. We're building exceptional products ready for the globe. And most of the existing"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:28",
                    "end": "00:19:57",
                    "text": " sort of barriers is getting over, capacity is building in the country, the tools are openly available, open models, we are aggressively supporting them also. That's becoming available. Capital is becoming available. I think although a lot more patient capital would really help. And like you were saying, the talent and the efficiency which is here is just incredible. So while we have you, can you also talk a little bit about DLI and Inception and what you're doing with founders in India?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:57",
                    "end": "00:20:21",
                    "text": " Sure. Let me start with Inception, per se, and then cover DLI in between. So like I said, Inception is a free program. We nurture the startups through their journey, routing to them all the benefits that the company can offer, starts from enabling them with the capacity which they can have for building what they want. First part of that is to give you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:21",
                    "end": "00:20:40",
                    "text": " tools to build which you want with credits and rubates on our compute infrastructure cloud credits from all our partners the hyperscalers plus local cloud capacity which we build in country from demo also in here in country like from the yota which is building gpu cloud and from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:40",
                    "end": "00:21:09",
                    "text": " all the hyperscalers which they have individual programs. So we route them to you so that you can leverage them to build what you want within the program framework that they're offering. So you have a choice and you can take advantage of what would be most useful and beneficial to you based on the program. We support that further with enabling you to build what you want with the tools which we provide. See the company offers, like I said, libraries, development toolkits, which is the real treasure which you..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:09",
                    "end": "00:21:37",
                    "text": " take advantage of to accelerate the workload which you are working on. At an aggregate, across any workload which you're working on, there is a lift which is possible and lift some of them is particularly in training and inference workload is really incredible. So we take advantage of the full technical backing of the company, route them to you in either a one is to one strategic manner or a one is to many continued effort manner. Towards that learning journey, this was enough."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:37",
                    "end": "00:22:01",
                    "text": " In the early days itself, it was a request from across our customer base that we build capacity for our customers in skilling them. So we built an IP called Deep Learning Institute, which has capsulized fundamental expertise in each area, like, for example, in the case of deep learning."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:01",
                    "end": "00:22:21",
                    "text": " fundamentals of deep learning, to what is natural language programming, to what is conversational AI, and then build them into a tight timeline, eight hour type of listening or two hour type of courses, which is mixed between the theory of that particular topic, like what is deep learning."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:21",
                    "end": "00:22:39",
                    "text": " immediately supported with a practical on it with the GPU enabled in the cloud and so that you can get to experiment on that with real data sets and that program enables you the compute required for that and then you move into the second level of it delivered to you either with an instructor"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:39",
                    "end": "00:22:59",
                    "text": " or in a self-paced manner. And then we scale that across the ecosystem in the country, leveraging this great number of engineering colleges and students in the country, giving it free of cost to them, making the faculty the ambassador for this, and getting them to deliver to their students."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:59",
                    "end": "00:23:22",
                    "text": " And honestly, I took a little bit of unfair advantage of it for India because we could scale it almost 10x of what we have done in the other regions. Over the last four years, we have trained around 300 or 1,000 people in the country with Deep Learning Institute. So across domains, we have now almost 30, 40 courses on that offered both in an instructor-led manner and also in..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:22",
                    "end": "00:23:44",
                    "text": " self-paced manner, so we offer that support to the startups, giving them credits on that. And then take your journey of development on your technical, what you wanting to build it, and once you're ready for that, we have a program within the inception environment, which we call the Capital Connect, which onboards venture capital companies like..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:44",
                    "end": "00:24:12",
                    "text": " pull these innovation capital into the program and then we enable connection between you and them. So if you want the connection to the capital we enable them to be exposed to you so that you can directly interact and we offer also networking opportunities in multiple forms and form factors which enables that interactions to go on. We also sort of table those opportunities to our investment arm called Inventure so that they find alignment to this."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:12",
                    "end": "00:24:38",
                    "text": " in region to region as a part of their journey, they would look at investing into it. And then once you're ready to go to the market, we enable you to that by connecting you to our ecosystem in that region, our friend and team in that region, so that we can get you to be in that market and enable you into that market. And then celebrate your success. If you have a booth here, I'm..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:38",
                    "end": "00:25:08",
                    "text": " hoping and requesting all of you to come and join us. We have put 28 of the top startups in the region out there, from model builders, to HNDKI use case builders across healthcare, to solving problems in legal, to department level problems in any of the use case areas, both on a cloud environment and also with what we have announced with DGX Spark, the GB10 based system."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:08",
                    "end": "00:25:36",
                    "text": " on an air gap mode also and then voice tech we are leading that across multiple use cases as many of them out there with the cost per inference at far lower than the world average currently which you can yourself see and then companies which we are which was working on entertainment to healthcare physical AI across AMRs to drones and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:36",
                    "end": "00:25:54",
                    "text": " and building disruptive technologies. I request you all to come and look at it and hopefully celebrate that. Thank you, Oni. I think that's been a pretty detailed explanation of what NVIDIA has to offer. So I think we spoke a little bit, we touched upon how on the supply side we are sort of sorted."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:54",
                    "end": "00:26:18",
                    "text": " cost of capital, we touched upon it, talent availability in India, and sort of awareness and readiness to sort of attack difficult problems in middleware. I want to spend a little bit of time on the demand side of enterprise AI, right? So which is, you know, what can startups do today? And I'll start with the purvi to sort of make sure that they're able to get traction with enterprises, right? So where's the gap?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:18",
                    "end": "00:26:45",
                    "text": " If I'm a young founder, it makes a lot of sense for me to build something that's B2C. But if I want to get attraction with enterprises, what could I do today, especially in the AI space, to show that, demonstrate that I'm mature and I'm a trusted partner. Yeah, sure. See, I think, I mean, to get into any enterprise, A, that's a tough thing. But just go and chat with people. There are thousands and thousands of problems in any enterprise today. Pick a sector that you like."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:45",
                    "end": "00:26:55",
                    "text": " and where you think you have some unfair advantage to be a founder who can build a big company. If you think you have worked in healthcare before, you understand some processes there."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:55",
                    "end": "00:27:16",
                    "text": " and you can actually have some ideas. You may not have all the ideas, but let's say you have some ideas, you understand certain processes, go speak to people in enterprises there. If you're a lawyer, if you've worked in legal before, go and spend time with law firms. If you've worked in manufacturing, actually go and spend time with manufacturing. The reason I'm saying all of this is that you need to start somewhere."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:16",
                    "end": "00:27:41",
                    "text": " Because from outside everyone can build customer support because that's an obvious problem. It's a big time But the real problems in enterprises are extremely messy today and a lot of sales is also very consultative And that's where people are ready and ready to talk to people who understand that domain a little bit more Because from day one you will not have the solution you may not even have the right perfect problem to solve for but if you can tell them that hey, this is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:41",
                    "end": "00:27:55",
                    "text": " this is AI, this is what is possible with AI and then starts working on a specific use case, people will give you their time, right? I think that's the part that people should start with, especially young founders that just understand that what is that?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:55",
                    "end": "00:28:25",
                    "text": " industry where you have some unfair advantage and everyone like I'm pretty sure this is a big room and full packed thankfully. There is someone who's coming from specific industries and if you go and think and look back that hey what is it that those industries have been doing which fundamentally can be automated with everything you know about AI so far and start thinking and building those solutions, start showing those solutions to those enterprises and you will start seeing some interest and once you have that interest then go deep and find that use case."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:25",
                    "end": "00:28:54",
                    "text": " understood with one enterprise what that looks like, go to second enterprise and start telling them what the first enterprise was doing, because that's what is consultation. And that's the best way to sell an enterprise. So I think enterprise business is always tough, but once you get in, it's super sticky, it makes money, and it's actually great to be building in that space. Thank you. And Arun, if I can invite some comments from you as well. For startups trying to do the India-US corridor sale, what do you think they should be doing, and what do they need to get right in order to succeed?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:55",
                    "end": "00:29:18",
                    "text": " First of all, I want to just reinforce what Puri said, because I came from the world that you're now living in, essentially dealing with enterprises. And I think today, there's a lot of appetite among enterprises to check out AI. So almost everybody will say, give me a proof of concept. So you've got to pick carefully so that it's somebody who can build deeper relationships. It's a really great point."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:18",
                    "end": "00:29:48",
                    "text": " So in terms of the US and India corridor but the question that keeps asking the people keep asking is Should you build for Indian customers first of a global customers? I see a lot of people saying, you know immediately go to San Francisco Start serving customers there because the margins are higher and so forth My own personal view is that you should look at India as a test bet There are very good sophisticated customers here and if you can you know make your solution successful here"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:48",
                    "end": "00:30:16",
                    "text": " then you're in much better shape to take it globally. Secondly, there are many global companies in India. They could be a good stepping stone to test something out and then take it to globally. So my feeling is, use India as a good testbed. It's much more efficient to do it here. And then start looking at how you can solve the same problem elsewhere. But having global markets in mind is really important."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:16",
                    "end": "00:30:42",
                    "text": " Again, it depends on the kind of product. If you're building a semiconductor chip, then you have to have global scale from the beginning. So from the beginning, you need to be solving a problem that's a global problem. In the enterprise case, you solve a problem here, most likely that solution will apply globally. Thank you, thank you. I just wanted to add one very small point. I think on that..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:42",
                    "end": "00:30:48",
                    "text": " point right where where should you start I think you should start where the customer pin point"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:48",
                    "end": "00:31:17",
                    "text": " you can actually see if you're actually building for healthcare in the US, actually healthcare in India and healthcare in the US is very different. It's very tough to start building from India, right? Versus if you're building for, let's say, DevTools, infrastructure, middleware, your customers, at least your pilots can be here because the persona is sitting here. The processes are very similar. So you have to be just closer to your customers and just start building. And one can always be in that conundrum of, you know, I just want to be in SF, but that's tone deaf. A lot of people cannot be in SF."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:17",
                    "end": "00:31:46",
                    "text": " from the get-go, so just start wherever you are, but just make sure that the personas are similar and the tools and processes that they are using are similar. Thank you. So Swarg, if I can just come back to you, right? So I think you touched upon this earlier in terms of focusing on AI solutions that stand the test of time and building for the long term, right? So can I just ask you for some of the problem statements that you've seen which you think make sense for people to focus on as startup founders for the long term?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:46",
                    "end": "00:32:07",
                    "text": " So I think if you look at the kinds of things that, again, coming back to what is selling or what is actually being used, the things that are actually the most impactful right now tend to be voice applications. Those are the things that a lot of corporations are actually putting in production. Because the number of dollars that are spent on what I'll call"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:07",
                    "end": "00:32:36",
                    "text": " proof of concept or play money, so to speak, is in the hundreds of billions of dollars right now. Because every corporation is trying to figure out what they do with their AI. So which means that there's lots of money being spent to try out different things to see what works. So really what you want to look for is the things that are actually moving into production, that are actually scaling to large revenues and taking big chunks of cost or revenue on any given company, any one customer."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:36",
                    "end": "00:32:53",
                    "text": " Voice tends to be one big area in that category. Developer Tools, of course, is another one that seems to see a lot of traction in that category. So there's certain areas that I think have a lot of traction. There's also, I think, infrastructural stuff that..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:53",
                    "end": "00:33:21",
                    "text": " is also starting to become more and more of an issue. So as people are trying to say, I'm going to be fine tuning models. There was a time when everybody was going to try and build their own model. I think now that's less the case. Now the people are trying to say, what do I fine tune to get to what I need to get to? So the tool sets and the tooling to kind of make those kinds of things work is of course also starting to get more interest and traction. There's also foundational elements that I think are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:21",
                    "end": "00:33:46",
                    "text": " Problems that are not fully solved. Context and memory are some of the things that are starting to become bigger and bigger issues. As the context windows, these elements are actually growing. The opportunity to use context more interestingly and more effectively is starting to change, which means there's a whole new slew of opportunities of what you can do at that layer. Memory, of course, of remembering."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:46",
                    "end": "00:34:16",
                    "text": " what has happened before to guide the future actions, all of which, again, as context windows keep growing, all of these different problems start to become bigger and bigger opportunities. So I'd say that the infrastructural opportunities are also quite rich as you look forward. Great, thank you for that. So just with an eye on the clock, what I would love to do is to invite each one of you to just sort of make us make some closing comments. What would you like founders in the room today to take away from the conversation today?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:16",
                    "end": "00:34:48",
                    "text": " What would you like to leave them with? Maybe I can start with Arun. Yeah. In the AI area, in the deep tech area, I'd say solve a difficult problem and solve it using defensible technology. Have an edge, a technological edge, and have it with a team that can execute. Thank you. Unni? Building on what Arun was saying, I think that becomes the foundation, right? From there, I think you're"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:48",
                    "end": "00:35:18",
                    "text": " your limitations are no more what your limitations was earlier. None of those, I think, exist today, right? With AI becoming possible and opening up markets the way it is doing. So, think global, think bigger, and make it a hard problem which will be difficult for others to follow through without that much more being crossed over, right? So, that's my thought."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:20",
                    "end": "00:35:46",
                    "text": " So I think I'll pick on something Arun said a little bit earlier. I think India has some very unique advantages. I think cost advantages being one of them. I think AI specifically has real ways you can leverage that in some making it a superpower. So leverage the cost differential that you have in India to your advantage as opposed to effectively just doing one more."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:46",
                    "end": "00:36:12",
                    "text": " Copycat solution of something else that's being done in a different vertical or different category You can in fact, you know train or fine-tune and do things because the labor that you might need in order to make that happen might be Impossible to do anywhere else in the world, but you can actually get that done here So there's some very interesting opportunities. You can solve tougher problems. You can start"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:12",
                    "end": "00:36:42",
                    "text": " looking at infrastructural kind of components that can be built. So India can actually look at building some real powerful technology. We're trying to do something like that here in Chennai. So I think that's something that I think is possible to be done out of India that everybody should take advantage of. It's tough to go last. But no, I think the only thing I'll say is that the fundamentals of business building has not changed. So attack a deep pain point."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:42",
                    "end": "00:37:12",
                    "text": " solve it in the most consistent way, while using AI as the technology enabler, and just be bold and ambitious. I think it's a golden time to be building in AI, and most of us should be building more things, whether as a company or just our internal stuff, but we should just be using AI a lot more. All right, thank you so much for that. I think we're sort of done with the formal round of panel discussions, but I think we're open for some audience questions perhaps. I don't know if someone can facilitate. Yeah, I can just hand you your mic."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:15",
                    "end": "00:37:41",
                    "text": " Actually, my question is for Arun sir and Puri ma'am. Do you suggest employment, then startup or direct startup? And do you invest in ARR based startup or pre-revenue based startup? And the final thing is, should the new generation startups use lean startup methodology set by Eric Rice or go ahead with user research, then pick a problem statement and go to the customers?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:41",
                    "end": "00:37:58",
                    "text": " One question at a time, man. So what is your first question? First, should we work as an employee in a company, have our own pockets full of money, then start our own startup, or directly start a startup? So I think maybe Purvi, you want to take that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:58",
                    "end": "00:38:25",
                    "text": " No, I can definitely answer that question. I think there is absolutely no right time to start up. I have seen founders, and if you look at YC, half of them are college dropouts. They didn't even finish their studies, leave about employment. So it doesn't really matter whether you've been employed, you've worked before or not, or if you have, maybe you have certain kind of experience that you can leverage. I think to be a founder, you have to learn and unlearn a lot of things."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:25",
                    "end": "00:38:50",
                    "text": " You need to have the courage and the risk appetite. So just get started if you have that today. And as a VC, do you invest in a startup that is building revenue or a pre-revenue-based startup who are ready for launch? Because early student founders don't have money to invest in their own startup, so they need to bootstrap. So for launching, they need, of course, capital."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:50",
                    "end": "00:39:19",
                    "text": " venture capital available at all stages. So there is venture capital available at a pre-seed stage, which is what you are saying, then at a seed stage, and then series A, series B, et cetera. So you will find investors, including ourselves, where we will invest in just ideas. If we really like the thesis, if we really like the founders, we will invest in them. They don't have revenue. Sometimes at the same stage, we will want a product, but no revenue. And while investing the same money in certain categories, we would want a product and a revenue."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:19",
                    "end": "00:39:43",
                    "text": " So it really matters, but you will find investors who are willing to deploy money without revenue. Sorry to take your time, my last question. Sorry, I think there are going to be others who want. You can definitely get in touch with her afterwards. Hello, good morning, panel. Thanks for the great session. My question is, when you are on the IDC stage, VC will fund you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:43",
                    "end": "00:40:13",
                    "text": " If you are not clear with the business models, then the business model will be changed and the market disrupts. What will be the further actions on that? Okay, I mean actually let me take that with my experience as a former entrepreneur. So I think all good ideas or all good battle plans get disrupted the first time you make contact with the enemy. So you can sort of have a great plan on paper. The moment you meet your first prospective client, their feedback is going to..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:13",
                    "end": "00:40:45",
                    "text": " Impact you you're going to have to evolve with that So while there might be startups who are officially saying that they pivoted and there are others who are not calling that out Every company in this world is continuously evolving, right? So I think that there's no there's no right answer to that You always have to continuously change yourself and what you want to do based on the feedback that the market's giving you to get stiff from your idea you have to be Constantly changing your idea business model and so I would say"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:45",
                    "end": "00:41:13",
                    "text": " The one advice that has been given to me which has stood the test of time is fall in love with your problem and not your solution, right? So if your customer problem is real, it's validated. You can keep changing the way you solve it. You could solve it using traditional automation. You could switch over to Gen.AI or AI. But as long as you sort of keep obsessing over that problem and you're clear that that's the problem you want to go after, I think that level of clarity must be there and it's probably immutable. But beyond that, I mean, be opportunistic."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:13",
                    "end": "00:41:52",
                    "text": " figure out where, if you're finding a better problem to work on, of course, you should consider making that switch. Hello, panel. My name is Sameer Nigam, and I am the founder of AI Society, which is a consulting and training company for startups. We consult people how to invest into startups. So recently, I got an offer from a founder. He's into a healthcare startup, and he's building an application for stress management. And I got an offer for a CTO."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:52",
                    "end": "00:42:22",
                    "text": " So I was, when I was looking at this deck and I was looking for the market, there I found there was a lot of compliances among, from government to keep the data within the borders, especially in the healthcare sector, in the field of AI. So, you know, how can, and we are also looking for the funding, obviously, and you know, how can we make this work into, you know, whenever we go to, you know, funding rounds, they always force a lot into, you know, keeping the data within the borders for the future perspective. So how?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:22",
                    "end": "00:42:37",
                    "text": " Can we do that? So let me try to reframe that question. I think we spoke a lot about India-U.S. corridor. So I think the question is really how does data localization rules impact that decision making of should you start up in India or abroad and how you frame your co-op structure."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:37",
                    "end": "00:43:07",
                    "text": " I'm happy to take that. See, there will be certain categories, especially regulatory sectors, BFSI, healthcare, etc., where you will have to be an Indian company because your customers, whether it's a bank or a healthcare company in this case, they would want the data residency to be there. You could be a global company and still have data residency in India. It's a little bit more complicated right now. A lot of global companies are trying to do that with India as well, where they are putting data residency in India. But I think it's a very"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:07",
                    "end": "00:43:34",
                    "text": " it's an important call to make whether you want to serve regulatory sectors in India or not, and if yes, then you should be an Indian company. Okay, thank you. Yeah, thank you. I think we have room for one last question. Yeah, please. Sorry, broader question. Are there segments that you think that there should be more AI startups in, but you don't see? We are building something in media and we don't see the same level of buying happening or aggressiveness in certain sectors, so is that? Sorry, what are you building in?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:34",
                    "end": "00:43:42",
                    "text": " Media media, so we didn't see the same level of funding going in in the sector. So are they sector you staying away from or you think startup should build more in?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:43",
                    "end": "00:44:11",
                    "text": " That's a great question here to be fair. I am seeing startups building across the spectrum I am like which is which is a great place to be in there is obviously a lot more startups and obvious use cases like customer support voice etc But would love to see a lot more in creative generation we're seeing a lot of obviously model companies coming up in video and video image generation, but do feel that the entire creative suit"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:11",
                    "end": "00:44:41",
                    "text": " fundamentally can be rethought of. Do feel that essentially the thesis personally I have is that any knowledge work, no matter who you are, whether you're a lawyer, you're a healthcare expert, you're a marketer, you're a videographer, et cetera, everyone needs to have an OS which they will start creating and they will actually start using their workspace, right? And that I think is an unsolved problem today. In certain categories in law, for example, it has started to happen, but it has not happened for a lot of other knowledge work categories. So that's an interesting space where"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:41",
                    "end": "00:45:01",
                    "text": " We've seen a few things, but not as much. But I'll let others also answer. Yeah, if anyone else wants to add to that. I think we're on the clock. So, you know, thank you so much. You've been a lovely panel and a lovely audience. Yeah, so the panel's available for questions afterwards as well, in case you want to get in touch with any of them. They're all here."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:03",
                    "end": "00:46:23",
                    "text": " I think we're over time, ma'am, so it's okay. You can get in touch with them right after. We're right here. Yeah, so thank you so much. You've been an amazing audience, and please enjoy the rest of your time at the India AI Impact Summit. Thank you. Thank you so much for joining us and sharing the valuable insights. There is a small momento from India AI. So this is the..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:23",
                    "end": "00:46:50",
                    "text": " Thank you."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
    {
    "_id": "69959681565455da1fc75404",
    "projectId": "806991",
    "projectName": "Fueling the Revolution Democratizing Compute for AI Startups _ Economic Growth - German",
    "projectDescription": "Fueling the Revolution Democratizing Compute for AI Startups _ Economic Growth - German",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T10:37:51.776Z",
    "projectUpdatedDate": "2026-02-18T10:37:51.776Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T10:39:02.189Z",
    "projectDeleted": false,
    "projectDataLength": 3,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "Fueling the Revolution_ Democratizing Compute for AI Startups _ Economic Growth(MP3_128K).docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "Fueling the Revolution_ Democratizing Compute for AI Startups _ Economic Growth(MP3_128K).docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "German"
},
{
    "_id": "69959a68565455da1fc754ca",
    "projectId": "129668",
    "projectName": "From Simulation to Reality The Rise of Physical AI - AI Impact Summit 2026 - English",
    "projectDescription": "From Simulation to Reality The Rise of Physical - AI Impact Summit 2026 - English",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T10:54:31.667Z",
    "projectUpdatedDate": "2026-02-18T10:54:31.667Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T11:43:24.988Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 2,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "From Simulation to Reality_ The Rise of Physical AI _ AI Impact Summit 2026(MP3_128K).mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771412061411.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:30",
                    "text": " part in this event. So I'm going to talk about physical AI, and it is what we think about as the next wave of AI. As you all recollect, AI started in 2012 with AlexNet, which you think about as the first contact. And the first applications that got transformed by AI were perception applications. We saw speech recognition."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:30",
                    "end": "00:00:55",
                    "text": " medical imaging, deep recommendation systems. But then the next wave of AI which took place was generative AI, and this is where LLMs came in, content creation, digital marketing, all of that got transformed. And then the big wave which is going on right now, which is agentic AI, where you have a bunch of these AIs working together to solve really hard problems."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:55",
                    "end": "00:01:26",
                    "text": " What you see is that with each wave of AI, the opportunity that it solves, the problems that it solves keeps increasing. The scale of AI keeps increasing. And we are on the cusp of the next big wave of AI, which is the physical AI, and where AI leaves the digital world and actually starts taking decisions and manifesting itself as autonomous vehicles or general robots."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:26",
                    "end": "00:01:53",
                    "text": " A question often I get asked is, why do we think physical AI is going to be bigger than digital AI? And the answer is pretty simple. If you look around us, what IT has done for software industry, our physical world has completely been left untouched by it. If you go to a hospital today, if you go to a manufacturing site, a factory, a retail store,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:53",
                    "end": "00:02:20",
                    "text": " or even inside our homes, we are still doing the same thing that we were doing 20 years ago. So the opportunity for AI when it comes to physical world is huge. The number of atoms in the world is way more than the number of bits. So these are trillions of dollars of industries when you look at manufacturing, when you look at logistics, when you look at retail, when you look at healthcare."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:20",
                    "end": "00:02:49",
                    "text": " which has not at all been touched by AI, and the improvements have been very incremental. And as we have seen with the IT world, when AI comes in, we get 10x, 100x improvements in what we can do with that technology. So just by first principle, the opportunity ahead of us when we bring AI into the physical world is going to be much bigger. And what exactly does separate, you know,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:49",
                    "end": "00:03:11",
                    "text": " digital AI from physical AI. So in physical AI, we think about, we have sensors that are inputs. In digital AI, most of that sensor is images, videos, or text. That's about it. But the physical world has a lot more sensing modalities. We have the notion of touch. We have..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:11",
                    "end": "00:03:41",
                    "text": " the notion of spatial and temporal understanding of the world. We have notion of speech. We have a lot more sensors. So the data modality itself for physical AI is different, right? So we have all the sensor inputs. And then we have, of course, an AI model that has to reason across these modalities. And at the output of it is an action. And we think about at NVIDIA, physical AI will manifest itself in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:41",
                    "end": "00:04:10",
                    "text": " kind of three different forms or embodiments. The most common one that you all might be familiar with that is actually getting deployed at scale today is autonomous cars. I live in San Francisco, I go to the city and I sit in a Waymo and it just drives me around. My commute to office at work is 99% autonomous. So we are starting to see that physically I..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:10",
                    "end": "00:04:33",
                    "text": " taking scale in autonomous vehicles. The second area that we think about for physical AI is robots. These are industrial collaborative robots that we see today, and autonomous mobile robots moving around in the logistics facilities. But the next wave of robots as well, which is going to be more"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:33",
                    "end": "00:05:02",
                    "text": " collaborate more human-like, so it could be humanoids. You saw several different robot form factors in the video that I shared earlier. So these are robots that are moving around, doing things, interacting with objects in the real world. The third physical AI system that we think of is factories and warehouses. And people ask me, like, how is a factory and a warehouse a physical AI? If you have cameras that are monitoring"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:02",
                    "end": "00:05:30",
                    "text": " and taking decisions on what should be routed where, what should be the speed of the conveyor belt, it's checking the quality of your products, and making decisions on when to stop the line. All of that may not be directly doing the action, but it's taking the same thing. It's taking the sensor data from your site, making sense of it with a model, and taking action on what you should do. So in a nutshell,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:30",
                    "end": "00:06:01",
                    "text": " You've got sensors input, you've got a model that understands that sensing input and take actions. And India is uniquely positioned to succeed in the physical AI and robotics world. And there are two big reasons for that. First is India has this natural resource of talent which comes from IT and computer science."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:01",
                    "end": "00:06:28",
                    "text": " physical things, as manufacturing, as all of these operations become a digital problem, it all becomes a software-defined problem, then the talent that has been used to build the IT infrastructure and the digital infrastructure of the world can be used to build the physical infrastructure for India. So that huge talent pool that we have from IT and computer science"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:28",
                    "end": "00:06:59",
                    "text": " can be applied to physical world and transform that world. The second big reason why India can succeed is the emerging manufacturing base. We all know there's a lot of changes that are happening in the manufacturing in the geopolitical environment. India is becoming the base of a lot of manufacturing that will be done in the future. So there's a huge opportunity from there. But India also is a big consumer."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:59",
                    "end": "00:07:29",
                    "text": " You have a big population that consumes these products. So why not build it for yourself? So combining that huge opportunity for manufacturing that is coming because of all the on-shoring that's happening and local push for Made in India combined with the demand puts another dimension to why India can succeed in robotics and physical AI. And if you combine these two things together, you have a virtuous flywheel that enables the success for India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:29",
                    "end": "00:08:00",
                    "text": " for India. Before talking to the solution that we are building, I want to give a little bit of overview of why physical AI is different from digital AI. You know, LLMs have made a lot of progress and they're solving really hard problems. But when you come to the physical world, there are something that is very unique. I talked earlier about multimodality. You have data is coming from lidars, cameras, microphones, touch, torque, IMUs."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:01",
                    "end": "00:08:29",
                    "text": " All of this modality of data is not understood by the models of today. And this data is continuous. And it's something that the models need to really understand. A human being, for example, can generate a lot more data through all his senses than all of internet in a matter of days. So when you think about physical AI, the multimodality becomes very important. The second challenge that..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:29",
                    "end": "00:08:59",
                    "text": " that is faced by physical AI is the data gap. You'll say, well, you just said there's so much data that gets generated, then what's the data gap? The data gap is for AI to learn that data needs to be digitized. You need to tokenize the data. Are we doing that today? Everything that's happening in a factory, in a retail store, in a hospital, none of that operational data is digitized today. So we have this gap where"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:59",
                    "end": "00:09:26",
                    "text": " Let's assume we have a model that can learn from all of the data, but is that data available for the AI to learn? So where are we going to get the data from? The third challenge that comes with physical AI is evaluation. Because this AI is going to live in real world and actually taking actions, the amount of evaluation and rigor that you need to put into testing and validation,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:26",
                    "end": "00:09:54",
                    "text": " is way different than you can do with a LLM that is a chatbot. What's the worst that can happen with a chatbot is you don't like the answer and you say, try again. But if your surgical robot made a mistake, there's no try again. You've got to do it right in the first shot. So evaluation of the AI is super important. And the last thing which is important to understand is the need for accelerated computing on the edge."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:54",
                    "end": "00:10:25",
                    "text": " When Chad GPD came out, it could reach millions and millions of people overnight. Because for distribution, all that you needed was access to a computer or a phone and a website, which is everywhere. But when you think about physical AI, the AI needs to run on the device. Because it has to be low latency, it has to be safe, it has to be for security reasons. You want the AI to be running on the edge. That means"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:25",
                    "end": "00:10:54",
                    "text": " we need to start building systems, we need to start building infrastructure where you have AI compute on the device to run the system models. So, you know, these are some of the big challenges that differentiate LLMs from what we see in the physical AI world. And, you know, the way we are, the biggest, if you're working in this domain, you may have heard a lot about, you know, the data problem that I mentioned earlier."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:54",
                    "end": "00:11:25",
                    "text": " But I think we have a path forward to solve that today. We have the technology to solve that today. And how that's being solved is through what we call the data pyramid. The first source of information that's being used to train physical AI is human videos and web data. India is playing a huge role in providing this data economy. I know of partners who are creating millions and millions of hours of egocentric data just observing people do things."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:25",
                    "end": "00:11:54",
                    "text": " and learning from that. And of course our YouTube and the videos that we have on the internet are great for general understanding of how the world around us operates. So that forms the basis, the foundation layer of how we are going to train the physical AI. But, you know, if everybody could just watch videos and learn, we would all be like feders and tiger, you know, be experts at our craft and sushi chef. It turns out you need to do things in the real world to become good at it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:54",
                    "end": "00:12:23",
                    "text": " And there we are working with simulation and synthetic data, creating highly fidelity models of the world, where actually the robots, once they've learned, they can practice their craft. But that's also not enough, because as we know, any time you try to create a model of something, you lose some fidelity. The real world is way more complex. The physics of the real world is way more complex. We cannot encode it all into simulation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:23",
                    "end": "00:12:53",
                    "text": " Once you have the base layer and a model in simulation where you can learn and practice, then the last yard becomes just a few attempts on the real world. And just with a few attempts, that sim-to-real gap can be closed, and you can have a system that is fully performing. So with this data pyramid, we believe that we can solve the data problem, which has been a big challenge for physical AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:53",
                    "end": "00:13:21",
                    "text": " We think about the transition in robotics is going to be two stages, right? First is the robots that we have today. These are specialist robots. They can use AI, but they can use AI to do only one task really well. If it's a robot that knows how to do welding, it just knows how to do welding. If a robot that knows how to do packaging, sorting, it just knows how to do sorting really well. If it's an AMR that is doing last mile delivery, it only knows that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:22",
                    "end": "00:13:52",
                    "text": " In order to get to the scalable solution, we need journalists. It does not make sense that every time you get a robot you have to start from scratch and teach it what does motion mean, how do you pick up an object. It should have the basic intelligence built into it. So that's where a lot of focus is right now is to create these pre-trained models and to create hardware that is multipurpose. You need to have a general purpose hardware."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:52",
                    "end": "00:14:22",
                    "text": " and a general purpose model that has a basic understanding. But the real unlock of all of that will happen when we have generalist specialists. So if you had just graduates in the world and they were allowed to run the whole country, the whole manufacturing, we would not be where we are. We need specialists who know the craft, who have the domain expertise. And so taking this generalist brain and building specialist apps through your data,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:22",
                    "end": "00:14:51",
                    "text": " for your own domain is where the real unlock happens. And that's where the real opportunity lives. So what's Nvidia doing to solve this problem? We build computers. And as I explained earlier, creating physical AI, deploying physical AI requires something more than what we have seen today. In the world today, for LLMs, you have a data center."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:51",
                    "end": "00:15:21",
                    "text": " That's pretty much all you need, and you need a lot of applications on top of that. You can train in that data center. You can run inference in the data center. That's great. But for physical AI, we need two more computers. We need a simulation computer where you can create data that doesn't exist, where you can have the robots practice and learn the skills that they have acquired by watching videos or by watching other egocentric data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:21",
                    "end": "00:15:50",
                    "text": " The third thing, so that's the simulation computer where you train, test, and create data for your robots. And the third computer you need is the edge computer where your brain is going to get deployed. So at NVIDIA we're building the NVIDIA Jetson which is, and today you can run 2,000 teraflops of compute on the robot brain itself. So tens of billion parameter models which could only be run on the cloud can now be deployed onto the edge."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:50",
                    "end": "00:16:19",
                    "text": " So if you're attending this show here, you want to go to NVIDIA booth and see a lot of our partners actually deploying really advanced models onto the edge. So that's what we do on the computer side. We're building these three computers that will help developers to build physical AI. But that's not sufficient, right? We need to build a software stack on top for the developers to actually leverage these computers."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:19",
                    "end": "00:16:45",
                    "text": " We think about the software NVIDIA is building as fully open platform. The reason we are doing this is, as I mentioned earlier, every domain, every country, every task will require specialization. And if these models and these tools are not open, then you cannot scale the deployment of physical AI and robots. So we are making our entire platform open."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:45",
                    "end": "00:17:14",
                    "text": " So developers in every region of the world can actually get access to it and build their own applications. So the core work on the software side can be bucketed into four categories. We have open models. We have frameworks and tools where developers can write their own applications. In robotics world, it turns out, in addition to AI, you need some more algorithms."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:14",
                    "end": "00:17:41",
                    "text": " that are non-AI, but they still need accelerated computing. So we have a lot of those model libraries for accelerated computing, which we call it the CUDA-X libraries. They are accelerated on the GPUs. And the last piece is these workflows, right? I have a model, I have a tool. How am I going to use those to build a navigation stack, right? So we provide workflows using our libraries and tools."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:41",
                    "end": "00:18:10",
                    "text": " that the developers can take in a tutorial and start building on top. I want to spend a little bit of time talking about simulation and why I feel it's super important. What you see in this picture is a variety of robots. If you went today in the market to go buy these robots, first of all, by the time you get them delivered to you, it's gonna be probably a year out, right?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:11",
                    "end": "00:18:41",
                    "text": " And the second thing is you would need millions of dollars of budget to actually go buy these robots. But what we have done with simulation is in order to get access to any of these robots today, you just need a computer with a GPU and internet access, and you have access to this robot. These are $100,000 robots that is available to a developer anywhere in the world to go build applications."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:41",
                    "end": "00:19:07",
                    "text": " Simulation is not just accelerating how we build robots, but it's really democratizing the access to robotics to everybody in the world. So whether it's a mobile robot, it could be a Funook or a Kawasaki industrial arm, a collaborative arm from UR, the Adverb robots will be soon available here. So all these robots are available at one click in your browser."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:07",
                    "end": "00:19:38",
                    "text": " Imagine what that does for the developer ecosystem and how you build applications. So we are investing heavily in simulation to make sure that the robotics can be democratized. But in addition to being accessible and democratized, simulation is at the heart of how AI robots are getting built. We have environments where robots can learn by practicing. We call that reinforcement learning."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:38",
                    "end": "00:20:07",
                    "text": " We have a product called Isaac Lab. What you can do with Isaac Lab is on a single GPU, you can create 3, 4, 5,000 copies of the same robot and let them learn and practice by themselves. Imagine if you had to do that in the physical world. It literally brings the cost of development down to zero for AI, for robotics. And also, what it does is..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:07",
                    "end": "00:20:37",
                    "text": " the time it takes to build applications reduces. So when you're doing robot learning, you would have to take years and years of data in the real world, which can now be compressed into minutes and hours, and you can have policies that can be deployed. The second thing you can do with simulation is create synthetic data. You can have 3D assets of the environment where you want to deploy your robot, and you can create variations of it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:37",
                    "end": "00:21:07",
                    "text": " You want a wooden version of it, marble version of it, metal version of it. All of that can be created just by programming and creating more data sets. An important thing for robots is testing. You need to be able to test them at scale. So instead of having to have hundreds of robots and breaking them and putting things at risk and safety hazard, you can test your entire robotic stack into simulator."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:08",
                    "end": "00:21:36",
                    "text": " Any robot that you see out today, if you are into robotics, you may have seen all these videos of humanoids doing amazing things from dances to kung fu. Each and every robot that you see there today that's doing those skills has learned it in a simulator that's powered by Nvidia. There is a huge opportunity there to really..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:36",
                    "end": "00:22:07",
                    "text": " jumpstart and accelerate the development of robotics through simulation. So with IsaacSim, which is a tool for where you can create environments, test your robots, to IsaacLab, where you can scale them out and let them learn on their own, the time from thinking about a problem and solving it has dramatically reduced in the world of robotics. One of the important models that we are building is called Cosmos."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:07",
                    "end": "00:22:36",
                    "text": " It's a world foundation model, and as with all other models that Nvidia builds, it's open source. Not just the model weights are open source, but the scripts for fine tuning that model is open sourced. Not just the scripts are open sourced, the data sets that we use for creating these models are open sourced. And why we do that is because we want to build the base layer on which everybody else can build their applications. We absolutely do not know how"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:36",
                    "end": "00:23:06",
                    "text": " a manufacturing site for Tata motors look like. We absolutely do not know what a specific retail store in a particular region looks like. So how is it possible that we can create one model that can serve them all? Instead what we do is we take the knowledge of the word, build it into these models called Cosmos Word Foundation models, and we make it all available so you can bring your own data, fine tune them, and customize it for your domain and application."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:07",
                    "end": "00:23:36",
                    "text": " And there are three categories of models that we have in Cosmos. One is Cosmos Predict. What it does is it takes the text and a video or an image as an input. And it has learned from seeing what happens in a typical world when somebody says something and sees something, what are the next things that can happen. So Cosmos predicts the future based on the prompt and the current sensor data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:37",
                    "end": "00:24:07",
                    "text": " Imagine if you had this capability running in real time. What it does tells the robot is, this is what you should do. So the planning part of the robot is already done. All that you need to do is okay, I know what should be done, the planning is done, I just need to do now the controls aspect of it. So Cosmos Predict is super powerful for, as a simulator, neural simulator, it's a great tool for planning purposes for the robots."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:07",
                    "end": "00:24:37",
                    "text": " Cosmos Transfer is a model that can take video in and video out. When we are collecting data, if you collect the data in a particular factory, in a particular lighting, with a particular environment, you go to a different factory, there's no chance that the environment is gonna be the same. So how do you take the data that was collected in one environment and multiply it so that the models can generalize it? That's where Cosmos Transfer comes in. It can take video in."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:37",
                    "end": "00:25:03",
                    "text": " And with prompts, it can generate different versions. So you can drive a car during the day. You can operate a robot in a specific lighting, collect the data, and then let the computers work on it to augment the data. From that one day driving, it will create six variations or hundreds of variations of it in different environments. And now you can use that to train your model and generalize it across scenarios."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:04",
                    "end": "00:25:32",
                    "text": " The last model in this family is called Cosmos Reason. It is a video language VLM. It takes the video as input and language and prompt as input, and it generates text as an output. I'm gonna dive a little bit deeper into it in the next slides, but Cosmos Reason gives you a lot of capabilities to understand video, which encodes physics, but it also encodes"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:32",
                    "end": "00:26:01",
                    "text": " spatial understanding, it also includes temporal understanding. So it goes a lot more than standard detection and segmentation-based models. So let's take a look at a few use cases where Cosmos Reason, which is a video language model, is used by physical AI developers today. The first one is what we call as data curation and annotation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:01",
                    "end": "00:26:32",
                    "text": " Earlier I was talking about the data being captured and being tokenized so that the AI can learn. Even if you have a car that's driving around the street and collecting the data, before it can be used to train an AI model, there's a lot of work that needs to happen. First of all, you don't want all the data that the car has been collecting because that's a huge amount of data. Half of it is just things that are normal and there's no signal in there."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:32",
                    "end": "00:27:01",
                    "text": " things that the model already knows. So you want to first identify interesting segments of the data that was being collected. How are you going to do that? If you're going to have a person look at all the logs, that's super inefficient. So you can pass them to video language models like Cosmos Reason and say, hey, why don't you just do all the annotation of whatever you saw, break down the video into 30 seconds chunks, and summarize it in text prompts."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:01",
                    "end": "00:27:31",
                    "text": " Once it's done the text prompt summarization of that video now you can go to a LLM, you know choose your favorite one whether it's crock or Gemini or You know Nemotron you can pass all this text and they say okay, this is what happened I'm looking for this type of events and now the LLMs can reason about the text they understand text They don't have to watch the video anymore. They will just look at the text annotations"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:31",
                    "end": "00:28:00",
                    "text": " identify the relevant videos, and boom, from your petabytes of data, you have interesting data that you can use for training of the models. The second one is robot planning and reasoning. Typically, if you look at a robot software stack today, it has atomic skills, move left, go down, pick up object, so it has a set of base commands so you say that it understands."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:01",
                    "end": "00:28:30",
                    "text": " As a person who's controlling the robot, you cannot give it prompt in that sense. You want to say, pick up this thing and put it in that particular bin. Somebody needs to go and translate that into robot-specific language. Robot language is all text. So what do you do? You give it a prompt, you give it a video, and it generates the step-by-step instructions now that robot understands, so it gives in language that the robot understands, and it can do the task."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:31",
                    "end": "00:28:60",
                    "text": " Another big area where we have a lot of partners who are using Cosmos Reason, and you can see sort of these applications at the NVIDIA booth, is in what we call video analytics. So you've seen a lot of cameras mounted at intersections inside facilities for security, safety reasons. But their ability has been very limited, because all that they do today is..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:60",
                    "end": "00:29:31",
                    "text": " They detect a car. That's interesting. But is it not interesting if they can detect a car that's moving at a certain speed? Or is it not interesting if they can detect a car that is going in the wrong direction? In order to do that, you need to not look at the single frame. You need to look at the video. You need to be able to understand, is it going left to right or right to left? What is the speed? So these are spatial and temporal understanding that the current detection and segmentation models lack."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:31",
                    "end": "00:30:00",
                    "text": " This is where Cosmo Reason comes in. It looks at a sequence of videos, understands it, and gives you an answer. And you can ask, hey, how many cars went through this intersection at this particular speed, or which were in this particular lane? That level of querying was just not possible in the past. But now with Vision Language Action with VLM models, you can build these applications really easily by just text prompting it and using a pre-trained model."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:01",
                    "end": "00:30:32",
                    "text": " So the possibilities are endless. I'm just giving you some examples of where developers are using it today. But if you are working with physical AI and robotics, I highly recommend you try the Cosmos models. They're super powerful. They're all open source for you to build on top of. This is another example of how we use Cosmos. Think about training a robot. In a previous session, we were talking about how we have to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:32",
                    "end": "00:30:55",
                    "text": " grab a particular can and put it away, and it seems very non-efficient if you have to do this one time with the right hand, one time with the left hand, and then move the object a little left and a little right. That's just super inefficient use of human capabilities. So what do we do? Well, we show it one time what needs to be done, and then Cosmos model can reason through it and create."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:55",
                    "end": "00:31:29",
                    "text": " variants of it. Now you can just prompt it. Instead of a can, you can say, hey, what if there was a cup here? Can you generate the data for me? What if there was an apple here? Can you generate the data for me? So the whole theme that you will see here is that we want to reduce the effort that it takes to create these AI models. If we were limited by scale of people and time, it will take us years to build these models. We have to rely on compute. Compute is a data strategy for physical AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:30",
                    "end": "00:31:57",
                    "text": " Another big model that we are investing in and we are making available to developers is the Groot model. This has been trained for humanoids, and why humanoids is important is tied to my earlier explanation of what's happening in robotics. The robots of today, they are all very purpose-built, single function, but with the humanoid, we finally have a platform shift that's happening where..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:57",
                    "end": "00:32:20",
                    "text": " you can have a general purpose brain and a general purpose hardware that can do multiple things. So Groot is NVIDIA's vision language action model. It can take the sensor inputs from the robot and then command from an operator of what they want the humanoid to do and it can generate the actions for the humanoid to perform that task. Again."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:20",
                    "end": "00:32:49",
                    "text": " We are not focusing on, there was a question earlier, how do we do that in a wet lab? That's not our focus. Our focus is to make the foundation skills available in a model that you can then use to post-train it for your specific application. And we don't build robots, and we don't build even the solutions for the robots. We build the computers and base infrastructure. And we are proud to have partners that use our technology"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:49",
                    "end": "00:33:17",
                    "text": " to build applications. And we have a pretty big ecosystem of partners here in India. I want to call out a few of them. Adverb is a leading provider of warehouse and logistics solutions. They are showing the robot quadruped here, and they also have a mobile manipulator that's built with NVIDIA technology that is capable of doing this generalized task, into and learning capabilities."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:17",
                    "end": "00:33:42",
                    "text": " ATI makes AMRs for factory and warehouses. E-Info Chips is a great partner of ours. They provide services. So if you have a company that is looking to use NVIDIA technology, you don't have the in-house R&D capability. You can go to E-Info Chips. They will build customized solutions for you. Autonomy."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:42",
                    "end": "00:34:11",
                    "text": " They're a partner that is doing, started off with last mile delivery, but they've expanded into a lot of different areas, like hospitals, and their robots are unique because they can do indoor and outdoor delivery operations, which really opens up the spectrum of what you can do. Ansir Robotics, they're building an interesting category of robots which we call as mobile manipulators. These are AGVs with industrial arms on top of them."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:12",
                    "end": "00:34:40",
                    "text": " What we saw was that once an AGV and AMR takes the payload to a specific location, it used to keep waiting there forever, because you need somebody to unload it. And that made the value of an AMR really less compared to if the robot could actually unload the stuff as well. So with mobile manipulators, the robot can not just move there, but actually do the unloading and loading task as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:40",
                    "end": "00:35:09",
                    "text": " AuraML is a great startup which is investing in the area of simulation. As I told you earlier, all modern robots are trained and built and tested in simulation, and AuraML provides the base infrastructure which can take your data, talk about your application, and give you a simulation environment in which you can go train your models. Orangewood is creating industrial robots."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:09",
                    "end": "00:35:37",
                    "text": " Remember I was telling you about how today the robots have to be programmed in a very old language where very few people know that? Orangewood is changing that. You can just talk to your robot, it'll understand what you want, break it down into tasks, break it down into a language that the robot can understand and perform the specific task. And general autonomy is building robots for developers so that we have a lot more hardware in which people can build applications."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:38",
                    "end": "00:36:05",
                    "text": " Last but probably one of our most important partners in India are GSIs, the system integrators. Robotics is pretty complex problem. It has many parts from design to develop and then operate. And our partners here in India, EY, Infosys, TCS, Wipro."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:05",
                    "end": "00:36:36",
                    "text": " They are bridging this serving the purpose of bringing the industries and the technology together by building customized solutions So they're all here at the event highly recommend you to go and check out what they're building so So we talked a lot about robots Which are you know doing things? I want to briefly talk a little bit about Factories and warehouses and how they are becoming robotics"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:36",
                    "end": "00:37:03",
                    "text": " What you see in this video is, on the left is the physical robots. This is where we are, this is one of our partners' manufacturing site. And as you can see, they're using a digital twin to train these robots how to actually do the server assembly for one of NVIDIA's GPU servers. But on the right side, this is where what we call as the outside in AI. So the first one is inside out AI. The robot sees something."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:03",
                    "end": "00:37:33",
                    "text": " and AI helps it to understand and act on it. Outside in AI is where you don't have actual control of something, but you are observing what the physical agents are doing. And based on that, you are making decisions, right? So this is AI observing other AI and making actions on it. And this particular example is of a forklift, which is manually driven in a warehouse, and there's cameras that are observing it, and you can ask,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:33",
                    "end": "00:37:59",
                    "text": " if the forklift was doing something wrong, if there was a security hazard, if it was operating with the right efficiency or not. So all of this is now possible also with the power of physical AI models. Adding to that, we are seeing several visual AI agents that are being deployed in manufacturing today. The first one is visual inspection."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:59",
                    "end": "00:38:26",
                    "text": " whether you are looking at a part and you want to see whether it's missing something of the label or something is in the wrong place. It's very easily doable using AI agents today. A second thing that is happening is what we call this operator co-pilots. Imagine what happens when you have to start manufacturing a new product. You have to train your entire workforce. You have to give them a manual of how to build. But..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:26",
                    "end": "00:38:56",
                    "text": " Oftentimes, people take wrong steps, even with the manual, and the way they get reported is at the end of the process. It did not work. Well, you trace it back and you saw the operator did something wrong. Wouldn't it be great if there was an AI that was watching the operator at each step and told them, hey, you're doing this wrong or you missed this particular step right when they did that? So with visual AI models and VLM models, because it can understand space and time,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:56",
                    "end": "00:39:25",
                    "text": " It gives you prompts to tell you, to the co-pilot, this is the next thing you need to do so that there is no guessing and you can have a better quality. And the last one is these higher-level orchestration agents. There are a lot of things that need to be orchestrated and coordinated when you're running a manufacturing site or a factory. And with AI agents today, if you can get all the information about"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:25",
                    "end": "00:39:54",
                    "text": " with the sensors of what's happening where all this orchestration can be automated with AI as well. One blueprint that NVIDIA has built that really accelerates the workflows that I talked about earlier is called video search and summarization. Again, it's available openly. You can download this blueprint, run it in the cloud, run it on the device, and actually build applications for computer vision. You can train multiple"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:54",
                    "end": "00:40:25",
                    "text": " you can send in multiple camera streams and do an optical inspection, do safety and security, and do operator assistance. So all of this using this one single blueprint. So with that, I would like to conclude my talk. We have a big event coming up. I would like to open up the invitation to all of you. It's called a GPU technology conference. It's happening in March."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:25",
                    "end": "00:40:55",
                    "text": " our CEO Jensen is going to be giving a keynote. So any of you who are interested in getting a glimpse of what's next for AI and the future of AI, highly recommend you all to join us for GDC. And in conclusion again, huge thanks to all our partners who have supported us in our journey. We could not have done it alone. So thank you so much for being a part of this event and being a part of NVIDIA ecosystem."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:04",
                    "end": "00:41:43",
                    "text": " I think we have a few minutes for questions or? Five minutes, okay. Five minutes, yeah. All right. The question of device or mobile, this robot identification and orchestration and fleet kind of management, do you support all the integrations part and?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:43",
                    "end": "00:42:18",
                    "text": " Suppose multiple robots from multiple vendors, multiple tech stacks, all these issues, how we are going to resolve? I'll just repeat the question. The question was about standardization and how do you have different people, operators, providing technology that can be integrated together. So that is one of the reasons we made our platform open. So if you look at Isaac Sim, it supports ROS, so any ROS robot can talk to it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:18",
                    "end": "00:42:48",
                    "text": " And we also make the APIs available, so if you don't support ROS, you can write a custom wrapper. So that's why the openness of the platform allows you to make a communication bridge. We, of course, support the standard things like MQTT or ROS and other interfaces, but you can extend it to your need. And the second thing we're doing is, I don't know if you're familiar with OpenUSD. This is a standard that we're working with partners around the world so that all the assets"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:48",
                    "end": "00:43:16",
                    "text": " Every physical object can have a digital copy of it, which can be used for simulation, which can be used for data generation for testing. So it's an open standard. It was started by Pixar, but Siemens, Intrinsic from Google, Foxconn, a lot of big industry players are all part of it. They're all making, contributing to it so that we all have a digital twin that we can use."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:16",
                    "end": "00:43:51",
                    "text": " And once you have a standard digital twin, then everything can be connected together to do the simulation. Another question. What would be your plan and what you call vision for quantum and this particular physical AI possibilities and integrations? I think quantum computing, of course, has a lot of benefits. And we are actively working with the ecosystem there to see how"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:51",
                    "end": "00:44:20",
                    "text": " accelerated computing combined with quantum computing and solve some of the problems that require more compute than what is possible today. So I think there are some really hard simulation problems. If you think about simulating the entire factory at particle scale, you cannot do that with today's technology. So I think we are looking at what it can unlock further as that technology becomes available. I wanted to ask, is current edge in compute?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:20",
                    "end": "00:44:45",
                    "text": " ready to support space robotics demand if you see there is a market? Because this was all on Earth applications. Yes, the answer is yes. And there are jets and devices in the International Space Station today that are doing multispectral imaging. And imagine if you're collecting all that satellite data and if you have to stream it down for insights, it's very inefficient. So we have partners."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:45",
                    "end": "00:45:14",
                    "text": " that have CubeSats today and who are doing a lot of analytics. The next wave is likely going to be infrastructure in the cloud. The infrastructure in the cloud is going to be infrastructure in the space for AI computing. Definitely on the edge side, we have deployments right now through these CubeSats and hardened environments in which you can do analytics."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:14",
                    "end": "00:45:39",
                    "text": " Hi, I'm with one question from here. I'm an investor in startups. One common question is, if we have to take physical AI robotics to common use cases. For example, I'm an investor in a company that does ironing, steam ironing. We're just trying to automate that. And how do I leverage your ecosystem to do that? What cost would it take? Is there an ROI on such a project? I would love to take some guidance."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:39",
                    "end": "00:46:09",
                    "text": " Yeah, I think we can talk about it in detail offline, but fundamentally the technology that we're building for physical AI Goes across domains right we don't we don't If you look at if you look at the cost of models They have been trained on a wide variety of data sources which includes some of you know What your domain is needed, but the good thing is that it can be fine-tuned with your own data so what you need is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:09",
                    "end": "00:46:37",
                    "text": " What you need is the data and for your specific domain and our open source models and you can build autonomy on top of that. In terms of ROI, I think often robotics ROI is thought of in a very silo way, which kills a lot of opportunity for robotics. In the past, robotics was thought of as something that's going to improve my cost."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:37",
                    "end": "00:47:06",
                    "text": " It's always like I can automate this thing and I don't need that much labor to do this and what is my ROI for cost. But the problems have completely changed in the last few years for manufacturing, for example. It's not about cost, it's about how quickly can I change my production line. You have new demand coming in from the customers. You have new tariffs and what require you to change how you manufacture things."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:06",
                    "end": "00:47:34",
                    "text": " If you had robotics and if you had digital AI and solutions, overnight you could change how your factory operated. It's not a question about cost anymore, it's about speed at which you can pivot. Another aspect is quality. The product that's coming out when you're building with the robots versus if you have a lot of human operations involved, the quality gets compromised."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:34",
                    "end": "00:48:04",
                    "text": " The ROI of automation, physical AI, is multifaceted. It starts from how quickly you can build, what is the opportunity that you can serve, how can you improve the quality, and, of course, bring down the cost along the way. Do you make non-humanized, general-purpose robots, like, from a first-principle perspective? The human body is humanized because of biological constraints, right? It's not the mechanical optimum. Yeah, so..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:04",
                    "end": "00:48:33",
                    "text": " If you look at the Groot model that we have right now, we tested on a humanoid form factor because it kind of gives us all the complexities of all kind of robots into one. But that does not mean that it cannot be used for a specific industrial robots. Absolutely, yes, it can be used there."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:33",
                    "end": "00:49:02",
                    "text": " You make software side looking really good, but what's on the manufacturing side? Because there is a huge demand on the market, where you are the one of the best provider for the GPU. So are there any new things are you going to innovate or going to do to respond to the demand in the future? Well, I think I gave you the example of Foxconn, what they're doing with the platform. We are digitizing our entire supply chain exactly for this reason."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:02",
                    "end": "00:49:31",
                    "text": " We want to, even before we background, we need to know what is going to be the throughput, how can we optimize and improve the efficiency of our manufacturing. So we are absolutely using all of the technology to improve our own operations as well. Hey Amit, I'm Arjun. Hey Amit, I'm Arjun. I generate and sell millions of hours of egocentric data to a lot of robotics companies. Everyone is buying egocentric data now, but there's such a big embodiment gap."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:31",
                    "end": "00:49:60",
                    "text": " You talked about a layered approach towards it, but there's still so much loss of data happening. So what do you think is the goal standard right now to get the most out of these egocentric or any form of scalable data? Yeah, I think the reason why researchers are pursuing egocentric data is just based on first principles, right? If in a world when we're going to have a lot of humanoids, the easiest way to bridge the same to, you know, the embodiment gap is to build a humanoid."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:60",
                    "end": "00:50:28",
                    "text": " and then assuming that there will be a lot of humanoid. If you can see what's happening in the world right now, last year alone, there were 20,000, 30,000 humanoid's built, and it's going to be order of magnitude more this year. So it's clear that if you are going to have humanoid hardware, the egocentric data, which although has some embodiment gap, is the best, closest data that can be scaled for learning purposes. Yeah, we can talk offline."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:28",
                    "end": "00:50:58",
                    "text": " Yeah, I have a lot of ideas about that. One short question. Okay, Amit, great presentation. My question is, you talked about democratizing AI, but look at how you have NVIDIA Cosmos, you already had Omniverse, and you're talking about ISAC platform. Are we creating an ecosystem which is like what Microsoft did 30 years ago and everybody started using Microsoft Windows and that was it?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:58",
                    "end": "00:51:26",
                    "text": " So, and I'm part of one of your strategic partners. We are really looking at, when we do this upscaling, how much of general training is needed, how much of physics is needed, how much of NVIDIA specific thing is getting into the blood here. Yeah, I think when I say democratize, I mean that all of these things are available to every developer, right? So why did NVIDIA have to build it? Because nobody else is building it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:26",
                    "end": "00:51:56",
                    "text": " And we see that this is a critical infrastructure that needs to be created so that the developers can build applications. There's no doubt in my mind that developers won't be programming computers anymore. They won't be writing code for computers. Everybody's going to be writing code for robots. And they can be of different forms, right? It could be digital robot or a physical robot, but in the future we're going to be writing applications all for robots."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:56",
                    "end": "00:52:36",
                    "text": " In order to do that, you need these base capabilities. So our mission is to make it available, make it open source. And if somebody else is doing something, we are more than happy to build on top. I think time is up. I'll be around for a little bit. Thank you so much. Those who want to exit, please exit from this gate. Thank you so much. It's been a pleasure meeting. Thank you so much."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6995a429565455da1fc75670",
    "projectId": "836968",
    "projectName": "From Simulation to Reality The Rise of Physical AI - AI Impact Summit - Spanish ",
    "projectDescription": "From Simulation to Reality The Rise of Physical AI - AI Impact Summit ",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T11:36:07.774Z",
    "projectUpdatedDate": "2026-02-18T11:36:07.774Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T11:42:16.135Z",
    "projectDeleted": false,
    "projectDataLength": 3,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "From Simulation to Reality_ The Rise of Physical AI _ AI Impact Summit 2026.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "From Simulation to Reality_ The Rise of Physical AI _ AI Impact Summit 2026.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Spanish"
},
{
    "_id": "6995a3ca565455da1fc7565a",
    "projectId": "567918",
    "projectName": "From Simulation to Reality The Rise of Physical AI - AI Impact Summit - Hindi ",
    "projectDescription": "From Simulation to Reality The Rise of Physical AI - AI Impact Summit ",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T11:34:32.758Z",
    "projectUpdatedDate": "2026-02-18T11:34:32.758Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T11:42:20.336Z",
    "projectDeleted": false,
    "projectDataLength": 3,
    "projectType": "DOCUMENT",
    "cfDataLength": 3,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "From Simulation to Reality_ The Rise of Physical AI _ AI Impact Summit 2026.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "From Simulation to Reality_ The Rise of Physical AI _ AI Impact Summit 2026.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Hindi"
},
    {
    "_id": "69957728565455da1fc75270",
    "projectId": "609787",
    "projectName": "How AI Is Transforming Military Sustainment and Readiness - Tamil",
    "projectDescription": "How AI Is Transforming Military Sustainment and Readiness",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T08:24:07.272Z",
    "projectUpdatedDate": "2026-02-18T08:24:07.272Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T08:24:54.104Z",
    "projectDeleted": false,
    "projectDataLength": 3,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "How AI Is Transforming Military Sustainment and Readiness.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "Other"
        },
        {
            "col1": "P1_Part_2",
            "col2": "How AI Is Transforming Military Sustainment and Readiness.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "Other"
        }
    ],
    "sourceLanguage": "Other",
    "targetLanguage": "Tamil"
},
    {
    "_id": "6995768d565455da1fc7525e",
    "projectId": "997853",
    "projectName": "How AI Is Transforming Military Sustainment and Readiness - Hindi ",
    "projectDescription": "How AI Is Transforming Military Sustainment and Readiness",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T08:21:31.347Z",
    "projectUpdatedDate": "2026-02-18T08:21:31.347Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T08:22:30.053Z",
    "projectDeleted": false,
    "projectDataLength": 3,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "How AI Is Transforming Military Sustainment and Readiness.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "Other"
        },
        {
            "col1": "P1_Part_2",
            "col2": "How AI Is Transforming Military Sustainment and Readiness.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "Other"
        }
    ],
    "sourceLanguage": "Other",
    "targetLanguage": "Hindi"
},
    {
    "_id": "69957537565455da1fc75240",
    "projectId": "486914",
    "projectName": "How AI Is Transforming Military Sustainment and Readiness - English",
    "projectDescription": "How AI Is Transforming Military Sustainment and Readiness",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T08:15:50.999Z",
    "projectUpdatedDate": "2026-02-18T08:15:50.999Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T08:20:45.884Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "How AI Is Transforming Military Sustainment and Readiness(MP3_128K).mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771402542404.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:05:17",
                    "end": "00:06:57",
                    "text": " جہاں سر جہاں جہاں جہاں جہاں جہاں جہاں جہاں جہاں جہاں جہاں جہاں جہاں"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:57",
                    "end": "00:07:27",
                    "text": " اور اگسٹ آرڈینس کو اپنے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے مجھے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:27",
                    "end": "00:07:57",
                    "text": " اور اپنے وارفیر کو تصویر کرنے کے لئے اگر ڈیٹا ہی نیو ایمینیشن ہے ڈیٹا ہی گنڈا ہے جنہوں نے اسے پیار کیا جاتا ہے جنہوں نے اسے پیار کیا جاتا ہے جنہوں نے اسے پیار کیا جاتا ہے جنہوں نے اسے پیار کیا جاتا ہے جنہوں نے اسے پیار کیا جاتا ہے جنہوں نے اسے پیار کیا جاتا ہے جنہوں نے اسے پیار کیا جاتا ہے جنہوں نے اسے پیار کیا جاتا ہے جنہوں نے اسے پیار کیا جاتا ہے جنہوں نے اسے پیار کیا جاتا ہے جنہوں نے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:57",
                    "end": "00:08:25",
                    "text": " اس کے بارے میں بہت زیادہ نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی نقصانی ن"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:25",
                    "end": "00:08:54",
                    "text": " دم ٹی وی کو سمارٹ ٹی وی کو سمجھتے ہیں۔ لہذا اسی طرح انویٹیف تکنیکل سلوشنوں کو ہمارے لگیسیس سسٹم کو تصور کرنے کے لئے ہمیں دیکھ رہے ہیں۔ سسٹینمنٹ بہت سمجھ گئی ہے کیونکہ سپلائی چین کوششنے کے لئے ہمارے لگنے کے لئے آئی کیا ہے۔ آئی کو کھوڑنے کے لئے ہمارے لگنے کے لئے ہمارے لگنے کے لئے ہمارے لگنے کے لئے ہمارے لگنے کے لئے ہمارے لگنے کے لئے ہمارے ل"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:54",
                    "end": "00:09:23",
                    "text": " اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:27",
                    "end": "00:09:53",
                    "text": " بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا کی بیوڈیٹا"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:53",
                    "end": "00:10:22",
                    "text": " انہوں نے اپنے ساتھ پلیسی میکنے کے ساتھ ایم ایڈریکٹرٹ اور نش تکنالوڈی ایم ایڈریکٹرٹ اور نش تکنالوڈی ایڈریکٹرٹ اور نش تکنالوڈی ایڈریکٹرٹ اور نش تکنالوڈی ایڈریکٹرٹ اور نش تکنالوڈی ایڈریکٹرٹ اور نش تکنالوڈی ایڈریکٹرٹ اور نش تکنالوڈی ایڈریکٹرٹ اور نش تکنالوڈی ایڈریکٹرٹ اور نش تکنالوڈی ایڈریکٹرٹ اور نش تک"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:22",
                    "end": "00:10:52",
                    "text": " اے آئی این ریسرچ اور انجینرنگ اکسلنس اور سنٹک مرسل بش جیت بشواز چیف ڈیٹا سائنٹیس تارٹا ایلیکسی ڈرائیونگ انوویشنس ان اے آئی اور ادوانس کمپیوٹنگ ڈاک مستہ شیرام انانتا سائنم پارٹنر ان دلوئٹ ڈرائیونگ لارڈ سکیل ڈیجٹرل ڈرائیونگ ڈرائیونگ ڈرائیونگ ڈرائیونگ ڈرائیونگ ڈرائیونگ ڈرائیونگ "
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:52",
                    "end": "00:11:20",
                    "text": " اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:20",
                    "end": "00:12:04",
                    "text": " ایمی اور انڈین آمی کے لئے نش تکنالوژیوں کی تصویر کرنے کے لئے مجھے نہیں مجھے ایڈو میں دیجی ایمی کو دلیور کرنے کے لئے مجھے ایڈو میں دلیور کرنے کے لئے مجھے ایڈو میں دلیور کرنے کے لئے مجھے ایڈو میں دلیور کرنے کے لئے مجھے ایڈو میں دلیور کرنے کے لئے مجھے ایڈو میں دلیور کرنے کے لئے مجھے ایڈو میں دلیور کرنے کے لئے مجھے ایڈو میں دلیور کرنے کے لئے مج"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:04",
                    "end": "00:12:36",
                    "text": " لہذا ایک دیڈیکیٹیڈ سیشن، ایک دیڈیکیٹیڈ پلیٹ فارم کے لئے انجینرنگ سپورٹ یا سسٹیننس یا کمبیٹ فورس ری جنریشن یا ہم اسے کہتے ہیں اس کی وجہ سے نہیں ہے کیونکہ کوئی ملیٹری پار کوئی ملیٹری پار کوئی ملیٹری پار کوئی ملیٹری پار کوئی ملیٹری پار کوئی ملیٹری پار کوئی ملیٹری پار کوئی ملیٹری پار کوئی ملیٹری پار کوئی ملیٹری پار کوئی ملیٹری پار کو"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:36",
                    "end": "00:13:05",
                    "text": " کمانڈرز کے ساتھ جانتے ہیں جیسے اپنے مقاموں کے ساتھ جانتے ہیں جیسے اپنے مقاموں کے ساتھ کمانڈرز کے ساتھ کمانڈرز کے ساتھ کمانڈرز کے ساتھ کمانڈرز کے ساتھ کمانڈرز کے ساتھ کمانڈرز کے ساتھ کمانڈرز کے ساتھ کمانڈرز کے ساتھ کمانڈرز کے ساتھ کمانڈرز کے ساتھ کمانڈرز کے ساتھ کمانڈرز کے ساتھ کمانڈرز کے ساتھ کمانڈرز کے ساتھ کمانڈرز کے ساتھ کمانڈر"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:05",
                    "end": "00:13:35",
                    "text": " جہاں ہم سمجھ سکتے ہیں لہذا ایک ایڈئیہ تھا کہ اس کیل چین کو سمجھ سکتی ہے اور اس نے ترین مطابقات اور ترین مطابقات جہاں ہم سمجھ سکتے ہیں جہاں ہم سمجھ سکتے ہیں جہاں ہم سمجھ سکتے ہیں جہاں ہم سمجھ سکتے ہیں جہاں ہم سمجھ سکتے ہیں جہاں ہم سمجھ سکتے ہیں جہاں ہم سمجھ سکتے ہیں جہاں ہم سمجھ سکتے ہیں جہاں ہم سمجھ سکتے ہیں جہاں ہم سمجھ سکتے ہیں جہاں ہم سمج"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:35",
                    "end": "00:14:06",
                    "text": " جو میرے اپکسی روشنے والدنی روشنے والدنی روشنے والدنی روشنے والدنی روشنے والدنی روشنے والدنی روشنے والدنی روشنے والدنی روشنے والدنی روشنے والدنی روشنے والدنی روشنے والدنی روشنے والدنی روشنے والدنی روشنے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:06",
                    "end": "00:14:32",
                    "text": " اگر میرے شوٹر کو پیدا کیا ہے تو میرے شوٹر کو پیدا کرنا چاہتا ہے جنہوں نے پیدا کیا ہے جنہوں نے پیدا کیا ہے جنہوں نے پیدا کیا ہے جنہوں نے پیدا کیا ہے جنہوں نے پیدا کیا ہے جنہوں نے پیدا کیا ہے جنہوں نے پیدا کیا ہے جنہوں نے پیدا کیا ہے جنہوں نے پیدا کیا ہے جنہوں نے پیدا کیا ہے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:32",
                    "end": "00:15:01",
                    "text": " اندرن آمی ہے جنہوں نے ہمارے لئے اپنے لئے اپنے لئے اپنے لئے اپنے لئے اپنے لئے اپنے لئے اپنے لئے اپنے لئے اپنے لئے اپنے لئے اپنے لئے اپنے لئے اپنے لئے اپنے لئے اپنے لئے اپنے لئے اپنے لئے اپنے لئے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:01",
                    "end": "00:15:32",
                    "text": " انجینرین ریسورس ایلیمینٹس ایڈی رائیٹ پلیس ایمینٹس ایمینٹس ایلیمینٹس ایلیمینٹس ایلیمینٹس ایلیمینٹس ایلیمینٹس"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:32",
                    "end": "00:15:42",
                    "text": " لہذا ہم یہاں پہلے مجھے پہلے ہماری نمبر پہلے ہماری نمبر پہلے ہماری نمبر پہلے ہماری نمبر پہلے ہماری نمبر پہلے ہماری نمبر پہلے ہماری نمبر پہلے ہماری نمبر پہلے ہماری نمبر پہلے ہماری نمبر پہلے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:42",
                    "end": "00:16:06",
                    "text": " دوسرے چیز یہ ہے کہ اپنی طور پر کنٹرول انیلیٹیکس کی تعلیم ہے جس طور پر کلیٹی اور پڑکشن کا تعلیم ہے اور دوسرے چیز یہ ہے کہ سیمولیشن کا تعلیم ہے جس طور پر کلیٹی اور پڑکشن کا تعلیم ہے جس طور پر کلیٹی اور پڑکشن کا تعلیم ہے جس طور پر کلیٹی اور پڑکشن کا تعلیم ہے جس طور پر کلیٹی اور پڑکشن کا تعلیم ہے جس طور پر کلیٹی اور پڑکشن کا تعلیم ہے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:06",
                    "end": "00:17:02",
                    "text": " لہذا اس کے ساتھ اس پلیٹ فارم کو تعلیم کرنا چاہتے ہیں۔ ہم نے اپنے سامنے کے لئے بہت سال کیا ہے۔ لہذا اس کے ساتھ اپنے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامن"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:03",
                    "end": "00:17:34",
                    "text": " دیڈی ایمی نے پہلے بات کیا ہے کہ سمارٹائیزیشن کے لیگیسی ویپن سسٹم اور ہمارے طریقے پر چلانجیز ہے لہذا ہم سب ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہم"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:34",
                    "end": "00:18:04",
                    "text": " درونوں کو سمجھا رہے ہیں۔ لیکن حقیقت یہ بھی ہے کہ معظم ہماری تصویرات ہماری تصویرات ہماری تصویرات ہماری تصویرات ہماری تصویرات ہماری تصویرات ہماری تصویرات ہماری تصویرات ہماری تصویرات ہماری تصویرات ہماری تصویرات ہماری تصویرات ہماری تصویرات ہماری تصویرات ہماری تصویرات ہماری تصویرات ہماری تصویرات ہماری تصویرات ہماری تصویرات"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:04",
                    "end": "00:18:34",
                    "text": " یہ ایک بہت بڑی طاقت ہے جو ہمارے سٹموں کے لئے ہمارے سٹموں کے لئے ہمارے سٹموں کے لئے ہمارے سٹموں کے لئے ہمارے سٹموں کے لئے ہمارے سٹموں کے لئے ہمارے سٹموں کے لئے ہمارے سٹموں کے لئے ہمارے سٹموں کے لئے ہمارے سٹموں کے لئے ہمارے سٹموں کے لئے ہمارے سٹموں کے لئے ہمارے سٹموں کے لئے ہمارے سٹموں کے لئے ہمارے سٹموں کے لئے ہمارے سٹموں کے لئے ہمارے سٹموں کے لئے ہمارے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:34",
                    "end": "00:19:04",
                    "text": " سمارٹائیزیشن کا مطلب ہے کہ ہمارے لئے لیگیسی پلیٹ فارم کو دیٹا اوئیر ایڈاپٹیف دیشن سپورٹ سسٹموں میں تعلق کرنا چاہتے ہیں. ہمارے لئے لئے لیگیسی سسٹموں میں تعلق کرنا چاہتے ہیں جس کے لئے ہمارے لئے سنسرز کو امبیڈ کرنا چاہتے ہیں. ہمارے لئے دیٹا پائپلائنز کو تصویر کرنا چاہتے ہیں۔ اس لیگیسی سسٹم کو ہمارے لئے دیٹا اوئیر سسٹم کو تصویر کرنا چاہ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:04",
                    "end": "00:19:35",
                    "text": " اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طرح اسی طر"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:35",
                    "end": "00:20:03",
                    "text": " پیس ٹائم کی طور پر کامل کرتے ہیں۔ لہذا ہم یہ کوئی پیس ٹائم کی طور پر کامل کرنا چاہتے ہیں۔ یہ کچھ ہے جو ایسی طور پر کامل کی طور پر کامل کی طور پر کامل کی طور پر کامل کی طور پر کامل کی طور پر کامل کی طور پر کامل کی طور پر کامل کی طور پر کامل کی طور پر کامل کی طور پر کامل کی طور پر کامل کی طور پر کامل کی طور پر کامل کی طور پر کامل کی طور پر کامل کی طور پر کامل کی طور پر کامل کی طور پر کامل کی طور پر کامل کی"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:03",
                    "end": "00:20:31",
                    "text": " ہمیں اچھا کرنا چاہتے ہیں کہ جو ہم کیا کرتے ہیں وہ آفلائن اور سکوری نیٹ ورکز ہیں. بہت اہم جب ہم ایک آئیی کو بنائیں گے ہمیں اچھا کرنا چاہتے ہیں کہ مصدر اور بہت اہم کیا ہے کیونکہ ہماری اندرلوپ کے بارے میں اہم ہے جب ہم اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ ا"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:31",
                    "end": "00:21:04",
                    "text": " اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:04",
                    "end": "00:21:35",
                    "text": " لہذا جب میں آرٹلری سسٹموں کو دیکھا جاتا ہوں تو ہمارے مطابقات کیا ہے؟ بسیطا ہمارے پیر مشن ایجائیشن ہمارے پیر مشن ایجائیشن ایجائیشن ایجائیشن ایجائیشن ایجائیشن ایجائیشن ایجائیشن ایجائیشن ایجائیشن ایجائیشن ایجائیشن ایجائیشن ایجائیشن ایجائیشن ایجائیشن ایجائیشن ایجائیشن ایجائیشن ایجائیشن ایجائیشن ایجائیشن ایجائی"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:35",
                    "end": "00:22:04",
                    "text": " لوگوں کے لئے ویڈیو ویڈیو ویڈیو ویڈیو ویڈیو ویڈیو ویڈیو ویڈیو ویڈیو ویڈیو"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:04",
                    "end": "00:22:32",
                    "text": " کمپیٹیشن، بلسٹک کمپیوٹرز، میٹڈیٹا اور آنباری کمپیٹیشن کے لئے آنباری کمپیٹیشن کے لئے آنباری کمپیٹیشن کے لئے آنباری کمپیٹیشن کے لئے آنباری کمپیٹیشن کے لئے آنباری کمپیٹیشن کے لئے آنباری کمپیٹیشن کے لئے آنباری کمپیٹیشن کے لئے آنباری کمپیٹیشن کے لئے آنباری کمپیٹیشن کے لئے آنباری کمپیٹیشن کے لئے آنباری"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:32",
                    "end": "00:22:59",
                    "text": " یہاں ہم ایک بارل انسپیکشن سیسٹم پر دیکھ رہے ہیں جس کا بارل بڑھ رہا ہے اور اس کا بارل بڑھ رہا ہے اور اس کا بارل بڑھ رہا ہے اور اس کا بارل بڑھ رہا ہے اور اس کا بارل بڑھ رہا ہے اور اس کا بارل بڑھ رہا ہے اور اس کا بارل بڑھ رہا ہے اور اس کا بارل بڑھ رہا ہے اور اس کا بارل بڑھ رہا ہے اور اس کا بارل بڑھ رہا ہے اور اس کا بارل بڑھ رہا ہے اور اس کا بارل بڑھ رہا ہے اور اس کا بارل بڑھ رہا ہے اور اس کا بارل بڑھ رہ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:59",
                    "end": "00:23:30",
                    "text": " بہت زیادہ ریزیڈوال کی بہت زیادہ ریزیڈوال کی بہت زیادہ ریزیڈوال کی بہت زیادہ ریزیڈوال کی بہت زیادہ ریزیڈوال کی بہت زیادہ ریزیڈوال کی بہت زیادہ ریزیڈوال کی بہت زیادہ ریزیڈوال کی بہت زیادہ ریزیڈوال کی بہت زیادہ ریزیڈوال کی بہت زیادہ ریزیڈوال کی بہت زیادہ ریزیڈوال کی بہت زیادہ ریزیڈوال کی بہت زیادہ ریزیڈوال کی"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:30",
                    "end": "00:24:09",
                    "text": " اور اسی طرح ہمارے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے والے بڑھنے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:09",
                    "end": "00:24:35",
                    "text": " پیس ڈی ڈی ڈی ایم اے ڈی ڈی ڈی ایم اے ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈی ڈ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:35",
                    "end": "00:25:02",
                    "text": " آئی ایسی طریقے کے لئے ہماری ملٹری قابلتیوں اور سٹریٹک ایڈوانٹیج کو تصویر کرنے کے لئے آئی ایسی طریقے کے لئے آئی ایسی طریقے کے لئے آئی ایسی طریقے کے لئے آئی ایسی طریقے کے لئے آئی ایسی طریقے کے لئے آئی ایسی طریقے کے لئے آئی ایسی طریقے کے لئے آئی ایسی طریقے کے لئے آئی ایسی طریقے کے لئے آئی ایسی طریقے کے لئے آئی ایسی طریقے کے لئے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:02",
                    "end": "00:25:31",
                    "text": " اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:31",
                    "end": "00:25:60",
                    "text": " ملٹری درونوں کو بہت زیادہ بہت زیادہ ملٹری درونوں کو بہت زیادہ ملٹری درونوں کو بہت زیادہ ملٹری درونوں کو بہت زیادہ ملٹری درونوں کو بہت زیادہ ملٹری درونوں کو بہت زیادہ ملٹری درونوں کو بہت زیادہ ملٹری درونوں کو بہت زیادہ ملٹری درونوں کو بہت زیادہ ملٹری درونوں کو بہت زیادہ ملٹری درونوں کو بہت زیادہ ملٹری درونوں کو بہت زیادہ ملٹری درونوں کو بہت زیادہ ملٹری درونوں کو"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:60",
                    "end": "00:26:29",
                    "text": " اس پیلڈ کے ساتھ اس کے ساتھ ڈرونوں کی تصویرات کے ساتھ ڈرونوں کی تصویرات کے ساتھ ڈرونوں کی تصویرات کے ساتھ ڈرونوں کی تصویرات کے ساتھ ڈرونوں کی تصویرات کے ساتھ ڈرونوں کی تصویرات کے ساتھ ڈرونوں کی تصویرات کے ساتھ ڈرونوں کی تصویرات کے ساتھ ڈرونوں کی تصویرات کے ساتھ ڈرونوں کی تصویرات کے ساتھ ڈرونوں کی تصویرات کے ساتھ ڈرونوں کی تصویرات کے ساتھ ڈرون"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:29",
                    "end": "00:26:58",
                    "text": " میں آپ کو بتا سکتا ہوں کہ اپلیگیشن اسی طور پر بہت زیادہ افیشنسی اور افیشنسی کے لئے افیشنسی اور افیشنسی کے لئے افیشنسی اور افیشنسی کے لئے افیشنسی کے لئے افیشنسی کے لئے افیشنسی کے لئے افیشنسی کے لئے افیشنسی کے لئے افیشنسی کے لئے افیشنسی کے لئے افیشنسی کے لئے افیشنسی کے لئے افیشنسی کے لئے افیشنسی کے لئے افیشنسی کے لئے اف"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:58",
                    "end": "00:27:27",
                    "text": " کسی مصطفیات کے لئے اسی طور پر ہمارے لئے اسی طور پر ہمارے لئے اسی طور پر ہمارے لئے اسی طور پر ہمارے لئے اسی طور پر ہمارے لئے اسی طور پر ہمارے لئے اسی طور پر ہمارے لئے اسی طور پر ہمارے لئے اسی طور پر ہمارے لئے اسی طور پر ہمارے لئے اسی طور پر ہمارے لئے اسی طور پر ہمارے لئے اسی طور پر ہمارے لئے اسی طور پر ہمارے لئے اسی طور پر ہمارے لئے اسی طور پر ہمارے لئے اسی ط"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:27",
                    "end": "00:27:54",
                    "text": " آئی ایسی تکنالوڈی و سپیس کا مطلب ہے کہ آئی ایسی تکنالوڈی و سپیس کا مطلب ہے کہ آئی ایسی تکنالوڈی و سپیس کا مطلب ہے کہ آئی ایسی تکنالوڈی و سپیس کا مطلب ہے کہ آئی ایسی تکنالوڈی و سپیس کا مطلب ہے کہ آئی ایسی تکنالوڈی و سپیس کا مطلب ہے کہ آئی ایسی تکنالوڈی و سپیس کا مطلب ہے کہ آئی ایسی تکنالوڈی و سپیس کا مطلب ہے کہ آئی ایسی تکنالوڈی اور"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:54",
                    "end": "00:28:24",
                    "text": " اس لئے اپنے سیبر سیکورٹی چیزوں کو سیبر سیکورٹی چیزوں کو ایسی طریقوں کو ایسی طریقوں کو ایسی طریقوں کو ایسی طریقوں کو ایسی طریقوں کو ایسی طریقوں کو ایسی طریقوں کو ایسی طریقوں کو ایسی طریقوں کو ایسی طریقوں کو ایسی طریقوں کو ایسی طریقوں کو ایسی طریقوں کو ایسی طریقوں کو ایسی طریقوں کو ایسی طریقوں کو ایسی طریقوں کو ایسی طریقوں کو ایسی طریقوں کو ایسی طریقوں کو ایسی طریقوں"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:24",
                    "end": "00:28:51",
                    "text": " اس کے لئے ہمارے بھی بہت سارے پروڈکٹس ہیں جنہوں نے اٹھنوموز درونوں اور اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز اٹھنوموز"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:51",
                    "end": "00:29:18",
                    "text": " ہمارے دنیا کے بارے میں بہت زیادہ تصویر ہوتا ہے اور سسٹیننس کے بارے میں سمارٹ سسٹیننس موڈلوں کے بارے میں سمارٹ سسٹیننس موڈلوں کے بارے میں سمارٹ سسٹیننس موڈلوں کے بارے میں سمارٹ سسٹیننس موڈلوں کے بارے میں سمارٹ سسٹیننس موڈلوں کے بارے میں سمارٹ سسٹیننس موڈلوں کے بارے میں سمارٹ سسٹیننس موڈلوں کے بارے میں سمارٹ سسٹیننس موڈلوں کے بارے میں سمارٹ سسٹیننس موڈلوں"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:18",
                    "end": "00:30:01",
                    "text": " ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:01",
                    "end": "00:30:29",
                    "text": " اور ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے میں ہمارے بارے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:29",
                    "end": "00:30:58",
                    "text": " ہمارے پیرویڈوں میں ہمارے پیرویڈوں میں ہمارے پیرویڈوں میں ہمارے پیرویڈوں میں ہمارے پیرویڈوں میں ہمارے پیرویڈوں میں ہمارے پیرویڈوں میں ہمارے پیرویڈوں میں ہمارے پیرویڈوں میں ہمارے پیرویڈوں میں ہمارے پیرویڈوں میں ہمارے پیرویڈوں میں ہمارے پیرویڈوں میں ہمارے پیرویڈوں میں ہمارے پیرویڈوں میں ہمارے پیرویڈوں میں"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:58",
                    "end": "00:31:25",
                    "text": " اس کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے دنوں کے بعد اپنے د"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:25",
                    "end": "00:31:58",
                    "text": " اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے لئے اس کے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:58",
                    "end": "00:32:28",
                    "text": " اسی طور پر کنفلکت کے ساتھ ہے یہ کچھ ہے جو ہم نے انیلائیس کیا ہے جو ہم نے اپنے ایکوپنڈ کو دیکھنے کے لئے ہم نے اس کی مطلب کیا ہے اگر نہیں تو ہم نے اس کی مطلب کیا ہے کہ ہم نے اس کی مطلب کیا ہے کہ ہم نے اس کی مطلب کیا ہے کہ ہم نے اس کی مطلب کیا ہے کہ ہم نے اس کی مطلب کیا ہے کہ ہم نے اس کی مطلب کیا ہے کہ ہم نے اس کی مطلب کیا ہے کہ ہم نے اس کی مطلب کیا ہے کہ ہم نے اس کی مطلب کیا ہے کہ ہم نے اس کی مطلب کیا ہے کہ ہم نے اس کی مطلب کیا ہے کہ ہم نے اس کی مطلب کیا ہے کہ ہم"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:28",
                    "end": "00:32:57",
                    "text": " جب آپ ایک دیکھنے کے لئے جانتے ہیں تو وہ آپ کی تاریخ دیکھتا ہے لیکن وہ آپ کو صرف اس تاریخ کی طرف دیکھنے کے لئے نہیں کرتا ہے وہ آپ کی حالی تاریخ دیکھتا ہے اور اس کی تاریخ دیکھتا ہے اور پھر ایک تاریخ دیکھتا ہے جہاں ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:57",
                    "end": "00:33:25",
                    "text": " اگر آپ نے ہمیں اپنی ریزیڈلی ویڈیویلی زندگی کو دیتے ہیں جس کیا ہے کہ ہمارے لئے پریڈکٹیو اور پریسکرپٹیو مانٹیننس کے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے لئے ہمارے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:25",
                    "end": "00:33:52",
                    "text": " ہمیں ایک دیٹرلوگر میں لگ کرنے کی ضرورت ہے۔ اسے بندل کرنے کی ضرورت ہے۔ ہمیں ایک پلیٹفونڈ جانتے ہیں جہاں ہم کنڈیشن بیس بنیٹرنگ کرنے کے لئے ہمیں ایک کنڈیشن بیس بنیٹرنگ کرنے کے لئے ہمیں ایک کنڈیشن بنیٹرنگ کرنے کے لئے ہمیں ایک کنڈیشن بنیٹرنگ کرنے کے لئے ہمیں ایک کنڈیشن بنیٹرنگ کرنے کے لئے ہمیں ایک کنڈیشن بنیٹرنگ کرنے کے لئے ہمیں ایک کن�"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:52",
                    "end": "00:34:21",
                    "text": " ہمارے پارٹنروں کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامن"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:21",
                    "end": "00:34:51",
                    "text": " ہمیں آپ کو جمعہ کرنا چاہتے ہیں کہ ہمارے بڈ میں جانتے ہیں کہ ہمارے بڈ میں جانتے ہیں کہ ہمارے بڈ میں جانتے ہیں کہ ہمارے بڈ میں جانتے ہیں کہ ہمارے بڈ میں جانتے ہیں کہ ہمارے بڈ میں جانتے ہیں کہ ہمارے بڈ میں جانتے ہیں کہ ہمارے بڈ میں جانتے ہیں کہ ہمارے بڈ میں جانتے ہیں کہ ہمارے بڈ میں جانتے ہیں کہ ہمارے بڈ میں جانتے ہیں کہ ہمارے بڈ میں جانتے ہیں کہ ہمارے بڈ میں جانتے ہیں کہ ہمارے بڈ میں جانتے ہیں کہ ہمارے بڈ میں جان"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:51",
                    "end": "00:35:29",
                    "text": " پن انڈین آرمی دیپلائیمنٹ اور اگر یہ سکسرل ہے اس پلیٹ فارم میں ہم اسے سوچنا چاہتے ہیں جنٹلمنٹ میں یہاں ہمارے ساتھ کھانے کے لئے یہاں ہمارے ساتھ کھانے کے لئے یہاں ہمارے ساتھ کھانے کے لئے یہاں ہمارے ساتھ کھانے کے لئے یہاں ہمارے ساتھ کھانے کے لئے یہاں ہمارے ساتھ کھانے کے لئے یہاں ہمارے ساتھ کھانے کے لئے یہاں ہمارے ساتھ کھانے کے لئے یہاں ہمارے ساتھ کھانے کے لئے یہاں"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:29",
                    "end": "00:35:58",
                    "text": " میرے ساتھ ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ایسے ای"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:58",
                    "end": "00:36:24",
                    "text": " یہ سب سے ہے کہ جو بہتر اور افیسینی بیپن ہے اور بہت اہم جو ایسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی افیسی ا"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:24",
                    "end": "00:36:46",
                    "text": " اور ہم نے تکلیم کیا تھا اور اس کے بارے میں اس کی تعلیم کیا ہے کہ ہم اس کے بارے میں اس کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم کی تعلیم"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:46",
                    "end": "00:37:11",
                    "text": " جیسے اپنے انتہائیوں میں جنہوں نے اپنے انتہائیوں اور انتہائیوں اور اس طرح انتہائیوں میں جنہوں نے اپنے انتہائیوں میں جنہوں نے اپنے انتہائیوں نے اپنے انتہائیوں نے اپنے انتہائیوں نے اپنے انتہائیوں نے اپنے انتہائیوں نے اپنے انتہائیوں نے اپنے انتہائیوں نے اپنے انتہائیوں نے اپنے انتہائیوں نے اپنے انتہائیوں نے اپنے انتہائیوں نے اپنے انتہائیوں نے اپنے انتہائیوں نے اپنے انتہائیوں نے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:11",
                    "end": "00:37:37",
                    "text": " ہراری کل اکسس ہے لہذا پہلے چیلنج اس کے بارے میں دیتا ہے لہذا ہمیں سیلوز میں دیتا ہوں گے یقینی طور پر ہمیں سیلوز میں دیتا ہوں گے جس طور پر ہمیں سیلوز میں دیتا ہوں گے جس طور پر ہمیں سیلوز میں دیتا ہوں گے جس طور پر ہمیں سیلوز میں دیتا ہوں گے جس طور پر ہمیں سیلوز میں دیتا ہوں گے جس طور پر ہمیں سیلوز میں دیتا ہوں گے جس طور پر ہمیں سیلوز میں دیتا ہوں گے جس طور پر ہ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:37",
                    "end": "00:38:04",
                    "text": " میں نے اپنے چھوٹے میں اپنے چھوٹے میں اپنے انتہائیوں کو اپنے انتہائیوں کو اپنے انتہائیوں کو اپنے انتہائیوں کو اپنے انتہائیوں کو اپنے انتہائیوں کو اپنے انتہائیوں کو اپنے انتہائیوں کو اپنے انتہائیوں کو اپنے انتہائیوں کو اپنے انتہائیوں کو اپنے انتہائیوں کو اپنے انتہائیوں کو اپنے انتہائیوں کو اپنے انتہائیوں کو اپنے انتہائیوں کو اپنے انتہائیوں کو اپنے انتہائیوں کو اپنے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:04",
                    "end": "00:38:30",
                    "text": " جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں جنہیں"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:30",
                    "end": "00:39:01",
                    "text": " یہ حقیقت ہے کہ ایک ایسی سلوشن کو اس پرڑھنے کی ضرورت نہیں ہے کہ ہر پرڑھنے کو ایک ایسی سلوشن کی ضرورت نہیں ہے لہذا یہ ایک ایسی سلوشن کی ضرورت ہے کہ یہ ایک ایسی سلوشن کی ضرورت ہے کہ یہ ایک ایسی سلوشن کی ضرورت ہے کہ یہ ایک ایسی سلوشن کی ضرورت ہے کہ یہ ایک ایسی سلوشن کی ضرورت ہے کہ یہ ایک ایسی سلوشن کی ضرورت ہے کہ یہ ایک ایسی سلوشن کی ضرورت ہے کہ یہ ایک ایسی سلوشن کی ضرورت ہے کہ یہ ایک ایس"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:01",
                    "end": "00:39:29",
                    "text": " جنہوں نے کہا کہ ہمارے ساتھ سلوک کرتے ہیں۔ لہذا یہ بہت اہم ہے اور جب ہم اسے بات کرتے ہیں کہ ہمارے سلوکوں کو سلوک کرتے ہیں اور جنہوں نے کہا کہ ہمارے سلوکوں کو سلوک کرتے ہیں اور جنہوں نے کہا کہ ہمارے سلوک کو سلوک کرتے ہیں اور جنہوں نے کہا کہ ہمارے سلوک کو سلوک کرتے ہیں اور جنہوں نے کہا کہ ہمارے سلوک کو سلوک کرتے ہیں اور جنہوں نے کہا کہ ہمارے سلوک کو سلوک کرتے ہیں اور جنہوں نے کہا کہ ہمارے سلوک کو سلوک کرتے ہیں"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:29",
                    "end": "00:39:55",
                    "text": " دیٹر لئیر بہت اہم ہے جیسا کہ اسے دیٹر کو کلکٹنگ کرنے کے لئے ایک آئی موڈل بننے کی ضرورت ہے جیسا کہ جب آپ اسے دیٹر کرنے کے لئے ہمیں امید کرنے کی ضرورت ہے کہ ہمارے پیپلائن کے ساتھ ہمارے پیچھروں کے ساتھ ہمارے پیچھروں کے ساتھ ہمارے پیچھروں کے ساتھ ہمارے پیچھروں کے ساتھ ہمارے پیچھروں کے ساتھ ہمارے پیچھروں کے ساتھ ہمارے پیچھروں کے ساتھ ہمارے پیچھروں کے ساتھ ہ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:55",
                    "end": "00:40:25",
                    "text": " وہاں اپنی اطلاعی ایسی سلوشن کو بنائیں گے۔ ایسی سلوشن کو بنائیں گے اور اگر ایسی سلوشن کو بنائیں گے وہاں ایسی سلوشن کو بنائیں گے اور ایسی سلوشن کو بنائیں گے اور ایسی سلوشن کو بنائیں گے اور ایسی سلوشن کو بنائیں گے اور ایسی سلوشن کو بنائیں گے اور ایسی سلوشن کو بنائیں گے اور ایسی سلوشن کو بنائیں گے اور ایسی سلوشن کو بنائیں گے اور ایسی سلوشن کو بنائیں گے اور ایسی سلوشن کو بنائیں گے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:25",
                    "end": "00:40:54",
                    "text": " پیس ٹائم ، نہیں وارڈ ٹائم ، لہذا اب یہ سوچا ہے کہ اپنی تعلیمات سے اپنی تعلیمات سے اپنی تعلیمات سے اپنی تعلیمات سے اپنی تعلیمات سے اپنی تعلیمات سے اپنی تعلیمات سے اپنی تعلیمات سے اپنی تعلیمات سے اپنی تعلیمات سے اپنی تعلیمات سے اپنی تعلیمات سے اپنی تعلیمات سے اپنی تعلیمات سے اپنی تعلیمات سے اپنی تعلیمات سے اپنی تعلیمات سے اپنی تعلیمات سے اپنی تعلیمات سے اپنی ت"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:54",
                    "end": "00:41:24",
                    "text": " بہت زیادہ مشکل کریں اس کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:31",
                    "end": "00:41:60",
                    "text": " شکریہ جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:60",
                    "end": "00:42:20",
                    "text": " لہذا یہ اپنے سوپٹر دیفنڈ کو آئی دیفنڈ ہوتا ہے لہذا سوپٹر دیفنڈ کا مطلب ہے کہ آپ کے ساتھ سوپٹر دیفنڈ کے ساتھ سوپٹر دیفنڈ ہوتا ہے اور اب یہ اپنے سوپٹر دیفنڈ کے ساتھ آئی دیفنڈ ہوتا ہے ہمان دیگر ہمان دیگر ہمان دیگر ہمان دیگر ہمان دیگر ہمان دیگر ہمان دیگر ہمان دیگر ہمان دیگر ہمان دیگر ہمان دیگر ہمان دیگر ہمان دیگر ہمان دیگر ہمان"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:20",
                    "end": "00:42:47",
                    "text": " سواری نیائی ہے لہذا آپ نے سواری نیائیوں کو دیکھا ہے جو ہوتا ہے جو ہوتا ہے وہ سواری نیائیوں کو بڑھتے ہیں اور ہیپر وار ہیپر وار ہیپر وار ہیپر وار ہیپر وار ہیپر وار ہیپر وار ہیپر وار ہیپر وار ہیپر وار ہیپر وار ہیپر وار ہیپر وار ہیپر وار ہیپر وار ہیپر وار ہیپر وار"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:47",
                    "end": "00:43:16",
                    "text": " اگر آپ کو اپنی ایسی ایسی کوشش کرنے کے لئے ایسی کوشش کرنے کے لئے ایسی کوشش کرنے کے لئے ایسی کوشش کرنے کے لئے ایسی کوشش کرنے کے لئے ایسی کوشش کرنے کے لئے ایسی کوشش کرنے کے لئے ایسی کوشش کرنے کے لئے ایسی کوشش کرنے کے لئے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:16",
                    "end": "00:43:44",
                    "text": " ہمارے بیٹلفیلڈ سینارئیوں میں رکھنے کے لئے نہیں آتے گا کیونکہ یہ کمیونکشن فیلیور ہے اور بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے ب"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:44",
                    "end": "00:44:10",
                    "text": " انہیں ایک دینامی کلی جنگ سیناڈیو میں آپ کو اسے نہیں کرنا چاہتے ہیں آپ کو ایک ایسی تعلیم کرنا چاہتے ہیں آپ کو کہتے ہیں کہ مجھے اپنی ترینیڈ موڈل میں آتے ہیں میں اپنی ترینیڈ موڈل میں اپنی ترینیڈ موڈل میں اپنی ترینیڈ موڈل میں اپنی ترینیڈ موڈل میں اپنی ترینیڈ موڈل میں اپنی ترینیڈ موڈل میں اپنی ترینیڈ موڈل میں اپنی ترینیڈ موڈل میں اپنی ترین"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:10",
                    "end": "00:44:38",
                    "text": " اس کے لئے کونٹیکسٹ ہاور ہے اس کے لئے ریال ٹائم انپوٹس ہے اس کے لئے کونٹیکسٹ ہے اس کے لئے کونٹیکسٹ ہے اس کے لئے کونٹیکسٹ ہے اس کے لئے کونٹیکسٹ ہے اس کے لئے کونٹیکسٹ ہے اس کے لئے کونٹیکسٹ ہے اس کے لئے کونٹیکسٹ ہے اس کے لئے کونٹیکسٹ ہے اس کے لئے کونٹیکسٹ ہے اس کے لئے کونٹیکسٹ ہے اس کے لئے کونٹیکسٹ ہے اس کے لئے کونٹیکسٹ ہے اس کے لئے کونٹیکسٹ ہے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:38",
                    "end": "00:45:04",
                    "text": " انفرنس کی طرح جانتا ہے جو انفرنس کی طرح جانتا ہے اور اس کی طرح جانتا ہے جو انفرنس کی طرح جانتا ہے اور اس کی طرح جانتا ہے اور اس کی طرح جانتا ہے اور اس کی طرح جانتا ہے اور اس کی طرح جانتا ہے اور اس کی طرح جانتا ہے اور اس کی طرح جانتا ہے اور اس کی طرح جانتا ہے اور اس کی طرح جانتا ہے اور اس کی طرح جانتا ہے اور اس کی طرح جانتا ہے اور اس کی طرح جانتا ہے اور اس کی طرح جانتا ہے اور اس کی طرح جانتا ہے اور اس کی طرح جانتا ہے اور اس کی طرح جانتا ہے اور اس"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:04",
                    "end": "00:45:34",
                    "text": " لہذا یہ بہت اہمید ہے اور بہت اہمید ہے جس طریقے میں جانتے ہیں لہذا یہ بہت اہمید ہے جس طریقے میں جانتے ہیں اور ہم نے دیکھا دیکھا ہے کہ آخری بارے میں بہت سارے بارے میں جانتے ہیں جنہوں نے جانتے ہیں جنہوں نے جانتے ہیں جنہوں نے جانتے ہیں جنہوں نے جانتے ہیں جنہوں نے جانتے ہیں جنہوں نے جانتے ہیں جنہوں نے جانتے ہیں جنہوں نے جانتے ہیں جنہوں نے جانتے ہیں جنہوں نے جانتے ہیں جنہوں نے جانتے ہیں جنہوں نے جانتے ہیں جنہوں نے جانت"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:34",
                    "end": "00:46:01",
                    "text": " سنس افیشن ہے اسی طور پر اہم کرنے کے لئے سنس افیشن کے لئے سنس افیشن کے لئے سنس افیشن کے لئے سنس افیشن کے لئے سنس افیشن کے لئے سنس افیشن کے لئے سنس افیشن کے لئے سنس افیشن کے لئے سنس افیشن کے لئے سنس افیشن کے لئے سنس افیشن کے لئے سنس افیشن کے لئے سنس افیشن کے لئے سنس افیشن کے لئے سنس افیشن کے لئے سنس افیشن کے لئے سنس افیش"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:01",
                    "end": "00:46:28",
                    "text": " لہذا آپ کو معلومی اوائی کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضرورت کرنے کی ضر"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:28",
                    "end": "00:46:59",
                    "text": " آہ آہ آہ آہ آہ آہ آہ آہ آہ آہ آہ آہ آہ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:59",
                    "end": "00:47:28",
                    "text": " اور ہم نے پہلے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے بہت سے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:28",
                    "end": "00:48:00",
                    "text": " اپنی شیلف اے آئی ایک سویسائیڈ ہے جو ہم سب جانتے ہیں ایسی طور پر ایسی طور پر ایسی طور پر ایسی طور پر ایسی طور پر ایسی طور پر ایسی طور پر ایسی طور پر ایسی طور پر ایسی طور پر ایسی طور پر ایسی طور پر ایسی طور پر ایسی طور پر ایسی طور پر ایسی طور پر ایسی طور پر ایسی طور پر ایسی طور پر ایسی طور پر ایسی طور پر ایسی طور پر ایسی طور پر ایسی ط"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:00",
                    "end": "00:48:31",
                    "text": " اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے لئے اندیا کے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:31",
                    "end": "00:48:57",
                    "text": " اس کا مطلب ہے لہذا ملطفہ دیلیمرات ہیں جو واقعی اور ترقیقی دیلیمرات ہیں جو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو واقعی کو"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:59",
                    "end": "00:49:28",
                    "text": " ایک آئی اینجین کوئی کہ کچھ ایک رفال کے بارے میں کوئی بہترین کوئی بہترین کوئی بہترین کوئی بہترین کوئی بہترین کوئی بہترین کوئی بہترین کوئی بہترین کوئی بہترین کوئی بہترین کوئی بہترین کوئی بہترین کوئی بہترین کوئی بہترین کوئی بہترین کوئی بہترین کوئی بہترین کوئی بہترین کوئی بہترین کوئی بہترین کوئی بہترین کوئی بہترین کوئی بہترین کوئی بہترین کوئی بہت"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:29",
                    "end": "00:49:55",
                    "text": " ایک پارٹنر سے کہتے ہیں کہ وہ اپنے ساتھ آئیوں کو اپنے ساتھ آئیوں کو اپنے ساتھ آئیوں کو اپنے ساتھ آئیوں کو اپنے ساتھ آئیوں کو اپنے ساتھ آئیوں کو اپنے ساتھ آئیوں کو اپنے ساتھ آئیوں کو اپنے ساتھ آئیوں کو اپنے ساتھ آئیوں کو اپنے ساتھ آئیوں کو اپنے ساتھ آئیوں کو اپنے ساتھ آئیوں کو اپنے ساتھ آئیوں کو اپنے ساتھ آئیوں کو اپنے ساتھ آئیوں کو اپنے ساتھ آئیوں کو اپنے سات"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:55",
                    "end": "00:50:25",
                    "text": " ہمارے لئے آنسرڈ ہونا چاہتے ہیں جب ہم اپلکیشن آئی کو دیکھ رہے ہیں لیکن کچھ اہلی سپیکرز نے کہا کہ بورومیٹر کو بڑھنے کے لئے بڑھنے کے لئے بڑھنے کے لئے بڑھنے کے لئے بڑھنے کے لئے بڑھنے کے لئے بڑھنے کے لئے بڑھنے کے لئے بڑھنے کے لئے بڑھنے کے لئے بڑھنے کے لئے بڑھنے کے لئے بڑھنے کے لئے بڑھنے کے لئے بڑھنے کے لئے بڑھن"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:25",
                    "end": "00:50:55",
                    "text": " سوورنڈ دفنس آئی پلیٹ فارم ہے جس طور پر ہمارے بارے میں جانتے ہیں کہ ہمارے بارے میں مکروز ہے ہمارے بارے میں تکنالوجی ہے ہمارے بارے میں انفیمیشن ہے جو موجود ہے ہمارے بارے میں کیا کرتے ہیں جس طور پر ہمارے بارے میں کچھ ملائنوں کے لئے کیا کرنا چاہتے ہیں کہ ہمارے بارے میں کچھ ملائنوں کے لئے کیا کرنا چاہتے ہیں کہ ہمارے بارے میں کچھ ملائنوں کے لئے کیا کرنا چاہتے ہیں کہ ہمارے بارے میں کچھ ملائنوں کے لئے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:55",
                    "end": "00:51:22",
                    "text": " اس کے ساتھ ایسا جانتے ہیں جس کے ساتھ ایسا جانتے ہیں جس کے ساتھ ایسا جانتے ہیں جس کے ساتھ ایسا جانتے ہیں جس کے ساتھ ایسا جانتے ہیں جس کے ساتھ ایسا جانتے ہیں جس کے ساتھ ایسا جانتے ہیں جس کے ساتھ ایسا جانتے ہیں جس کے ساتھ ایسا جانتے ہیں جس کے ساتھ ایسا جانتے ہیں جس کے ساتھ ایسا جانتے ہیں"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:22",
                    "end": "00:51:52",
                    "text": " جہاں ہم اپنی ویڈیو پارٹنیشپ کے لئے بڑھ رہے ہیں؟ ہمارے اندیر میں بہت زیادہ ایناویشن ایکو سسٹم ہے جہاں ہم کیا کرتے ہیں؟ ہم کیسی طرح افنسیو آئی کابلٹیوں کو بڑھ رہے ہیں؟ جہاں ہم جانتے ہیں کہ ہم جانتے ہیں کہ ہم افنسیو آئی کابلٹیوں کو بڑھ رہے ہیں اور بہت اہمی طور پر ہم کیسی طرح افنسیو آئی کابلٹیوں کو بڑھ رہے ہیں؟ اور بہت اہمی طور پر ہم کیسی طرح"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:52",
                    "end": "00:52:19",
                    "text": " اس کے ساتھ ہمارے پیچھوں کے ساتھ ہمارے پیچھوں کے ساتھ ہمارے پیچھوں کے ساتھ ہمارے پیچھوں کے ساتھ ہمارے پیچھوں کے ساتھ ہمارے پیچھوں کے ساتھ ہمارے پیچھوں کے ساتھ ہمارے پیچھوں کے ساتھ ہمارے پیچھوں کے ساتھ ہمارے پیچھوں کے ساتھ ہمارے پیچھوں کے ساتھ ہمارے پیچھوں کے ساتھ ہمارے پیچھوں کے ساتھ ہمارے پیچھوں کے ساتھ ہمارے پیچھوں کے ساتھ ہمارے پیچھوں کے ساتھ ہمارے پیچھوں کے س"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:19",
                    "end": "00:52:39",
                    "text": " ہمارے بارے بارے پہلے ہمارے پہلے ہمارے پہلے ہمارے پہلے ہمارے پہلے ہمارے پہلے ہمارے پہلے ہمارے پہلے ہمارے پہلے ہمارے پہلے ہمارے پہلے ہمارے پہلے ہمارے پہلے ہمارے پہلے ہمارے پہلے ہمارے پہلے ہمارے پہلے ہمارے پہلے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:39",
                    "end": "00:53:05",
                    "text": " ہمارے پیشانوں اور ہمارے پیشانوں کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:05",
                    "end": "00:53:35",
                    "text": " ہمارے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ کاملہ کے ساتھ ک"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:35",
                    "end": "00:54:07",
                    "text": " انہوں نے ہمارے مطابقوں میں اپنے مطابقوں میں اپنے مطابقوں میں اپنے مطابقوں میں اپنے مطابقوں میں اپنے مطابقوں میں اپنے مطابقوں میں اپنے مطابقوں میں اپنے مطابقوں میں اپنے مطابقوں میں اپنے مطابقوں میں اپنے مطابقوں میں اپنے مطابقوں میں اپنے مطابقوں میں اپنے مطابقوں میں"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:07",
                    "end": "00:54:41",
                    "text": " اور اسی طرح میں ہمارے ساتھ یہ کہنا چاہتا ہوں کہ فیوچر نہیں ہے اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں اسی طرح میں"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:41",
                    "end": "00:55:21",
                    "text": " ہمارے دوسرے ساتھ چیزوں کو سمجھا رہا ہمارے دوسرے ساتھ چیزوں کو سمجھا رہا ہمارے دوسرے ساتھ چیزوں کو سمجھا رہا ہمارے دوسرے ساتھ چیزوں کو سمجھا رہا ہمارے دوسرے ساتھ چیزوں کو سمجھا رہا ہمارے دوسرے ساتھ چیزوں کو سمجھا رہا"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:21",
                    "end": "00:55:46",
                    "text": " اس کا مطلب ہے کہ یہ اندیہ کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے سامنے کے س"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:47",
                    "end": "00:56:21",
                    "text": " کیا شب نام آپ کا"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:21",
                    "end": "00:57:03",
                    "text": " بہت سوچا ہے لیکن اندرن آمی اور اندرن آمی مطابقات کو اپنے مطابقات کو اپنے مطابقات کو اپنے مطابقات کو اپنے مطابقات کو اپنے مطابقات کو اپنے مطابقات کو اپنے مطابقات کو اپنے مطابقات کو اپنے مطابقات کو اپنے مطابقات کو اپنے مطابقات کو اپنے مطابقات کو اپنے مطابقات کو اپنے مطابقات کو اپنے مطابقات کو اپنے مطابقات کو اپنے مطابقات کو اپنے مطابقات کو اپنے مطابقات کو اپ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:03",
                    "end": "00:57:42",
                    "text": " سو سو دیرزب دیر ہز بین اڈیکویٹیلی بین بیلٹن فائن سو وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وینہور وین"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:42",
                    "end": "00:58:11",
                    "text": " اگر آپ اپنے بارے میں شاملہ رکھتے ہیں تو اگر آپ اپنے بارے میں شاملہ رکھتے ہیں تو اگر آپ اپنے بارے میں شاملہ رکھتے ہیں تو اگر آپ اپنے بارے میں شاملہ رکھتے ہیں تو اگر آپ اپنے بارے میں شاملہ رکھتے ہیں تو اگر آپ اپنے بارے میں شاملہ رکھتے ہیں تو اگر آپ اپنے بارے میں شاملہ رکھتے ہیں تو اگر آپ اپنے بارے میں شاملہ رکھتے ہیں تو اگر آپ اپنے بارے میں شاملہ رکھتے ہیں تو اگر آپ اپنے بارے میں شاملہ ر"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:11",
                    "end": "00:58:41",
                    "text": " جنرال مہد کاندی صرف اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ اپنے ساتھ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:41",
                    "end": "00:59:16",
                    "text": " اگلی سیشن کیسے ہمارے پیچھے پیچھے پیچھے پیچھے پیچھے پیچھے پیچھے پیچھے پیچھے پیچھے پیچھے پیچھے پیچھے پیچھے پیچھے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:16",
                    "end": "00:59:43",
                    "text": " جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے جیسے"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:43",
                    "end": "01:00:12",
                    "text": " آپ کے سوال جانتے ہیں جب ہمارے ایک ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ ملتفہ م"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:12",
                    "end": "01:00:40",
                    "text": " اور جیسے دیس ایسے مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مطمئنے میں مط"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:40",
                    "end": "01:02:26",
                    "text": " اور سرٹیوکیٹز پیس سر میں سوال کرتا ہوں کہ سلوکیوں کو اپنے ناموں کے بارے میں اپنے ناموں کے بارے میں اپنے ناموں کے بارے میں اپنے ناموں کے بارے میں اپنے ناموں کے بارے میں اپنے ناموں کے بارے میں اپنے ناموں کے بارے میں اپنے ناموں کے بارے میں اپنے ناموں کے بارے میں اپنے ناموں کے"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:02:26",
                    "end": "01:03:26",
                    "text": " آنانتا سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہنم سہن"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:03:26",
                    "end": "01:03:29",
                    "text": " ہاں!"
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
    {
    "_id": "699577d3565455da1fc7528a",
    "projectId": "331985",
    "projectName": "AI for Defence Insights on Strategy and National Security- English",
    "projectDescription": "AI for Defence Insights on Strategy and National Security",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T08:26:58.343Z",
    "projectUpdatedDate": "2026-02-18T08:26:58.343Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T08:30:05.292Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "AI for Defence_ Insights on Strategy and National Security(MP3_128K).mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771403187592.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:30",
                    "text": " this event and most of the sessions have had very well turned out and while there's a lot of excitement around AI in India and in almost every dimension and in every angle including the consumers and all but one of the very important topics is where the artificial intelligence operates or enhances"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:30",
                    "end": "00:01:00",
                    "text": " or impacts the defense ecosystem and as much as we would have wanted to deliberate on it in the 55 minutes that we have we would try and cover most of the grounds with the questions to the panelists but before that the address by the respected deputy chief ISNT."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:00",
                    "end": "00:01:24",
                    "text": " you know we are at a point of time where in the country we are trying to look at infusion of technology on so many areas and even beyond what we are doing in the country we are also looking at all the global linkages and the partnerships yesterday the French president Emmanuel Macron was in Mumbai and if you saw the 21 outcome statements"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:24",
                    "end": "00:01:47",
                    "text": " five of them were related to defense and then of course there is one very specifically on the critical and emerging technologies and in the defense ecosystem. Same if you look at the conversations with US that has been happening AI remains at the forefront of all the cutting-edge technologies and the areas where we want to cooperate."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:47",
                    "end": "00:02:07",
                    "text": " and nevertheless with all these efforts and what we are trying to do here we are also trying to see that how responsively we use AI in all our activities and that is very very important when you look even at the defense ecosystem so AI being a force multiplier AI trying to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:07",
                    "end": "00:02:36",
                    "text": " you know improve and maybe modernize the battlefield tactics, look at the doctrines, look at how do you manage the supplies and of course enabling the equipments to be performing much more optimally or in newer horizons. So the changing battlefield that we all see will have a significant impact with the usage of artificial intelligence and you know the emerging technologies. So this is where"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:36",
                    "end": "00:03:02",
                    "text": " I think, you know, we have a fantastic panel. We are very lucky that this panel is confined to experts and the number is less. In many sessions, I've had a huge number of experts who actually cannot do justice in the time of 55 minutes, that's that. With all of that, and I would like to invite the respected Deputy Chief of the Army Staff Information Systems Technology, General..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:02",
                    "end": "00:03:41",
                    "text": " Thank you so much and good to be here in this panel in the keynote discussion like I was saying just previously that they are the experts and I am the generalist that is why I am called a general and you know one more thing for senior officer like us addressing gathering like this is very daunting"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:41",
                    "end": "00:04:07",
                    "text": " because normally we are addressing gathering which are uniform people and by my service they are generally all junior to me so they have no option but to firstly clap at my punch line laugh at my jokes and so on here that is not a necessity so we will have a rule when I give a punch line I will raise my left hand so all of you clap when I give a joke I will raise my right hand then all of you laugh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:07",
                    "end": "00:04:36",
                    "text": " Because you know that's the only way that I'll be able to interact anyway. That's on lighter weight Let me start distinguished guests the lead leaders innovators from the AI community my colleagues in uniform Some of whom are also innovators in the AI field students ladies and gentlemen a very good morning to you all It is a privilege and an honor to be addressing this important forum in this path-breaking India AI summit impact AI summit"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:36",
                    "end": "00:05:06",
                    "text": " and they are coverage professionals who are going to be speaking about AI and its impact on warfare and conflict So allow me to focus on a different aspect which concerns us all that is military leadership and the responsibility of ethics regulations in AI and what is that military leaders need to look at when they are applying AI enabled systems in conflict Let me begin with a brief operational moment"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:07",
                    "end": "00:05:38",
                    "text": " During a recent high tempo military operation a senior commander was presented with a machine generated recommendation Based on multiple sensor threats and AI analysis to engage a target immediately The system was confident the probability score was high the decision window was measured in seconds But the commander paused Something was amiss So he asked the question What does the machine not know?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:40",
                    "end": "00:06:13",
                    "text": " The pause revealed something the area that the sensors were looking at had the adversary troops But there was also a civilian evacuation underway, which had just started which had not yet been fed into the data It is possible that there were some troops also of the adversary with the civil evacuation however, the commander excised judgment and restraint stop the strike and prevented loss of innocent lives the mission later on was completed and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:13",
                    "end": "00:06:45",
                    "text": " and also one but at that point we excise judgment this brings us to a fundamental truth AI can inform decisions but only humans can excise judgment and bear responsibility for them AI is today having an unprecedented speed, scale and efficiency those attributes are real and are very well understood in this room what deserves deeper reflection is something else the new burdens AI places on leadership"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:45",
                    "end": "00:07:17",
                    "text": " command responsibility and strategic stability. I use the word burden with care. Autonomous and AI enabled systems advance at extraordinary speed. Therefore four questions deserve our attention. Firstly, which decisions must remain with the commanders and which decisions can be delegated to data or algorithms. No matter how accurate or efficient they become. When must human judgment"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:17",
                    "end": "00:07:50",
                    "text": " moral responsibility and command authority remain paramount Secondly have our rules of engagement and safety frameworks kept pace with systems that rely on data Do we fully understand how these systems are trained the quality of their data and the assumptions that they make It is also called the black box phenomena Today even the creator of an AI does not know what is happening inside the black box So you can let alone how can commanders trust it?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:50",
                    "end": "00:08:21",
                    "text": " Thirdly are we subjecting AI enabled systems to the same rigor that we subject new weapon systems to trial evaluating them fielding them in contested conditions and then applying them After all an AI enabled system is not a software it's a weapon Thirdly or fourthly are we preparing our commanders and staff to lead in a world where machines can recommend predict and act faster than humans"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:21",
                    "end": "00:08:50",
                    "text": " Without allowing speed to substitute judgment command and control By commanders will exercise life and death decisions can then be given to machines if a machine recommends a strike with 90% accuracy and The commander takes a decision and it's the wrong decision it gives you a moral buffer To say that it was a machine that recommended it, but that is that correct. These are questions that we need to answer"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:51",
                    "end": "00:09:21",
                    "text": " It is also important to understand that these concerns are not in the military India has released the India AI governance guidelines which is a path breaking document on a pragmatic approach to AI governance duly supported by a techno legal framework The framework identifies that since AIs are probabilistic, generative and adaptive they can lead to unintended consequences These are lessons for military planners too The second thing that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:21",
                    "end": "00:09:51",
                    "text": " the framework talks about are risks and within that the sixth risk is national security one of them is which identifies that AI facilitated disinformation campaigns cyber attacks on critical infrastructure and the use of lethal autonomous weapon systems as looming threats to public safety policy makers and defense planners need to be cognizant of these threats and also make people aware of them so that we can be resilient to them to remain relevant I must say"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:51",
                    "end": "00:10:22",
                    "text": " Military leaders must stay on the learning curve Understand technology both its challenges pitfalls and uses while also ensuring that technology is absorbed and adopted by our units and troops That is our responsibility Having said that let me be very clear that we as the Indian Armed Forces and the Indian Army are fully cognizant of the transformative power of AI to increase our operational efficiency and We are making every effort the direction to ensure"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:22",
                    "end": "00:10:53",
                    "text": " that AI is incorporated into our decision support systems, into our surveillance, Reiki and all the other functions they do. We are actively working with industry leaders, some of them over here, startups and academic institutions to harness AI for military applications, drawing strength from India's vibrant innovation ecosystem and our own growing band of uniformed innovators. In fact, to give push to this direction, the Indian army has declared this year as the year of data centricity and networking."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:53",
                    "end": "00:11:24",
                    "text": " by which we are acknowledging that data is the new oil and data centric warfare is what we need to fight. In fact we have set up a booth in the expo as well and I will urge all of you to go and encourage and see it and encourage our young innovators there. For India therefore the question is never as to whether we should adopt AI enabled systems. But how? And we are clear that this transition must be undertaken responsibly. So what does responsible and effective approach look like?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:25",
                    "end": "00:11:57",
                    "text": " Firstly we must institutionalize human control not as a slogan but as law This requires clearly defining which functions may be assisted by AI which may be recommended by AI and which may must always remain human decisions Secondly like I said earlier AI enabled systems need to be treated as a weapon system and be tested accordingly Remember that the most chaotic data environment is the battlefield"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:57",
                    "end": "00:12:27",
                    "text": " An AI trained on very clear satellite images given to it in a computer lab will fail when it sees grainy mud soaked smoke deception like images in a battlefield and can produce a wrong decision and therefore there needs to be the same certification same red teaming and same tile evaluation of AI enabled systems Thirdly is about sovereignty and trust How does a commander"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:27",
                    "end": "00:12:57",
                    "text": " Trust the data that is being fed into the AI enabled system to give him the decision support There needs to be more clarity on that there are systems in place which are doing it and commanders themselves must learn about it and Lastly leadership development must involve must evolve so that Commanders who are now going to be interrogating systems going to be you know, not only leading troops was also going to be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:57",
                    "end": "00:13:33",
                    "text": " Would say interacting with systems and taking decision on them should know as to what are the pitfalls? What are the challenge and how best to use them? Moral responsibility remains with the commander, but ultimately I would like to say that we must Remember a historical truth. I personally believe in the wisdom humanity Whenever faced with a new crisis we have responded well the rules governing the use of NBC weapons The rules the framework that has come into place for governing landmines"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:33",
                    "end": "00:14:02",
                    "text": " The Geneva Convention itself which talks about conduct with prisoners of war have all stood the test of time even in conflict. So I am confident that AI as the new looming technology on the forefront will also find ways and means to be regulated. In fact as we speak the United Nations Convention on certain conventional weapons is already on. There are many governments, many states which are part of the discussion."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:02",
                    "end": "00:14:30",
                    "text": " Consensus is complex but it will come India today stands at the cusp of three powerful realities We are a major military power We are a rapidly growing AI country or AI ecosystem and we are a civilization that has long understood that power must be governed by restraint Shakti must go hand-in-hand with Dharma or righteousness This particular thing or particular"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:30",
                    "end": "00:14:59",
                    "text": " You know ethos gives us both capability and credibility to lead the world in using AI responsibly in conflict as a line from popular culture Which says that with great power comes great responsibility? I think the time has come how wisely we carry the responsibility will shape not only future was but the moral leadership we choose to project The character of war may change our conscience cannot with that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:59",
                    "end": "00:15:33",
                    "text": " I would end here. I look forward to the insights by the accomplished panelists on subjects related more closely to AI and its use in conflict. Thank you and Jai Hind. Thank you so much for bringing out many perspectives, but the most important point that you make is that the humans must be on top. AI cannot dictate humans and that moral responsibility of the leadership and how do you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:33",
                    "end": "00:16:01",
                    "text": " cultivate the new leadership towards that and how the Geneva Convention has stood the test of time. I think these are very, very significant and this gives all of us a guidance about the ethos and the direction with which you are taking the AI enablement in the Indian Army forward. Now let me go to the panel so that we get more ideas and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:01",
                    "end": "00:16:32",
                    "text": " what they are doing in their areas because I am left with I think about 40 minutes. So we'll try and apportion seven minutes to each speaker, 35 minutes and then five minutes we'll keep for questions and if maybe the organizers allow maybe we'll go a little bit more. But let me start with you General Chipper."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:32",
                    "end": "00:17:01",
                    "text": " The AI disruption in military operations have already gained the grounds in terms of conversations that happen at strategic level as well as now even in media circles also. Basically, how the integration of AI into military operations is aligned to the changing character of warfare, including ISR."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:01",
                    "end": "00:17:25",
                    "text": " cybersecurity information operations and decision support systems. So my first question to you would be how can military forces effectively integrate AI enabled ISR and decision support systems while maintaining human judgment in the OODA loop, particularly in time compressed operational environments where the speed advantage of AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:25",
                    "end": "00:17:51",
                    "text": " may conflict with the principles of proportionality, distinction and command responsibility. In fact that question flows very naturally from what the deputy chief just mentioned. So I would like you to respond to that sir."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:52",
                    "end": "00:18:23",
                    "text": " The requirement actually is to make better decisions and not bad decisions faster. In this context let me highlight where AI can help us make better decisions. AI can in fact help us to overcome what Herbert Simon calls it bounded rationality. That human beings tend to make a satisfying decision rather than an optimum or a best decision and that is because of four factors. The first factor being the information is never complete."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:23",
                    "end": "00:18:47",
                    "text": " and humans have to make a decision second is there is a cognitive limitation if say multiple factors have to be considered multi criteria human mind can easily decide between two factor analysis if there are 20 factors which have to be taken into consideration for a decision then it is challenging then you tend to choose only one or two factors and make your decision that is the second limitation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:47",
                    "end": "00:19:03",
                    "text": " Third is the time is always short. You have to take a decision in a very quick time frame. Therefore time to analyze all factors is not available. Fourth factor of bounded rationality is the human attention span. It is very difficult to keep attention for a very long time."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:03",
                    "end": "00:19:32",
                    "text": " All these limitations artificial intelligence can very easily overcome. So therefore the decision making in all the domains what you talked about can be superior to what a human would make without AI enablement. So that is the advantage of artificial intelligence. There are also some limitations and let me highlight some limitations were highlighted by Deputy Chief in his address. There are certain limitations. When we are talking about artificial intelligence today we are primarily"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:32",
                    "end": "00:20:01",
                    "text": " talking about GPT large language models. We are talking about generative pre-trained transformer. So as the name suggests it is pre-trained. There is a training cycle and there is an inference cycle. There is no real time learning which is happening. If additional learning has to take place it has to go back to the training cycle. Whereas in a battlefield you will have to learn as you go. There will be many factors which will emerge where you have to learn as you go."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:01",
                    "end": "00:20:26",
                    "text": " the transformer technology itself is actually a probabilistic model of finding the next token the next word or part of word it is very good at that but it fails wherever there are computational kind of algorithm to be run so therefore to rely on an LLM for such factor and other limitation AI also starts to fail the moment context changes"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:26",
                    "end": "00:20:55",
                    "text": " If it has been trained in one context, the moment you change the context, the result will be erroneous. That is another limitation. And the major limitation of an AI model is that it lacks abductive reasoning. Reasoning when the facts are not complete. Reasoning when the new situation has come up. That kind of decision making if it is left to the artificial intelligence will be disastrous. So these are the limitations. I would highlight another issue, Vimal."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:55",
                    "end": "00:21:16",
                    "text": " we normally when we discuss this you know challenge our discussion is either in the technology domain the reasons which I or the argument which I highlighted or it goes into ethical domain what you have said to personality or distinction these two are fine I think the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:16",
                    "end": "00:21:43",
                    "text": " Discussion should also be at a deeper level at a foundational level at a psychological or philosophical level how should this technology be dealt with specially in the warfare. In that context let me highlight and I am sure those other two that technology and ethical part is important and there are people on the panel who are better experts than me. Let me deal this foundational aspect. Let me quote two examples."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:43",
                    "end": "00:22:12",
                    "text": " We recently had and it has been reported in the media that IDF used Lavender database to identify the low level operatives in Hamas and then they eliminated them and the authorization was done that if the AI engine was more than 90% accurate you can mark that person as a target and then eliminate him. 37,000 people were identified by this AI engine and eliminated as per reported in open source."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:13",
                    "end": "00:22:40",
                    "text": " So if it is a 10% accepted error there were 3,700 people who were actually innocent as per the logic itself and if there was a collateral damage that can be put on one side. So can we rely on that? Let me now relate it to another incident and most of you have learnt it during Vietnam War the incident of My Lai when a whole village was massacred in revenge."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:40",
                    "end": "00:23:07",
                    "text": " ladies, children, old people because able bodied men would have run away it was massacred. There was no AI engine there. So who was guiding that kind of justification or decision making? So which is better? I think we can address it in this manner. In Mailai incident there was a human agency therefore attribution. You know William Callay could be fixed."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:07",
                    "end": "00:23:26",
                    "text": " you could put checks and balances that a new military decision is done by any person in future you could put deterrence in a human mind because there was attributability whereas in case of later or the earlier example where we are using an AI engine where is the attributability that 10% innocent people"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:26",
                    "end": "00:23:51",
                    "text": " Who can you attribute those data to? Is it attributable to the algorithm? Is it the person who made the database? Is it the person who wrote the code? That attributability part is missing. So if we have to deal with this technology into military domain I think we need to address this foundational question and that brings me to the last part of your query on the command responsibility."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:51",
                    "end": "00:24:21",
                    "text": " Who's responsible? Is it the coder? Is it the person who authorized? Is it the decision maker who laid down that 10% error is acceptable? Who's responsible? So as far as military leader is concerned, I think command responsibility can be defined in two ways. One is it involves a commander need to have competence that is how to do, what to do and second is the human skill. Why to do it? These two functions of command make a commander a good or a bad commander."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:21",
                    "end": "00:24:49",
                    "text": " and command responsibility in military parlance is absolute. It cannot be delegated. You cannot do your cognitive offloading to a machine. It's not expected in the present circumstances to another human being and in the future warfare maybe even to a machine. So one is that. Second, should we authorize a lethal kind of decision or causing violence?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:49",
                    "end": "00:25:18",
                    "text": " or divorce it from human agency I don't think that would be a wise decision AI is a tool we need to treat it as a tool we need to use it effectively any decision which affects application of lethality will remain with the human agency so if I have to conclude I would say that the entire proportionality distinction and command responsibility in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:18",
                    "end": "00:25:45",
                    "text": " ethos of the Indian Armed Forces would remain with the human agency. The accountability, the responsibility cannot be delegated to even another human, let alone a machine. Thank you very much."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:45",
                    "end": "00:26:13",
                    "text": " building a lot of solutions and getting the you know forces also involved in the AI age. Now you have been of course tracking the global landscape around how the whole AI ecosystem is evolving and particularly if the AI integration in defense. So you know how would you see and what are what have the major military powers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:13",
                    "end": "00:26:34",
                    "text": " Have been implementing AI in the command and control systems and intelligence gathering operations over the last over the next past three five years and what specific capabilities have been demonstrated in real-world conflicts or exercises you see the information not always is accurate and you know"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:34",
                    "end": "00:27:01",
                    "text": " look at our adversaries and some people what they do now they spring surprises very regularly when it comes to the AI space but even if you look with the friendly countries there is different level there is a skewed approach around this whole understanding and all. So what are the competitive advantages and limitations that you have observed in different national approaches to AI centric warfare and how do you think you know we are also aligning ourselves. Thank you for that question."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:02",
                    "end": "00:27:31",
                    "text": " I think no command center will trust AI as a black box. That's key. Everyone here is on the same page there. What AI can be used for is situation summarization, pattern detection at scale. What machines are good at, what AI is good at is recall. We are very good at precision. Domain experts are good at precision."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:31",
                    "end": "00:28:00",
                    "text": " also the capability to correlate across multiple sensors again AI brings that big advantage it's a force multiplier there and finally recommending course of action right or taking you from what Dr. Rajeev Bell the DGICMR always tells me he says that it's basically you are presenting scenarios here's a one scenario I mean and that also you know reminds us of war games but basically"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:00",
                    "end": "00:28:30",
                    "text": " multiple scenarios and what are recommendations so alternatives I think what AI can give you is alternatives technology alternatives procedure alternatives and present with each alternative the provenance now I will use some more technical keywords here provenance has originated with databases if you write a database query you have the execution plan and you know why a particular result came up"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:30",
                    "end": "00:28:59",
                    "text": " So that providence is very key. When I was at IBM Research before joining IIT Bombay, we were actually quite frustrated with AI over-promising and under-delivering. And we were so inspired by databases, we said, let's extend a SQL language, a structured query language, to annotated query language. We actually created an annotated query language which uses a bit of AI. And a product was launched when I was at IBM Research. And it had its own"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:59",
                    "end": "00:29:26",
                    "text": " algebra just like SQL has its own algebra very interesting exercise why because we can get provenance so that's one provenance to sources when we met the honorable PM on 8th Jan long conversation he said we also need attribution so there is provenance in terms of sources but there is attribution in terms of experts again if we can get to the attribution by lying alternatives that will be great"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:26",
                    "end": "00:29:56",
                    "text": " and third is the most technical part which is observability stack when processes get executed and today when training happens at Bharat Gen we have our deputy director Prasad Tre and we have a section at company at IIT Bombay called Bharat Gen where we are building models on thousands of GPUs training models even inference you need to ultimately delegate to a cluster of GPUs observability becomes very important"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:56",
                    "end": "00:30:26",
                    "text": " what is being delegated to CPUs, GPUs versus the network switch because network switch is also kind of used as a virtual memory today that's how you are able to run computation across multiple GPUs Observability is there a data leakage so having recipes, having these models as glass boxes with observability will ensure that you are running in very secure environments in air gap situations without leaving any footprint at all so I think this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:26",
                    "end": "00:30:55",
                    "text": " The question you also brought up was specific examples of what other countries are doing From an academic perspective Stanford University released this very interesting open source platform called snorkel Sometimes programs are more reliable than AI models What snorkel does? It lets humans design programs Complex programs Sometimes you annotate data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:55",
                    "end": "00:31:26",
                    "text": " and you have annotative guidelines what they say is translate all those annotative guidelines into programs these programs co-exist with your model can co-exist with your model now what Stanford snorkel did was they actually use these programs to annotate data learn parameters of these programs so these programs are parameterized then these programs are thrown away what we realize is India India we like to always compliment our UPI complimented our savings economy"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:26",
                    "end": "00:31:55",
                    "text": " Can we complement what exists? So we tried to differ from them. We said programs are important but programs cannot be used and through. Programs are not just to annotate data. These data programs which you mentioned should have the ability to override the ML model output. So we actually created an open source platform called Sphere Semi Supervised Data Programming. My PhD student Janvi is also here in the audience. She worked very closely with Sama with DGIS. The idea there is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:55",
                    "end": "00:32:24",
                    "text": " our program should be able to override the models because as you pointed out ultimately trust is a problem and it took quite some time for the academic community to appreciate because anything that comes from India is not necessarily viewed as state of the art but finally the community accepted that this is a very good paradigm I think we will also have to innovate technologically to make sure that the human the domain experts are the command center"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:24",
                    "end": "00:32:53",
                    "text": " Is it the driver's wheel and not AI? I am really enjoying this conversation. Thank you for bringing this up. Lovely. I think you hit the nail when you said attribution because that's possibly the biggest issue with cyber security for decades now and we still don't have any solution in the horizon. Let me turn to General Anand at this stage. You know and look at the again the governance regulation and the policy enablement for responsible AI. I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:53",
                    "end": "00:33:14",
                    "text": " Deputy Chief's speech was very much clear and both the panelists have also spoken around that. You know that that responsibility is very very crucial. You know maybe in a civilian domain we really don't bother much about responsibility and etc. But here it has to be absolutely the command."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:14",
                    "end": "00:33:43",
                    "text": " remains with an individual and it's not delegated. So what are the regulatory approaches that defense organizations adopt to enable rapid AI deployment and development for maintaining strategic advantage while ensuring adequate testing, validation and human oversight mechanisms are in place? And how can policy framework manage the tensions between operational urgency in military context and the precautionary principles"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:43",
                    "end": "00:34:13",
                    "text": " Required for high stakes AI systems. I'm sure sir, you know, there are a lot of precedents that we are Sort of building up. So where do you see this whole landscape? Yeah, it's you know, it's it's a pretty complex situation And it's probably as complex as high how new technologies come in and then need to be regulated So we've been facing it earlier on also. It's not that we haven't faced such situations there was a very"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:13",
                    "end": "00:34:43",
                    "text": " potent technology which came in in terms of nuclear technology and people were wondering how to control it and then you did read certain conclusions as was mentioned by General Shingle as well but the only change I see here is that you know that was a technology which was very few people had and AI is going to be much more democratic so it will be approachable by a lot of people the second thing is the pace at which it is changing nuclear technology didn't change as fast so we have to keep up with it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:43",
                    "end": "00:35:13",
                    "text": " So the policies that we bring into this will have to take both these factors into consideration. Mainly I would say that AI with its civilian users as compared to what we have in the military is characterized by two or three differences. So in our case, it's as complex perhaps as AI anywhere. But the complexity then comes in terms of the fielding of or the deployment of that technology. The second is the urgency. There are constraints of time."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:13",
                    "end": "00:35:40",
                    "text": " And the third is the risk. The risk is very high. It is lethal. It costs lives. It's not just a system going off course. So with these three constraints, we still have to function in the fog of war. So while there is noise and there is physicality and there is dust, there is also communication which has to take place with the machine which can be interdicted and probably even jammed completely."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:40",
                    "end": "00:36:09",
                    "text": " So at one point it will have to go autonomous or at one point completely human controlled by human physical means. Right? So again it gets very complex over there. But yet we have to use it in an environment where we can be sure that it is delivering results. So it has to be absolutely reliable. And that is where it becomes very different from civilian applications. It has to be absolutely reliable."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:09",
                    "end": "00:36:39",
                    "text": " if it has to kill it has to kill and the second thing is we have to make sure that with all this we still keep in mind responsibility and international human law and international human laws will have to be adhered to otherwise humanity is lost so somewhere in between this when you use a probabilistic technology but you are getting very deterministic outcomes there is a fundamental kind of a difference which we are bringing in here"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:39",
                    "end": "00:37:07",
                    "text": " And that's what I wanted to highlight in the first place and the complexity comes in here. Now how are we going to do this? I think when it comes to governance of AI generally we may still accept that governance lags behind how AI begins to advance faster and faster. But in the military perhaps we can't do that. We don't have that luxury. So therefore AI responsibility will have to be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:07",
                    "end": "00:37:35",
                    "text": " Inserted at every stage every stage of the life cycle of the development of the AI system itself in the military and Therefore right from the ideation stage to the design stage to the development stage to the deployment stage But not only that I would say even as you spirally develop the systems you have to keep putting the trust factor and the rename the responsible AI factor into it and then finally"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:35",
                    "end": "00:38:02",
                    "text": " you even have to ensure that you destroy it at the right time because before it mutates and goes amok you will have to make sure that it comes out to control if you've lost use of it you have to destroy it and its completion so at every stage of the life cycle and that these later parts I find are completely missing from our debates so as you go into agentic AI I think that the later part becomes more and more important now"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:04",
                    "end": "00:38:32",
                    "text": " As we go from the lab into the simulation stage and into the deployment stage What is it that I would say we should have in our phases? How do we identify each phase itself is a question mark? They are not very well defined So you may try to define each phase, but even within that there are various Spaces which you need to ensure that you have inserted responsibility into it. The second is bias mitigation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:32",
                    "end": "00:39:03",
                    "text": " Biases are bound to creep in How do you keep detecting those biases and that requires a very high amount of alertness So while we develop a system the responsible aspect also we need to be alert to all along and the third aspect is safety when it comes to critical use especially when you're talking about lethal autonomous weapon systems Because that is where you're bringing in You know the drift which needs to be checked and kept controlled"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:03",
                    "end": "00:39:32",
                    "text": " Drift detection is very important, but I also feel that you have to create certain bounded envelopes under which you use your weapon systems. So those bounded operational envelopes will need to be clearly defined as to where those weapon systems will be used and that once defined by us will have to be adhered to by our adversaries and therefore we need to bring in these policy decisions. Of course we have to have national policies and then we'll have to start looking at the international policies."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:32",
                    "end": "00:39:59",
                    "text": " Work is happening at both ends. We need to plug them together, but there are differences in capabilities between the US and China and us, and maybe Europe. And how will we balance that out? That's the kind of discussion that we did in the RE-AIM Summit where I just came back from, that's Spain. And I had suggested that our confidence building measures amongst ourselves would have to be not only tiered,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:59",
                    "end": "00:40:28",
                    "text": " Between various between the advanced paws and AI and the middle paws, but you'll also have to be tailored according to the requirement of each So and these will have to be acceptable to everyone. So I'll pause here and then Go on if you get time to get thank you, sir I sometimes rise to see what's going there so that I'm still able to come for a second question Vikram Jairam and founder Neuralix you are already"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:28",
                    "end": "00:40:57",
                    "text": " working with the forces and particularly you have a wider industry perspective at this stage, how things are happening in this country. So what mechanisms and institutional frameworks you feel are required to foster effective collaboration between the defense professional, academic researchers, technology startups, established industry players to create a vibrant AI ecosystem in the defense applications here in the country? Thank you, First of all."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:58",
                    "end": "00:41:27",
                    "text": " I think everybody is able to hear me OK. First of all, thanks for the distinguished panels for inviting us to be here. This is one of a kind. And we're also at the inflection point of how AI is getting adopted in India. And it's a great time to be present as part of this engagement here. What I wanted to first talk about is how"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:27",
                    "end": "00:42:03",
                    "text": " we are able to essentially bring these clusters together and when I say cluster it is the combination of deep tech startups the nodal agencies, the military coming together the system integrator something that we don't pay a lot of attention to and finally the domain experts"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:04",
                    "end": "00:42:33",
                    "text": " So it is one of those things that we have to ensure that such clusters come together and they are moving forward in terms of developing systems that are providing what is called as an ROI. Now ROI for defense is not profitability. ROI for defense is security. ROI for defense is getting intelligence at its faster pace."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:33",
                    "end": "00:43:02",
                    "text": " ROI for defenses, building effective decision support systems. So if that has to happen, there are certain plumbing system that has to come in place. And these plumbing systems are nothing but AI models being deployed, being served. There is constant incoming of the data that we often build these sovereign models on and other things."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:02",
                    "end": "00:43:32",
                    "text": " Now, what is really important here is how is it that something like this comes apart. And I want to take an example of my past life as a postdoctoral researcher at MD Anderson Cancer Center. Again, very similar things. It's a question of life and death when it comes to cancer as concerned. How is it that we bring together the domain experts? There is certain policies set by"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:32",
                    "end": "00:44:02",
                    "text": " the government, especially in this case, on the health side of things, whenever we build new instrumentation, how is it that they go through what is called as basic research? Basic research goes into clinical research. Clinical research then goes applied on what is called as imaging technologies, and you try them out on small animal imaging. You take FDA approval to try it on trials on humans."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:02",
                    "end": "00:44:31",
                    "text": " And then finally, it becomes a technology that moves forward. I think defense is also very similar to the medical industry, where lives matter. And when lives matter, such decisions have to be taken in this clustering approach that the other industries have taken. So there's a lot of things that is drawn from all of these different avenues that are all there. Secondly, there has to be a form wherein the founders"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:31",
                    "end": "00:44:58",
                    "text": " are now embedded within the army in terms of how we are able to understand the domain. I was privileged enough to be born to the olive green so I have seen this from the very beginning but more importantly how is it that the teams are able to understand the various nuances, the limitations also of putting such systems together."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:58",
                    "end": "00:45:21",
                    "text": " Proceeding with what is called an iterative approach is you build something fast enough, you test it out, you see the outcomes of it, then go back and build again. You cannot come up with something which is the greatest form of it at the very first stage itself. You've got to go back and keep iterating on it to do that. The finest models that you see today"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:21",
                    "end": "00:45:51",
                    "text": " In the world whether it is Gemini whether it is anthropic whether it is chat GPT or any of this have gone through these these these phases from where we were where they are today at this point and Then more importantly it is the embed of the other side also It is the the the members of the armed forces should also embed themselves into the startup foundation how do they think about it because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:51",
                    "end": "00:46:11",
                    "text": " As a founder some of us have left 8 figure jobs In US we have left what is called as a tenure position I was a university professor at University of Oklahoma Left tenure jobs and we have come here and we are doing this kind of stuff The only way to be able to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:11",
                    "end": "00:46:38",
                    "text": " have that passion in your bone to doing innovation is when the other side also understands and have empathy for what it takes to go build technologies. And if that happens in a very bilinear and fashion, you'll see the great outcomes coming out of it. I think this is something has also been adopted very recently, although there is with DARPA in the US and how they're doing it at the same time."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:38",
                    "end": "00:47:07",
                    "text": " So this is something that is of a great monumental value of people understanding each other and how we go ahead and perfectly be able to apply technologies going further. The other area that I wanted to also mention is when I come back to the plumbing system is making sure that they are effective. We test them out very quickly. I'm grateful to the IDEX program."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:07",
                    "end": "00:47:35",
                    "text": " 15, 20 years ago, something like this never existed wherein startups could come and innovate so fast. The IDEX program has given us the opportunity to be able to come and very quickly produce something and be able to apply it to the various agencies and see what the outcomes could come out. The earlier we see the outcomes, the earlier the impact of all these technologies are happening. So in short,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:35",
                    "end": "00:48:01",
                    "text": " What I wanted to kind of conclude again is the plumbing system has to be together, making sure that we have an iterative process for testing. And GenAI, if you look at it, is only a very small portion of the entire AI umbrella. About 15 years back, 97, when I actually came out, when I took my first grad course in UT, in AI, people were laughed at it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:01",
                    "end": "00:48:31",
                    "text": " that you're taking neural networks, it's a theoretical subject. Nobody is even going to ask anything about it. You're not going to find a job. You're going to be jobless. And if you look at it, fast forward so many number of years now, there's not a sentence that does not start with it. But as a matter of fact, what is really important is there is a whole area of AI and data science that also needs attention at the same time. There's so many other areas."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:31",
                    "end": "00:49:00",
                    "text": " wherein especially some of my own employees they are PhDs from UCLA and they're data scientists they wouldn't know D of deep learning today because they're very good at certain things of data science and they want to make sure that the deterministic algorithms are also part of this understanding. How do we use deterministic algorithms? How do we utilize game theory concepts? How do we make sure that we are learning from data at the same time?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:00",
                    "end": "00:49:30",
                    "text": " and all of these aspects have to come together to understand it. It's not being AI cannot be at the fashion of what comes in and what goes out. I think that is one thing that we have to realize that the moment something else becomes more important, we can't be on top of that. We have to make sure that we build discipline based upon what are the outcomes of it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:30",
                    "end": "00:49:44",
                    "text": " What Madhumita is to speak about because she is already into this whole seeing the military side as well as the Industry and particularly one of the areas supply chain, you know, that's again area where a lot of transformation happens but"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:44",
                    "end": "00:50:13",
                    "text": " I just wanted to mention that almost the same period of time I was a student of mathematics pursuing my post-graduation and when I had artificial intelligence and special paper people laughed at me and today it's changed so much you know and the moment you looked at Python programming and looked at everything of statistics through that lens everything has changed."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:13",
                    "end": "00:50:41",
                    "text": " mentioned about supply chain and the defense consulting but maybe you know what I can tell a little bit that how are the defense organizations adapting as far as the supply chain management is concerned, deploying AI particularly from that purpose so that you know the demand forecasting, inventory management, route optimization and real-time etc. I would want you to speak on that. That was a large question. Thank you so much and firstly thank you everyone for having me here."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:41",
                    "end": "00:50:60",
                    "text": " Indeed my privilege to be a part of such an august panel and such a lovely audience. So to answer your question I think globally defense supply chain is now moving from what it used to be like forecasting, stocking and then reacting to actually sensing predicting"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:60",
                    "end": "00:51:29",
                    "text": " and then adapting. So we have demand forecasting embedded in AI embedded across demand forecasting, inventory routing, and all sorts of, you know, through the entire supply chain, not as a pilot, but as a mission critical systems tied to readiness. Why this is important is because we, why it's very specific to India is because we do not have a linear supply chain. We are operating"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:29",
                    "end": "00:51:49",
                    "text": " How we should be able to sustain combat and be ready for combat is something that our supply chain networks have to address."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:49",
                    "end": "00:52:14",
                    "text": " So to give an example in terms of you know sustained supply chain and demand forecasting that has been implemented worldwide and because I'm a consultant is you know if you could draw from the US defend logistics organization which has which uses machine learning to predict spares failure and they have a very efficient inventory management system currently at play."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:14",
                    "end": "00:52:40",
                    "text": " And the NATO as well uses this predictive analytics to ensure that they take into ingesting a lot of algorithms on operational tempo, terrain, and weather, and do not only rely on past data and past orders. So you see a very, very effective inventory management system at play here. And there is something that we can learn from how they have taken this beyond."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:40",
                    "end": "00:53:10",
                    "text": " What also is very, very essential is effective warehouse and distribution. And AI has specifically played a large role here. Globally, I think we have seen AI-driven warehouse automation in UK and US defense depots, wherein they utilize robotics, vision systems, and intelligence lotting in making sure they have an efficient warehousing automation system. We also see the Ministry of Defense"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:10",
                    "end": "00:53:31",
                    "text": " in UK wherein they have taken the onus of implementing supply chain digitization which focuses clearly on AI enabled demand forecasting and of course automated fulfillment for defense forces that are deployed. These are some of the good best practices that we can see the world over."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:31",
                    "end": "00:53:59",
                    "text": " And of course, the Russia and Ukraine war has taught us in terms of how route optimization and how they are actually operating and showcasing resilience under fire, in which they're utilizing AI-enabled technologies for route optimization to plan their convoys. So these are some of the examples that I can think of. And in closing, I'd like to say for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:59",
                    "end": "00:54:23",
                    "text": " India specifically, I think AI and defense logistics is not about copying global markets. It's about building a resilient and indigenous system which will actually help adapt our supply chain networks to ensure that we are turning the Atman Nirbhar Bharat initiative into our operational advantage. Thank you Madhumita."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:23",
                    "end": "00:54:52",
                    "text": " I would be unfair to the audience if I don't take your question. I know we are out of time So just one question, you know if anyone has Just just give the mic. Can you come and tell only? The adversary was using the AI at a large space"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:52",
                    "end": "00:55:15",
                    "text": " and we do have indigenous capabilities but we have developed certain softwares but for the hardware part we are still dependent on other countries so what are your perspective how we are going ahead in building capabilities in that?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:16",
                    "end": "00:55:42",
                    "text": " Alright so probably you are referring to the hardware supply chain issues and primarily because artificial intelligence present technology relies on the GPUs and we are reliant for imports for GPUs. So this vulnerabilities are known they can be addressed through multiple means one is you have multiple sources"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:42",
                    "end": "00:56:07",
                    "text": " then is you have alternate technologies and we are working towards that only. Can we today develop instead of a large language model which runs on a GPU a small language model which runs even on a CPU and then have a large this agentic garden which small language models are running on a form of CPUs and then you can serve the same purpose which a GPU based large language model will give."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:07",
                    "end": "00:56:35",
                    "text": " That is another. Third is which is related to a longer time span is having an indigenous capability to develop this hardware. So that again that activity is being taken up very seriously. There are four lead agencies which are working on producing these semiconductors within India. We lost that opportunity once in the 80s when our semiconductor limited got burned down in 1986 and we became import reliant."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:35",
                    "end": "00:56:59",
                    "text": " TSMC came up in 1987, one year later. Today controls 75% of the market. So we are starting late, but as I say that there is advantages of the first movers, they get a lead, but there are advantages of the last mover, you get the clarity. So we are moving in all three directions."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:59",
                    "end": "00:57:28",
                    "text": " to do better hardware design For example the new architecture the mixture of experts we just launched a 17 billion parameter mixture of experts with just two shared mixture experts Now the advantage is you can do very quick inference even at the level of CPUs How can we use the workload? I think how do you turn your software into hardware advantage? How do you turn your software workloads to design better hardware? That's an advantage you can get through the Lakemore story"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:28",
                    "end": "00:57:57",
                    "text": " Fans are still there, but you know, we are extremely short of time I wanted to just acknowledge mr. Pradeep Gupta who wanted to raise a quite he has worked with EAM RM finance minister all of them But you know you are seeing all the changes that are happening. So we'll have a conversation outside. Mr. Gupta Thank you to all of you the deputy chief sir for giving a great context and motivating everyone to speak so well"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:57",
                    "end": "00:58:18",
                    "text": " I know we were short of time even at the start. General Chhibwar, so nice to hear your thoughts and how the DGIS is proceeding at pace around all of the army efforts around AI. General Anand, Vikram."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:18",
                    "end": "00:58:46",
                    "text": " Ganesh and Madhumita, I think you all have done a fantastic job. And thank you to all of you in the audience and also to the organizers for your support. I'd like to just thank everybody. I'd like to thank everybody for being here as also the panel. We put it together in a quick time and I think you saw that there were experts in the field who could have gone on till the evening. And I'd like to assure you both the industry, the academia and all the others"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:46",
                    "end": "00:59:05",
                    "text": " And of course there will be people in the industry that we are committed to you know Committed to committed to ensuring that the AI ecosystem that is there in the country Whether it be startups or large players and innovators that we incorporate them into building AI enabled systems"
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
    {
    "_id": "699578e5565455da1fc752a6",
    "projectId": "151712",
    "projectName": "AI for Defence Insights on Strategy and National Security - Hindi",
    "projectDescription": "AI for Defence Insights on Strategy and National Security",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T08:31:31.356Z",
    "projectUpdatedDate": "2026-02-18T08:31:31.356Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T08:32:31.520Z",
    "projectDeleted": false,
    "projectDataLength": 3,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "AI for Defence_ Insights on Strategy and National Security.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "AI for Defence_ Insights on Strategy and National Security.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Hindi"
},
    {
    "_id": "6995794d565455da1fc752c2",
    "projectId": "715367",
    "projectName": "AI for Defence Insights on Strategy and National Security - Spanish",
    "projectDescription": "AI for Defence Insights on Strategy and National Security",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-18T08:33:15.906Z",
    "projectUpdatedDate": "2026-02-18T08:33:15.906Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-18T08:34:15.503Z",
    "projectDeleted": false,
    "projectDataLength": 3,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "AI for Defence_ Insights on Strategy and National Security.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "AI for Defence_ Insights on Strategy and National Security.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Spanish"
},
    {
    "_id": "69945b1525c9f4e0d7e70c95",
    "projectId": "840130",
    "projectName": "Consumers at the Core_ Building AI With Consumer-Centric Approach - English",
    "projectDescription": "Consumers at the Core_ Building AI With Consumer-Centric Approach - English",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T12:12:10.395Z",
    "projectUpdatedDate": "2026-02-17T12:12:10.395Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T12:17:58.029Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "Consumers at the Core_ Building AI With Consumer-Centric Approach.mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771330313535.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:01",
                    "end": "00:00:54",
                    "text": " Thank you once again everyone for joining us today and in fact thank you Government of India and Ministry of Electronics and IT to give us this opportunity to host this panel. I think the most unique thing about this panel we were talking about"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:54",
                    "end": "00:01:24",
                    "text": " all of our co-panelist days. This is the one session which is completely focused on consumer empowerment, which is somehow is an important element and needs to be there when we discuss any new and emerging technology as part of it. So before I jump into inviting our speakers to share their views and perspective, I just wanted to share a few elements for everyone to think around those aspects when we are looking at how AI helps"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:24",
                    "end": "00:01:53",
                    "text": " on consumer empowerment and how AI also creates risks. And I'll try to give smaller one or two examples which are very specific to India at this point of time. How AI helps to be consumer centric? It's basically it helps you to have personal recommendations and services. It increases the decision making time over there. It provides you a 24-7 digital assistance which increases the accessibility part of it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:53",
                    "end": "00:02:22",
                    "text": " There is a real-time risk monitoring which increases because of AI if you look from a consumer perspective and also there's the new language AI which has come which helps you to go beyond English specifically looking at the regional languages. Two very recent examples is how the banks are using AI specifically from a fraud detection and monitoring perspective and how effective they have been. The second example is how telcos and all are using the voice-based"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:22",
                    "end": "00:02:47",
                    "text": " AI chatbot ecosystem, which are more regional into nature, if you look at it over there. But if I quickly look at what the risks that are emerging from AI from a consumer empowerment perspective, the most crucial part is there's opaque automated decision making, which the consumer is difficult to challenge or kind of question those over there."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:47",
                    "end": "00:03:13",
                    "text": " There is deep fakes, audios and videos which has the impersonation and I'll talk about an example also about it. There is hyper-personalized targeting which knows too much about the person and tries to use those emotional questions to influence your buying behavior. And lastly is again how do you hallucinate AI for giving you the wrong decisions as part of it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:13",
                    "end": "00:03:43",
                    "text": " examples again if you look at the kind of Impersonation has happened on fake defects AI videos for people to take financial decisions using finance ministers Pictures and others also as part of that part over there. So with that background, I think I like to first invite our keynote speaker Mr. Ravit Kumar Singh former Secretary government of India and he was also a member of NCDRC and I think sir you led the entire AI part at Consumer Affairs at the initial so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:43",
                    "end": "00:04:28",
                    "text": " We wanted to hear from you more in terms of where do you see the risk, where do you see the answers to those risks and also specific recommendation that we should look at as a follow up from the impact summit specifically which we should take as a recommendation to the government. Sir. That's old school. Good evening, everyone. First of all, let me thank the organizers for calling me here. I retired from the government one and a half years back."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:28",
                    "end": "00:05:00",
                    "text": " So it's good to be back in a suit and a tie after a while. My last posting was Secretary Department of Consumer Affairs. And then I moved to Member National Consumer Dispute Redressal Commission, NCDRC. And then I quit for some personal reasons. I'm a techie by education and have done many assignments in government of India and state government of Rajasthan on technology."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:01",
                    "end": "00:05:33",
                    "text": " I'm amazed to see the energy at this summit. And this summit is different. I follow what is happening in global AI and I followed what happened at Bletchley Park and then Paris. But this one is different. I think what this has done is it has been able to put people at the center. So while US and China can fight over who has the most efficient LLM,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:33",
                    "end": "00:06:01",
                    "text": " whether it is Claude or Gemini or Chad GPT or DeepSeek or whatever hidden is being developed in China. I think when it comes to shaping responsible AI, I see India has a role to play and India should play it very aggressively. As in the times of Mrs. Indira Gandhi, we headed something called the non-aligned movement."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:02",
                    "end": "00:06:30",
                    "text": " Russia was this side and US was the western powers were on the other side and we played the middle game and we were good at it. So I think the time has come to use that principle. Now coming to consumers and revolutions whether technology or other revolutions, any revolution it changes the balance of power."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:30",
                    "end": "00:07:00",
                    "text": " When we were buying from the corner stores, the mom-and-pop shops, the kirana dukan, there was a different dynamic between the buyer and the seller. The whole equation was different. And then 10-15 years back, the Amazons and the Flipkarts started coming and grabbing that space. And the whole power equation changed. So consumer became less and less powerful and the seller became more and more powerful."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:01",
                    "end": "00:07:30",
                    "text": " Now with the advent of AI, this balance, this power distribution, redistribution is going to further aggravate and we need to be very careful because unlike the other technologies, this is transformative. This is going to be a way of life. This is not like digital revolution. This is an age of intelligence. So it's going to make a lot of difference which are already visible."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:30",
                    "end": "00:07:60",
                    "text": " So we have to, all of us, all stakeholders have to take a call that when this redistribution happens, whether we are able to put some in favor of the consumer or everything will be governed by opaque systems which consumers neither understand nor control. So AI is very powerful in the consumer game as the previous speaker said. You know, it can..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:00",
                    "end": "00:08:30",
                    "text": " either exploit the consumer or empower the consumer. It can do partly both. It will depend on the systems, on the regulators, the government, the developers, and most importantly the consumers as to how this is going to play out. If it empowers, it's going to democratize access, which it has done, although the $20 bill of Chad GBT still hurts when you convert it into rupees, but"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:30",
                    "end": "00:09:00",
                    "text": " It has democratized. It has reduced costs. It has improved access. But when it comes to exploitation, the manipulative algorithms which are leading to unfair trading practices are abundant. When I was Secretary of Consumer Affairs two years back, we did an analysis as to how these online platforms are heavily indulging in what we call the dark patterns."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:00",
                    "end": "00:09:30",
                    "text": " They are manipulating the consumer behavior. They are nudging the consumer to go in a direction without your intent and it is harming the relationship which was fair between the consumer and the seller. Then there is of course algorithmic discrimination. If you're looking for a credit score and your residence is not very hip, you will get a bad credit score, your insurance premiums or everything. So there is algorithmic discrimination which is mostly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:30",
                    "end": "00:10:01",
                    "text": " beyond our control. There is privacy invasion. Starting from a simple, you go to any shop. I went to my barber yesterday, and he said, not the regular one. And the lady said, sir, can I have your phone number? I said, no. Because the moment it gets into that system, and when I was in the government, we had issued guidelines that no seller can force you to give your telephone number unless you want to."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:01",
                    "end": "00:10:28",
                    "text": " I had a little tip at a store which I will name because without naming you can't shame them. Decathlon. I went to the Anshal Plaza Decathlon and the lady there, I wanted to buy a basketball and said sir you can't buy unless you give your phone number. So I said no I will not give you a phone number but I still want to buy. Of course I didn't tell her who I was. At that time I was a secretary. So I could have, I had more weight. I could have thrown it around. I didn't."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:28",
                    "end": "00:10:55",
                    "text": " So she said, sir, the way our software is there in the POS, the terminal, I have to put a number. So I said, you put your number. And then she talked to her manager and put her number. That's how strangulating this whole thing is about privacy. We don't realize that every day we are sharing so much information, and it's all being embedded there. Nothing is being deleted in the digital world."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:55",
                    "end": "00:11:24",
                    "text": " If you think by deleting your account, by deleting a file, you are deleting anything, you are not. It's all there. So people, some people don't bother about privacy. They will put every half an hour something on Facebook or Twitter or Instagram. I ate this, I did this, I did that. So that's good for them. But people who want to vehemently protect their privacy, which I think many of us should, we should be careful. And this..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:24",
                    "end": "00:11:55",
                    "text": " AI-driven technology is going to take it away without your knowing about it. Of course, there is a lot of fraud. So all these conferences that are taking place, they must realize that at the center of AI governance is the consumer. So ideally, this room should have been the most crowded, which I hope it will be when you have the next session. Because everywhere you want to buy something,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:55",
                    "end": "00:12:26",
                    "text": " your insurance premium, job, medical decisions, online content. Everything is driven by this. Again, I will name them. We ran an investigation on Uber. If you are buying the ride, it depends what phone are you using. If you are using an expensive phone versus a cheap phone, the rates will be different. Same thing happens when you're shopping online."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:26",
                    "end": "00:12:55",
                    "text": " I worked under Mr. Piyush Goyal, who was my minister, and we did it together. We sat next to each other in his office and we bought something from Amazon. I said, sir, you use the other phone. He, of course, had the expensive phone. And then we realized that the rates we were getting from the same portal sitting next to each other were different. More, if your battery is running low, your Uber cost will come higher because the algorithm knows that you are desperate."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:55",
                    "end": "00:13:23",
                    "text": " And if your battery is less than 20%, your cost will be higher. So these are the things which are happening unintentionally, and we should be very, very wary. So where all this is leading to? I think the currency of AI is trust. So if we demolish directly, indirectly, partly the trust, then we are going to destroy"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:23",
                    "end": "00:13:53",
                    "text": " this relationship between consumer and the seller which is going to be very very bad and lot of people say lot of techies including my son they say that you know if you put too much into regulation and trust you are stifling innovation partly true so we have to see which point on the development curve we are if we are very new"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:53",
                    "end": "00:14:24",
                    "text": " And things are just, for example, let's take three years back when maybe chart GPT was just launched two and a half years back. On that point, if you regulate too much, of course, you will stifle innovation. But three years hence, when it is becoming part of everybody's life, you have to regulate. Otherwise, it will cause so much harm. So since I had to do some government kind of stuff, so I wrote that what should be the five pillars"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:24",
                    "end": "00:14:52",
                    "text": " of consumer-centric AI. First and foremost, of course, is transparency. When I am using AI, I should know what is it doing? Am I talking to a person? Am I talking to a bot? Does he know everything about my background, or is it the first interface, or they have information? Plus, is it just pattern matching, which a lot of LLMs do? Or is there a reasoning which some of them started doing?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:52",
                    "end": "00:15:22",
                    "text": " So the consumer should know how the interface is being driven by AI or not. The second is accountability. If I do get harmed, if the consumer gets harmed, then who is responsible? If chat GBT sessions are leading someone to a suicide, or I read just this morning that somebody made a bomb out of a cloud prompt. So who's responsible?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:22",
                    "end": "00:15:50",
                    "text": " If you're getting into that extreme situation, is the user solely responsible or is there responsibility on the system, on the algorithm or whatever is being used? So that has to be clear. Third of course is fairness and non-discrimination. We just spoke about the kind of phones that you use and you get. Even otherwise, fairness will depend on what algorithms, what data was it trained on?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:50",
                    "end": "00:16:17",
                    "text": " If it is trained on Western data and an Indian guy is looking at a solution, obviously he or she will get answers biased towards the Western world. So those biases have to be removed. If not removed, have to be reduced to a minimum. Then we spoke about privacy and data protection, which is the fourth pillar. And the fifth is accessibility and inclusion. Sitting in the heart of Delhi in air-conditioned Bharat Mandapam,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:17",
                    "end": "00:16:47",
                    "text": " We are chatting about AI, but what about the guy 500 kilometers away, 5,000 kilometers away, 3,000 kilometers away? Of course, a lot of use cases. I was very happy to read yesterday's Indian Express, maybe. I still read newspapers, old school. There was very good examples of how AI use cases are being promoted in three of my favorite areas, education, health care, and agriculture."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:47",
                    "end": "00:17:17",
                    "text": " And if you're pushing that, then the people who are supposed to use that, they have to be included in terms of access to technology, regional languages, the local context, et cetera, et cetera. So in the end, I want to say that technology is changing exponentially. Human capability is not. So people who talked about AGI and super intelligence, it is true. It will happen."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:17",
                    "end": "00:17:54",
                    "text": " and we will be left behind because human capability is limited. And in the end, all of us stakeholders including those sitting the high and mighty on the panel, we have to decide where consumers are just data points or empowered participants. Thank you very much. Thank you, sir. I think I'll just take your permission to add one more pillar to that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:54",
                    "end": "00:18:23",
                    "text": " connecting to your point in terms of the audience numbers is I think that consumer empowerment has to change to consumer awareness and knowing your rights I think that is still a very limited element it is still available maybe in this room or maybe not in this room itself but that has to go to a large so knowing your rights why you're giving those data points and what you get is a most important thing that has to come from AI as principles as we speak"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:23",
                    "end": "00:18:43",
                    "text": " So we have exact 27 minutes, two minutes for me for asking the questions and five minutes each speaker. So I'll request everyone to be very to the point and to the specific recommendation that they want for either the industry or the government or academia or civil society to work together or specific actions."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:43",
                    "end": "00:19:12",
                    "text": " So I'll start with Sarita ma'am. I think you're looking at the entire DPRG part and I think DPRG has been very actively using AI now. What we have been seeing specifically addressing public grievances and also improving the functioning governance part of it. So could you give an overview of what are these initiatives and how it has helped the ministry specifically because you handle the biggest number of consumers that even a big tech company can also handle."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:27",
                    "end": "00:19:55",
                    "text": " Good evening to all. Yes, the Department of Administrative Reforms and Public Governances has been in the forefront of creating an enabling ecosystem for development of e-services. As we all know that digitization is the prerequisite for enhancement of services through the application of specific AI models."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:55",
                    "end": "00:20:25",
                    "text": " This ecosystem is developed with the help of the developers, the scientists, the technologists and also the administrators in the field who are the first or a level of consumers who will develop these platforms, the digital public infrastructure platforms with the use of the latest technology for the consumers which are mostly the citizens and the consumers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:25",
                    "end": "00:20:50",
                    "text": " in case of the government is of course the whole of citizenry different target groups are of course there and one of the important aspects that is handled which is very important for the government is of course the grievances so as the government provides the services the feedback of the citizens and the grievances of the citizens are centric"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:50",
                    "end": "00:21:18",
                    "text": " to the success of any governance. And therefore, the present government working on the policy of the Government of India is maximum governance, minimum government, and at the same time, sabka sadh, sabka vikas, sabka prayas, as we know. So basically, governance for the people truly can happen with the help of the next generation administrative reforms which are embarked upon through digitization."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:18",
                    "end": "00:21:42",
                    "text": " As a young DC 20 years back, we really struggled to have that ecosystem, trying to know from where to get the knowledge partners, from where to have the funds, how to converge them, and of course the developers themselves. Today we are seeking out people from the DRPG who want to develop such infrastructures."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:42",
                    "end": "00:22:08",
                    "text": " for the century. The objective being obviously transparency in the delivery of services, inclusion to the last mile. No one is to be left out. The holistic development is another goal towards which the AI models can be developed. Of course, the predictive analytics, analytical models and the generative AI models."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:08",
                    "end": "00:22:36",
                    "text": " are used to a large extent and the DRPG encourages the dissemination also of the innovations in the field of digital public infrastructure through the recognitions through national award for e-governance which is given since 2003. The very development of the categories and the award indicates how the evolution of these digitization and adoption of technology has happened."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:36",
                    "end": "00:23:02",
                    "text": " from the very basic digitizations for citizen services to now using AI for even the pandemic responses and mitigation of disasters. The awards, if you see, there is a specific category for the use of AI and emerging technologies. And therein, there are some of the remarkable ones which include areas of bhasini, you must have heard."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:02",
                    "end": "00:23:18",
                    "text": " Yatra, all of you must have heard, but I wonder if you've heard about Nadris. This is an AI and ML trained model developed by the scientists, by the ICAR for predicting the diseases in livestock."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:19",
                    "end": "00:23:37",
                    "text": " So the way they work on the data depending on what are the climatic conditions depending on what are the Of course the samples are also drawn because livestock is a very important area of livelihood For our country and is the fastest growing sector also here"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:37",
                    "end": "00:24:06",
                    "text": " The loss of livestock is a loss of livelihood. It's a loss of loss to the economy, which is many a times than what you find in agriculture, purely sector. So similarly, at the state level, if we see, you would find use of AI in cases such as Voka Sati. This is again an initiative which is developed in Assam in the Northeast. This is helping the citizens through the chatbots for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:06",
                    "end": "00:24:35",
                    "text": " easy access at one place to all the services and also faster grievances. These are the use case models, the best practices from across the country which you see, which the DRPG is encouraging through its schemes. Thank you ma'am. I think some interesting stories and examples that you have shared. I think we need to promote these more to understand especially the stock part."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:35",
                    "end": "00:25:04",
                    "text": " I'll come to Irene and Irene because Consumer International works a lot around a lot of jurisdictions around 100 plus. So my primary question to you is do you see across governments or industries present in these geographies AI by design keeping consumer centric approach is there? Is it more of a compliance mechanism or is it more of a self-regulation approach that they're seeing? And a follow up to that is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:04",
                    "end": "00:25:23",
                    "text": " What are the common minimum framework that is required irrespective of jurisdiction because always companies use those aspects of this is what has followed in US this is followed in UK you So how do we come through a common minimum program on that from a consumer centric perspective? Thank you a lot to cover in five minutes"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:23",
                    "end": "00:25:52",
                    "text": " Hi everyone, just want to say thank you for being here. Thanks for inviting us to be on this panel. I am with Consumers International as Kaushal said. Consumers International is a federation of consumer advocacy groups. So there are different advocacy organizations in different countries. We have the pleasure of having two on this panel today. And we represent more than 200 different consumer advocacy organizations across 100 different countries."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:52",
                    "end": "00:26:01",
                    "text": " The majority of these consumer advocacy groups are in the global south."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:01",
                    "end": "00:26:31",
                    "text": " visions that we see with these different members and we're able to represent them at the international level and in global forums like this one today. So what's so interesting about the case of AI is that we're seeing so many different things play out in different countries. So we just had a meeting last week with some different members and they come forward from different countries and represent basically that they're having very different experiences in these."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:31",
                    "end": "00:27:00",
                    "text": " different countries. And while there might be one consumer experience in one country, it's very different than another. So when we start talking about what are the baseline things we might need to put consumers at the center, when we talk about consumer rights at Consumers International, we're often talking about the United Nations guidelines for consumer protection, which are basically the bedrock of consumer protection at the UN and which consumers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:00",
                    "end": "00:27:29",
                    "text": " international was involved in drafting. So when we talk about these consumer rights and how they're acting in different countries, we are basically seeing that the experience is very varied across different countries. But our members do a lot of work and then they come to us to look into what can be the global trends and what can be the main principles that we take forward. And when it comes to different principles for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:29",
                    "end": "00:27:36",
                    "text": " AI and consumer-centric AI, a lot of what we're talking about now is kind of three things."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:36",
                    "end": "00:28:05",
                    "text": " One is transparency. So I think this has come up a couple of times already in this panel, but consumers need to know when they're using an AI and what this means for them. Consumers often end up interacting with AI systems and they're not aware that they're doing so or what it means for their data. So when we talk about consumers, we're often assuming that a consumer has an informed choice to use a product or service."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:05",
                    "end": "00:28:34",
                    "text": " but when you have an AI system where they're not aware of it, then how can we call them a consumer that knows what's happening with that product? Another important aspect that we're really looking into is data privacy. We've seen that there are the limits of individualized consent when it comes to data privacy, and we've seen all the issues that consumers face when it comes to evaluating what data they will share."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:34",
                    "end": "00:29:02",
                    "text": " often they don't know what it means for them to share that data. So I encourage everyone to look forward to some of the work we're doing this year, which will focus on data privacy and AI, and thinking about how we can see it as a public good, actually, as something that data is very social, we interact with each other, we share data. It's no longer an individual thing to be looked at in such an individualistic manner."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:02",
                    "end": "00:29:04",
                    "text": " And then the other thing we're talking about is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:04",
                    "end": "00:29:33",
                    "text": " adoption and quality. So there's so much movement towards getting people to adopt and participate in new AI products and services. We've seen this a lot, I think in the last months with agentic AI, for example. And at Consumers International, we're really concerned with not just adoption that we have consumers using new products and services, but that the quality is high and that that leaves consumers in a place where they"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:33",
                    "end": "00:29:42",
                    "text": " feel empowered and where they're resilient when they're using a given product or service. So these are kind of three."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:42",
                    "end": "00:30:12",
                    "text": " three points that I would say kind of can underpin such vastly different experiences. But what's so great about the work that we do at Consumers International is that even though we have all of these different members with such different experiences and different emerging opportunities and risks that they're seeing in their country, we're able to kind of bring those together, see those global trends and represent the consumer voice in places like this one. So thank you so much."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:12",
                    "end": "00:30:38",
                    "text": " Thank you, Irene. I think just to add to your point, I think one thing which the government is very looking is democratizing the AI resources. So in a similar DPI format, the entire GPU framework is being looked at, which is around 38,000, which has been allocated right now. So it's, again, in the DPI framework itself. I'll move to Asim, sir, and then a very specific question over here is, if I look at a consumer pattern, the entire lifecycle of a consumer,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:38",
                    "end": "00:31:06",
                    "text": " Which are the very specific areas? Sorry, there's a question to a scene. So if you look at a consumer behavior, entire pattern of a consumer decision making, which are the very specific areas where you think AI influences the decision or empowers the consumer a lot? And what are the risks associated in that entire journey from an identification to the final purchase and the services?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:06",
                    "end": "00:31:33",
                    "text": " can consumer organization like voice specifically work on those pillars? Great. So I'll come to the conclusion. Thank you very much for inviting me to this discussion. I think I'm going to start with the conclusion first. Very strange, but then that explains most of your question as an answer. I'll try to explain it. Without mandatory accountability, bias audits, stricter enforcement."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:33",
                    "end": "00:32:01",
                    "text": " under the Consumer Protection Act, and I'm sticking to my domain of Consumer Protection Act, especially for high-risk AI in finance and e-commerce. These risks could deepen the exclusion of vulnerable consumer, turning potential empowerment into exploitation. Now, this is something of a mindset. We are all slaves of AI. We all know that as consumers. We travel with completely"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:01",
                    "end": "00:32:29",
                    "text": " AI enabled systems, we buy e-commerce or generally nowadays to UPI and other modes which are again AI enabled. So most of the things we are doing as consumers are UPI enabled or AI enabled. Question here is basically is the consumer awareness or what he's doing and how he's using AI relevant and if it is relevant."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:29",
                    "end": "00:32:58",
                    "text": " then how much it is being practiced. And who's the practitioner here? It is the companies who are the AI enablers. And as consumers, we will say that we are dependent on you. We are fully confident that you will serve us best with the AI. It's very good. It actually is a turning point for many government schemes. From our consumer affairs, if you look at the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:58",
                    "end": "00:33:26",
                    "text": " research that has been done in terms of a report, 2025 end report. The Grahak Nya chat box have facilitated rupees 36, 46 crore in refunds, which is fantastic. We couldn't dream that earlier on through the manual process. AI has enabled that. And because the system operates on AI. We obviously also have the language factor. This is..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:26",
                    "end": "00:33:53",
                    "text": " Taking the same example, it is in 17 languages. So language is well covered here. AI enabled, in fact. And also, AI assisted complaint filing. Again, helping the consumers. These are very, very good things. But again, these are run by government after due diligence. Are the corporates doing the same? Are the brands doing the same? We have our own doubt and risk management, therefore."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:53",
                    "end": "00:34:22",
                    "text": " is paramount on this particular factor as far as consumer mindset is concerned. When you travel, it has been said that it has been observed that you share your phone number 46 times, you share your photographs, think about it, when you travel next. You share your photograph 11 times and you share your personal data to an extent, date of birth and birth and other things, at least 24 times."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:22",
                    "end": "00:34:52",
                    "text": " This is something which is troubling, because AI takes over that, and what is happening to your data, we don't know. So in conclusion of this five minute unfortunate break, I must say that AI amplified dark patterns. Rohitji already said that dark patterns are today into our lives as a bad patch. Everything is dark pattern, and for those of you who don't know dark pattern, it is those numbers that crop up two seats left."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:52",
                    "end": "00:35:20",
                    "text": " Last two boxes for sale. I'm just giving an example for understanding in fact. Algorithms. We are just fighting for algorithm understanding. Why algorithms are developed without consumer consent or without consumer participations? They can be told to participate and probably we can contribute. Manipulative personalization."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:20",
                    "end": "00:35:49",
                    "text": " is something that is happening every day. I think Kaushal himself gave an example. Deepfakes, as we know that, AI is now totally responsible for deepfakes and misleading the purchases. So the pros and cons are well balanced out. As Rohitji said, this is the time to act for the government to take a decision. Does it need a regulation or let it free flow?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:49",
                    "end": "00:36:07",
                    "text": " In fact, that's why I wanted to come next to you is a follow-up question is which are the sectors where you see actually consumer gating impacted because you work very actively in consumer grievance management. And second thing is we have seen in a lot of sectors specifically banking and financial where SRO framework has worked."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:07",
                    "end": "00:36:38",
                    "text": " a self-regulatory organization where consumer groups think tanks they come or they form SRO and they are taking the burden from the regulator but the benefit is because they work with industry they are better at that part. So do you need something of that framework is required and not maybe a policy which also creates the balance between innovation and regulations? Yeah, so good evening and thank you for having me here. So three areas of vulnerability which the consumer faces."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:38",
                    "end": "00:36:59",
                    "text": " One is the fintech sector and we've talked about it extensively till now. All the algorithmic biases and e-commerce, dark patterns, surveillance pricing and everything. But I would like to focus on what's working at the consumer grievance redressal."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:59",
                    "end": "00:37:29",
                    "text": " systems. At the NCH, the National Consumer Helplines, we now have excellent mechanisms, as Ashim mentioned, that we have 17 languages and voice inputs, AI-powered speech recognition and voice processing, and refunds of 4.5 crore till November 2025. That's the data of refunds that the consumers have received."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:29",
                    "end": "00:37:52",
                    "text": " Secondly, this Grahak Naya in collaboration with Metta and that guides consumers through the consumer grievance, filing notices and everything and so helps draft legal notices and the pilot looks very promising. Similarly, we have something called eJagriti."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:52",
                    "end": "00:38:12",
                    "text": " which is an integrated platform now for complaints, filing, and virtual hearings. So are these really working, is the question. Now, at the end of it, the consumer courts are clogged. Lytic pendency in consumer courts like"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:12",
                    "end": "00:38:42",
                    "text": " The AI is helping power generate complaints in a bit speedy manner. But unless for consumers, it is a second punishment when he gets into litigation. First punishment, of course, he's scorned. Second is when he gets into this litigation. So unless those things are taken care of by, say, giving teeth to class action suits, which now are, you know,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:42",
                    "end": "00:38:57",
                    "text": " gets into a lot of logistical and procedural issues. Also giving power to consumer organizations to take up such issues. So these are certain areas."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:57",
                    "end": "00:39:19",
                    "text": " area of vulnerability is the service sector. I'll just give you an example. Just about a week back, in fact, I had this experience of going to an ATM and the vending machine gave me less of cash that I had asked for. And of course, I..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:19",
                    "end": "00:39:48",
                    "text": " connected to the service center. And the same feedback loop, they did not have that specific number for my kind of complaints. So we got into that frustrating feedback loop. So that's what happens with most of the consumers who are faced with a chatbot. And the issue is that most companies, they don't have a hybrid mode. Like you cannot escalate your issue to a human interface. It is the same."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:48",
                    "end": "00:40:06",
                    "text": " And it becomes very frustrating. And at the end of it, me being from a consumer organization, I have kind of given up on my complaint. So you can imagine what must be happening to the common man on the ground."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:06",
                    "end": "00:40:35",
                    "text": " Black box inefficiency AI chatbots often fail to understand the nuanced or responses which needs customized input. So those are the issues. Then loss of accountability is another issue. Where do I go next from here? Who is accountable? And, of course, the lack of empathy. So these are certain issues that we are facing on the ground as consumers. Thank you. Thank you, Linda."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:35",
                    "end": "00:41:04",
                    "text": " Sudeep, I've preserved the best for the last because you are an AI expert, you work with industry, you hear the perspectives. Yes. Now, your views on how to address it, to kind of come to a part where consumer is as empowered, as well as it also helps the industry to grow at much more faster speed and make much more consumer-centric products. Yeah, I think the whole discourse here is like something, it's also on every industry person's mind. It's not just like, you know..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:04",
                    "end": "00:41:33",
                    "text": " one-sided story. I've been in AI for a long time. Right now, I head the AI org in DoorDash. Before that, I was at Netflix for seven years. One thing you see in these companies is that they're actually very consumer-obsessed. In the good companies, and not all of them maybe, but basically the guiding force is the consumer is right. You're building this product at the end of the day to actually have the consumer get..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:33",
                    "end": "00:42:01",
                    "text": " the best experience in the product. That leads to long-term value of the consumer. That actually leads to better revenue for you. So it's actually a self-sustaining cycle. But as industry leaders, the responsibility we have is what I call setting the floor and also raising the ceiling of this. And that comes from the point of, in your company's KPIs, in the metrics that you're tracking,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:01",
                    "end": "00:42:29",
                    "text": " safety, the consumer rights, the consumer transparency are not options. They actually must be built into. It's like a part of the infrastructure itself. We talked a lot about these algorithmic biases. I'll touch on them a little bit. Some of these biases are by construction. These big LLMs, they learn most of the stuff from English and from the Western world. They will actually fail and falter when you..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:29",
                    "end": "00:42:56",
                    "text": " try to serve this very tail language that is not represented there. So it's on us who are developing these products to actually go and build those models in a way that can be unbiased from day one. There's the other kind of bias, like the one you talked about, which is like Uber charges more. And this is because the algorithms are taught to maximize profit or something, right? And then it's actually learning from the patterns."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:56",
                    "end": "00:43:24",
                    "text": " unintentionally it's picking up all these signals and this is where we as industry practitioners we work very very closely with academics who are looking into these problems because this de-biasing of AI is a very hard problem by itself but the best companies out there they're actually working with the experts in the field to see how to de-bias these algorithms because at the end of the day as I said you lose the consumer's trust once you lose that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:24",
                    "end": "00:43:48",
                    "text": " you actually lose the long-term value of the consumer. The other thing that industry is very front and center in our minds is the accessibility of the technology. It has also been talked about. That's also part of the fairness. A device that's not your best iPhone is like..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:48",
                    "end": "00:44:18",
                    "text": " very slow. It's probably in a geographical region where the signal is very low. A company like Netflix or YouTube has to think of how to get the best experience even there. The technology needs to also be built in a way from ground up. That's actually fair. That just doesn't apply to technologies like streaming services or delivery services. It's like all of AI, essentially."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:18",
                    "end": "00:44:47",
                    "text": " I talk to a lot of my peers in the industry, and there are various things that all these companies are thinking about, very front and center in their minds, is how to rein in these systems. One thing I would say is, as we are getting into more agentic, more of this generative AI, the risks are larger. One of the ways that these companies are dealing with this is by setting up what is called red teaming. Essentially, you build a system,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:47",
                    "end": "00:45:17",
                    "text": " The idea is to help the consumer. You build an agentic system that deals with consumer complaints. You build an agentic system that helps you order food or plan your party tonight. But you can also ask the system, how much of X should I take to kill myself? So essentially, how do you prevent those things is by these systems called rate teaming, where you adversely attack these systems very, very comprehensively to figure out where the problems are at the edges."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:17",
                    "end": "00:45:44",
                    "text": " So just to elevate and to conclude my thing, I feel like, as someone said, in the early stages of this product development, you cannot put too much of regulation because that will stifle innovation. But once it's semi-mature to mature, you have to think of these things. Because otherwise, at the end of the day, you are going to lose the consumer. Thank you. Thank you, everyone, for being a patient audience. And thank you for the expert speakers."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:44",
                    "end": "00:46:53",
                    "text": " I think only one line I would like to add at the end is we need to explain to the consumers that AI is not free. I think the feeling is AI is free, it's not free. It is taking a lot from you to kind of grow and then it will again sell back to you again at a higher price. So thank you again, thank you all our speakers today and thank you again Ministry for giving us this opportunity. Sorry, we can take questions now or after the session outside because there's a next panel which is waiting."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:53",
                    "end": "00:47:13",
                    "text": " Check. We have a next panel discussion starting from Indian Institute of Management Lucknow, IIM Lucknow. The next session starts in five minutes. We request you to please take your seats and be part of the panel discussion. I repeat, the next panel discussion by IIM Lucknow is about to start. We request you to please stay and join for the panel discussion."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "69945cde25c9f4e0d7e70cb1",
    "projectId": "210118",
    "projectName": "Consumers at the Core_ Building AI With Consumer-Centric Approach - Hindi",
    "projectDescription": "Consumers at the Core_ Building AI With Consumer-Centric Approach - Hindi",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T12:19:43.251Z",
    "projectUpdatedDate": "2026-02-17T12:19:43.251Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T12:20:29.804Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Consumers at the Core_ Building AI With Consumer-Centric Approach.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Hindi"
},
    {
    "_id": "69945dc125c9f4e0d7e70cc7",
    "projectId": "250092",
    "projectName": "Consumers at the Core_ Building AI With Consumer-Centric Approach - German",
    "projectDescription": "Consumers at the Core_ Building AI With Consumer-Centric Approach - German",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T12:23:33.482Z",
    "projectUpdatedDate": "2026-02-17T12:23:33.482Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T12:24:25.871Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Consumers at the Core_ Building AI With Consumer-Centric Approach.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "German"
},
   {
    "_id": "699457d125c9f4e0d7e70c11",
    "projectId": "415665",
    "projectName": "Founders & Funders The India AI Capital Ecosystem - Spanish",
    "projectDescription": "Founders & Funders The India AI Capital Ecosystem - Spanish",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T11:58:13.826Z",
    "projectUpdatedDate": "2026-02-17T11:58:13.826Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T11:59:00.079Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Founders & Funders The India AI Capital Ecosystem.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Spanish"
},
{
    "_id": "699456a825c9f4e0d7e70bbd",
    "projectId": "164756",
    "projectName": "Founders & Funders The India AI Capital Ecosystem - Hindi",
    "projectDescription": "Founders & Funders The India AI Capital Ecosystem - Hindi",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T11:53:16.922Z",
    "projectUpdatedDate": "2026-02-17T11:53:16.922Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T11:54:02.011Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "Founders & Funders The India AI Capital Ecosystem.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Hindi"
},
{
    "_id": "699461ca25c9f4e0d7e70d60",
    "projectId": "704559",
    "projectName": "Democratizing AI Resources in India",
    "projectDescription": "Democratizing AI Resources in India",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T12:40:39.626Z",
    "projectUpdatedDate": "2026-02-17T12:40:39.626Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-17T12:48:02.259Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "Democratizing AI Resources in India.mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771332001833.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:01:48",
                    "text": " Can I request Mr. Sunil Gupta, Chairing our DC Council, to please moderate the session. Sir, over to you. Good afternoon, everybody. Can I just request somebody to close the door? There's a lot of noise around. Sir, before we start the session, can I request all for a group photograph? Okay. Good afternoon. I think energy is palpable. हिंदी में बोलू तो माहुल बना हुआ है।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:48",
                    "end": "00:02:17",
                    "text": " आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:17",
                    "end": "00:02:43",
                    "text": " आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:43",
                    "end": "00:03:12",
                    "text": " इंडिया के लिए अप्लाइइद पार्ट के लिए अपने अप्लाइइद पार्ट के लिए अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:12",
                    "end": "00:03:39",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:39",
                    "end": "00:04:09",
                    "text": " And the real topic of this panel discussion is that business led AI inclusion. From pilots to production to real impacts. Whether it is in B2B space, whether it is in B2C space or whether it is in G2C space. Across board, how do we lead adoption of AI at a mass scale. So on this panel, you will see practitioners who actually are working on across the whole value layer of AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:09",
                    "end": "00:04:38",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:38",
                    "end": "00:05:06",
                    "text": " credit scoring for fraud detection you know and taking it on a very very large global scale so according to you how can AI driven financial you can call it decision making how it can expand credit access especially when I am talking about the in a context of India where you know you are dealing with so many of people how can AI led"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:06",
                    "end": "00:05:35",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:35",
                    "end": "00:05:51",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:51",
                    "end": "00:06:21",
                    "text": " तो मैं अच्छाली अच्छाली अच्छाली अच्छाली अच्छाली अच्छाली अच्छाली अच्छाली अच्छाली अच्छाली अच्छाली अच्छाली अच्छाली"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:21",
                    "end": "00:06:48",
                    "text": " आपका आपके लिए में क्या है? आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपक"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:48",
                    "end": "00:07:10",
                    "text": " अच्छाना अच्छाना अच्छाना अच्छाना अच्छाना अच्छाना अच्छाना अच्छाना अच्छाना"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:10",
                    "end": "00:07:39",
                    "text": " यह आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:39",
                    "end": "00:08:07",
                    "text": " आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:07",
                    "end": "00:08:33",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:33",
                    "end": "00:09:01",
                    "text": " So now when your company is scaling very big through your Azure AI initiatives and you are scaling it to the remotest corners of the country, you are taking it to startups, you are taking it to SMEs. My question to you is regarding SMEs. When you take it to production cases in SMEs, how are you ensuring compliance and safety and security? Because if that is not there, nothing will succeed. How are you planning to do that? How are you doing that as a company?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:01",
                    "end": "00:09:28",
                    "text": " आप आप आप आप आप आप आप आप आप आप आप आप आप आप"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:28",
                    "end": "00:09:51",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:51",
                    "end": "00:10:19",
                    "text": " और अपने लाइन्स के लिए हम आपने लिए नहीं हुआप हुआप हुआप हुआप हुआप हुआप हुआप हुआप हुआप हुआप हुआप हुआप हुआप हुआप हुआप हुआप हुआप हुआप हुआप हुआप"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:19",
                    "end": "00:10:41",
                    "text": " So, what started as a framework has got much more deeper now. So, it has got tools, it has got standards, it has got dashboards, all of that. And all of that is open source. So, if you go to responsible AI, search for that on GitHub, you can basically get the entire toolbox. For example, if you want to understand"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:41",
                    "end": "00:11:09",
                    "text": " आर्टिफ्याइंटिलिजन्स के लिए अपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:09",
                    "end": "00:11:38",
                    "text": " ताइज ताइज ताइज ताइज ताइज ताइज ताइज ताइज"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:38",
                    "end": "00:12:08",
                    "text": " अपने स्मी का आप्लिगेश्यों के लिए अपने स्मी के लिए अपने स्मी के लिए अपने स्मी के लिए अपने स्मी के लिए अपने स्मी के लिए अपने स्मी के लिए।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:08",
                    "end": "00:12:27",
                    "text": " आप्लिकेश्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:27",
                    "end": "00:12:58",
                    "text": " इंदियास्प्रोब्लें के लिए इंदियास्प्रोब्लें के लिए इंदियास्प्रोब्लें के लिए इंदियास्प्रोब्लें के लिए इंदियास्प्रोब्लें के लिए इंदियास्प्रोब्लें के लिए इंदियास्प्रोब्लें के लिए इंदियास्प्रोब्लें के लिए"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:58",
                    "end": "00:13:27",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:27",
                    "end": "00:13:55",
                    "text": " संदीप, क्या आप आपके लिए? आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:55",
                    "end": "00:14:24",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:24",
                    "end": "00:14:53",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:53",
                    "end": "00:15:21",
                    "text": " अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:21",
                    "end": "00:15:46",
                    "text": " I will put a voice bot and it will talk to my customers. In fact, I would put that the last because while AI can scale advantages, it can also scale mistakes. So I would put that as the last. But in understanding my customers needs better. If a customer is reaching out to me, being able to anticipate why would they be reaching out and can I have a solution ready? Can I use AI and machine learning in that? That's one kind of an example. If I look at internal processes,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:46",
                    "end": "00:16:10",
                    "text": " तुड़े अपपर्चिनिटी यूजिंग आई टूल्स यूजिंग यूजिंग यूजिंग यूजिंग यूजिंग यूजिंग यूजिंग यूजिंग यूजिंग यूजिंग यूजिंग यूजिंग यूजिंग यूजिंग यूजिंग यूजिंग यूजिंग यूजिं"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:10",
                    "end": "00:16:38",
                    "text": " वालियोट नार्ट पर वालियोट नार्ट पर वालियोट पर वालियोट पर वालियोट पर वालियोट पर वालियोट पर वालियोट पर वालियोट पर वालियोट पर वालियोट पर वालियोट पर वालियोट पर वालियोट पर वालियोट पर वालियोट पर वालियोट पर वालिय"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:38",
                    "end": "00:17:04",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:04",
                    "end": "00:17:27",
                    "text": " I try to solve a problem, but I start thinking what all constraints I have. I don't have data. Even if I data, I don't have access to it in silos. So what will I do? So better let the business go as it's going. So what Sandeep is saying, instead of getting just too much zest up or complex or confused by this technologies and all the jargons which people like us keep on putting it, just think of your regular business problem, which you want to solve."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:27",
                    "end": "00:17:51",
                    "text": " And then come to practitioners like Fractal who will actually tell you and guide you what you already have in terms of your data which is lying in your company itself and then they will start building solutions for you for the problem which you are facing in your business. Thanks Adib. So Manoj coming to you, Manoj represents largest or one of the largest data center company in the world Equinix and they have data centers across the world."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:51",
                    "end": "00:18:16",
                    "text": " इंडिया में इंडिया में इंडिया में इंडिया में इंडिया में इंडिया में इंडिया में इंडिया में इंडिया में इंडिया में इंडिया में इंडिया में इंडिया में इंडिया में इंडिया में इंडिया में इंडिया में इंडिया में इंडिया में इंडिया में इं�"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:16",
                    "end": "00:18:45",
                    "text": " अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:45",
                    "end": "00:19:12",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:12",
                    "end": "00:19:40",
                    "text": " आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:40",
                    "end": "00:20:04",
                    "text": " all that we need is very high and good quality of internet as long as we have good internet with good latency the resources whether it is data center or the compute frankly speaking can be anywhere but it's very well known that a video conferencing where you are interacting with a AI agent is nothing but a video conference you say something and the other person responds to you can be handled"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:04",
                    "end": "00:20:26",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:26",
                    "end": "00:20:53",
                    "text": " अवार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्व"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:53",
                    "end": "00:21:14",
                    "text": " एक्विनिक्स के लिए आपके लिए अपने आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:14",
                    "end": "00:21:43",
                    "text": " और आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:43",
                    "end": "00:22:02",
                    "text": " आपका वाइबल आपका वाइबल आपका वाइबल आपका वाइबल आपका वाइबल आपका वाइबल आपका वाइबल आपका वाइबल"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:02",
                    "end": "00:22:32",
                    "text": " आपका आपका आपका आपका आपका आपका।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:32",
                    "end": "00:22:58",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:58",
                    "end": "00:23:28",
                    "text": " सब का साज सब का विकाश"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:28",
                    "end": "00:23:43",
                    "text": " अपने आई लेड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्ड्�"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:43",
                    "end": "00:24:12",
                    "text": " आपने सचकर दिखाये कि सब्रुक का फल वाकही मीथा होता है।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:12",
                    "end": "00:24:22",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:22",
                    "end": "00:24:51",
                    "text": " इं इंडिया, हम इंडिया के लिए नहीं है, हम इंडिया के लिए लिए नहीं है, और वॉल्ड के लिए नहीं है, और तोपिक के लिए नहीं है, और तोपिक के लिए नहीं है, और तोपिक के लिए नहीं है, और तोपिक के लिए नहीं है, और तोपिक के लिए नहीं है, और तो"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:51",
                    "end": "00:25:16",
                    "text": " और आधार के नाम से जानते हैं।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:16",
                    "end": "00:25:43",
                    "text": " प्रवार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:43",
                    "end": "00:26:13",
                    "text": " आद्टेक काम्पनियों के लिए लिए लिए लिए लिए लिए लिए लिए लिए"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:13",
                    "end": "00:26:42",
                    "text": " आपका आपका इंडिया AI मिशिन क्यों क्यों क्यों क्यों क्यों क्यों क्यों क्यों क्यों क्यों क्यों क्यों क्यों क्यों क्यों"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:42",
                    "end": "00:26:54",
                    "text": " प्रूर्डक्तिविटी साइकल्स आप चेंजिंग। प्रूर्डक्तिविटी साइकल्स आप चेंजिंग। प्रूर्डक्तिविटी साइकल्स"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:54",
                    "end": "00:27:23",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:23",
                    "end": "00:27:46",
                    "text": " विश्गौल्जी के लिएडिशिप के लिएडिशिप के लिएडिशिप के लिएडिशिप के लिएडिशिप के लिएडिशिप के लिएडिशिप के लिएडिशिप के लिएडिशिप के लिएडिशिप के लिएडिशिप के लिएडिशिप के लिएडिशिप के लिएडिशि�"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:46",
                    "end": "00:28:08",
                    "text": " If we don't do that, we are leaving 11,000 billion US dollars on the table. It's now larger than the combined GDP of India and China. It's no longer a choice. But how are you going to do it in manufacturing? How are you going to do it in education, remedial education? How is it that an enterprise solution will look at AI is now going to be a policy overlap?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:08",
                    "end": "00:28:31",
                    "text": " And the last thing that I want to leave you with is companies are no longer competing with each other. Countries are competing with each other. So we have to make sure that we are behind this ambition of the India IIM mission. We are going to work with the government. ASUCHAM will make sure that sectoral committees work and unlock this for you so that every single sector, whether it's healthcare, agriculture, ad tech,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:31",
                    "end": "00:28:56",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:57",
                    "end": "00:29:25",
                    "text": " क्यों क्यों क्यों क्यों क्यों क्यों क्यों"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:25",
                    "end": "00:29:55",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:55",
                    "end": "00:30:23",
                    "text": " आपका आपका आपका आपका आपका। आपका आपका आपका आपका आपका।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:23",
                    "end": "00:30:53",
                    "text": " आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:53",
                    "end": "00:31:23",
                    "text": " And it's making a measurable significant impact on how do they do things, how is it improving their life. So diffusion and impact will determine AI is real. For me AI for all really becomes meaningful if it starts consuming much lesser computer resources, much lesser power, much lesser water and becomes sustainable and also in some way becomes net job creative. Thank you and I think we are lucky to have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:23",
                    "end": "00:31:53",
                    "text": " या या या या या या या या या या या या या या या या या या या या या या या या।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:53",
                    "end": "00:32:20",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:20",
                    "end": "00:32:50",
                    "text": " अच्छानी विष्टित्ते हैं। अच्छानी विष्टित्ते हैं। अच्छानी विष्टित्ते हैं।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:50",
                    "end": "00:33:19",
                    "text": " आप आप अपने अपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:19",
                    "end": "00:33:43",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:43",
                    "end": "00:34:03",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका।"
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "699454cc25c9f4e0d7e70b57",
    "projectId": "698502",
    "projectName": "Founders & Funders The India AI Capital Ecosystem - English",
    "projectDescription": "Founders & Funders The India AI Capital Ecosystem - English",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T11:45:21.347Z",
    "projectUpdatedDate": "2026-02-17T11:45:21.347Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T11:50:16.259Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "Founders & Funders The India AI Capital Ecosystem.mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771328702417.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:28",
                    "text": " Can you use the mic? Yeah. So Vinod has braved all of Bharatmandapam to make it over here. Super excited to do this, what feels now a closed-door session, after a lot of braving the crowd. But very excited to have so many people in the room, founders, investors. The session is founders and funders, India's AI capital ecosystem."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:28",
                    "end": "00:00:56",
                    "text": " And when I think Rajan and I and Mohit and I first thought of this session, we only had one person in mind to kick this off for us, which was Vinod, and he gladly agreed. So very excited to have him here today. Guys, few rules. This is a closed-door, off-the-record session. You can take pictures, but please don't post anything from the session. In case there's media, please, this is like a saturated mouse, off-the-record."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:56",
                    "end": "00:01:33",
                    "text": " Maybe Vinod's available offline after this. Anything I say can be on the record. I don't care much. All right. Sorry. That's an edit. Okay. So, Mohith. So with that, I'd like to welcome Mohith to kick this off with Vinod. Mohith is MD at Peak 15 Partners. This is great. It's great to see the excitement with all the people outside."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:33",
                    "end": "00:01:52",
                    "text": " It's amazing how much interest there is in the subject nobody had heard about five years ago. AI can solve anything other than Delhi traffic. Well, AI can solve that too. It's just we haven't implemented it the right way. Hey, but before we jump into AI, we just want to say on behalf of everyone,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:52",
                    "end": "00:02:20",
                    "text": " Just incredible inspiration to the rest of us. You've just done wonders, not just being a founder that we've all looked up to, but then crossed over to the dark side, like I call it, and become an incredibly influential and impactful investor. So you make it hard for the rest of us, but thank you for inspiring. Thank you. OK. So we have a good group. How many founders in the room, if we can just get a read of the room? Wow. Awesome. Awesome. And how many investors?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:20",
                    "end": "00:02:47",
                    "text": " are willing to put their hands up. It sounds like the same people put their hands up. I don't know if you noticed. Well, you can be an investor and a founder. The best investor is somebody who's been a founder. Okay, I got lots of questions for you and then we'll open it up to the team as well. But we know we are sitting in Delhi, a city that's very special to you. It's where it started from you and you're probably sitting at the most important influential AI conference that India has ever had."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:47",
                    "end": "00:03:13",
                    "text": " And I want to start with that, which is your impression of India's AI sovereign journey, India's strategy to have a bank of GPUs, its own foundational models, clearly at the application layer. Take a step back. You're sitting with Prime Minister Modi, you're sitting with the ministers, and you're trying to opine with them around what should India's AI strategy be over the next decade?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:13",
                    "end": "00:03:42",
                    "text": " Well, clearly they're doing a good job. This summit seems to be unbelievably successful. I had one of my nieces wanting to register and she said 300,000 people registered. She couldn't register, which is surprising that many people interested. So it's very exciting. All the interest will cause more people to participate in the AI economy. So at the policy level, they're doing the right thing."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:42",
                    "end": "00:04:07",
                    "text": " The idea of a sovereign model and we are investors in Sarvam is a very good idea. In fact, I've long believed countries should have sovereign AI and that's why we invest in Sarvam here in India. And we invest in Sakana because Japan should have its own sovereign model and different countries should have their own models. The small countries, it's very hard to do."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:07",
                    "end": "00:04:35",
                    "text": " But India has done a good job of embracing this idea. Clearly for national defense and cyber security and all that, you cannot have foreign models. Every country will want its own models. So sovereign models are important. But I think the most important thing to do with AI is not so much in business. The first applications, and I'll be talking about this in the plenary session on the 19th."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:35",
                    "end": "00:05:05",
                    "text": " is very much it is possible today and within the next year or two in India to provide a near free AI doctor to every Indian 24-7 as an AI doctor. It's possible to provide every Indian child, all 250 million children in India, an AI personal tutor. That would be far better than if they could afford to pay"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:05",
                    "end": "00:05:35",
                    "text": " for a human tutor. Why? Because it can judge a person's level of knowledge, know where the gaps are, and cheat at the level of personalization nobody could. Those are two very fundamental things. I was just researching an agronomist. Every farmer in India, small or large, can have an AI agronomist available at the PhD level to them in their locality."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:35",
                    "end": "00:06:04",
                    "text": " speaking their language. If you're in Magalia, you can speak Kashi. And know the diseases and crops and pests in that area. So that level of personalization, I've just mentioned three that are absolutely essential and they should be part of the Aadhaar system. Like three, imagine UPI, primary care, AI tutors, AI agronomy agents."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:04",
                    "end": "00:06:34",
                    "text": " Those take care of essential needs for the bottom half of the Indian population. There's people who can afford personal tutors and frankly, they will have poor education relative to people who use AI tutors for free. Makes sense. Makes a lot of sense. Vinod, if I may just push on that a little bit though. But I want to add one more thing. You talked about you can't fix Delhi traffic. That's not true."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:34",
                    "end": "00:07:04",
                    "text": " That's a failure of imagination. We have a company that's building public transit based on self-driving vehicles. The interesting thing about public transit based on self-driving vehicles, it has 10 times the throughput. Why do you have congestion? Because you don't have enough throughput. Passengers per hour per meter of street width is how these things are measured. With AI-based..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:04",
                    "end": "00:07:32",
                    "text": " public transit pods. Service can be like Uber or Ola on demand. It can be point to point. It can be safe for women even at two o'clock in the morning. And it can be cheaper than today's public transit if you do the math. So these fundamental structural problems in the country can be addressed first before"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:32",
                    "end": "00:08:03",
                    "text": " We talk about the business side and there's lots of business opportunity. I suspect all of you are interested in that. I wrote a document that I published last month, but I actually wrote it last August on what AI can do for India. I forget the exact title. I can get it to you if you're interested. It is stunning the level of services that can be done in the next two, three years. Then of course, there's the longer term where"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:03",
                    "end": "00:08:32",
                    "text": " You know, it was in 2008, I looked at the question because my wife was working in education. I asked myself the question, what could you do for medicine or enough doctors in the country? I said, if I personally had a trillion dollars to spend, a trillion dollars, I could not to get the doctor patient ratio the same in India as it is in the US."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:32",
                    "end": "00:08:57",
                    "text": " With a trillion dollars and 30 years, I couldn't start enough medical schools, find enough professors for medical schools. It just was not doable. That's why in 2008 I decided the only solution was an AI doctor. That's how I came to this conclusion as I was looking at doing something nonprofit in this country in medicine."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:57",
                    "end": "00:09:20",
                    "text": " And in 2012, I wrote my first blog called Do We Need Doctors. I wrote a second blog called Do We Need Teachers, both in January of 2012 in TechCrunch, so you can find them for the record. So these are massive opportunities. And of course, you probably want to talk about the business side of things. No, but I agree with you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:20",
                    "end": "00:09:49",
                    "text": " The story of India has been that of access and getting distribution and the only way you can do it is if you're incredibly low cost, whether it's your mobile phone and lowest data cost in the world or even at e-commerce, the average ticket sizes that we do e-commerce within India and so on and so forth. So I agree with you that technology enables us to serve the next 50 million, the next 100 million because it allows us to drop the cost. Well, it's the next 500, 700 billion very easily."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:49",
                    "end": "00:10:18",
                    "text": " But I have a question for you. What do you think it would cost to provide daily primary care to 700 million people? My bet is less than a billion dollars or two billion a year. Even in India's context, that's a very small part of what the Ministry of Health spends. It'd be 10 or 20 cents per day at max."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:19",
                    "end": "00:10:49",
                    "text": " Separate question coming up later what we do with all our teachers, but but we'll handle that next okay Let's flip back a little bit Globally because you know a lot of people here are immersed into our day in day out in the AI world of India if there were three Important trends that you think are happening globally that are not as obvious to folks yet But is gonna hit them over the head in the next 12 months well, it's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:49",
                    "end": "00:11:10",
                    "text": " very clear to me that people in India don't believe that the whole idea of IT services will go away. By 2030, there will be no such thing as IT services. There will be no such thing as BPO. Those are gone, right? There will be new kinds of services based on AI that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:10",
                    "end": "00:11:35",
                    "text": " Indian companies can form and bring to the rest of the world because India has the best engineers and the best talent and education and all that. So there are opportunities, but those will be very, very disruptive to the Indian economy. And I don't think people are paying enough attention. Just somebody at Infosys is hating me right now. But they have to make the transition."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:35",
                    "end": "00:12:04",
                    "text": " And I've heard you say this before, and then you talk about the export for transformative services. What do you mean by that? So AI knowledge, so there's a massive gap between AI capability. There's a blog that was just in the last two or three days, I was reading it on the plane last night, called Something Big is Happening. It was an interesting blog."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:05",
                    "end": "00:12:33",
                    "text": " You know, I forget where he wrote it on Twitter or someplace else. And within a day or two, this blog had 80 million views. Even entertainment products don't get that kind of readership. In a couple of days, 80 million views. I think I'd recommend people read that. So the magnitude of the change, some of these areas are not."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:33",
                    "end": "00:13:04",
                    "text": " not really recognized by people. So imagine the world had 10 to 100 times more scientists in material science, in biology, in energy or power. We will have that capability for near free. So what do we do with that? There's lots of social problems. You pay a construction worker the same you pay an oncologist."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:04",
                    "end": "00:13:25",
                    "text": " who spent 10 years in medical school because they have the same knowledge, which is the knowledge of the AI. So everybody's up level to being a PhD in almost every subject that they can express in English language. That'll be the hard skill. What do I want from an AI? The specification of need is the hard part."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:25",
                    "end": "00:13:52",
                    "text": " It's a very different world. And by 2050, one way or another, that will happen. Go back to the trends that are shaping tomorrow. So one is you're saying ID services is in deep trouble. That's one entire industry that's going to not. But I want to understand, with all the data center investments and all the downstream memory and everything else with it, what is it that is getting built out that is going to impact us in 12 months later? Well, obviously, I think compute."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:52",
                    "end": "00:14:19",
                    "text": " usage will keep going up and it should go up. And if you give a billion people free education and free medical services or at least physician services, heart surgery is a little different. So there are things, but even, you know, we are starting to design drugs for one patient. If they have a cancer, we can design a drug for one patient."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:19",
                    "end": "00:14:50",
                    "text": " It's not an approved drug. You don't even have a regulatory process because you don't have a patient population. It's a population of one. So really amazing stuff why the AI pops it out. It's better than a five-year development process from one of the major pharma companies. So very interesting stuff happening. So look, has the AI let you down? What do you mean?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:50",
                    "end": "00:15:17",
                    "text": " As in, you're such a proponent, it's going to change everything. You probably are an early adopter of almost any of these, including drug discovery. Has it let you down? Look, I take the long view. I think it was in the year 2000 I first said AI will have us redefine what it means to be human. I was vague enough and sounded weird enough that I didn't elaborate more. I was very clear on my view."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:17",
                    "end": "00:15:47",
                    "text": " But 2012, I wrote two blogs. Do we need doctors, and do we need teachers? And I've been steadfast that's going to happen. And I think just this year, certain states in the US, Utah as an example, letting AI prescribe medicine, that'll happen this year. The American Medical Association doesn't love it because you don't have to see a doctor to get a prescription."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:47",
                    "end": "00:16:16",
                    "text": " It'll roll out very slowly in small increments. Start with the less dangerous drugs first. So of the 2,000 drugs in the market, 500 have very few side effects. So you start with those, start with the simpler, less complex cases, but you go up in complexity very, very rapidly. Let AI diagnose you and then maybe get a second opinion if you really care to. So almost certainly, I only use AI for diagnosis."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:16",
                    "end": "00:16:43",
                    "text": " And if a doctor tells me something, I get a second opinion from my AI, including dosage and stuff. In half the cases, and I probably have the best doctors available, I pay an ungodly amount for a personal doctor. Almost always the AI refines the things and will usually result in a change of drugs. That's consequential."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:43",
                    "end": "00:17:05",
                    "text": " Okay, you've got a dial in your hand. On this end is the entire Wild West as you're painting it out. Goodbye, doctors. Goodbye, educators. Goodbye, IT services. And on this end is... By the way, we should add the other business areas. So I talked about what I think can be government services. Yeah, please. Like medicine, doctors, tutors."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:05",
                    "end": "00:17:35",
                    "text": " I think legal services should be free so people have access to the law because most people who are poor can't access their legal rights and agronomy, so some of those. But entertainment will be free. It is so easy to have great entertainment. It should be free. You know, TikTok without creators is entirely possible if you use Sora 2 today."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:35",
                    "end": "00:17:58",
                    "text": " They don't have enough chips to keep serving all the people who want to use it. But otherwise it'd be all over. So entertainment will generally be free. We are seeing a lot of movies being made in one minute increments by AI. So if entertainment is free, all these services are near free."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:58",
                    "end": "00:18:28",
                    "text": " And I think transportation will be near free. The hard area is something like housing, which I haven't seen yet, how AI impacts, though I have plenty of ideas I'm working on. But these are fundamental human needs. If they all become hugely lower cost, it's a pretty interesting world. Sorry, where I was going with that was the policy makers, the government's got to balance this, look after the citizens, give them a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:28",
                    "end": "00:18:54",
                    "text": " call it a path in this new journey. And on the other end, we don't want to hold back and not restrain innovation. On which parts do you sort of completely dial right? Which ones do you sort of be a little bit more careful? Look, whenever there's large social change involved, there's large politics involved. So the biggest issue to be managed very, very carefully is politics."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:54",
                    "end": "00:19:22",
                    "text": " My general view is the adaption of AI will be very different in different countries, because the politics are different in different countries. In politics, a technical capability will be the key determinant of what is adapted and what's not adapted. Give me an example in the US where they've... Well, let me give you an idea. This silliest idea, I wouldn't even have imagined it's possible."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:22",
                    "end": "00:19:48",
                    "text": " Germany does not have retail workers on Sundays. So you can't open a shop and work in retail in Germany on Sunday. They just banned robots from working on Sundays. How ridiculous. That's an example of what not to do. I don't know the stupidity of whoever legislated that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:48",
                    "end": "00:20:18",
                    "text": " Alright, I'm an investor too and valuations are always challenging us in this new world. Sometimes we say to each other as early stage investors, our job is not to value companies as much as to pick the right companies because if it's going to get big, it's going to not matter. But the flip is you just get started and through a little bit of AI washing and you can get to hundreds of millions of dollars before you know it. Somebody walks into your boardroom or your IC, how are you able to tell?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:18",
                    "end": "00:20:43",
                    "text": " What's real, what's not. So we don't have an IC. So we are one of the venture firms that doesn't have an IC. We discuss every investment. We discuss it rare monthly. But after that, any of the managing directors can make their own decision. And I will disagree with something, and they can still go ahead. That happens all the time."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:43",
                    "end": "00:21:12",
                    "text": " We should try that, Rajan. Well, you have to have confidence in your partners. And you want to encourage diversity of views. At the academic level, everybody says you should have a diversity of views. But if one person's approving the investment, or if you have consensus, you're going to go for mediocrity. Consensus always means you missed outliers. So we maximize our process for outliers."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:12",
                    "end": "00:21:43",
                    "text": " There's no way everybody would agree to invest in OpenAI in 2018 when it was a non-profit with no product plan, no revenue plan, no notion of when the technology would work. It took conviction. It took guts. Big time. What did you see? Very simply, in Sam and the team, a really great team, and I was pretty sure..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:43",
                    "end": "00:22:13",
                    "text": " When AI is successful, not if. When it's successful, it'd be massive. It wouldn't matter if you made your return in three years or 15 years. The IRR would still be awesome. I was right on the magnitude of the AI impact. Nobody would have imagined companies worth near trillion dollars as startups. But it's entirely possible today. I'll give you my favorite story."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:13",
                    "end": "00:22:42",
                    "text": " this Indian company called Emergent. Anybody from Emergent here by any chance? They're all selling in the US. Approach me just like four or five months ago and they were a small company. They just launched their product in I think in July or August. They just sent me a note yesterday. Already up to a hundred million dollar run rate in a few months."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:42",
                    "end": "00:23:07",
                    "text": " In India, for an Indian company, that's just stunning. It was a controversial company because there's Replet, Cursor, and Lovable, and you name it. We bet on the founders. They're great founders. And it's going to be a massive success. Obviously, Sarvam has dominated."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:07",
                    "end": "00:23:35",
                    "text": " in the models. Yeah, let's talk about that when we're investors too. That would be a good one. Well, you know, you could easily have say Ambani or you know, one of the other big efforts with a lot more money behind it might be the successful one. That's not the case. It's great entrepreneurs. Entrepreneurs like Prathush. Yeah, amazing. Sorry, just one more example if I may."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:35",
                    "end": "00:24:05",
                    "text": " in your discussions with your partners, when you guys are now looking at new AI investments to make, beyond the founders, is there something structural? For example, in RIC, we talk about, do you believe in a multi-model world or do you believe in a single-model world? And if you believe in a multi-model world, then you have to take a set of certain directions. So one of the things we have, which is pretty different from most other firms, is we are deeply technical."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:05",
                    "end": "00:24:28",
                    "text": " I can have a detailed discussion. My partners can have a very detailed discussion. In fact, one of my partners who happens to be German and loves my robot joke about Germans, his self-driving car that he was the chief engineer of is in the Smithsonian Museum as the first..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:28",
                    "end": "00:24:58",
                    "text": " self-driving car that achieved certain milestones that won the DARPA competition. So we have that level of technical depth to be able to have those discussions. What do we believe? We've had intense discussions on robotics. We have five different approaches to robot models. I am my favorite, but you know, we will bet on different things and we will bet against each other."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:58",
                    "end": "00:25:26",
                    "text": " There's no competition in our firm, because nobody gets paid on the investments they make. Nobody has a quota, and nobody owns the investment. We have a very unusual process. And if I say to somebody, let me take over your investment, they'll say, happily, I have too many other things to do. You can take my workload. OK, I have one question of founders, and maybe I can take one or two questions from there, and then we'll carry on."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:26",
                    "end": "00:25:54",
                    "text": " It's a question you've probably answered before, but it's hard to be a founder in this very fast-changing world right now in terms of... Oh, it's so much easier. No, tell me more. Tell me more. Things didn't move quickly when I was a founder in 1980, my first startup, or my second startup in 1982. Nobody believed in anybody under the age of 40 doing anything interesting. Like, it was age."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:54",
                    "end": "00:26:24",
                    "text": " People were very ageist. It was like, hey, people with funny accents don't wear suits and ties, and I've really not worn suits and ties. Startups didn't have credibility. You always trusted IBM over a startup. Now, you always trust a startup more than any IBM company. If somebody's worked at Cisco for 15, 20 years,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:24",
                    "end": "00:26:55",
                    "text": " I consider them unemployable in the real economy. Anybody from Cisco or IBM or General Electric? Look, if you're at these big companies for 15, 20 years, you're so obsolete and so ossified, don't say anything. No, no, no. Now GE fits in that category. So get out of there. Now I have very strong views. You get ossified in big companies."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:55",
                    "end": "00:27:21",
                    "text": " You have to be at the edge of learning. I'm a big believer. I'm 71. I don't think I've ever learned as fast as I'm learning today, ever. I'm literally doing years of learning every month in every area. I can talk to you about AI models in antibody design for a particular cancer or about fusion."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:21",
                    "end": "00:27:42",
                    "text": " I'm happy to talk about any of these talks because I'm always learning and that's all I do. That's incredible. I want to eat with you. I don't talk to financial people. I don't ever talk at financial conferences. I talk to entrepreneurs. Wonderful. Anybody else? I mostly just end up insulting and they hate me. So a question."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:48",
                    "end": "00:28:17",
                    "text": " Hi Vinod, thank you for sharing. Hi, my name is Fargun and I'm from the UK. I studied medicine and neuroscience for years in England and then I became a film. Actually good, digital health startups in the UK. Yeah, no, and so now I'm, thank you, and now I'm actually a filmmaker, I turned into film. I'm actually directing a documentary on AI, but my question to you is... And I would bet if you asked an AI to write the script for a documentary about AI... Yeah."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:17",
                    "end": "00:28:42",
                    "text": " It would do better. Actually, I'm very religious. Yeah, I wrote my own and then I asked AI and yes, I emerged them, right? Okay, that's a good way. But my question to you is, so you mentioned we can make now one minute videos, right? So when do you think we can make a 90 minute, full on 90 minute Hollywood movie?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:42",
                    "end": "00:29:12",
                    "text": " full lip-sync everything, like from, for one person in one room instead of like 100 people. How, what's your timeline? Yeah, I think it's in probably the next two, three years. That will almost certainly happen. And there was a filmmaker that just got pinned because he used AI to build pieces of his film. I forget the name of the person who did it. But the question you have to ask is, do people want to watch a two-hour movie?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:12",
                    "end": "00:29:35",
                    "text": " The answer is probably not. China is a good indicator. What is the fastest growing category of media that people watch? The number of hours in China today, one minute shorts. But do you learn from one minute compared to like, do you watch documentaries, 90 minute documentaries?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:35",
                    "end": "00:30:04",
                    "text": " You still do, right? I don't. I like to have Chad GPT watch it and summarize it for me. I'm gonna keep moving. But this evening there is a session at I think 6 to 9 p.m. at the Qutb Minar where they're talking film and AI. You will not believe what's going on in Bombay and how many studios are actually trying to create a Mahabharat just using AI and AI only. There's a lot of stuff going on. My friend Makin right at the back. You can scream Makin."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:15",
                    "end": "00:30:44",
                    "text": " You know, that's a hard question. First thing I say is in the entrepreneurial world, you have to be really curious. And that's what I mean by I'm always learning. Any professor anywhere wants to come educate me on any topic, which they're often ready to, I will sit down with them over any business person always and say educate me in your area. And it doesn't matter whether it's fusion or"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:44",
                    "end": "00:31:14",
                    "text": " interplanetary science or geoscience, I will, you know, I'll read a book on how you do carbon dating. And my kids make fun of me, you know, reading of that. Or an encyclopedia of grasses. I have my favorite encyclopedia of grasses. I read about everything. So curiosity is number one to your question. I think the second most important question is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:14",
                    "end": "00:31:43",
                    "text": " agency. People who have agency. There's people who think that's how the world is and there's people who create the world they want. You know, I joined IIT Delhi in 1971. They want the single IIT with a programming class. So you couldn't really, people said you can't get the administration to do something in 1971. We started a programming club with four people and one junior professor."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:43",
                    "end": "00:32:06",
                    "text": " we made programming happen. A few years later, I worked with Professor Goa at IIT Delhi in those days to make a biomedical engineering program happen. Why? I got, we created the program and we created programming because I was curious about it and I had the agency to say, hey, let's do it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:06",
                    "end": "00:32:33",
                    "text": " We started working with IIT Delhi's electrical engineering department and All India Institute of Medical Sciences. Agency is very important and then anybody who's done startups knows startups are really hard. You have to have persistence. You don't give up when you fail. I always say my willingness to fail is what allows me to succeed. I'll bang my head against the wall if I think something's a good idea."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:33",
                    "end": "00:33:09",
                    "text": " till it succeeds. So persistence is really key. So those are three quick things. That's great. He's doing a great job, by the way, with many Delhi schools and inserting entrepreneurship as part of the curriculum and seeing great results. We'll take one here. You know, so if you ask me 15 years from now, I don't see any need for universities and colleges. I think there will be other ways to learn that are much better on your time."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:09",
                    "end": "00:33:38",
                    "text": " without the expense of a university and the time dedication. So I don't see a huge future for universities in these universities with 100 year endowments, just they'll have their endowment left, not much else. Yeah, that's a good use, but there's other ways to do that. And I think in high school,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:38",
                    "end": "00:34:05",
                    "text": " That is the principal, or even in kindergarten, that's the principal purpose of school is socialization. And I think that's very important. Whether universities do that or are as needed for that function, I think by 18, most people are independent enough, there's other ways to socialize. For an eight or 10-year-old,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:05",
                    "end": "00:34:32",
                    "text": " That's much harder and so schools are important for that. Is there any industry, you said housing was one, that you were still trying to figure out how it gets impacted. How about healthcare? Is there any one or two industries that are more insulated? You know, first, every time I think of an industry, I find that somebody finds a way to obsolete it. You know, take minerals and mining. I thought, you know, we need..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:32",
                    "end": "00:35:01",
                    "text": " Everybody in development, you need more steel, you need more copper, you need more lithium. Almost certainly, over the next 15 years, we will discover more minerals than we will consume. Why? Because AI is helping discover. Today, how do you find a copper deposit? It's on the surface in some part of the planet you start digging and refining. If you could look a kilometer under the Earth's surface,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:01",
                    "end": "00:36:01",
                    "text": " AI lets you do special kinds of sensors and I can get into every kind of sensor that's used in geoscience and I actually know it reasonably well. You'll find resources everywhere. So I don't view those things as a constraint. Great answer. Yeah, we have an entrepreneur who cannot sit down. He really wants to get asked the question. Go for it. I feel sorry for you five years from now."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:01",
                    "end": "00:36:31",
                    "text": " But keep going, sorry. Sorry, did you have a question or was there more for comment? Oh, okay, that's not here, not here. I have the Mike Anna question. Yes, yes. And it's a non-technical question, you know, that's a military brat. My name is Seema Jatavedi, I run a venture fund, big bullish on the AI. But non-technical question, what are your views on the future of humanity because of the future of AI? What are all these people going to do when they have no jobs? What are your views on this, especially if they don't have agency?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:31",
                    "end": "00:37:02",
                    "text": " Okay, the question you have to ask is will anybody need jobs, right? I think by 2050 it'll be very clear nobody needs jobs because enough production of goods and services that are near free, you won't need jobs. Now that's not to say people won't do jobs. People have much more time to spend with their kids, with their parents. So caregiving is a really"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:02",
                    "end": "00:37:27",
                    "text": " interesting use. Competitions, the Winter Olympics is going on right now. People love competition. People love singing. People need art. If you're a painter, I used to tell kids, go do that, but make sure you don't need to support your, need a house or to send your kids to college. Well, you'll be able to do any of those things."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:27",
                    "end": "00:37:57",
                    "text": " whether it's painting, or surfing, or skiing, or Olympics, and dedicate your life to whatever passion you have, and the number of things people have passion for will explode. So let me give you my favorite example. If you have a five or six year old child, maybe a 10 year old child, you say, go to school, study hard, get into good college, you'll get a good job. In the future,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:57",
                    "end": "00:38:29",
                    "text": " you will tell that kid, learn a passion and explore it. From age five, they will know it's not about getting life. Education isn't about getting a job. It's about developing curiosity and having passions. And I think people, humanity will generally pursue passions, not jobs. I think all those, so I suspect the minimum level of services, healthcare, education,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:29",
                    "end": "00:38:56",
                    "text": " Legal food, food will be very cheap because of bipedal robots and not needing labor. And I can go into much detail. Housing is the one I mentioned. I have a trouble with saying how it gets to near freeze levels. But the basic level of services afforded, everybody will be very high."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:56",
                    "end": "00:39:19",
                    "text": " by governments that do this right by 2050. And this is only 25 years. It's in my working life. I hope I'm working at 95. I hope it's in your working lives, most of you are younger than me. I think we are looking at a very different world now. Very unpredictable how the path is from here to there."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:19",
                    "end": "00:39:49",
                    "text": " People will object to unemployment and lots of other things and social causes. So that's why I say politics will play a role, especially in democratic countries. And I think China has a huge advantage there. Last question, please. What are your thoughts on AI potentially starting a nuclear war? And if yes, if it's possible, is there a way to protect against it? I'm much more worried about Trump than AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:49",
                    "end": "00:40:16",
                    "text": " Now that is the final word, really. I have the mic. Are we actually done? Yes, unless you have one last question. I have one last question. Prediction. Which vintage do you think AGI actually becomes a reality? Best of your knowledge. So it depends on how you define AGI. People keep moving the goal post. I define AGI very specifically as when AI can do"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:16",
                    "end": "00:40:43",
                    "text": " 80% of 80% of all the jobs that have economic value. So whether you're a structural engineer, farm worker, assembly line worker, a doctor, an accountant, if AI can do 80% of 80% of all jobs, that's AGI. I think we are there in the next two years. Wow. Awesome. Thank you. Thank you, Mohit. Thank you, Vinod. Thanks, Vinod."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:45",
                    "end": "00:41:19",
                    "text": " We're not going to do more sessions. We're out of time. But why don't we all take a picture? Pratyush, Sharad, just one second. Come, come."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
 {
    "_id": "6994508925c9f4e0d7e709ea",
    "projectId": "145745",
    "projectName": "Governing Autonomy_ Trust in Agentic AI, Multi-Agent Systems and the Infrastructure of Trust - Spanish",
    "projectDescription": "Governing Autonomy_ Trust in Agentic AI, Multi-Agent Systems and the Infrastructure of Trust ",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T11:27:02.238Z",
    "projectUpdatedDate": "2026-02-17T11:27:02.238Z",
    "projectStatus": "Active",
    "companyId": "69453bf2e603a0a401b6f35b",
    "projectCreatedBy": "69453bf2e603a0a401b6f35c",
    "projectUpdatedBy": "69453bf2e603a0a401b6f35c",
    "projectLastAccessDate": "2026-02-17T11:27:02.238Z",
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "Content",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "Governing Autonomy_ Trust in Agentic AI, Multi-Agent Systems and the Infrastructure of Trust.mp3.docx",
            "col3": "We'll say probably a few words about what they do. I'm sorry, but Eddie Sahai from Google. Go ahead, Eddie. Hi, everyone. I'm Eddie Sahai. I am policy manager working on AI and emerging tech, particularly agent AI and robotics. Hello, and good afternoon. This is Mahesh. Thank you all, and thanks to Amir and Pauline for having us here. And also, they are generous enough to put my younger version. in the picture. And so I'm from Project Nanda. We are pioneering the building and foundational infrastructure for the internet of AI agents. It's a new kind of internet is born, or getting born, from last year onwards. And one of the most foundational issues, the way I say that, trust, that is me. That's my photo. The same thing, how we are gonna solve this problem using in this internet of AI agents, and those primitives need to be solved at the foundational layer. It's not like after the fact or after the thought solutioning. It has to happen as a first principle. Yeah, thank you. Alpesh. Hello, I'm Alpesh Shah. I am the Managing Director of the IEEE Standards Association and a member of the Management Council at IEEE. Thank you. Thank you. Hey, guys. Hi. My name is Apoorv Goyal. I'm a principal at Insight Partners. I help lead a lot of our US-India investing efforts. Just as context, we're one of the leading global technology investment funds. We manage close to $90 billion in AUM investing out of a $12.5 billion fund into leading global software companies around the world, including a bunch of companies based out of India. Wonderful. Thank you so much. I'll just be getting to Praveen. He's my co-moderator. Praveen is the head of AI Lab. but before we start talking about it, with that, as a raise of hand, how many of you have built an agent or played with an agent? How many of you are scared of what's gonna happen with agents? One is probably like 5%, 10%. So it thinks that 90 or 80 or 90% of you don't feel that agents are gonna be dangerous or anything else. Have you followed what happened with Open Cloud recently and Notebook? Do you feel this is cool or exciting or energizing or fun? Who's excited about what happened? It's a beautiful experiment, right? So we're learning about all these things and as we talk about agentic world, the concept of autonomy scares a lot of people. Why? Because we don't fully know how they're built. We don't fully know how they're going to react, if their objective is going to be changing, if they're going to be... mixing with each other, including new babies, we don't know what's gonna happen. And then in that world of autonomy, a lot of decisions that enterprises take and societies functioning may not be able to be put in the hands of agents. So the concept of governance, which is a concept which is very encompassing about how we control the outcome of systems and procedures and ownership and risk management. becomes central again in the world of agent, which was not the case about even a year ago. So today we're going to be talking about this concept of governing autonomy, how we can actually make sure that this world, this future world, we're going to be fully autonomous, not only is our hand to some extent, but we can somehow trust it, and we don't have to really extend. So to get started, maybe we can have a quick demo. What do you think, Praveen? Questions? Okay, perfect, that's a good thing. So maybe I can start with Eli. Eli, in your vantage point, can you help us understand what's different between multi-agent environment and systems and traditional AI that we've been used to? How should we think about it? So with the raise of hand that I saw in this room, I think we are talking with a group of experts. From my perspective, and many of you have already worked with ModBug and have seen OpenClaw and have built agents, so you know better than me that the way that we think about agents, it's not a specific line or a specific point in time that these group of things are agentic and this group of things are non-agentic, rather thinking about them as a spectrum of agency or autonomy going from basic chatbots that can do a little bit of research and at the end present the result. For example, Google Deep Research does a research given your prompt and finds the information and presents that information. That has some agentic features, but doesn't take actions. compared to let's say going all the way towards the end of the autonomy that is, for example, autonomous cars that are completely end to end and they take action in the real world. So I think seeing agents as a spectrum or a continuum of autonomy across different dimensions, including memory, including planning and long-term planning, short-term planning, planning horizon that they can plan. execute things and also autonomy itself. So what I'm trying to say is that we don't want to say agents are the ones that take actions. We want to say agents live on a continuum from very basic autonomy all the way to things like autonomous vehicles that are fully end-to-end autonomous. Mahesh, you started talking about interesting issues, but let's follow what Ellie said. How do we know agents could make an error? How can we actually even know? Yeah, so she used a very nice word. It's the continuum, right? So it's not like a single point responses that you can keep monitoring, you know, like after the agent took an action. And another thing that she mentioned, they become network actors. They will be acting at every single stage. visibly or invisibly. And that makes it mandatory for us to make it, I would say that it has to be a runtime governance, rather than wait for, let it happen, and then we reverse. So the traditional governance models that we have or exist were very much evolved from our understanding of machine-to-machine interaction. Then it evolved into a SaaS model and a cloud model. Then we evolved to the microservices. architecture and we kind of assumed that you know like somebody else is taking care of those governance after we reach at a scale you know like so when autonomy will reach at a scale then we cannot think of like okay we need to solve this problem now. Now we are thinking of like agent as like okay Chad GPT is one of the agents I'm just looking for the search answers. rather than, and slowly we're saying that, oh, okay, it can just go and do some action, it can crawl some web pages. Okay, then we say that, okay, then there will be multi-agent systems, agent to agent interaction, collaboration, that will happen, and just we are adopting it without addressing the foundational issue that the safety and the governance has to be addressed, that's one, that, along with the trust, but also to make it, like, how it can become a deployable. across the enterprises or the public services or the civil society without sacrificing, which is a very important factor, without sacrificing the openness of this execution and the accountability. So those are the two points I would like to mention. So you gave us more information than I asked for. Thank you. incident that happens, of course, we're going to remove bad actors, right? People that are using agents for bad purposes. We're not talking about that, which is a different issue, security, cyber security, or control. We talk about agent themselves making an error or having a mistake. Some of you know that we're talking recently about agentic identity as a way to make sure that who are these agents? Where are they coming from? What's their training? What data have been used? Who created them? who they belong, they have their passport, how they're gonna talk, what language they speak, can we talk with each other in some language and so forth, which brings now the concept of certifying agents. And I'm gonna ask to Alpesh to question, because Alpesh in the IEEE stands, and we're working a lot on many, many concepts ahead of his time on energetic, how do you think about certifying agents in the world of autonomy? How should we think about it? I think this works out. I guess that's one of the most important points, right? Knowing when it's on and when it's not. I say that in jest, but this is also part of when we're thinking about these various models and how they interact with each other. Not having the ability to understand what is the end result or what good looks like creates a difficulty in understanding the parameters by which you can state this is working as it should. Which means that very much, and already we've seeing this done quite successfully in a number of the use cases we've been involved in and prototypes and implementations around this that we've done is that First making sure you have a very strong governance structure in place is critical second is making sure that you have a very clear process by which you are determining what does and doesn't go out or is ready to go into the sandbox or what you're ready to even put into that runtime state, as you were sharing earlier. It also requires ensuring that you have an iterative approach to it. These are not one-time exercises because you won't know necessarily when there might be a shift or where things may go in a different direction, right? And so this requires sort of a constant monitoring approach. Adding to this is also the need to contextualize a lot of what you're looking at. some of the reasons why you may see the behaviors of the multi-agents may be because there was a trigger that you typically do not see or are unaware of, which means that the bounds that you currently had in mind may not be the right bounds. And therefore, it's important to also ensure that you have transparent, accountable, not only logging, but people in that process. Ultimately, at the end, what it is that you can evaluate, let's say, at a much more level of confidence is the means by which it achieves the output. You won't know 100% if the thing that is out there is perfect. But over iterative series, you begin to clearly understand, right? This, out of the Monte Carlo simulation or whatever model you choose to apply here, at some point you have enough data to know. what good should be, right? So our certified program allows for this. At the same time, we have a series of standards that Amir was instrumental in being involved, and a number of folks in the front row here as well, which are focused on data transparency, focused on age-appropriate design, focused on also the critical elements around what does it mean to be accountable, transparent. What does data privacy mean in the context of systems that we don't know what they were to be, right? And what we found the greatest power in is people. It was the community of people that worked on this. They were heavily representative of a mix of lawyers, doctors, artists were even involved with, engineers, and technical people that really understood the contextual problem so well. that it allowed for us to better understand and appreciate what some of those variances might have been. Thank you, Alpesh. Apurva, you invest in many, many startups and companies, and you are in the first line to evaluate them and see if the company is going to have solid technical background and the teams is right. But because Alpesh mentioned the terms of governance, we're talking about governance and governing AI, how do you imagine because we talk about the technical stack, we're all familiar with the technical stack, but what would be a governance stack look like when everything is autonomous? I think the way we think about governance stack in this AI world is a 5K clear model. If I kind of go first is like the build time. Around the build time the idea is okay, how is the company architected around the idea of data governance, model versioning. when you kind of come to the next step, which is basically deploy time. How do you think about the idea of policy, tracking, permissioning, secrets management? When we come to the third piece of it, we think about this whole idea of runtime. How are you ensuring that your architecture allows for real time observability? How do you ensure that there is an idea of kill switch? If things go bad, how do you ensure that you're able to cut it off at the right point of time? Then we think about the whole idea of How do you drive remediation? So do you have the right audit trails to be able to do a post-mortem? Do you have the right incidence response architecture to be able to attack that? And then how can you drive a whole accountability layer, which is around the whole idea of who drives accountability? How do you have the right reporting structures? How can you conduct right post-mortem? Do you have the right set of tools to be able to do a compliance mapping there? So I think we think of governance stack in today's world in a holistic 5K clear model and anybody who thinks about this holistically is where we feel comfortable about the fact that okay, this is the team that is truly thinking about governance in a holistic way and as you think about the best AI native teams, governance is becoming less of a compliance issue but more of like an issue which is new product in GTM because whoever is embedding governance into product is truly winning on the GTM side. And that is becoming a competitive advantage, and so it's becoming a key part of how we evaluate a lot of the companies. Some people jokingly said a few years ago, 10, 20 years ago, we talk about intern insight when the PC era was that. Now we're talking about governance insight. Would you agree with that? 100%. I feel like with so much of uncertainty, I feel like every enterprise that is procuring today, like the upfront conversation before they take on any business through multiple layers of evaluation is talking through a bunch of questions around what is your auditability, what's your traceability, what's your data handling practices, are the kills which is in place because the costs of things going wrong are so high that people are willing to spend millions of dollars in ensuring the right governance while even buying a half a million ACV contract. So I feel like it's becoming a core part of... So how startup, sorry this is a question I didn't plan to ask but how startups that don't have all the funding and all the resources necessary could comply with this level of scrutiny or requirement? I think the way I think about it is when you're starting an agentic AI business in today's world, I call it like you need to have a minimal viable trust stack. You need to be at a high level, you need to be able to tell what is this agent supposed to do? Is it actually doing what it's supposed to do? And if not, can you actually have someone stop doing what it's supposed to do on a real-time basis? And that becomes critical and for that you basically need to have what you call as a clearly defined agent identity registry. You need to have the right set of guardrails at the orchestration layer. You need to have a clear real-time observability architecture. And lastly, you need to have very clear set of defined oversight. I think if you don't have any of this, I don't think you should be launching any Gentic ecosystem into production because you're bound to meet more failures than success and it's not that expensive to do these four basic things and of course as you scale you'll have a lot of compliance start building into your product but this is the minimal viable thing that any startup that's looking to start in the Gentic AI needs to do to be able to go into production. I think you should write an article about that because a lot of us would need that little four stack layer to get started. To use the term orchestration, I would like to pass on to Praveen as we talk about orchestration and coordination. Because as we define multi-agent systems, they need to be coordinated in some ways to do certain things, but also to not do certain things. So Praveen, is you ready for a demo, quick demo? I think it's very tiny, so maybe you can explain. I truly believe that the next phase of AI is going to be Or the agent system. Am I audible? Sorry about that. I thought this was working. So we at Cognizant and Cognizant AI lab truly believe that the next frontier is going to be multi-agent systems. And in that endeavor, sometime in May 2025, we sort of open-sourced our multi-agent framework, which is called Neuro AI Multi-Agent Accelerator, which you can see on the screen here. It's a low-code, no-code framework. that's designed to accelerate the prototype development to scaling. It's a production-grade environment. Currently, it's available under Apache 2.0 license on GitHub. There's an active community that is sort of working on building and enhancing this framework as we speak. Some of the features and the capabilities of this accelerator is that it's a little bit agnostic, right? It's cloud agnostic. So you as a user will have the control over which LLMs that you want the agent network to talk to. You have a control on which environment that you want to deploy on that. And it's interoperable. I mean, currently, we support multiple protocols from an agent to agent communication, our own OSR protocol, as well as A2N MCP server. And most importantly, as we heard about the governance, autonomy, and the guardrails, security is built into it. So we build this from security at the core of it, not as an afterthought. So in the next one minute, and most importantly, we've made it so simple that even a seasoned AI ML engineer to a CXO who has minimal or no coding experience can start building their agent network in a matter of minutes. And that's all. The whole reason is that we want to make this as a de facto standard. and we want everyone sort of adopt it and sort of, you know, not just experiment it, actually build those agent networks at scale and deploy them, right? So let me just do a quick demonstration. And today, what I'm going to do is we'll quickly wipe code an agent network in front of you guys. So this is a wiping environment. So let me just, anybody wants to give me a prompt? Anybody? Anything? Anything you want prompted? Yeah. All right. Can you read, because it's not easy to read. Yeah. I'm saying create a multi-agent network. I'm saying create a multi-agent network for India AI Summit New Delhi, right? That's what they asked. OK. So what you can really see is just give it a few more seconds. What you will see is that there's an hour. multi-agent architect and a designer which is working behind just to understand the intent. And the intent that I gave was just create a multi-agent network for India AI Summit. New Delhi, that's currently happening. First thing that you will see is that the large language model understood my requirement and it created a set of agents. It's defined the agents. Now what it's actually doing is that it's actually connecting those agents to an orchestrator. It's still building through it. And then as a next step, what will happen really is that basically, now every agent needs to be defined. It needs to have its own roles, responsibilities, and what it should do and what it should not do. So currently, that's a step that we are actually going through it, where you can see that the matter of moment, all these individual agents will be populated with descriptors. You can see that, right? So those descriptions are nothing but its definitions, what it should be doing at the end of the day, right? It's still finalizing the agent network. The two things that you need to see at this here is that there are agents which have been specifically created for that agent network. And then there are a few of them which are under the leaf nodes, orange in color. So basically what the designers understood that there are certain agents which are already created and they were available in my environment, so it just connected them directly. So instead of rebuilding them, reinvesting in creating the agent network, it just connected them so you could use it. So it also allows as a designer of the agent network is that you have the control, auditability and traceability of it. At any point you think that that leaf node is not required in that agent, you can just delete it and save it, the server understands the changes and your new agent network is ready to be played around with. So now that the agent network is ready, What I'm really going to do is launch it. You can see that the agent network is made available. Again, just let me blow it a little bit here. And before I fire a query, and again, this is a development environment. The reason we are showing a development environment is just to showcase a couple of capabilities. What you see on the right-hand side panel is basically your chat window. And then you have an internal chat. which basically shows in real time the communication between the agents, what kind of information is being transmitted, what the nature of it is. So you as a designer have the control over it as well. So you can take a dump of it later on and you can sort of go through it, review it to see if any confidential data is being leaked through a large language model got access to it, right? And then you have the logs. The logs are pretty granular in nature in the sense that it gives you details of a level of the token usage and the cost associated with each of that communication. And last but not the least, I mean, most of the folks here are pro ESD metrics, right? So you also get a score of energy usage, carbon footprint, and the cost associated with each of the prom that you are filing. So let me just, so I'm asking the agent, it goes, can you provide a schedule for the keynote speeches at the India AI Summit, right? So I think it's. OK, let me see if I can get. So currently, it's not grounded. It's just talking to a large language model behind. So every agent network that we create here, by default, it talks to GPT-4. But you, as a user, have a control. As I said, it's LLM agnostic. So if you in your organization have a qualified large language model that you want this agent network to, we can. So globally, each and every agent network has access to one large language model. But the framework has the coolest capability where Every leaf node agent also can be empowered with two specialized or general purpose agents, LLMs as well. So they act as a fallback. And if you have a specialized agent, like for auditing or governance or compliance, you want higher accuracy, you can sort of connect a specific large language model just to that leaf node agent as well. So that's sort of the, and today it's available on GitHub. It's been there for the last five months now. We have an active community. The GitHub repository gives you access to the code base under the Apache 2.0 license. And this framework has met the highest standard of GitHub security as well today. So that's an example of a... You have no queries to run? Yeah. So that's an example of a coordination which basically takes into account all the considerations that was discussed. And you said there's open source, right? Yes, it's open source. So here is the QR code. And if you need more, here is a QR code that you can sort of scan. It will directly take you to the GitHub repository that we have. We encourage you to fork it, clone it, and if you like it, give us a star. And if you don't like it, let us know why. And we are here to work with you on that. I just have a question. If you have multiple agents running at the same time, All of them have different contexts and different memories. How do you manage that in a single platform? Good question. In the interest of time, let's come back to it offline. But we'll get into that. We'll get into it. Great question. So that was just a demo, too, basically. Because the audience is diverse, choosing what these things look like. But this has been certified ISO 4 as a pneumonia compared with this framework, as well. So there's a lot of that behind the scene. But I think what I was talking about has been taken into account in terms of observability, and monitoring, and so forth. Going back to the issues of incident and tracking incident and danger that could happen, there are examples in other areas such as aviation and nuclear power, et cetera. Maybe Ellie can give us a view of how the other industries have dealt with these issues because this issue is, today we talk about agentic autonomy. We have these issues before in other sectors. How have we dealt with them? Absolutely. The technology may be new, but the concept of safety, especially in safety critical industries, is not new. We have seen that before and we've solved for it. Let's take the example of aviations and drones, for example. So drones are a newer technology and a lot of regulatory bodies are looking at regulating drones. And the interesting thing about looking at drones and comparing them with agents is that, when you look at, let's say, the regulations that, at least in the U.S., aviation industry is considering for drones, it was originally based on what FAA or Federal Aviation Administration calls VLOS, meaning visual line of sight, meaning that a pilot has all these is in command and they have to keep visual line of sight with the drone. And that is how they are accountable for managing that drone. However, as the safety and as the system safety based on using AR technology develops, now the VLAS regulation is moving to beyond visual line of sight. And that means that now the safety of an AI detect and avoid system has gone above what a human can do. And based on this advantage of detect and avoid based on AI system, now the regulation is moving from a pilot has to always keep visual line of sight to beyond visual line of sight. And using that as... as a way to maintain safety. So drawing a comparison to agents, if you think of, when we talk about human in the loop, human in the loop, we are basically talking about the loss. The human has to keep approving every step of the way that agents are taking. However, that undermines the utility of the agent. So as agents become safer and more reliable, similar to what detect and avoid in the aviation industry, and got better and better and better over time, we have to be moving from human in the loop to pilot in command in aviation industry. In this case, we'll be human in command. So human will be doing the supervision, but not in the loop, approving everything that agent is supposed to do. So drawing this kind of parallels with respect to other industries that are moving from safety of human. making those decisions make a lot of sense to apply to agentic AI because similar to aviation, in this case we see that maybe keeping the human always in the loop is not the best thing. Similar to detect and avoid, there could be better safety systems that AI can provide that keeps agents even safer. I'm not saying that data-wise today, I'm saying similar to other industries that moved from one side to the other, agentic AI would also move. Now we are human in the loop, and then we have to move towards human in command. Thank you very much. As you talked about the similarity between drones and agents and FAA as a regulation, I'm going back to the rest of the audience, the two of you, to ask a question. In your opinion, what's the balance between engineering guidelines and technical design at the core versus regulations and norms and standards, right? Because both of them are needed, we're not discussing that. But today with what's going on with agents, in your opinion, what's the balance of that? Should it be 80% engineering design and good practices, like observability, measurement, et cetera, or should it be probably a little bit less controlled in engineering, let innovation go, but also provide a framework of certification and standards. How do you balance these two? So I'm having an analogy with the early days of internet, like how internet even evolved at first place, being sure that it's not owned by individual player or anything. So it was all inclusive and very much focused on solving the foundational primitives like identity, discovery, trust. and that was in a fragmented system. It was never a centralized system. So in this context, like when you said, okay, so compliance, it should be, heavy compliance will kind of sabotage the innovation, like how the tear-off we will do, and my opinion on that is we should keep the innovation open, and that remains as a data plane. But the governance has to be the control plane. So they have to build in parallel. We cannot choose one versus other. So yeah, that's my thought. I'm curious to hear Apish's and Apuva's opinion on that as well. Yeah, I mean, I think, Amir, if we took some real world examples, there's some policies regionally that some have claimed were too early in the days And that have led to unnecessary regulation. And the effects of that unnecessary regulation have compromised, perhaps, the innovation that could have emerged as a result. I'm not so sure whether or not I agree with that. I think I'll let time claim that one. But what I would say is that it really just depends what you're trying to do. If you're in early days, you're just trying to understand what is possible and the bounds within your own sandbox, you should be able to test it out and understand. You need the chance to even observe it before you can jump to the conclusion that it's something that's very good to market. At the same time, when you get to the proper maturation point, as my colleagues have shared, this is when regulation starts to make a little bit more sense from a requirements standpoint. Policy can also function, though, in an earlier stage to begin framing. And the framing is important to know just so you have a sense of where is too far, right? But framing is different than regulation. And at a certain point, as I'm sure Pruva knows as well, is that there comes a point where there's a market acceptance of what is normalized. And at that point in time, That's when the market differentiation really starts to play a role. That's when you see the major uptick. Prior to that, you're not necessarily going to see everyone buy into paying $100,000 or $150,000 to get something certified. The costs become a barrier to entry. But this also means that there is perhaps a different way of approaching this. I go back to the comment Amir and Apurva were speaking to earlier of the mini trust stack. You know, in many ways, we can also consider the economic problem of how you make it easier for startups to come into play. And one of those ways could be, let's say theoretically, but maybe some of us are working on this, is the integration of the required standards and policies baked into the fabric from the get-go. And then as you build and stage get beyond that, what you've already done is you've bought time. for those startups. You've also given a level of trust for the governments as they're playing here. And you've given regulators at least something much clearer to measure against. It's very, very difficult to ask someone to measure something and hold you accountable when they don't know what they're supposed to hold you accountable against. And it's the same problem for companies. They don't want to waste time on something that makes no sense in all this market play. So Amir, I hope that helps answer your question, but that's my take on it. Thank you. Can I ask a follow-up? Please go ahead. I may want to answer too. In your experience... Alpesh captured most of what I said, so I would save the time. Actually, I wanted to get your perspective on how you see founders actually balancing this. Again, I feel like the idea is everybody's working with the certain idea of what that framework looks like. And I feel like... there is not a lot of mutual tension between deploying responsibly AI and what leads to high commercial deployment. It's like if you need to be a half a billion ARR business today, you need to be able to sell into enterprises in mid-market. And those mid-market solutions are selling into either government institutions, large-scale enterprises, and all of them today have a much larger business to protect. And so if you're experimenting with a new technology that is evolving very, very fast, and if you're not able to provide for certain frameworks of those responsible guidelines, it's like, hey, I wanna make sure my data is being handled well. I wanna make sure my secrets are being managed in the right way. The agents have the right set of permissions and policies. Those guidelines are being, and I think today those frameworks are very individual. are localized to the company that you're trying to serve. But if you kind of distill that, those will broadly come down to like 15, 20 large principles. And most of the innovation that is happening is happening within the bound of those 20, 25 frameworks. I think to Altesha's point, that is going to start solidifying a lot more as the technology matures. But I think people always try and frame being responsible and commercial innovation as a trade-off, I feel like with the way the world is moving, it's not a tension. It's actually very, very aligned. And that's how even most of the board conversations happen is because if you can't do that, you would not be able to scale. So on that note, staying with you, Apurva, right? You talked about the trade-offs. You talked about the investment, right? If we switch gears into the economics of it, right? So when do you see the governance actually becoming a competitive advantage and not a cost burden to the organization? To be very honest, in very early stages of the company, we have a lot of our portfolio companies at seed series A stage start selling into enterprises day one. So if you are selling into the cognizance of the world or if you're selling into the Googles of the world and all of that, a bunch of these companies today and just how innovative they are and how fast they're moving, like day one they sell into. a half a million, a million ARR contract into a very large enterprise. And when you're doing that, you can't do that without governance. Like SOCTU used to be that version, SOCTU compliance for like your old age SAS framework there. So I feel like most of these conversations become pretty much very, very embedded in your seed series. The moment the company's about to sell, if governance is not an embedded conversation in your product, it will not be a competitive edge for you in the GTN. And to be honest, I meet 100 AI companies a month. The best AI native teams today have actually embedded it so deeply that, like security, it's part of the weekly operating cadence. They'll talk about the idea around evals. How solid are your eval reviews? Are you doing it on a weekly basis? Are you making sure that... meeting the standards that you want to espouse serving into an enterprise customer. They're thinking about red teaming in a very solid way. They're doing very high quality post-mortem reviews if things don't work the way they are supposed to work. So I feel like governance is like security today and they will start playing out pretty much in a series A conversation if you're selling into an enterprise. I feel like that conversation becomes later if you're a PLG led business. that is serving individual creators, individual designers. I don't think if you are trying to buy a $50 per month contract as a designer and using that to create videos, you're thinking as much about, hey, is this meeting those governance standards that I need to? But anyone selling into a mid-market enterprise solution, this becomes a very early on conversation. I have just a follow-up thought on that, right? So our India's digital infrastructure story, like when we built Aadhaar, we built UPI payment system, we solved identity problem, we solved the payment problem, then we were like, okay, what about the data privacy, then we have the DG Locker, then we have OCD, and then, now what is happening, that's very important success story, and very much aligned with this agentic AI, because, the government took a stand saying that, hey, we are going to standardize the interfaces, okay? And what we really want to do, the rest of the people or the companies, they can do any application that they want on top of it. So that government plays a very important role in this agentic AI evolution, I would say. Thank you. Thank you. At the beginning, we talked about governing autonomy and we discussed about, autonomy talks about governance to some extent, I think. What would be, in your opinion, well, there are two aspects to that question. One is if we count about trust, what is infrastructure for trust? Because it's so fragmented between different pieces. And then how should we think about the one single thing that would enable us to go to the right direction? Because we talked about, again, many things, but in your opinion, what would be the one single thing that we have to do today, given what you expect? What you mentioned is just that we need to give innovation open, we need to experiment, we need to have some general guidance and framing, of course. But besides that, what would be one little thing that we need to really be pushing collectively, not as individuals but collectively, to make everybody progress, besides open sourcing it, of course? Eileen, do you want to start? Sure. Yeah, I can start. Maybe one of the things that we can think about today is how can we evolve testing methods and evaluation benchmarks for multi-agent systems. And the reason I say that is that we have a paper called Distributional AGI Safety, meaning that the paper discusses the fact that AGI may not emerge as one monolithic, big, powerful system, but it may actually emerge as a lot of sub. top AGI agents, specialized agents, working with each other in a multi-agent system. So we may be talking about multi-agent systems that reach AGI level capabilities. And for that to be successful, what should we be thinking about today when it arrives? In that paper, we discussed four different categories of safety, including market design, including baseline safety for each agent itself, including inter-agent communication, et cetera, et cetera. Please feel free to refer to that paper. But I think what we need to do today is making sure that we build evaluation benchmarks that are relevant to multi-agent systems. Frontier Labs that are developing these models that are core to the agents, it's important for them to, and by them I mean Google Live Mind being one of them, to report, test against these benchmarks, report transparently about the limitations, report transparently about the capabilities, and I think that is one of the goals that. Google DeepMind has been pursuing, and a lot of that information is available on the website. Please feel free to read our frameworks on AGI safety as well. So I'll stop there. So I'll just, again, follow up on that, right? One of the important factors that we're missing in the multi-agent system, we are assuming that it's a single model agent, right? Across the multi-agent system, I'm using one LLM model. But when we extend that thought, that what will happen when I'm using models provided by different players, open source or non-open source, you know, like all of those models will try to... So, single-matter responses are easy to verify. We can have the metrics or all the telemetry associated with it to verify whether the agent is claiming. But when it goes across those boundaries... And it becomes like a multimodal system, multi-agentic system. And every single company is saying, hey, trust us. So one of the thought experiments, or the way we try to solve it, is get every voice heard. Meaning that let's hear from all the academia how they think about this problem. Get all the LLM or a big. per companies getting involved in this discussion. The same thing with the system integrators and hyperscalers also participate in it. And not just them, but including the policy makers as",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "Governing Autonomy_ Trust in Agentic AI, Multi-Agent Systems and the Infrastructure of Trust.mp3.docx",
            "col3": "well as the younger generation who is gonna be the agentic AI native. So how are we preparing for that? That's very important. So the problem doesn't... get solved with the single model responses. So the only solution that we kind of lean towards to make it enough decentralized or enough integrated system so that we can have those test parameters across every single thing. Yeah, thank you. I think for me, largely, the thing which I think can solve a lot of it because that kind of distills enough information to people to start working from is the idea around auditability and traceability to be able to capture and accurate reporting in an immutable way of a lot of what is happening. And as long as there are more open source platforms that allow for a lot of those, you know, anonymized way of auditability and traceability and the incidents that's coming to come up, I think it drives for more collective learning at a more global platform level. I don't know if there are... And this is a conversation where when we meet a lot of startups, they're actually learning a lot from talking to each other. And they keep doing about the fact that there are not a lot of global platforms that just talk about just learning around auditable trails of what is happening, how is it happening, where do systems fail at what scale and level of infrastructure. And I think something that solves for just a collective platform to learn from. will go a long way. It's a very innovative environment around the world and what's happening. So as long as you provide the right infrastructure for even knowing what's going wrong, you know, I think there's a lot that can actually accelerate the learning and development there. That's a cue for our session. Oh, I thought that was where you programmed the agenda. Yeah, so I think there's two. Two key points I'd like to just leave everyone with. The first, Amir runs a very innovative organization called AI Commons that many of us have been a part of over the years. And really the critical nature of what they have been looking at, which very much aligned with a lot of what we were looking at as well, was the idea of having these technologies treated as a commons, a public utility. Because if you don't do that, you can't do exactly what everyone here is speaking to. Only those that have access to that can do that. And the point on education is very critical. But the point, I would just read into this then, is the thing to do today, we have to spend a lot more time getting rid of the jargon. Even the title of today's session. isn't something that's accessible readily to the same people we're asking to make use of these technologies, right? And so a lot of the effort needs to be spent on demystifying what we're talking about. Show the value of it. Because if we don't do that, then you're going to have a lot of fear. And a lot of the fear is unnecessary. And I think these would be the critical things. I would also love to use your app to turn off that alarm. So maybe you can teach us later about that. Thanks. Thank you very much. Thank you. Thank you everyone for coming.",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "projectDeleted": false,
    "codeFrameColumn": [
        {
            "col1": "col1",
            "col2": "col3",
            "researchObjective": "",
            "filterColumn": [],
            "netSetting": true,
            "subNetSetting": true,
            "categorySetting": true,
            "sentimentSetting": true,
            "thinkingBudgetSetting": 2000,
            "questionType": "OPENEND",
            "otherBrand": false,
            "existingCodeFrame": [],
            "isList": false,
            "codeFrameWithDataList": [],
            "codeFrameList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 10362,
                        "CandidatesTokenCount": 2823,
                        "TotalTokenCount": 14600,
                        "thoughtsTokenCount": 1415,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Net": "Conceptos y Características de los Agentes de IA",
                                "Subnets": [
                                    {
                                        "Subnet": "Definición y Alcance",
                                        "Categories": [
                                            {
                                                "Category": "Continuo de Agentes Autónomos",
                                                "Summary": "Los agentes de IA existen en un espectro desde chatbots básicos hasta vehículos completamente autónomos, realizando diversas funciones sin una línea de demarcación clara."
                                            },
                                            {
                                                "Category": "Sistemas Multiagente",
                                                "Summary": "Los sistemas multiagente implican la interacción, colaboración y potencial interconexión de múltiples agentes de IA para lograr objetivos complejos."
                                            }
                                        ],
                                        "Summary": "Exploración de la naturaleza y el rango de la autonomía y las capacidades de los agentes de IA."
                                    },
                                    {
                                        "Subnet": "Desafíos de la Autonomía",
                                        "Categories": [
                                            {
                                                "Category": "Comportamiento Incierto de los Agentes",
                                                "Summary": "La falta de comprensión de cómo se construyen o reaccionan los agentes genera incertidumbre y preocupaciones de seguridad."
                                            },
                                            {
                                                "Category": "Cambio de Objetivos e Interacción de Agentes",
                                                "Summary": "La preocupación de que los objetivos de los agentes puedan cambiar o mezclarse sin control, lo que lleva a resultados impredecibles."
                                            },
                                            {
                                                "Category": "Preocupaciones de Seguridad y Control",
                                                "Summary": "La autonomía de los agentes plantea desafíos para mantener el control y garantizar la seguridad, requiriendo nuevas estrategias de gobernanza."
                                            }
                                        ],
                                        "Summary": "Identificación de los problemas inherentes y las preocupaciones que surgen de la operación autónoma de los agentes de IA."
                                    }
                                ],
                                "Summary": "Categorización de las discusiones sobre la naturaleza fundamental, las capacidades y los riesgos intrínsecos de los agentes de IA."
                            },
                            {
                                "Net": "Marcos de Gobernanza y Confianza",
                                "Subnets": [
                                    {
                                        "Subnet": "Principios de Gobernanza",
                                        "Categories": [
                                            {
                                                "Category": "Gobernanza en Tiempo de Ejecución (Proactiva)",
                                                "Summary": "La necesidad de modelos de gobernanza que actúen en tiempo real y no como soluciones posteriores al hecho, dada la naturaleza dinámica de los agentes de IA."
                                            },
                                            {
                                                "Category": "Responsabilidad y Transparencia",
                                                "Summary": "La importancia de tener mecanismos claros para la rendición de cuentas, la transparencia y el monitoreo constante de las acciones y el rendimiento de los agentes."
                                            },
                                            {
                                                "Category": "Enfoque Iterativo y Monitoreo Continuo",
                                                "Summary": "La gobernanza de la IA requiere un enfoque adaptativo, con monitoreo constante y mejora de los sistemas a medida que evolucionan y operan."
                                            }
                                        ],
                                        "Summary": "Discusión sobre los fundamentos y métodos para establecer control y supervisión sobre los agentes de IA."
                                    },
                                    {
                                        "Subnet": "Identidad y Certificación de Agentes",
                                        "Categories": [
                                            {
                                                "Category": "Identidad Digital de los Agentes",
                                                "Summary": "La propuesta de dotar a los agentes de una "
                                            }
                                        ],
                                        "Summary": "Categorización de las discusiones sobre la naturaleza fundamental, las capacidades y los riesgos intrínsecos de los agentes de IA."
                                    },
                                    {
                                        "Subnet": "Identidad y Certificación de Agentes",
                                        "Categories": [
                                            {
                                                "Category": "Identidad Digital de los Agentes",
                                                "Summary": "La propuesta de dotar a los agentes de una "
                                            }
                                        ],
                                        "Summary": "Discusión sobre la necesidad de establecer una identidad clara para los agentes y mecanismos para su verificación y certificación."
                                    },
                                    {
                                        "Subnet": "Pila de Gobernanza y Aplicación",
                                        "Categories": [
                                            {
                                                "Category": "Modelo de Gobernanza de 5 Capas",
                                                "Summary": "Un marco que desglosa la gobernanza en cinco etapas: tiempo de construcción, tiempo de implementación, tiempo de ejecución, remediación y responsabilidad, cubriendo todo el ciclo de vida del agente."
                                            },
                                            {
                                                "Category": "Pila de Confianza Mínimamente Viable",
                                                "Summary": "Directrices esenciales y requisitos básicos de gobernanza que las startups y los nuevos proyectos de IA deben implementar desde el principio para generar confianza y cumplir con los estándares."
                                            },
                                            {
                                                "Category": "Barreras de Protección en la Capa de Orquestación",
                                                "Summary": "La implementación de controles y medidas de seguridad en la capa donde se coordinan los sistemas multiagente para prevenir acciones no deseadas."
                                            }
                                        ],
                                        "Summary": "Detalle de los modelos estructurados y los requisitos prácticos para implementar la gobernanza de la IA."
                                    }
                                ],
                                "Summary": "Compilación de marcos teóricos y prácticos para garantizar la seguridad, el control y la confianza en los sistemas de agentes de IA."
                            },
                            {
                                "Net": "Desarrollo y Herramientas Técnicas",
                                "Subnets": [
                                    {
                                        "Subnet": "Marcos Multiagente",
                                        "Categories": [
                                            {
                                                "Category": "Acelerador Multiagente Neuro AI",
                                                "Summary": "Un marco de código bajo/sin código y de código abierto diseñado para acelerar el desarrollo y la implementación de sistemas multiagente, con características de seguridad integradas."
                                            },
                                            {
                                                "Category": "Protocolos de Comunicación Agente-Agente",
                                                "Summary": "La infraestructura y las normas que permiten a los diferentes agentes de IA comunicarse e interactuar de forma segura y eficiente."
                                            }
                                        ],
                                        "Summary": "Exploración de plataformas y estándares para la construcción y el despliegue de sistemas multiagente."
                                    },
                                    {
                                        "Subnet": "Características Técnicas Clave",
                                        "Categories": [
                                            {
                                                "Category": "Observabilidad y Registro en Tiempo Real",
                                                "Summary": "Funcionalidades que permiten monitorear las acciones y comunicaciones de los agentes de IA en tiempo real, con registros detallados para auditoría."
                                            },
                                            {
                                                "Category": "Funcionalidad de Interruptor de Apagado",
                                                "Summary": "La capacidad de detener o desconectar un agente o sistema de agentes si se comporta de manera inesperada o peligrosa."
                                            },
                                            {
                                                "Category": "Seguridad Integrada",
                                                "Summary": "La implementación de medidas de seguridad desde la fase de diseño y desarrollo, en lugar de añadirlas como una ocurrencia tardía."
                                            },
                                            {
                                                "Category": "Métricas de ESD",
                                                "Summary": "El seguimiento y la presentación de informes sobre el uso de energía, la huella de carbono y los costos asociados con las operaciones de los agentes de IA."
                                            }
                                        ],
                                        "Summary": "Identificación de componentes técnicos esenciales para la funcionalidad, seguridad y monitoreo de los sistemas de agentes de IA."
                                    }
                                ],
                                "Summary": "Descripción de las plataformas de software, los marcos y las características técnicas que facilitan la construcción y gestión de los sistemas de agentes de IA."
                            },
                            {
                                "Net": "Perspectiva de la Industria y Regulación",
                                "Subnets": [
                                    {
                                        "Subnet": "Paralelismos con Otras Industrias",
                                        "Categories": [
                                            {
                                                "Category": "Aviación y Drones",
                                                "Summary": "Comparación con la evolución de la regulación en la aviación, donde la seguridad de los sistemas de IA ha llevado a pasar de "
                                            }
                                        ],
                                        "Summary": "Exploración de las lecciones aprendidas de otros sectores con tecnologías críticas para la seguridad y cómo se aplican a la IA."
                                    },
                                    {
                                        "Subnet": "Equilibrio entre Ingeniería y Regulación",
                                        "Categories": [
                                            {
                                                "Category": "Innovación vs. Cumplimiento",
                                                "Summary": "Discusión sobre cómo equilibrar la promoción de la innovación en IA con la necesidad de regulación y cumplimiento normativo."
                                            },
                                            {
                                                "Category": "Políticas Marco vs. Regulación Restrictiva",
                                                "Summary": "La diferencia entre establecer directrices y marcos generales que permitan la experimentación y la regulación estricta que podría sofocar la innovación."
                                            },
                                            {
                                                "Category": "Estandarización y Aceptación del Mercado",
                                                "Summary": "La importancia de los estándares para la maduración del mercado y cómo la aceptación de la industria impulsa la adopción de prácticas de gobernanza."
                                            }
                                        ],
                                        "Summary": "Análisis del debate sobre el nivel adecuado de control técnico frente a la supervisión externa en el desarrollo de la IA."
                                    },
                                    {
                                        "Subnet": "Gobernanza como Ventaja Competitiva",
                                        "Categories": [
                                            {
                                                "Category": "No es un Costo, sino un Activo Estratégico",
                                                "Summary": "La perspectiva de que la gobernanza de la IA, especialmente para las ventas empresariales, se ha convertido en un diferenciador competitivo y no solo en un gasto de cumplimiento."
                                            },
                                            {
                                                "Category": "Integración Temprana en el Desarrollo de Productos",
                                                "Summary": "La necesidad de incorporar la gobernanza en las primeras etapas del desarrollo de productos de IA para cumplir con los requisitos del mercado y los clientes."
                                            },
                                            {
                                                "Category": "Pila Mínima de Confianza Viable",
                                                "Summary": "Los componentes esenciales de confianza que una startup debe tener para asegurar contratos empresariales, haciendo de la gobernanza un requisito previo para la entrada en el mercado."
                                            }
                                        ],
                                        "Summary": "Discusión sobre cómo las prácticas robustas de gobernanza de la IA pueden impulsar el éxito comercial y la diferenciación."
                                    }
                                ],
                                "Summary": "Visión general de cómo los diferentes sectores y las regulaciones existentes informan el desarrollo de políticas y estándares para la IA."
                            },
                            {
                                "Net": "Futuro y Acción Colectiva",
                                "Subnets": [
                                    {
                                        "Subnet": "Evolución de Pruebas y Evaluación",
                                        "Categories": [
                                            {
                                                "Category": "Puntos de Referencia para Sistemas Multiagente",
                                                "Summary": "La necesidad de desarrollar métodos de prueba y evaluación específicos para sistemas multiagente, reconociendo su complejidad y potencial para capacidades de nivel AGI."
                                            },
                                            {
                                                "Category": "Transparencia en la Informes de Capacidades y Limitaciones",
                                                "Summary": "La importancia de que los laboratorios de desarrollo informen de forma transparente sobre las capacidades y limitaciones de sus modelos y agentes de IA."
                                            }
                                        ],
                                        "Summary": "La discusión sobre la necesidad de métodos avanzados y estandarizados para evaluar la seguridad y el rendimiento de los sistemas multiagente."
                                    },
                                    {
                                        "Subnet": "Colaboración y Aprendizaje Colectivo",
                                        "Categories": [
                                            {
                                                "Category": "Infraestructura de Confianza Fragmentada",
                                                "Summary": "El desafío de construir una infraestructura de confianza coherente cuando los componentes son proporcionados por diferentes jugadores y operan en entornos diversos."
                                            },
                                            {
                                                "Category": "Plataformas Globales para la Aprendizaje de Auditoría",
                                                "Summary": "La propuesta de crear plataformas donde las organizaciones puedan compartir de forma anónima experiencias sobre auditoría, trazabilidad e incidentes para un aprendizaje colectivo."
                                            },
                                            {
                                                "Category": "Participación de Múltiples Partes Interesadas",
                                                "Summary": "La necesidad de incluir a la academia, las grandes empresas tecnológicas, los integradores de sistemas, los formuladores de políticas y las nuevas generaciones en la discusión sobre el futuro de la IA."
                                            }
                                        ],
                                        "Summary": "Exploración de la necesidad de esfuerzos coordinados y plataformas compartidas para abordar los desafíos de la IA."
                                    },
                                    {
                                        "Subnet": "Accesibilidad y Comprensión Pública",
                                        "Categories": [
                                            {
                                                "Category": "Desmitificar la Jerga de la IA",
                                                "Summary": "La importancia de simplificar el lenguaje técnico y la terminología compleja para hacer que los conceptos de la IA sean accesibles al público en general."
                                            },
                                            {
                                                "Category": "Mostrar el Valor de la IA",
                                                "Summary": "La necesidad de comunicar los beneficios y el impacto positivo de la IA para reducir el miedo y fomentar la aceptación pública."
                                            },
                                            {
                                                "Category": "Tratar la IA como un Servicio Público",
                                                "Summary": "La idea de que las tecnologías de IA, especialmente los fundamentos, deberían tratarse como un bien común o una utilidad pública para garantizar el acceso y la equidad."
                                            }
                                        ],
                                        "Summary": "Discusión sobre la importancia de hacer que la IA sea comprensible y aceptable para un público más amplio."
                                    }
                                ],
                                "Summary": "Proyección hacia las próximas etapas de la IA, incluyendo el desarrollo de métodos de evaluación, la colaboración de la industria y el compromiso público."
                            },
                            {
                                "Net": "996 > TEMA no específico",
                                "Subnets": [
                                    {
                                        "Subnet": "No Categorable",
                                        "Categories": [
                                            {
                                                "Category": "Contenido No Temático",
                                                "Summary": "Respuestas que son irrelevantes, no pertenecen a ninguno de los temas principales de la conversación, o que no pueden ser fácilmente codificadas en las categorías existentes y pueden requerir una codificación ascendente."
                                            }
                                        ],
                                        "Summary": "Respuestas que no encajan fácilmente en las categorías temáticas predefinidas y que podrían requerir una revisión posterior."
                                    }
                                ],
                                "Summary": "Respuestas que no se ajustan a las categorías temáticas predefinidas pero que pueden contener información relevante que no se ha categorizado explícitamente."
                            },
                            {
                                "Net": "997 > No sé/No puedo decir",
                                "Subnets": [
                                    {
                                        "Subnet": "Respuestas Vagas/No Concluyentes",
                                        "Categories": [
                                            {
                                                "Category": "Declaraciones No Conclusivas",
                                                "Summary": "Respuestas que son no comprometidas, evasivas, o que expresan una falta de conocimiento o incapacidad para emitir un juicio."
                                            }
                                        ],
                                        "Summary": "Respuestas que no ofrecen una opinión o información clara y son de naturaleza ambigua o incierta."
                                    }
                                ],
                                "Summary": "Respuestas que expresan incertidumbre, falta de conocimiento o una negativa a opinar."
                            },
                            {
                                "Net": "998 > No aplicable",
                                "Subnets": [
                                    {
                                        "Subnet": "Irrelevante para la Pregunta/Proyecto",
                                        "Categories": [
                                            {
                                                "Category": "Contenido Fuera de Alcance",
                                                "Summary": "Respuestas que son irrelevantes para la pregunta específica planteada o el contexto general del proyecto, o que se consideran completamente ajenas al tema."
                                            }
                                        ],
                                        "Summary": "Respuestas que no tienen relación directa con la pregunta o el propósito del estudio."
                                    }
                                ],
                                "Summary": "Respuestas que son inaplicables o no pertinentes al contexto de la pregunta o el proyecto."
                            },
                            {
                                "Net": "999 > Nada/Ninguno",
                                "Subnets": [
                                    {
                                        "Subnet": "Contenido Sin Sentido",
                                        "Categories": [
                                            {
                                                "Category": "Gibberish/Caracteres no Lingüísticos",
                                                "Summary": "Respuestas que consisten en galimatías, caracteres sin sentido, o que no transmiten ningún significado inteligible."
                                            }
                                        ],
                                        "Summary": "Respuestas que no contienen información significativa o son incomprensibles."
                                    }
                                ],
                                "Summary": "Respuestas que carecen completamente de contenido significativo o son uninteligibles."
                            }
                        ],
                        "Summary": "Esta transcripción es de un panel de discusión centrado en la gobernanza, seguridad, confianza y regulación de los sistemas de agentes de IA y multiagentes autónomos. Los ponentes, incluyendo expertos de Google, Project Nanda, IEEE Standards Association e Insight Partners, abordan los desafíos de la autonomía de la IA, la necesidad de nuevos modelos de gobernanza (incluido un "
                    }
                }
            ],
            "onlyCodeFrame": true,
            "errorList": [],
            "codeFrameGenerateType": "CODE_FRAME_ONLY",
            "requestType": "CODEFRAME",
            "requestMode": "DEVELOPMENT",
            "sourceLanguage": "English",
            "targetLanguage": "Spanish",
            "response": {
                "status": "OK",
                "data": {
                    "status": "OK",
                    "PromptTokenCount": 10362,
                    "CandidatesTokenCount": 2823,
                    "TotalTokenCount": 14600,
                    "thoughtsTokenCount": 1415,
                    "finishReason": "STOP",
                    "data": [
                        {
                            "Net": "Conceptos y Características de los Agentes de IA",
                            "Subnets": [
                                {
                                    "Subnet": "Definición y Alcance",
                                    "Categories": [
                                        {
                                            "Category": "Continuo de Agentes Autónomos",
                                            "Summary": "Los agentes de IA existen en un espectro desde chatbots básicos hasta vehículos completamente autónomos, realizando diversas funciones sin una línea de demarcación clara.",
                                            "code": 3
                                        },
                                        {
                                            "Category": "Sistemas Multiagente",
                                            "Summary": "Los sistemas multiagente implican la interacción, colaboración y potencial interconexión de múltiples agentes de IA para lograr objetivos complejos.",
                                            "code": 4
                                        }
                                    ],
                                    "Summary": "Exploración de la naturaleza y el rango de la autonomía y las capacidades de los agentes de IA.",
                                    "code": 2
                                },
                                {
                                    "Subnet": "Desafíos de la Autonomía",
                                    "Categories": [
                                        {
                                            "Category": "Comportamiento Incierto de los Agentes",
                                            "Summary": "La falta de comprensión de cómo se construyen o reaccionan los agentes genera incertidumbre y preocupaciones de seguridad.",
                                            "code": 6
                                        },
                                        {
                                            "Category": "Cambio de Objetivos e Interacción de Agentes",
                                            "Summary": "La preocupación de que los objetivos de los agentes puedan cambiar o mezclarse sin control, lo que lleva a resultados impredecibles.",
                                            "code": 7
                                        },
                                        {
                                            "Category": "Preocupaciones de Seguridad y Control",
                                            "Summary": "La autonomía de los agentes plantea desafíos para mantener el control y garantizar la seguridad, requiriendo nuevas estrategias de gobernanza.",
                                            "code": 8
                                        }
                                    ],
                                    "Summary": "Identificación de los problemas inherentes y las preocupaciones que surgen de la operación autónoma de los agentes de IA.",
                                    "code": 5
                                }
                            ],
                            "Summary": "Categorización de las discusiones sobre la naturaleza fundamental, las capacidades y los riesgos intrínsecos de los agentes de IA.",
                            "code": 1
                        },
                        {
                            "Net": "Marcos de Gobernanza y Confianza",
                            "Subnets": [
                                {
                                    "Subnet": "Principios de Gobernanza",
                                    "Categories": [
                                        {
                                            "Category": "Gobernanza en Tiempo de Ejecución (Proactiva)",
                                            "Summary": "La necesidad de modelos de gobernanza que actúen en tiempo real y no como soluciones posteriores al hecho, dada la naturaleza dinámica de los agentes de IA.",
                                            "code": 11
                                        },
                                        {
                                            "Category": "Responsabilidad y Transparencia",
                                            "Summary": "La importancia de tener mecanismos claros para la rendición de cuentas, la transparencia y el monitoreo constante de las acciones y el rendimiento de los agentes.",
                                            "code": 12
                                        },
                                        {
                                            "Category": "Enfoque Iterativo y Monitoreo Continuo",
                                            "Summary": "La gobernanza de la IA requiere un enfoque adaptativo, con monitoreo constante y mejora de los sistemas a medida que evolucionan y operan.",
                                            "code": 13
                                        }
                                    ],
                                    "Summary": "Discusión sobre los fundamentos y métodos para establecer control y supervisión sobre los agentes de IA.",
                                    "code": 10
                                },
                                {
                                    "Subnet": "Identidad y Certificación de Agentes",
                                    "Categories": [
                                        {
                                            "Category": "Identidad Digital de los Agentes",
                                            "Summary": "La propuesta de dotar a los agentes de una ",
                                            "code": 15
                                        }
                                    ],
                                    "Summary": "Categorización de las discusiones sobre la naturaleza fundamental, las capacidades y los riesgos intrínsecos de los agentes de IA.",
                                    "code": 14
                                },
                                {
                                    "Subnet": "Identidad y Certificación de Agentes",
                                    "Categories": [
                                        {
                                            "Category": "Identidad Digital de los Agentes",
                                            "Summary": "La propuesta de dotar a los agentes de una ",
                                            "code": 17
                                        }
                                    ],
                                    "Summary": "Discusión sobre la necesidad de establecer una identidad clara para los agentes y mecanismos para su verificación y certificación.",
                                    "code": 16
                                },
                                {
                                    "Subnet": "Pila de Gobernanza y Aplicación",
                                    "Categories": [
                                        {
                                            "Category": "Modelo de Gobernanza de 5 Capas",
                                            "Summary": "Un marco que desglosa la gobernanza en cinco etapas: tiempo de construcción, tiempo de implementación, tiempo de ejecución, remediación y responsabilidad, cubriendo todo el ciclo de vida del agente.",
                                            "code": 19
                                        },
                                        {
                                            "Category": "Pila de Confianza Mínimamente Viable",
                                            "Summary": "Directrices esenciales y requisitos básicos de gobernanza que las startups y los nuevos proyectos de IA deben implementar desde el principio para generar confianza y cumplir con los estándares.",
                                            "code": 20
                                        },
                                        {
                                            "Category": "Barreras de Protección en la Capa de Orquestación",
                                            "Summary": "La implementación de controles y medidas de seguridad en la capa donde se coordinan los sistemas multiagente para prevenir acciones no deseadas.",
                                            "code": 21
                                        }
                                    ],
                                    "Summary": "Detalle de los modelos estructurados y los requisitos prácticos para implementar la gobernanza de la IA.",
                                    "code": 18
                                }
                            ],
                            "Summary": "Compilación de marcos teóricos y prácticos para garantizar la seguridad, el control y la confianza en los sistemas de agentes de IA.",
                            "code": 9
                        },
                        {
                            "Net": "Desarrollo y Herramientas Técnicas",
                            "Subnets": [
                                {
                                    "Subnet": "Marcos Multiagente",
                                    "Categories": [
                                        {
                                            "Category": "Acelerador Multiagente Neuro AI",
                                            "Summary": "Un marco de código bajo/sin código y de código abierto diseñado para acelerar el desarrollo y la implementación de sistemas multiagente, con características de seguridad integradas.",
                                            "code": 24
                                        },
                                        {
                                            "Category": "Protocolos de Comunicación Agente-Agente",
                                            "Summary": "La infraestructura y las normas que permiten a los diferentes agentes de IA comunicarse e interactuar de forma segura y eficiente.",
                                            "code": 25
                                        }
                                    ],
                                    "Summary": "Exploración de plataformas y estándares para la construcción y el despliegue de sistemas multiagente.",
                                    "code": 23
                                },
                                {
                                    "Subnet": "Características Técnicas Clave",
                                    "Categories": [
                                        {
                                            "Category": "Observabilidad y Registro en Tiempo Real",
                                            "Summary": "Funcionalidades que permiten monitorear las acciones y comunicaciones de los agentes de IA en tiempo real, con registros detallados para auditoría.",
                                            "code": 27
                                        },
                                        {
                                            "Category": "Funcionalidad de Interruptor de Apagado",
                                            "Summary": "La capacidad de detener o desconectar un agente o sistema de agentes si se comporta de manera inesperada o peligrosa.",
                                            "code": 28
                                        },
                                        {
                                            "Category": "Seguridad Integrada",
                                            "Summary": "La implementación de medidas de seguridad desde la fase de diseño y desarrollo, en lugar de añadirlas como una ocurrencia tardía.",
                                            "code": 29
                                        },
                                        {
                                            "Category": "Métricas de ESD",
                                            "Summary": "El seguimiento y la presentación de informes sobre el uso de energía, la huella de carbono y los costos asociados con las operaciones de los agentes de IA.",
                                            "code": 30
                                        }
                                    ],
                                    "Summary": "Identificación de componentes técnicos esenciales para la funcionalidad, seguridad y monitoreo de los sistemas de agentes de IA.",
                                    "code": 26
                                }
                            ],
                            "Summary": "Descripción de las plataformas de software, los marcos y las características técnicas que facilitan la construcción y gestión de los sistemas de agentes de IA.",
                            "code": 22
                        },
                        {
                            "Net": "Perspectiva de la Industria y Regulación",
                            "Subnets": [
                                {
                                    "Subnet": "Paralelismos con Otras Industrias",
                                    "Categories": [
                                        {
                                            "Category": "Aviación y Drones",
                                            "Summary": "Comparación con la evolución de la regulación en la aviación, donde la seguridad de los sistemas de IA ha llevado a pasar de ",
                                            "code": 33
                                        }
                                    ],
                                    "Summary": "Exploración de las lecciones aprendidas de otros sectores con tecnologías críticas para la seguridad y cómo se aplican a la IA.",
                                    "code": 32
                                },
                                {
                                    "Subnet": "Equilibrio entre Ingeniería y Regulación",
                                    "Categories": [
                                        {
                                            "Category": "Innovación vs. Cumplimiento",
                                            "Summary": "Discusión sobre cómo equilibrar la promoción de la innovación en IA con la necesidad de regulación y cumplimiento normativo.",
                                            "code": 35
                                        },
                                        {
                                            "Category": "Políticas Marco vs. Regulación Restrictiva",
                                            "Summary": "La diferencia entre establecer directrices y marcos generales que permitan la experimentación y la regulación estricta que podría sofocar la innovación.",
                                            "code": 36
                                        },
                                        {
                                            "Category": "Estandarización y Aceptación del Mercado",
                                            "Summary": "La importancia de los estándares para la maduración del mercado y cómo la aceptación de la industria impulsa la adopción de prácticas de gobernanza.",
                                            "code": 37
                                        }
                                    ],
                                    "Summary": "Análisis del debate sobre el nivel adecuado de control técnico frente a la supervisión externa en el desarrollo de la IA.",
                                    "code": 34
                                },
                                {
                                    "Subnet": "Gobernanza como Ventaja Competitiva",
                                    "Categories": [
                                        {
                                            "Category": "No es un Costo, sino un Activo Estratégico",
                                            "Summary": "La perspectiva de que la gobernanza de la IA, especialmente para las ventas empresariales, se ha convertido en un diferenciador competitivo y no solo en un gasto de cumplimiento.",
                                            "code": 39
                                        },
                                        {
                                            "Category": "Integración Temprana en el Desarrollo de Productos",
                                            "Summary": "La necesidad de incorporar la gobernanza en las primeras etapas del desarrollo de productos de IA para cumplir con los requisitos del mercado y los clientes.",
                                            "code": 40
                                        },
                                        {
                                            "Category": "Pila Mínima de Confianza Viable",
                                            "Summary": "Los componentes esenciales de confianza que una startup debe tener para asegurar contratos empresariales, haciendo de la gobernanza un requisito previo para la entrada en el mercado.",
                                            "code": 41
                                        }
                                    ],
                                    "Summary": "Discusión sobre cómo las prácticas robustas de gobernanza de la IA pueden impulsar el éxito comercial y la diferenciación.",
                                    "code": 38
                                }
                            ],
                            "Summary": "Visión general de cómo los diferentes sectores y las regulaciones existentes informan el desarrollo de políticas y estándares para la IA.",
                            "code": 31
                        },
                        {
                            "Net": "Futuro y Acción Colectiva",
                            "Subnets": [
                                {
                                    "Subnet": "Evolución de Pruebas y Evaluación",
                                    "Categories": [
                                        {
                                            "Category": "Puntos de Referencia para Sistemas Multiagente",
                                            "Summary": "La necesidad de desarrollar métodos de prueba y evaluación específicos para sistemas multiagente, reconociendo su complejidad y potencial para capacidades de nivel AGI.",
                                            "code": 44
                                        },
                                        {
                                            "Category": "Transparencia en la Informes de Capacidades y Limitaciones",
                                            "Summary": "La importancia de que los laboratorios de desarrollo informen de forma transparente sobre las capacidades y limitaciones de sus modelos y agentes de IA.",
                                            "code": 45
                                        }
                                    ],
                                    "Summary": "La discusión sobre la necesidad de métodos avanzados y estandarizados para evaluar la seguridad y el rendimiento de los sistemas multiagente.",
                                    "code": 43
                                },
                                {
                                    "Subnet": "Colaboración y Aprendizaje Colectivo",
                                    "Categories": [
                                        {
                                            "Category": "Infraestructura de Confianza Fragmentada",
                                            "Summary": "El desafío de construir una infraestructura de confianza coherente cuando los componentes son proporcionados por diferentes jugadores y operan en entornos diversos.",
                                            "code": 47
                                        },
                                        {
                                            "Category": "Plataformas Globales para la Aprendizaje de Auditoría",
                                            "Summary": "La propuesta de crear plataformas donde las organizaciones puedan compartir de forma anónima experiencias sobre auditoría, trazabilidad e incidentes para un aprendizaje colectivo.",
                                            "code": 48
                                        },
                                        {
                                            "Category": "Participación de Múltiples Partes Interesadas",
                                            "Summary": "La necesidad de incluir a la academia, las grandes empresas tecnológicas, los integradores de sistemas, los formuladores de políticas y las nuevas generaciones en la discusión sobre el futuro de la IA.",
                                            "code": 49
                                        }
                                    ],
                                    "Summary": "Exploración de la necesidad de esfuerzos coordinados y plataformas compartidas para abordar los desafíos de la IA.",
                                    "code": 46
                                },
                                {
                                    "Subnet": "Accesibilidad y Comprensión Pública",
                                    "Categories": [
                                        {
                                            "Category": "Desmitificar la Jerga de la IA",
                                            "Summary": "La importancia de simplificar el lenguaje técnico y la terminología compleja para hacer que los conceptos de la IA sean accesibles al público en general.",
                                            "code": 51
                                        },
                                        {
                                            "Category": "Mostrar el Valor de la IA",
                                            "Summary": "La necesidad de comunicar los beneficios y el impacto positivo de la IA para reducir el miedo y fomentar la aceptación pública.",
                                            "code": 52
                                        },
                                        {
                                            "Category": "Tratar la IA como un Servicio Público",
                                            "Summary": "La idea de que las tecnologías de IA, especialmente los fundamentos, deberían tratarse como un bien común o una utilidad pública para garantizar el acceso y la equidad.",
                                            "code": 53
                                        }
                                    ],
                                    "Summary": "Discusión sobre la importancia de hacer que la IA sea comprensible y aceptable para un público más amplio.",
                                    "code": 50
                                }
                            ],
                            "Summary": "Proyección hacia las próximas etapas de la IA, incluyendo el desarrollo de métodos de evaluación, la colaboración de la industria y el compromiso público.",
                            "code": 42
                        }
                    ],
                    "Summary": "Esta transcripción es de un panel de discusión centrado en la gobernanza, seguridad, confianza y regulación de los sistemas de agentes de IA y multiagentes autónomos. Los ponentes, incluyendo expertos de Google, Project Nanda, IEEE Standards Association e Insight Partners, abordan los desafíos de la autonomía de la IA, la necesidad de nuevos modelos de gobernanza (incluido un "
                }
            },
            "summaryList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 10023,
                        "CandidatesTokenCount": 3142,
                        "TotalTokenCount": 14524,
                        "thoughtsTokenCount": 1359,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Summary": "<html><body><h2>Resumen Detallado y Exhaustivo: Gobernando la Autonomía en la Era de los Agentes de IA</h2><p>El presente informe sintetiza una discusión integral sobre la gobernanza y la autonomía en el contexto de los sistemas de IA agéntica, destacando la evolución, los desafíos y las estrategias para fomentar la confianza y la innovación. Participaron expertos de Google, Project Nanda, IEEE Standards Association, Insight Partners y Cognizant AI Lab, cubriendo perspectivas desde la política, la infraestructura, la estandarización, la inversión y el desarrollo tecnológico.</p><h3>1. Introducción al Mundo de los Agentes de IA y la Autonomía</h3><h4>1.1. Percepciones Iniciales y el Espectro de la Agencialidad</h4><p>La sesión comenzó con una exploración de la familiaridad y las preocupaciones del público sobre los agentes de IA. Se observó un bajo porcentaje de personas que sienten que los agentes serán peligrosos (5-10%), lo que contrasta con el entusiasmo general por los experimentos recientes como OpenClaw y Notebook. Eli Sahai de Google contextualizó la definición de agentes, no como una distinción binaria, sino como un <b>espectro o <i>continuum</i> de autonomía</b>. Este espectro abarca desde chatbots básicos con capacidades de investigación y presentación de resultados (ej., Google Deep Research) hasta sistemas completamente autónomos que toman acciones en el mundo real, como los coches autónomos. La autonomía se define en diversas dimensiones, incluyendo memoria, planificación (a corto y largo plazo) y la capacidad de ejecución.</p><h4>1.2. El Reto Central: La Gobernanza de la Autonomía</h4><p>El concepto de autonomía genera inquietud debido a la falta de conocimiento sobre cómo se construyen los agentes, cómo reaccionarán, si sus objetivos cambiarán o cómo interactuarán entre sí. Esto plantea un dilema para las empresas y la sociedad, ya que muchas decisiones cruciales no pueden dejarse enteramente en manos de los agentes. Por lo tanto, el <b>concepto de gobernanza</b>, que abarca el control de los resultados, los procedimientos, la propiedad y la gestión de riesgos, se vuelve fundamental, algo que no era tan evidente hace apenas un año. El objetivo principal de la discusión fue cómo asegurar que este futuro autónomo sea confiable y manejable.</p><h3>2. Desafíos y Consideraciones Clave en Sistemas Multiagente</h3><h4>2.1. Errores y la Necesidad de Gobernanza en Tiempo Real</h4><p>Mahesh de Project Nanda enfatizó que, dado el <i>continuum</i> de autonomía y la naturaleza de los agentes como actores en red, la <b>gobernanza debe ser en tiempo de ejecución (<i>runtime governance</i>)</b>. Los modelos de gobernanza tradicionales, evolucionados desde las interacciones máquina a máquina, SaaS y microservicios, no son suficientes para los sistemas agénticos escalables. La seguridad y la gobernanza deben abordarse como un primer principio, no como una solución posterior, para garantizar la implementabilidad, la apertura de la ejecución y la rendición de cuentas (referencia: Mahesh).</p><h4>2.2. Identidad, Certificación y Transparencia</h4><p>Se planteó la idea de una <b>identidad agéntica</b> o \"pasaporte\" para los agentes, que especifique su origen, datos de entrenamiento, creador, propietario e idioma. Esto lleva a la necesidad de <b>certificar agentes</b>, un área en la que la IEEE Standards Association, representada por Alpesh Shah, está trabajando activamente. Alpesh destacó la importancia de una estructura de gobernanza sólida, un proceso claro para la implementación y un enfoque iterativo para el monitoreo constante. También subrayó la necesidad de contextualizar el comportamiento de los agentes y asegurar <b>registros transparentes y responsables (<i>logging</i>)</b>, así como la participación humana en el proceso. La certificación del IEEE busca evaluar los medios por los cuales un agente logra un resultado, entendiendo que la perfección absoluta es inalcanzable, pero la confianza aumenta con la iteración y los datos. Estándares sobre transparencia de datos, diseño apropiado para la edad, responsabilidad y privacidad de datos son cruciales. La colaboración de una comunidad diversa (abogados, médicos, artistas, ingenieros) ha sido clave para comprender los problemas contextuales (referencia: Alpesh Shah).</p><h3>3. El \"Governance Stack\" y su Impacto Económico</h3><h4>3.1. Un Modelo de Gobernanza de Cinco Fases</h4><p>Apoorv Goyal de Insight Partners presentó un \"governance stack\" de cinco fases, vital para evaluar empresas de IA:<ol><li><b>Tiempo de Construcción (Build-time):</b> Gobernanza de datos, control de versiones del modelo.</li><li><b>Tiempo de Despliegue (Deploy-time):</b> Gestión de políticas, seguimiento, permisos, gestión de secretos.</li><li><b>Tiempo de Ejecución (Runtime):</b> Observabilidad en tiempo real, \"kill switch\" (interruptor de emergencia) si las cosas van mal.</li><li><b>Remediación:</b> Pistas de auditoría para análisis post-mortem, arquitectura de respuesta a incidentes.</li><li><b>Capa de Responsabilidad (Accountability):</b> Definición de roles, estructuras de informes, post-mortem, herramientas de mapeo de cumplimiento.</li></ol>Las mejores empresas nativas de IA están integrando la gobernanza no como un problema de cumplimiento, sino como una ventaja competitiva de producto y GTM (Go-To-Market), lo que las hace más atractivas para los inversores (referencia: Apoorv Goyal).</p><h4>3.2. \"Minimal Viable Trust Stack\" para Startups</h4><p>Para las startups con recursos limitados, Apoorv sugirió un <b>\"minimal viable trust stack\"</b> que incluye:<ol><li>Un registro de identidad de agente claramente definido.</li><li>Barreras de seguridad (<i>guardrails</i>) en la capa de orquestación.</li><li>Una arquitectura de observabilidad en tiempo real clara.</li><li>Supervisión definida y transparente.</li></ol>Estos cuatro elementos básicos son esenciales antes de lanzar cualquier ecosistema agéntico a producción, ya que su costo no es excesivo y previenen fallos mayores (referencia: Apoorv Goyal).</p><h3>4. Orquestación y Coordinación de Agentes</h3><h4>4.1. Neuro AI Multi-Agent Accelerator</h4><p>Praveen de Cognizant AI Lab presentó <b>Neuro AI Multi-Agent Accelerator</b>, un <i>framework</i> de código abierto (Apache 2.0) lanzado en mayo de 2025. Se describe como una herramienta de bajo/sin código para acelerar el desarrollo de prototipos y escalado de sistemas multiagente. Sus características principales son:<ol><li><b>Agnóstico:</b> A la nube y a los LLMs, permitiendo al usuario controlar qué modelos y entornos utilizar.</li><li><b>Interoperable:</b> Soporta múltiples protocolos de comunicación entre agentes (OSR, A2N MCP).</li><li><b>Seguridad Integrada:</b> Construido con la seguridad como un elemento central.</li><li><b>Facilidad de Uso:</b> Permite a usuarios de todos los niveles crear redes de agentes en minutos.</li></ol>La demostración mostró la creación de una red de agentes, la asignación de roles y responsabilidades, y la conexión de agentes existentes. El <i>framework</i> incluye paneles de chat internos (para monitorear la comunicación entre agentes), logs detallados (uso de tokens, costos) y métricas ESG (uso de energía, huella de carbono) (referencia: Praveen).</p><h3>5. Lecciones de Otras Industrias</h3><h4>5.1. La Evolución de la Seguridad en la Aviación</h4><p>Eli Sahai trazó paralelismos con la industria de la aviación, donde el concepto de seguridad en sistemas críticos no es nuevo. Mencionó la evolución de las regulaciones para drones en EE. UU., pasando de la \"línea de visión visual\" (VLOS) a \"más allá de la línea de visión visual\" (BVLOS). Esta transición se basa en el desarrollo de sistemas de detección y evasión impulsados por IA que superan las capacidades humanas. Esto implica un cambio del \"humano en el circuito\" (<i>human in the loop</i>), donde el humano aprueba cada paso, a un \"humano al mando\" (<i>human in command</i>), donde el humano supervisa pero no interviene en cada acción. Este modelo es aplicable a la IA agéntica a medida que los agentes se vuelven más seguros y confiables (referencia: Eli Sahai).</p><h3>6. Balance entre la Innovación y la Regulación</h3><h4>6.1. \"Data Plane\" vs. \"Control Plane\"</h4><p>Se debatió el equilibrio entre las directrices de ingeniería y diseño técnico frente a las regulaciones, normas y estándares. Mahesh argumentó que la innovación debe mantenerse abierta (el <b>\"data plane\"</b>), mientras que la gobernanza debe ser el <b>\"control plane\"</b>; ambos deben desarrollarse en paralelo para evitar sofocar la innovación (referencia: Mahesh).</p><h4>6.2. Regulación en Etapas Tempranas y Maduración del Mercado</h4><p>Alpesh Shah señaló que una regulación prematura puede comprometer la innovación. En las primeras etapas, la experimentación en entornos controlados es crucial para comprender las posibilidades. La regulación tiene más sentido en la etapa de maduración de la tecnología, pero las políticas pueden servir como un \"marco\" inicial para delimitar los límites. La aceptación del mercado normaliza ciertas prácticas. Para startups, la integración de estándares y políticas desde el principio puede ser una solución, proporcionando una base de confianza para gobiernos y reguladores (referencia: Alpesh Shah).</p><h4>6.3. La Gobernanza como Facilitador de Negocios</h4><p>Apoorv Goyal enfatizó que la responsabilidad en la IA y el despliegue comercial no son contradictorios; de hecho, están muy alineados. Las empresas que buscan vender a grandes clientes necesitan marcos de gobernanza robustos (manejo de datos, gestión de secretos, permisos) desde el principio. La gobernanza se ha convertido en una parte central de las conversaciones en las juntas directivas, esencial para la escalabilidad. Es como la seguridad: una parte intrínseca de la cadencia operativa semanal de los equipos nativos de IA (referencia: Apoorv Goyal).</p><h4>6.4. El Ejemplo de la Infraestructura Digital de la India</h4><p>Mahesh hizo un paralelo con la infraestructura digital de la India (Aadhaar, UPI), donde el gobierno estandarizó las interfaces (identidad, pagos), permitiendo que las empresas construyeran aplicaciones sobre ellas. Este modelo, donde el gobierno define los estándares fundacionales y permite la innovación en la capa superior, podría ser relevante para la evolución de la IA agéntica (referencia: Mahesh).</p><h3>7. Conclusiones y Recomendaciones Clave</h3><h4>7.1. Conclusiones de Alto Nivel</h4><ul><li>La IA agéntica opera en un continuo de autonomía, no como una entidad binaria, lo que requiere enfoques de gobernanza dinámicos y en tiempo de ejecución.</li><li>La gobernanza no es solo un requisito de cumplimiento, sino una ventaja competitiva crucial que impulsa la confianza del mercado y la adopción empresarial.</li><li>La seguridad y la responsabilidad deben integrarse desde la fase de diseño y construcción, no como una ocurrencia tardía.</li><li>La colaboración intersectorial y la estandarización son esenciales para construir una infraestructura de confianza global.</li><li>La desmitificación de la IA y la educación pública son fundamentales para superar el miedo y fomentar una adopción informada y responsable.</li></ul><h4>7.2. Acciones y Recomendaciones Detalladas</h4><p>Para avanzar colectivamente en la gobernanza de la autonomía en la IA, se proponen las siguientes acciones y responsabilidades:</p><ul><li><b>Evolucionar Métodos de Prueba y Benchmarks:</b><br>Eli Sahai (Google DeepMind) recomienda desarrollar y utilizar benchmarks de evaluación específicamente para sistemas multiagente (particularmente en el contexto de AGI distribuida). Los laboratorios de IA de frontera deben informar de forma transparente sobre las limitaciones y capacidades de sus modelos frente a estos benchmarks.</li><li><b>Fomentar la Descentralización e Integración de Sistemas:</b><br>Mahesh (Project Nanda) propone crear sistemas suficientemente descentralizados e integrados para permitir parámetros de prueba consistentes en modelos proporcionados por diferentes proveedores (fuente abierta y cerrada). Esto implica dar voz a la academia, las grandes empresas de LLM, los integradores de sistemas, los hiperescaladores y los responsables políticos, así como a la próxima generación de nativos de IA.</li><li><b>Mejorar la Auditabilidad y Trazabilidad:</b><br>Apoorv Goyal (Insight Partners) sugiere la creación de plataformas de código abierto que permitan la captura y el reporte preciso e inmutable de la auditabilidad y la trazabilidad de los eventos de los agentes. Esto facilitaría el aprendizaje colectivo a nivel global sobre fallos del sistema, la infraestructura y la escala.</li><li><b>Tratar la Tecnología como \"Bienes Comunes\" (Public Utility):</b><br>Alpesh Shah (IEEE Standards Association) y Amir (AI Commons) enfatizan la importancia de considerar estas tecnologías como bienes comunes o servicios públicos. Esto asegura que la accesibilidad no sea una barrera y permite la integración de estándares y políticas en la \"estructura\" desde el principio, facilitando a las startups cumplir y proporcionando a los reguladores un marco más claro para la medición.</li><li><b>Desmitificar la IA y Eliminar la Jerga:</b><br>Alpesh Shah y Amir resaltan la necesidad de dedicar más esfuerzo a eliminar la jerga y desmitificar la IA agéntica para el público en general. Mostrar el valor real de la tecnología es crucial para reducir el miedo innecesario y fomentar una comprensión y adopción más amplias.</li><li><b>Adoptar un \"Minimal Viable Trust Stack\" (para Startups):</b><br>Apoorv Goyal aconseja a las startups establecer un registro de identidad de agente, barreras de seguridad en la orquestación, arquitectura de observabilidad en tiempo real y supervisión clara como requisitos mínimos antes de la producción.</li><li><b>Integrar la Gobernanza como Ventaja Competitiva (para Empresas):</b><br>Las empresas deben integrar la gobernanza profundamente en sus productos y estrategias de salida al mercado desde las primeras etapas, considerándola un diferenciador clave y no una carga.</li></ul></body></html>"
                            }
                        ]
                    }
                }
            ]
        }
    ],
    "headerCols": {
        "col1": "Part Id",
        "col2": "FileName",
        "col3": "Content"
    },
    "fileName": "",
    "projectType": "DOCUMENT",
    "sourceLanguage": "English",
    "targetLanguage": "Spanish",
    "__v": 0
},
    {
    "_id": "699450e025c9f4e0d7e70a14",
    "projectId": "924221",
    "projectName": "Governing Autonomy_ Trust in Agentic AI, Multi-Agent Systems and the Infrastructure of Trust - Hindi",
    "projectDescription": "Governing Autonomy_ Trust in Agentic AI, Multi-Agent Systems and the Infrastructure of Trust.",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T11:28:29.415Z",
    "projectUpdatedDate": "2026-02-17T11:28:29.415Z",
    "projectStatus": "Active",
    "companyId": "69453bf2e603a0a401b6f35b",
    "projectCreatedBy": "69453bf2e603a0a401b6f35c",
    "projectUpdatedBy": "69453bf2e603a0a401b6f35c",
    "projectLastAccessDate": "2026-02-17T11:28:29.415Z",
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "Content",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "Governing Autonomy_ Trust in Agentic AI, Multi-Agent Systems and the Infrastructure of Trust.mp3.docx",
            "col3": "We'll say probably a few words about what they do. I'm sorry, but Eddie Sahai from Google. Go ahead, Eddie. Hi, everyone. I'm Eddie Sahai. I am policy manager working on AI and emerging tech, particularly agent AI and robotics. Hello, and good afternoon. This is Mahesh. Thank you all, and thanks to Amir and Pauline for having us here. And also, they are generous enough to put my younger version. in the picture. And so I'm from Project Nanda. We are pioneering the building and foundational infrastructure for the internet of AI agents. It's a new kind of internet is born, or getting born, from last year onwards. And one of the most foundational issues, the way I say that, trust, that is me. That's my photo. The same thing, how we are gonna solve this problem using in this internet of AI agents, and those primitives need to be solved at the foundational layer. It's not like after the fact or after the thought solutioning. It has to happen as a first principle. Yeah, thank you. Alpesh. Hello, I'm Alpesh Shah. I am the Managing Director of the IEEE Standards Association and a member of the Management Council at IEEE. Thank you. Thank you. Hey, guys. Hi. My name is Apoorv Goyal. I'm a principal at Insight Partners. I help lead a lot of our US-India investing efforts. Just as context, we're one of the leading global technology investment funds. We manage close to $90 billion in AUM investing out of a $12.5 billion fund into leading global software companies around the world, including a bunch of companies based out of India. Wonderful. Thank you so much. I'll just be getting to Praveen. He's my co-moderator. Praveen is the head of AI Lab. but before we start talking about it, with that, as a raise of hand, how many of you have built an agent or played with an agent? How many of you are scared of what's gonna happen with agents? One is probably like 5%, 10%. So it thinks that 90 or 80 or 90% of you don't feel that agents are gonna be dangerous or anything else. Have you followed what happened with Open Cloud recently and Notebook? Do you feel this is cool or exciting or energizing or fun? Who's excited about what happened? It's a beautiful experiment, right? So we're learning about all these things and as we talk about agentic world, the concept of autonomy scares a lot of people. Why? Because we don't fully know how they're built. We don't fully know how they're going to react, if their objective is going to be changing, if they're going to be... mixing with each other, including new babies, we don't know what's gonna happen. And then in that world of autonomy, a lot of decisions that enterprises take and societies functioning may not be able to be put in the hands of agents. So the concept of governance, which is a concept which is very encompassing about how we control the outcome of systems and procedures and ownership and risk management. becomes central again in the world of agent, which was not the case about even a year ago. So today we're going to be talking about this concept of governing autonomy, how we can actually make sure that this world, this future world, we're going to be fully autonomous, not only is our hand to some extent, but we can somehow trust it, and we don't have to really extend. So to get started, maybe we can have a quick demo. What do you think, Praveen? Questions? Okay, perfect, that's a good thing. So maybe I can start with Eli. Eli, in your vantage point, can you help us understand what's different between multi-agent environment and systems and traditional AI that we've been used to? How should we think about it? So with the raise of hand that I saw in this room, I think we are talking with a group of experts. From my perspective, and many of you have already worked with ModBug and have seen OpenClaw and have built agents, so you know better than me that the way that we think about agents, it's not a specific line or a specific point in time that these group of things are agentic and this group of things are non-agentic, rather thinking about them as a spectrum of agency or autonomy going from basic chatbots that can do a little bit of research and at the end present the result. For example, Google Deep Research does a research given your prompt and finds the information and presents that information. That has some agentic features, but doesn't take actions. compared to let's say going all the way towards the end of the autonomy that is, for example, autonomous cars that are completely end to end and they take action in the real world. So I think seeing agents as a spectrum or a continuum of autonomy across different dimensions, including memory, including planning and long-term planning, short-term planning, planning horizon that they can plan. execute things and also autonomy itself. So what I'm trying to say is that we don't want to say agents are the ones that take actions. We want to say agents live on a continuum from very basic autonomy all the way to things like autonomous vehicles that are fully end-to-end autonomous. Mahesh, you started talking about interesting issues, but let's follow what Ellie said. How do we know agents could make an error? How can we actually even know? Yeah, so she used a very nice word. It's the continuum, right? So it's not like a single point responses that you can keep monitoring, you know, like after the agent took an action. And another thing that she mentioned, they become network actors. They will be acting at every single stage. visibly or invisibly. And that makes it mandatory for us to make it, I would say that it has to be a runtime governance, rather than wait for, let it happen, and then we reverse. So the traditional governance models that we have or exist were very much evolved from our understanding of machine-to-machine interaction. Then it evolved into a SaaS model and a cloud model. Then we evolved to the microservices. architecture and we kind of assumed that you know like somebody else is taking care of those governance after we reach at a scale you know like so when autonomy will reach at a scale then we cannot think of like okay we need to solve this problem now. Now we are thinking of like agent as like okay Chad GPT is one of the agents I'm just looking for the search answers. rather than, and slowly we're saying that, oh, okay, it can just go and do some action, it can crawl some web pages. Okay, then we say that, okay, then there will be multi-agent systems, agent to agent interaction, collaboration, that will happen, and just we are adopting it without addressing the foundational issue that the safety and the governance has to be addressed, that's one, that, along with the trust, but also to make it, like, how it can become a deployable. across the enterprises or the public services or the civil society without sacrificing, which is a very important factor, without sacrificing the openness of this execution and the accountability. So those are the two points I would like to mention. So you gave us more information than I asked for. Thank you. incident that happens, of course, we're going to remove bad actors, right? People that are using agents for bad purposes. We're not talking about that, which is a different issue, security, cyber security, or control. We talk about agent themselves making an error or having a mistake. Some of you know that we're talking recently about agentic identity as a way to make sure that who are these agents? Where are they coming from? What's their training? What data have been used? Who created them? who they belong, they have their passport, how they're gonna talk, what language they speak, can we talk with each other in some language and so forth, which brings now the concept of certifying agents. And I'm gonna ask to Alpesh to question, because Alpesh in the IEEE stands, and we're working a lot on many, many concepts ahead of his time on energetic, how do you think about certifying agents in the world of autonomy? How should we think about it? I think this works out. I guess that's one of the most important points, right? Knowing when it's on and when it's not. I say that in jest, but this is also part of when we're thinking about these various models and how they interact with each other. Not having the ability to understand what is the end result or what good looks like creates a difficulty in understanding the parameters by which you can state this is working as it should. Which means that very much, and already we've seeing this done quite successfully in a number of the use cases we've been involved in and prototypes and implementations around this that we've done is that First making sure you have a very strong governance structure in place is critical second is making sure that you have a very clear process by which you are determining what does and doesn't go out or is ready to go into the sandbox or what you're ready to even put into that runtime state, as you were sharing earlier. It also requires ensuring that you have an iterative approach to it. These are not one-time exercises because you won't know necessarily when there might be a shift or where things may go in a different direction, right? And so this requires sort of a constant monitoring approach. Adding to this is also the need to contextualize a lot of what you're looking at. some of the reasons why you may see the behaviors of the multi-agents may be because there was a trigger that you typically do not see or are unaware of, which means that the bounds that you currently had in mind may not be the right bounds. And therefore, it's important to also ensure that you have transparent, accountable, not only logging, but people in that process. Ultimately, at the end, what it is that you can evaluate, let's say, at a much more level of confidence is the means by which it achieves the output. You won't know 100% if the thing that is out there is perfect. But over iterative series, you begin to clearly understand, right? This, out of the Monte Carlo simulation or whatever model you choose to apply here, at some point you have enough data to know. what good should be, right? So our certified program allows for this. At the same time, we have a series of standards that Amir was instrumental in being involved, and a number of folks in the front row here as well, which are focused on data transparency, focused on age-appropriate design, focused on also the critical elements around what does it mean to be accountable, transparent. What does data privacy mean in the context of systems that we don't know what they were to be, right? And what we found the greatest power in is people. It was the community of people that worked on this. They were heavily representative of a mix of lawyers, doctors, artists were even involved with, engineers, and technical people that really understood the contextual problem so well. that it allowed for us to better understand and appreciate what some of those variances might have been. Thank you, Alpesh. Apurva, you invest in many, many startups and companies, and you are in the first line to evaluate them and see if the company is going to have solid technical background and the teams is right. But because Alpesh mentioned the terms of governance, we're talking about governance and governing AI, how do you imagine because we talk about the technical stack, we're all familiar with the technical stack, but what would be a governance stack look like when everything is autonomous? I think the way we think about governance stack in this AI world is a 5K clear model. If I kind of go first is like the build time. Around the build time the idea is okay, how is the company architected around the idea of data governance, model versioning. when you kind of come to the next step, which is basically deploy time. How do you think about the idea of policy, tracking, permissioning, secrets management? When we come to the third piece of it, we think about this whole idea of runtime. How are you ensuring that your architecture allows for real time observability? How do you ensure that there is an idea of kill switch? If things go bad, how do you ensure that you're able to cut it off at the right point of time? Then we think about the whole idea of How do you drive remediation? So do you have the right audit trails to be able to do a post-mortem? Do you have the right incidence response architecture to be able to attack that? And then how can you drive a whole accountability layer, which is around the whole idea of who drives accountability? How do you have the right reporting structures? How can you conduct right post-mortem? Do you have the right set of tools to be able to do a compliance mapping there? So I think we think of governance stack in today's world in a holistic 5K clear model and anybody who thinks about this holistically is where we feel comfortable about the fact that okay, this is the team that is truly thinking about governance in a holistic way and as you think about the best AI native teams, governance is becoming less of a compliance issue but more of like an issue which is new product in GTM because whoever is embedding governance into product is truly winning on the GTM side. And that is becoming a competitive advantage, and so it's becoming a key part of how we evaluate a lot of the companies. Some people jokingly said a few years ago, 10, 20 years ago, we talk about intern insight when the PC era was that. Now we're talking about governance insight. Would you agree with that? 100%. I feel like with so much of uncertainty, I feel like every enterprise that is procuring today, like the upfront conversation before they take on any business through multiple layers of evaluation is talking through a bunch of questions around what is your auditability, what's your traceability, what's your data handling practices, are the kills which is in place because the costs of things going wrong are so high that people are willing to spend millions of dollars in ensuring the right governance while even buying a half a million ACV contract. So I feel like it's becoming a core part of... So how startup, sorry this is a question I didn't plan to ask but how startups that don't have all the funding and all the resources necessary could comply with this level of scrutiny or requirement? I think the way I think about it is when you're starting an agentic AI business in today's world, I call it like you need to have a minimal viable trust stack. You need to be at a high level, you need to be able to tell what is this agent supposed to do? Is it actually doing what it's supposed to do? And if not, can you actually have someone stop doing what it's supposed to do on a real-time basis? And that becomes critical and for that you basically need to have what you call as a clearly defined agent identity registry. You need to have the right set of guardrails at the orchestration layer. You need to have a clear real-time observability architecture. And lastly, you need to have very clear set of defined oversight. I think if you don't have any of this, I don't think you should be launching any Gentic ecosystem into production because you're bound to meet more failures than success and it's not that expensive to do these four basic things and of course as you scale you'll have a lot of compliance start building into your product but this is the minimal viable thing that any startup that's looking to start in the Gentic AI needs to do to be able to go into production. I think you should write an article about that because a lot of us would need that little four stack layer to get started. To use the term orchestration, I would like to pass on to Praveen as we talk about orchestration and coordination. Because as we define multi-agent systems, they need to be coordinated in some ways to do certain things, but also to not do certain things. So Praveen, is you ready for a demo, quick demo? I think it's very tiny, so maybe you can explain. I truly believe that the next phase of AI is going to be Or the agent system. Am I audible? Sorry about that. I thought this was working. So we at Cognizant and Cognizant AI lab truly believe that the next frontier is going to be multi-agent systems. And in that endeavor, sometime in May 2025, we sort of open-sourced our multi-agent framework, which is called Neuro AI Multi-Agent Accelerator, which you can see on the screen here. It's a low-code, no-code framework. that's designed to accelerate the prototype development to scaling. It's a production-grade environment. Currently, it's available under Apache 2.0 license on GitHub. There's an active community that is sort of working on building and enhancing this framework as we speak. Some of the features and the capabilities of this accelerator is that it's a little bit agnostic, right? It's cloud agnostic. So you as a user will have the control over which LLMs that you want the agent network to talk to. You have a control on which environment that you want to deploy on that. And it's interoperable. I mean, currently, we support multiple protocols from an agent to agent communication, our own OSR protocol, as well as A2N MCP server. And most importantly, as we heard about the governance, autonomy, and the guardrails, security is built into it. So we build this from security at the core of it, not as an afterthought. So in the next one minute, and most importantly, we've made it so simple that even a seasoned AI ML engineer to a CXO who has minimal or no coding experience can start building their agent network in a matter of minutes. And that's all. The whole reason is that we want to make this as a de facto standard. and we want everyone sort of adopt it and sort of, you know, not just experiment it, actually build those agent networks at scale and deploy them, right? So let me just do a quick demonstration. And today, what I'm going to do is we'll quickly wipe code an agent network in front of you guys. So this is a wiping environment. So let me just, anybody wants to give me a prompt? Anybody? Anything? Anything you want prompted? Yeah. All right. Can you read, because it's not easy to read. Yeah. I'm saying create a multi-agent network. I'm saying create a multi-agent network for India AI Summit New Delhi, right? That's what they asked. OK. So what you can really see is just give it a few more seconds. What you will see is that there's an hour. multi-agent architect and a designer which is working behind just to understand the intent. And the intent that I gave was just create a multi-agent network for India AI Summit. New Delhi, that's currently happening. First thing that you will see is that the large language model understood my requirement and it created a set of agents. It's defined the agents. Now what it's actually doing is that it's actually connecting those agents to an orchestrator. It's still building through it. And then as a next step, what will happen really is that basically, now every agent needs to be defined. It needs to have its own roles, responsibilities, and what it should do and what it should not do. So currently, that's a step that we are actually going through it, where you can see that the matter of moment, all these individual agents will be populated with descriptors. You can see that, right? So those descriptions are nothing but its definitions, what it should be doing at the end of the day, right? It's still finalizing the agent network. The two things that you need to see at this here is that there are agents which have been specifically created for that agent network. And then there are a few of them which are under the leaf nodes, orange in color. So basically what the designers understood that there are certain agents which are already created and they were available in my environment, so it just connected them directly. So instead of rebuilding them, reinvesting in creating the agent network, it just connected them so you could use it. So it also allows as a designer of the agent network is that you have the control, auditability and traceability of it. At any point you think that that leaf node is not required in that agent, you can just delete it and save it, the server understands the changes and your new agent network is ready to be played around with. So now that the agent network is ready, What I'm really going to do is launch it. You can see that the agent network is made available. Again, just let me blow it a little bit here. And before I fire a query, and again, this is a development environment. The reason we are showing a development environment is just to showcase a couple of capabilities. What you see on the right-hand side panel is basically your chat window. And then you have an internal chat. which basically shows in real time the communication between the agents, what kind of information is being transmitted, what the nature of it is. So you as a designer have the control over it as well. So you can take a dump of it later on and you can sort of go through it, review it to see if any confidential data is being leaked through a large language model got access to it, right? And then you have the logs. The logs are pretty granular in nature in the sense that it gives you details of a level of the token usage and the cost associated with each of that communication. And last but not the least, I mean, most of the folks here are pro ESD metrics, right? So you also get a score of energy usage, carbon footprint, and the cost associated with each of the prom that you are filing. So let me just, so I'm asking the agent, it goes, can you provide a schedule for the keynote speeches at the India AI Summit, right? So I think it's. OK, let me see if I can get. So currently, it's not grounded. It's just talking to a large language model behind. So every agent network that we create here, by default, it talks to GPT-4. But you, as a user, have a control. As I said, it's LLM agnostic. So if you in your organization have a qualified large language model that you want this agent network to, we can. So globally, each and every agent network has access to one large language model. But the framework has the coolest capability where Every leaf node agent also can be empowered with two specialized or general purpose agents, LLMs as well. So they act as a fallback. And if you have a specialized agent, like for auditing or governance or compliance, you want higher accuracy, you can sort of connect a specific large language model just to that leaf node agent as well. So that's sort of the, and today it's available on GitHub. It's been there for the last five months now. We have an active community. The GitHub repository gives you access to the code base under the Apache 2.0 license. And this framework has met the highest standard of GitHub security as well today. So that's an example of a... You have no queries to run? Yeah. So that's an example of a coordination which basically takes into account all the considerations that was discussed. And you said there's open source, right? Yes, it's open source. So here is the QR code. And if you need more, here is a QR code that you can sort of scan. It will directly take you to the GitHub repository that we have. We encourage you to fork it, clone it, and if you like it, give us a star. And if you don't like it, let us know why. And we are here to work with you on that. I just have a question. If you have multiple agents running at the same time, All of them have different contexts and different memories. How do you manage that in a single platform? Good question. In the interest of time, let's come back to it offline. But we'll get into that. We'll get into it. Great question. So that was just a demo, too, basically. Because the audience is diverse, choosing what these things look like. But this has been certified ISO 4 as a pneumonia compared with this framework, as well. So there's a lot of that behind the scene. But I think what I was talking about has been taken into account in terms of observability, and monitoring, and so forth. Going back to the issues of incident and tracking incident and danger that could happen, there are examples in other areas such as aviation and nuclear power, et cetera. Maybe Ellie can give us a view of how the other industries have dealt with these issues because this issue is, today we talk about agentic autonomy. We have these issues before in other sectors. How have we dealt with them? Absolutely. The technology may be new, but the concept of safety, especially in safety critical industries, is not new. We have seen that before and we've solved for it. Let's take the example of aviations and drones, for example. So drones are a newer technology and a lot of regulatory bodies are looking at regulating drones. And the interesting thing about looking at drones and comparing them with agents is that, when you look at, let's say, the regulations that, at least in the U.S., aviation industry is considering for drones, it was originally based on what FAA or Federal Aviation Administration calls VLOS, meaning visual line of sight, meaning that a pilot has all these is in command and they have to keep visual line of sight with the drone. And that is how they are accountable for managing that drone. However, as the safety and as the system safety based on using AR technology develops, now the VLAS regulation is moving to beyond visual line of sight. And that means that now the safety of an AI detect and avoid system has gone above what a human can do. And based on this advantage of detect and avoid based on AI system, now the regulation is moving from a pilot has to always keep visual line of sight to beyond visual line of sight. And using that as... as a way to maintain safety. So drawing a comparison to agents, if you think of, when we talk about human in the loop, human in the loop, we are basically talking about the loss. The human has to keep approving every step of the way that agents are taking. However, that undermines the utility of the agent. So as agents become safer and more reliable, similar to what detect and avoid in the aviation industry, and got better and better and better over time, we have to be moving from human in the loop to pilot in command in aviation industry. In this case, we'll be human in command. So human will be doing the supervision, but not in the loop, approving everything that agent is supposed to do. So drawing this kind of parallels with respect to other industries that are moving from safety of human. making those decisions make a lot of sense to apply to agentic AI because similar to aviation, in this case we see that maybe keeping the human always in the loop is not the best thing. Similar to detect and avoid, there could be better safety systems that AI can provide that keeps agents even safer. I'm not saying that data-wise today, I'm saying similar to other industries that moved from one side to the other, agentic AI would also move. Now we are human in the loop, and then we have to move towards human in command. Thank you very much. As you talked about the similarity between drones and agents and FAA as a regulation, I'm going back to the rest of the audience, the two of you, to ask a question. In your opinion, what's the balance between engineering guidelines and technical design at the core versus regulations and norms and standards, right? Because both of them are needed, we're not discussing that. But today with what's going on with agents, in your opinion, what's the balance of that? Should it be 80% engineering design and good practices, like observability, measurement, et cetera, or should it be probably a little bit less controlled in engineering, let innovation go, but also provide a framework of certification and standards. How do you balance these two? So I'm having an analogy with the early days of internet, like how internet even evolved at first place, being sure that it's not owned by individual player or anything. So it was all inclusive and very much focused on solving the foundational primitives like identity, discovery, trust. and that was in a fragmented system. It was never a centralized system. So in this context, like when you said, okay, so compliance, it should be, heavy compliance will kind of sabotage the innovation, like how the tear-off we will do, and my opinion on that is we should keep the innovation open, and that remains as a data plane. But the governance has to be the control plane. So they have to build in parallel. We cannot choose one versus other. So yeah, that's my thought. I'm curious to hear Apish's and Apuva's opinion on that as well. Yeah, I mean, I think, Amir, if we took some real world examples, there's some policies regionally that some have claimed were too early in the days And that have led to unnecessary regulation. And the effects of that unnecessary regulation have compromised, perhaps, the innovation that could have emerged as a result. I'm not so sure whether or not I agree with that. I think I'll let time claim that one. But what I would say is that it really just depends what you're trying to do. If you're in early days, you're just trying to understand what is possible and the bounds within your own sandbox, you should be able to test it out and understand. You need the chance to even observe it before you can jump to the conclusion that it's something that's very good to market. At the same time, when you get to the proper maturation point, as my colleagues have shared, this is when regulation starts to make a little bit more sense from a requirements standpoint. Policy can also function, though, in an earlier stage to begin framing. And the framing is important to know just so you have a sense of where is too far, right? But framing is different than regulation. And at a certain point, as I'm sure Pruva knows as well, is that there comes a point where there's a market acceptance of what is normalized. And at that point in time, That's when the market differentiation really starts to play a role. That's when you see the major uptick. Prior to that, you're not necessarily going to see everyone buy into paying $100,000 or $150,000 to get something certified. The costs become a barrier to entry. But this also means that there is perhaps a different way of approaching this. I go back to the comment Amir and Apurva were speaking to earlier of the mini trust stack. You know, in many ways, we can also consider the economic problem of how you make it easier for startups to come into play. And one of those ways could be, let's say theoretically, but maybe some of us are working on this, is the integration of the required standards and policies baked into the fabric from the get-go. And then as you build and stage get beyond that, what you've already done is you've bought time. for those startups. You've also given a level of trust for the governments as they're playing here. And you've given regulators at least something much clearer to measure against. It's very, very difficult to ask someone to measure something and hold you accountable when they don't know what they're supposed to hold you accountable against. And it's the same problem for companies. They don't want to waste time on something that makes no sense in all this market play. So Amir, I hope that helps answer your question, but that's my take on it. Thank you. Can I ask a follow-up? Please go ahead. I may want to answer too. In your experience... Alpesh captured most of what I said, so I would save the time. Actually, I wanted to get your perspective on how you see founders actually balancing this. Again, I feel like the idea is everybody's working with the certain idea of what that framework looks like. And I feel like... there is not a lot of mutual tension between deploying responsibly AI and what leads to high commercial deployment. It's like if you need to be a half a billion ARR business today, you need to be able to sell into enterprises in mid-market. And those mid-market solutions are selling into either government institutions, large-scale enterprises, and all of them today have a much larger business to protect. And so if you're experimenting with a new technology that is evolving very, very fast, and if you're not able to provide for certain frameworks of those responsible guidelines, it's like, hey, I wanna make sure my data is being handled well. I wanna make sure my secrets are being managed in the right way. The agents have the right set of permissions and policies. Those guidelines are being, and I think today those frameworks are very individual. are localized to the company that you're trying to serve. But if you kind of distill that, those will broadly come down to like 15, 20 large principles. And most of the innovation that is happening is happening within the bound of those 20, 25 frameworks. I think to Altesha's point, that is going to start solidifying a lot more as the technology matures. But I think people always try and frame being responsible and commercial innovation as a trade-off, I feel like with the way the world is moving, it's not a tension. It's actually very, very aligned. And that's how even most of the board conversations happen is because if you can't do that, you would not be able to scale. So on that note, staying with you, Apurva, right? You talked about the trade-offs. You talked about the investment, right? If we switch gears into the economics of it, right? So when do you see the governance actually becoming a competitive advantage and not a cost burden to the organization? To be very honest, in very early stages of the company, we have a lot of our portfolio companies at seed series A stage start selling into enterprises day one. So if you are selling into the cognizance of the world or if you're selling into the Googles of the world and all of that, a bunch of these companies today and just how innovative they are and how fast they're moving, like day one they sell into. a half a million, a million ARR contract into a very large enterprise. And when you're doing that, you can't do that without governance. Like SOCTU used to be that version, SOCTU compliance for like your old age SAS framework there. So I feel like most of these conversations become pretty much very, very embedded in your seed series. The moment the company's about to sell, if governance is not an embedded conversation in your product, it will not be a competitive edge for you in the GTN. And to be honest, I meet 100 AI companies a month. The best AI native teams today have actually embedded it so deeply that, like security, it's part of the weekly operating cadence. They'll talk about the idea around evals. How solid are your eval reviews? Are you doing it on a weekly basis? Are you making sure that... meeting the standards that you want to espouse serving into an enterprise customer. They're thinking about red teaming in a very solid way. They're doing very high quality post-mortem reviews if things don't work the way they are supposed to work. So I feel like governance is like security today and they will start playing out pretty much in a series A conversation if you're selling into an enterprise. I feel like that conversation becomes later if you're a PLG led business. that is serving individual creators, individual designers. I don't think if you are trying to buy a $50 per month contract as a designer and using that to create videos, you're thinking as much about, hey, is this meeting those governance standards that I need to? But anyone selling into a mid-market enterprise solution, this becomes a very early on conversation. I have just a follow-up thought on that, right? So our India's digital infrastructure story, like when we built Aadhaar, we built UPI payment system, we solved identity problem, we solved the payment problem, then we were like, okay, what about the data privacy, then we have the DG Locker, then we have OCD, and then, now what is happening, that's very important success story, and very much aligned with this agentic AI, because, the government took a stand saying that, hey, we are going to standardize the interfaces, okay? And what we really want to do, the rest of the people or the companies, they can do any application that they want on top of it. So that government plays a very important role in this agentic AI evolution, I would say. Thank you. Thank you. At the beginning, we talked about governing autonomy and we discussed about, autonomy talks about governance to some extent, I think. What would be, in your opinion, well, there are two aspects to that question. One is if we count about trust, what is infrastructure for trust? Because it's so fragmented between different pieces. And then how should we think about the one single thing that would enable us to go to the right direction? Because we talked about, again, many things, but in your opinion, what would be the one single thing that we have to do today, given what you expect? What you mentioned is just that we need to give innovation open, we need to experiment, we need to have some general guidance and framing, of course. But besides that, what would be one little thing that we need to really be pushing collectively, not as individuals but collectively, to make everybody progress, besides open sourcing it, of course? Eileen, do you want to start? Sure. Yeah, I can start. Maybe one of the things that we can think about today is how can we evolve testing methods and evaluation benchmarks for multi-agent systems. And the reason I say that is that we have a paper called Distributional AGI Safety, meaning that the paper discusses the fact that AGI may not emerge as one monolithic, big, powerful system, but it may actually emerge as a lot of sub. top AGI agents, specialized agents, working with each other in a multi-agent system. So we may be talking about multi-agent systems that reach AGI level capabilities. And for that to be successful, what should we be thinking about today when it arrives? In that paper, we discussed four different categories of safety, including market design, including baseline safety for each agent itself, including inter-agent communication, et cetera, et cetera. Please feel free to refer to that paper. But I think what we need to do today is making sure that we build evaluation benchmarks that are relevant to multi-agent systems. Frontier Labs that are developing these models that are core to the agents, it's important for them to, and by them I mean Google Live Mind being one of them, to report, test against these benchmarks, report transparently about the limitations, report transparently about the capabilities, and I think that is one of the goals that. Google DeepMind has been pursuing, and a lot of that information is available on the website. Please feel free to read our frameworks on AGI safety as well. So I'll stop there. So I'll just, again, follow up on that, right? One of the important factors that we're missing in the multi-agent system, we are assuming that it's a single model agent, right? Across the multi-agent system, I'm using one LLM model. But when we extend that thought, that what will happen when I'm using models provided by different players, open source or non-open source, you know, like all of those models will try to... So, single-matter responses are easy to verify. We can have the metrics or all the telemetry associated with it to verify whether the agent is claiming. But when it goes across those boundaries... And it becomes like a multimodal system, multi-agentic system. And every single company is saying, hey, trust us. So one of the thought experiments, or the way we try to solve it, is get every voice heard. Meaning that let's hear from all the academia how they think about this problem. Get all the LLM or a big. per companies getting involved in this discussion. The same thing with the system integrators and hyperscalers also participate in it. And not just them, but including the policy makers as",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "Governing Autonomy_ Trust in Agentic AI, Multi-Agent Systems and the Infrastructure of Trust.mp3.docx",
            "col3": "well as the younger generation who is gonna be the agentic AI native. So how are we preparing for that? That's very important. So the problem doesn't... get solved with the single model responses. So the only solution that we kind of lean towards to make it enough decentralized or enough integrated system so that we can have those test parameters across every single thing. Yeah, thank you. I think for me, largely, the thing which I think can solve a lot of it because that kind of distills enough information to people to start working from is the idea around auditability and traceability to be able to capture and accurate reporting in an immutable way of a lot of what is happening. And as long as there are more open source platforms that allow for a lot of those, you know, anonymized way of auditability and traceability and the incidents that's coming to come up, I think it drives for more collective learning at a more global platform level. I don't know if there are... And this is a conversation where when we meet a lot of startups, they're actually learning a lot from talking to each other. And they keep doing about the fact that there are not a lot of global platforms that just talk about just learning around auditable trails of what is happening, how is it happening, where do systems fail at what scale and level of infrastructure. And I think something that solves for just a collective platform to learn from. will go a long way. It's a very innovative environment around the world and what's happening. So as long as you provide the right infrastructure for even knowing what's going wrong, you know, I think there's a lot that can actually accelerate the learning and development there. That's a cue for our session. Oh, I thought that was where you programmed the agenda. Yeah, so I think there's two. Two key points I'd like to just leave everyone with. The first, Amir runs a very innovative organization called AI Commons that many of us have been a part of over the years. And really the critical nature of what they have been looking at, which very much aligned with a lot of what we were looking at as well, was the idea of having these technologies treated as a commons, a public utility. Because if you don't do that, you can't do exactly what everyone here is speaking to. Only those that have access to that can do that. And the point on education is very critical. But the point, I would just read into this then, is the thing to do today, we have to spend a lot more time getting rid of the jargon. Even the title of today's session. isn't something that's accessible readily to the same people we're asking to make use of these technologies, right? And so a lot of the effort needs to be spent on demystifying what we're talking about. Show the value of it. Because if we don't do that, then you're going to have a lot of fear. And a lot of the fear is unnecessary. And I think these would be the critical things. I would also love to use your app to turn off that alarm. So maybe you can teach us later about that. Thanks. Thank you very much. Thank you. Thank you everyone for coming.",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "projectDeleted": false,
    "codeFrameColumn": [
        {
            "col1": "col1",
            "col2": "col3",
            "researchObjective": "",
            "filterColumn": [],
            "netSetting": true,
            "subNetSetting": true,
            "categorySetting": true,
            "sentimentSetting": true,
            "thinkingBudgetSetting": 2000,
            "questionType": "OPENEND",
            "otherBrand": false,
            "existingCodeFrame": [],
            "isList": false,
            "codeFrameWithDataList": [],
            "codeFrameList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 10362,
                        "CandidatesTokenCount": 3838,
                        "TotalTokenCount": 15605,
                        "thoughtsTokenCount": 1405,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Net": "एजेंट AI और स्वायत्तता की अवधारणा",
                                "Subnets": [
                                    {
                                        "Subnet": "एजेंट की परिभाषा और दायरा",
                                        "Categories": [
                                            {
                                                "Category": "स्वायत्तता का स्पेक्ट्रम",
                                                "Summary": "एजेंटों को बुनियादी चैटबॉट से लेकर पूरी तरह से स्वायत्त वाहनों तक स्वायत्तता के एक निरंतरता के रूप में देखना।"
                                            },
                                            {
                                                "Category": "मल्टी-एजेंट सिस्टम",
                                                "Summary": "एजेंटों का एक-दूसरे के साथ बातचीत करना और सहयोग करना।"
                                            },
                                            {
                                                "Category": "AI एजेंटों के प्रकार",
                                                "Summary": "चैटबॉट, रिसर्च एजेंट, एक्शन-टेकिंग एजेंट।"
                                            }
                                        ],
                                        "Summary": "यह उप-विषय AI एजेंटों की मौलिक परिभाषाओं, उनके कार्य क्षेत्र और विभिन्न प्रकार के एजेंटों को स्पष्ट करता है, उन्हें स्वायत्तता के एक स्पेक्ट्रम पर रखता है।"
                                    },
                                    {
                                        "Subnet": "एजेंट AI के साथ चुनौतियाँ",
                                        "Categories": [
                                            {
                                                "Category": "परिणामों की अनिश्चितता",
                                                "Summary": "यह न जानना कि एजेंट कैसे प्रतिक्रिया देंगे या उनके उद्देश्य कैसे बदलेंगे।"
                                            },
                                            {
                                                "Category": "सुरक्षा और खतरे",
                                                "Summary": "एजेंटों के खतरनाक या अप्रत्याशित होने की आशंका।"
                                            },
                                            {
                                                "Category": "नियंत्रण और स्वामित्व",
                                                "Summary": "यह तय करना कि सिस्टम का नियंत्रण किसके हाथ में होगा।"
                                            }
                                        ],
                                        "Summary": "यह उप-विषय AI एजेंटों के उपयोग से जुड़ी आंतरिक चुनौतियों, विशेष रूप से उनके अप्रत्याशित व्यवहार और सुरक्षा जोखिमों पर प्रकाश डालता है।"
                                    },
                                    {
                                        "Subnet": "एजेंट AI का विकास",
                                        "Categories": [
                                            {
                                                "Category": "इंटरनेट ऑफ AI एजेंट्स",
                                                "Summary": "AI एजेंटों के लिए बुनियादी ढांचे का निर्माण।"
                                            },
                                            {
                                                "Category": "अगली पीढ़ी का AI",
                                                "Summary": "मल्टी-एजेंट सिस्टम को AI का अगला चरण मानना।"
                                            }
                                        ],
                                        "Summary": "यह उप-विषय AI एजेंट प्रौद्योगिकी के वर्तमान और भविष्य के विकास को संबोधित करता है, जिसमें AI एजेंटों के लिए नए बुनियादी ढांचे और मल्टी-एजेंट सिस्टम की उभरती भूमिका पर जोर दिया गया है।"
                                    }
                                ],
                                "Summary": "यह विषय AI एजेंटों की पहचान, उनकी बुनियादी परिभाषाओं, कार्यक्षमताओं, और उनसे जुड़ी प्राथमिक चुनौतियों को समाहित करता है। यह AI के इस उभरते क्षेत्र के सैद्धांतिक और व्यावहारिक पहलुओं पर केंद्रित है।"
                            },
                            {
                                "Net": "एजेंट AI की शासन, सुरक्षा और विश्वास",
                                "Subnets": [
                                    {
                                        "Subnet": "शासन का महत्व",
                                        "Categories": [
                                            {
                                                "Category": "रनटाइम शासन",
                                                "Summary": "एजेंट के कार्यों के बाद प्रतिक्रिया देने के बजाय वास्तविक समय में शासन।"
                                            },
                                            {
                                                "Category": "आधारभूत सुरक्षा और ट्रस्ट",
                                                "Summary": "AI एजेंटों को तैनात करने से पहले सुरक्षा और भरोसे के मुद्दों को संबोधित करना।"
                                            },
                                            {
                                                "Category": "जवाबदेही और पारदर्शिता",
                                                "Summary": "यह सुनिश्चित करना कि एजेंटों के कार्यों के लिए जवाबदेही हो और वे पारदर्शी रूप से कार्य करें।"
                                            }
                                        ],
                                        "Summary": "यह उप-विषय AI एजेंटों के लिए प्रभावी शासन संरचनाओं की महत्वपूर्ण आवश्यकता पर केंद्रित है, जिसमें वास्तविक समय की निगरानी और मौलिक सुरक्षा व भरोसे के सिद्धांतों को एकीकृत करने पर जोर दिया गया है।"
                                    },
                                    {
                                        "Subnet": "शासन स्टैक की अवधारणा",
                                        "Categories": [
                                            {
                                                "Category": "बिल्ड टाइम शासन",
                                                "Summary": "डेटा शासन और मॉडल संस्करण के आसपास कंपनी की वास्तुकला।"
                                            },
                                            {
                                                "Category": "डिप्लॉय टाइम शासन",
                                                "Summary": "नीति ट्रैकिंग, अनुमतिकरण और रहस्य प्रबंधन पर विचार।"
                                            },
                                            {
                                                "Category": "रनटाइम शासन (तकनीकी)",
                                                "Summary": "वास्तविक समय की अवलोकनीयता और किल स्विच सुनिश्चित करना।"
                                            }
                                        ],
                                        "Summary": "यह उप-विषय AI सिस्टम के जीवनचक्र में शासन को एकीकृत करने के लिए एक संरचित दृष्टिकोण का विवरण देता है, जिसमें विकास से लेकर परिनियोजन तक के विभिन्न चरणों को शामिल किया गया है।"
                                    },
                                    {
                                        "Subnet": "त्रुटियों और जोखिमों का प्रबंधन",
                                        "Categories": [
                                            {
                                                "Category": "त्रुटियों की पहचान",
                                                "Summary": "यह समझना कि एजेंट कब गलती कर सकते हैं और उनका पता कैसे लगाया जाए।"
                                            },
                                            {
                                                "Category": "खराब अभिनेताओं को हटाना",
                                                "Summary": "बुरे उद्देश्यों के लिए एजेंटों का उपयोग करने वाले व्यक्तियों को हटाना।"
                                            },
                                            {
                                                "Category": "घटना ट्रैकिंग और प्रतिक्रिया",
                                                "Summary": "त्रुटियों के लिए ऑडिट ट्रेल और घटना प्रतिक्रिया वास्तुकला का होना।"
                                            }
                                        ],
                                        "Summary": "यह उप-विषय AI एजेंटों के साथ संभावित त्रुटियों और सुरक्षा उल्लंघनों की पहचान करने, ट्रैक करने और प्रतिक्रिया देने के लिए प्रक्रियाओं पर केंद्रित है, जिससे प्रभावी जोखिम प्रबंधन सुनिश्चित होता है।"
                                    },
                                    {
                                        "Subnet": "कानूनी और नियामक विचार",
                                        "Categories": [
                                            {
                                                "Category": "नवाचार बनाम विनियमन",
                                                "Summary": "प्रौद्योगिकी नवाचार को बढ़ावा देने के लिए विनियमन और मानकीकरण के बीच संतुलन।"
                                            },
                                            {
                                                "Category": "क्षेत्रीय नीतियां",
                                                "Summary": "कुछ क्षेत्रीय नीतियों का नवाचार पर संभावित प्रभाव।"
                                            },
                                            {
                                                "Category": "बाजार स्वीकार्यता",
                                                "Summary": "विनियामक ढांचे के साथ बाजार स्वीकार्यता का संबंध।"
                                            }
                                        ],
                                        "Summary": "यह उप-विषय AI एजेंटों के आसपास के कानूनी और नियामक परिदृश्य की पड़ताल करता है, जिसमें नवाचार को बाधित किए बिना प्रभावी नीतियों को तैयार करने की चुनौतियों और बाजार अपनाने पर उनके प्रभाव पर विचार किया गया है।"
                                    }
                                ],
                                "Summary": "यह विषय AI एजेंटों के संचालन में विश्वास स्थापित करने, जोखिमों को कम करने और नैतिक उपयोग सुनिश्चित करने के लिए शासन, सुरक्षा प्रोटोकॉल और विश्वसनीय सिस्टम की आवश्यकता को समाहित करता है। यह तकनीकी और नीतिगत दोनों पहलुओं को कवर करता है।"
                            },
                            {
                                "Net": "एजेंट AI के लिए मानक और प्रमाणन",
                                "Subnets": [
                                    {
                                        "Subnet": "मानक और प्रमाणन की आवश्यकता",
                                        "Categories": [
                                            {
                                                "Category": "एजेंट पहचान",
                                                "Summary": "एजेंटों की उत्पत्ति, प्रशिक्षण और स्वामित्व को ट्रैक करना।"
                                            },
                                            {
                                                "Category": "एजेंट प्रमाणन",
                                                "Summary": "AI एजेंटों के प्रदर्शन, सुरक्षा और विश्वसनीयता के लिए औपचारिक प्रमाणीकरण प्रक्रिया।"
                                            },
                                            {
                                                "Category": "IEEE मानक",
                                                "Summary": "AI में डेटा पारदर्शिता, आयु-उपयुक्त डिज़ाइन और जवाबदेही के लिए मानक।"
                                            }
                                        ],
                                        "Summary": "यह उप-विषय AI एजेंटों के लिए स्पष्ट पहचान, सत्यापन और गुणवत्ता आश्वासन के लिए मानकों और प्रमाणन की आवश्यकता पर जोर देता है, जिससे उनके सुरक्षित और विश्वसनीय संचालन को सुनिश्चित किया जा सके।"
                                    },
                                    {
                                        "Subnet": "प्रमाणीकरण प्रक्रियाएं",
                                        "Categories": [
                                            {
                                                "Category": "शासन संरचना",
                                                "Summary": "प्रमाणीकरण के लिए एक मजबूत शासन संरचना का होना।"
                                            },
                                            {
                                                "Category": "पुनरावृत्तीय दृष्टिकोण",
                                                "Summary": "निरंतर निगरानी और समायोजन के साथ एक पुनरावृत्तीय प्रमाणीकरण प्रक्रिया।"
                                            },
                                            {
                                                "Category": "संदर्भित मूल्यांकन",
                                                "Summary": "विशेष व्यवहारों को समझने के लिए एजेंटों के संदर्भ में मूल्यांकन करना।"
                                            }
                                        ],
                                        "Summary": "यह उप-विषय AI एजेंटों के लिए प्रमाणीकरण को सफल बनाने के लिए आवश्यक प्रक्रियाओं और दृष्टिकोणों को रेखांकित करता है, जिसमें निरंतर सुधार और अनुकूलनशीलता पर जोर दिया गया है।"
                                    },
                                    {
                                        "Subnet": "सामुदायिक भागीदारी",
                                        "Categories": [
                                            {
                                                "Category": "बहु-विषयक समुदाय",
                                                "Summary": "कानूनविदों, डॉक्टरों, कलाकारों और इंजीनियरों सहित विविध समूहों की भागीदारी।"
                                            },
                                            {
                                                "Category": "सामुदायिक ज्ञान",
                                                "Summary": "विशेषज्ञों के विविध समुदाय से अंतर्दृष्टि का लाभ उठाना।"
                                            }
                                        ],
                                        "Summary": "यह उप-विषय मानकों और प्रमाणन प्रक्रियाओं के विकास में व्यापक सामुदायिक भागीदारी के महत्व पर प्रकाश डालता है, यह सुनिश्चित करता है कि विभिन्न दृष्टिकोणों को एकीकृत किया जाए।"
                                    }
                                ],
                                "Summary": "यह विषय AI एजेंटों की विश्वसनीयता, सुरक्षा और जवाबदेही सुनिश्चित करने के लिए मानकों, दिशानिर्देशों और प्रमाणन कार्यक्रमों के विकास और कार्यान्वयन की पड़ताल करता है। यह इस क्षेत्र में उद्योग निकायों और विशेषज्ञों की भूमिका पर प्रकाश डालता है।"
                            },
                            {
                                "Net": "तकनीकी कार्यान्वयन और उपकरण",
                                "Subnets": [
                                    {
                                        "Subnet": "एजेंटों के लिए बुनियादी ढाँचा",
                                        "Categories": [
                                            {
                                                "Category": "न्यूनतम व्यवहार्य विश्वास स्टैक",
                                                "Summary": "किसी भी एजेंटिक AI उत्पाद के लिए एक बुनियादी विश्वास ढांचा।"
                                            },
                                            {
                                                "Category": "ऑर्केस्ट्रेशन लेयर गार्ड्रेल",
                                                "Summary": "एजेंटों के बीच संचार और समन्वय के लिए सुरक्षित नियम।"
                                            },
                                            {
                                                "Category": "वास्तविक समय अवलोकनीयता",
                                                "Summary": "एजेंटों के व्यवहार की वास्तविक समय में निगरानी करने की क्षमता।"
                                            }
                                        ],
                                        "Summary": "यह उप-विषय AI एजेंटों के सुरक्षित और प्रभावी संचालन के लिए आवश्यक तकनीकी बुनियादी ढांचे और न्यूनतम आवश्यकताओं पर केंद्रित है, जिसमें विश्वास और नियंत्रण सुनिश्चित करने के लिए महत्वपूर्ण घटकों को रेखांकित किया गया है।"
                                    },
                                    {
                                        "Subnet": "ओपन-सोर्स फ्रेमवर्क",
                                        "Categories": [
                                            {
                                                "Category": "न्यूरो AI मल्टी-एजेंट एक्सेलेरेटर",
                                                "Summary": "कोग्निजेंट द्वारा विकसित एक लो-कोड/नो-कोड मल्टी-एजेंट फ्रेमवर्क।"
                                            },
                                            {
                                                "Category": "क्लाउड और LLM अज्ञेयवादी",
                                                "Summary": "विभिन्न क्लाउड वातावरण और बड़े भाषा मॉडल के साथ संगतता।"
                                            },
                                            {
                                                "Category": "अंतर्निहित सुरक्षा और प्रोटोकॉल",
                                                "Summary": "सुरक्षा को एक मूलभूत विशेषता के रूप में और एजेंट-से-एजेंट संचार के लिए विभिन्न प्रोटोकॉल का समर्थन करना।"
                                            }
                                        ],
                                        "Summary": "यह उप-विषय ओपन-सोर्स मल्टी-एजेंट फ्रेमवर्क के विकास और क्षमताओं का विवरण देता है, जो AI एजेंटों के परिनियोजन और प्रबंधन को गति देने के लिए डिज़ाइन किए गए हैं।"
                                    },
                                    {
                                        "Subnet": "परीक्षण और मूल्यांकन",
                                        "Categories": [
                                            {
                                                "Category": "परीक्षण विधियों का विकास",
                                                "Summary": "मल्टी-एजेंट सिस्टम के लिए परीक्षण और मूल्यांकन बेंचमार्क का उन्नयन।"
                                            },
                                            {
                                                "Category": "फ्रंटियर लैब रिपोर्टिंग",
                                                "Summary": "बड़े मॉडल की सीमाओं और क्षमताओं पर पारदर्शी रिपोर्टिंग।"
                                            },
                                            {
                                                "Category": "वितरणात्मक AGI सुरक्षा",
                                                "Summary": "AGI के एक मोनोलिथिक सिस्टम के बजाय उप-एजेंटों के रूप में उभरने की संभावना।"
                                            }
                                        ],
                                        "Summary": "यह उप-विषय AI एजेंटों, विशेष रूप से मल्टी-एजेंट सिस्टम के लिए प्रभावी परीक्षण और मूल्यांकन रणनीतियों की आवश्यकता पर जोर देता है, ताकि सुरक्षा और विश्वसनीयता सुनिश्चित की जा सके।"
                                    }
                                ],
                                "Summary": "यह विषय AI एजेंटों के विकास, परिनियोजन और प्रबंधन के लिए आवश्यक तकनीकी समाधानों, उपकरणों और फ्रेमवर्क पर केंद्रित है, जिसमें व्यावहारिक कार्यान्वयन और परीक्षण के तरीकों पर प्रकाश डाला गया है।"
                            },
                            {
                                "Net": "उद्योग तुलना और आर्थिक प्रभाव",
                                "Subnets": [
                                    {
                                        "Subnet": "अन्य उद्योगों से सीख",
                                        "Categories": [
                                            {
                                                "Category": "विमानन और ड्रोन सुरक्षा",
                                                "Summary": "विमानन से सुरक्षा अवधारणाओं को AI एजेंटों पर लागू करना।"
                                            },
                                            {
                                                "Category": "लूप में मानव से कमांड में मानव",
                                                "Summary": "सुरक्षा प्रणालियों में मानव भागीदारी के स्तर का विकास।"
                                            },
                                            {
                                                "Category": "सुरक्षा-महत्वपूर्ण उद्योग",
                                                "Summary": "अन्य सुरक्षा-महत्वपूर्ण क्षेत्रों में सामना की गई समान चुनौतियाँ।"
                                            }
                                        ],
                                        "Summary": "यह उप-विषय AI एजेंटों के लिए सुरक्षा और शासन रणनीतियों को सूचित करने के लिए विमानन जैसे अच्छी तरह से स्थापित, सुरक्षा-महत्वपूर्ण उद्योगों से प्राप्त सबक की पड़ताल करता है।"
                                    },
                                    {
                                        "Subnet": "शासन का आर्थिक मूल्य",
                                        "Categories": [
                                            {
                                                "Category": "प्रतिस्पर्धात्मक लाभ",
                                                "Summary": "शासन को एक लागत बोझ के बजाय एक प्रतिस्पर्धी लाभ के रूप में देखना।"
                                            },
                                            {
                                                "Category": "उद्यम स्वीकार्यता",
                                                "Summary": "उद्यमों में AI एजेंटों को अपनाने के लिए शासन की आवश्यकता।"
                                            },
                                            {
                                                "Category": "स्टार्टअप चुनौतियाँ",
                                                "Summary": "सीमित संसाधनों वाले स्टार्टअप के लिए नियामक आवश्यकताओं का पालन करने की चुनौती।"
                                            }
                                        ],
                                        "Summary": "यह उप-विषय AI एजेंटों के लिए शासन के आर्थिक निहितार्थों पर विचार करता है, जिसमें यह दर्शाया गया है कि यह बाजार में प्रवेश और उद्यम अपनाने के लिए एक महत्वपूर्ण कारक कैसे हो सकता है।"
                                    },
                                    {
                                        "Subnet": "भारतीय डिजिटल अवसंरचना",
                                        "Categories": [
                                            {
                                                "Category": "आधार और UPI",
                                                "Summary": "भारत के डिजिटल पहचान और भुगतान प्रणालियों की सफलता की कहानी।"
                                            },
                                            {
                                                "Category": "डेटा गोपनीयता और मानकीकरण",
                                                "Summary": "डेटा गोपनीयता और इंटरफेस के मानकीकरण में सरकार की भूमिका।"
                                            },
                                            {
                                                "Category": "सरकार की भूमिका",
                                                "Summary": "AI एजेंट विकास को मानकीकृत करने और सुविधा प्रदान करने में सरकार का महत्वपूर्ण योगदान।"
                                            }
                                        ],
                                        "Summary": "यह उप-विषय भारतीय डिजिटल अवसंरचना के विकास को एक उदाहरण के रूप में प्रस्तुत करता है, जिसमें AI एजेंटों के लिए मानकीकरण और सरकारी समर्थन के लिए सबक पर जोर दिया गया है।"
                                    }
                                ],
                                "Summary": "यह विषय AI एजेंटों के विकास और अपनाने के निहितार्थों को समझने के लिए अन्य उद्योगों से समानताएं खींचता है। यह शासन, सुरक्षा और नवाचार के बीच आर्थिक संतुलन की भी पड़ताल करता है।"
                            },
                            {
                                "Net": "भविष्य की दिशा और सामूहिक प्रयास",
                                "Subnets": [
                                    {
                                        "Subnet": "सामूहिक रूप से आगे बढ़ना",
                                        "Categories": [
                                            {
                                                "Category": "खुला नवाचार",
                                                "Summary": "नवाचार को खुला रखना और विनियमन के साथ संतुलन बनाना।"
                                            },
                                            {
                                                "Category": "सामान्य मार्गदर्शन",
                                                "Summary": "प्रयोग को प्रोत्साहित करने के लिए सामान्य मार्गदर्शन और रूपरेखा।"
                                            },
                                            {
                                                "Category": "सामूहिक सीख",
                                                "Summary": "ऑडिटबिलिटी और ट्रेसबिलिटी पर खुले प्लेटफार्मों के माध्यम से सामूहिक सीख।"
                                            }
                                        ],
                                        "Summary": "यह उप-विषय AI एजेंट प्रौद्योगिकी को आगे बढ़ाने के लिए एक सहयोगी दृष्टिकोण की वकालत करता है, जिसमें खुला नवाचार और साझा ज्ञान को बढ़ावा देने के लिए मार्गदर्शन और प्लेटफार्मों की स्थापना पर जोर दिया गया है।"
                                    },
                                    {
                                        "Subnet": "जटिलता को दूर करना",
                                        "Categories": [
                                            {
                                                "Category": "शब्दावली को हटाना",
                                                "Summary": "AI एजेंटों से संबंधित तकनीकी शब्दावली को सरल बनाना।"
                                            },
                                            {
                                                "Category": "मूल्य का प्रदर्शन",
                                                "Summary": "तकनीक के मूल्य को स्पष्ट रूप से संवाद करना।"
                                            },
                                            {
                                                "Category": "भय को कम करना",
                                                "Summary": "गलतफहमी के कारण होने वाले अनावश्यक भय को कम करना।"
                                            }
                                        ],
                                        "Summary": "यह उप-विषय AI एजेंटों के बारे में व्यापक समझ और स्वीकार्यता को बढ़ावा देने के लिए स्पष्ट संचार और जन जागरूकता के महत्व पर प्रकाश डालता है, जिससे अनावश्यक भय को दूर किया जा सके।"
                                    },
                                    {
                                        "Subnet": "विश्वास के लिए बुनियादी ढाँचा",
                                        "Categories": [
                                            {
                                                "Category": "खंडित विश्वास",
                                                "Summary": "विभिन्न घटकों में खंडित विश्वास को एकीकृत करना।"
                                            },
                                            {
                                                "Category": "बहु-आवाज भागीदारी",
                                                "Summary": "शिक्षाविदों, कंपनियों, नीति निर्माताओं और युवा पीढ़ी से सुनना।"
                                            },
                                            {
                                                "Category": "विकेंद्रीकृत प्रणाली",
                                                "Summary": "परीक्षण मानकों के साथ विकेंद्रीकृत और एकीकृत प्रणाली की आवश्यकता।"
                                            }
                                        ],
                                        "Summary": "यह उप-विषय AI एजेंटों के लिए एक मजबूत और एकीकृत विश्वास बुनियादी ढांचे के निर्माण की पड़ताल करता है, जिसमें विविध हितधारकों की भागीदारी और एक विकेंद्रीकृत दृष्टिकोण के माध्यम से जटिलता को दूर करने पर जोर दिया गया है।"
                                    }
                                ],
                                "Summary": "यह विषय AI एजेंटों के भविष्य को आकार देने के लिए आवश्यक सामूहिक प्रयासों, सहयोग और रणनीतियों पर केंद्रित है, जिसमें ज्ञान साझा करने, संचार और सामुदायिक भागीदारी के महत्व पर प्रकाश डाला गया है।"
                            },
                            {
                                "Net": "996 गैर-विशिष्ट विषय",
                                "Subnets": [
                                    {
                                        "Subnet": "गैर-विशिष्ट",
                                        "Categories": [
                                            {
                                                "Category": "अप्रासंगिक प्रतिक्रियाएँ",
                                                "Summary": "ऐसी प्रतिक्रियाएँ जो पूछे गए प्रश्न या परियोजना के लिए प्रासंगिक नहीं हैं और आसानी से कोड नहीं की जा सकतीं।"
                                            }
                                        ],
                                        "Summary": "इस उप-विषय में ऐसी प्रतिक्रियाएँ शामिल हैं जो किसी भी परिभाषित विषय में स्पष्ट रूप से फिट नहीं होती हैं और उन्हें आगे विश्लेषण या वर्गीकरण की आवश्यकता हो सकती है।"
                                    }
                                ],
                                "Summary": "यह विषय उन सभी प्रतिक्रियाओं को समाहित करता है जो चर्चा के मुख्य विषयों से संबंधित नहीं हैं और जिन्हें किसी विशिष्ट विषयगत श्रेणी में आसानी से कोड नहीं किया जा सकता है।"
                            },
                            {
                                "Net": "997 पता नहीं/कह नहीं सकते",
                                "Subnets": [
                                    {
                                        "Subnet": "गैर-प्रतिबद्ध",
                                        "Categories": [
                                            {
                                                "Category": "अनिश्चित या अस्पष्ट प्रतिक्रियाएँ",
                                                "Summary": "ऐसी प्रतिक्रियाएँ जो गैर-प्रतिबद्ध, अस्पष्ट या अनिर्णायक हैं।"
                                            }
                                        ],
                                        "Summary": "इस उप-विषय में ऐसी प्रतिक्रियाएँ शामिल हैं जो कोई निश्चित राय या जानकारी व्यक्त नहीं करती हैं, जो उत्तरदाता की अनिश्चितता को दर्शाती हैं।"
                                    }
                                ],
                                "Summary": "यह विषय उन प्रतिक्रियाओं को समाहित करता है जहां उत्तरदाता को प्रश्न का उत्तर पता नहीं है, या वे कोई निश्चित जवाब देने में असमर्थ हैं।"
                            },
                            {
                                "Net": "998 लागू नहीं",
                                "Subnets": [
                                    {
                                        "Subnet": "अप्रासंगिक",
                                        "Categories": [
                                            {
                                                "Category": "प्रश्न पर लागू नहीं",
                                                "Summary": "ऐसी प्रतिक्रियाएँ जो प्रश्न के लिए सीधे तौर पर लागू नहीं हैं।"
                                            }
                                        ],
                                        "Summary": "इस उप-विषय में ऐसी प्रतिक्रियाएँ शामिल हैं जो प्रदान किए गए प्रश्न या संदर्भ के लिए प्रासंगिक नहीं हैं।"
                                    }
                                ],
                                "Summary": "यह विषय उन प्रतिक्रियाओं को समाहित करता है जो प्रश्न या सर्वेक्षण के दायरे से बाहर हैं और इसलिए लागू नहीं होती हैं।"
                            },
                            {
                                "Net": "999 कुछ नहीं/खाली",
                                "Subnets": [
                                    {
                                        "Subnet": "व्यर्थ/निरर्थक",
                                        "Categories": [
                                            {
                                                "Category": "अस्पष्ट/गैर-भाषा वर्ण",
                                                "Summary": "ऐसी प्रतिक्रियाएँ जो निरर्थक, व्यर्थ, या गैर-भाषा वर्णों वाली हैं।"
                                            }
                                        ],
                                        "Summary": "इस उप-विषय में ऐसी प्रतिक्रियाएँ शामिल हैं जिनमें कोई सार्थक जानकारी नहीं है, जैसे कि निरर्थक पाठ या खाली प्रतिक्रियाएँ।"
                                    }
                                ],
                                "Summary": "यह विषय उन प्रतिक्रियाओं को समाहित करता है जो पूरी तरह से व्यर्थ हैं, समझ से बाहर हैं, या जिनमें कोई सार्थक सामग्री नहीं है।"
                            }
                        ],
                        "Summary": "यह चर्चा AI एजेंटों और स्वायत्तता के विकसित होते परिदृश्य पर केंद्रित है। Google, Project Nanda, IEEE और Insight Partners के वक्ताओं ने स्वयं और अपने काम का परिचय दिया। प्रमुख विषयों में AI एजेंटों को स्वायत्तता के एक स्पेक्ट्रम के रूप में परिभाषित करना, मल्टी-एजेंट सिस्टम में वास्तविक समय के शासन और विश्वास की महत्वपूर्ण आवश्यकता, मौलिक सुरक्षा का महत्व, और जवाबदेही व पारदर्शिता सुनिश्चित करने में प्रमाणन व मानकों (जैसे IEEE) की भूमिका शामिल है। AI के लिए \"शासन स्टैक\" की अवधारणा पर भी प्रकाश डाला गया है, जिसमें अवलोकनीयता (observability), किल स्विच, ऑडिट ट्रेल और जवाबदेही पर जोर दिया गया है। एक ओपन-सोर्स मल्टी-एजेंट फ्रेमवर्क (Neuro AI मल्टी-एजेंट एक्सेलेरेटर) का प्रदर्शन किया गया, जिसमें एजेंट ऑर्केस्ट्रेशन, संचार और अंतर्निहित सुरक्षा की क्षमताओं को उजागर किया गया। विमानन जैसे सुरक्षा-महत्वपूर्ण उद्योगों के साथ समानताएं खींची गईं, जिसमें \"लूप में मानव\" से \"कमांड में मानव\" में बदलाव की वकालत की गई। नवाचार और विनियमन के बीच संतुलन पर बहस की गई, जिसमें स्टार्टअप के लिए प्रारंभिक चरण के \"न्यूनतम व्यवहार्य विश्वास स्टैक\" और शुरुआत से ही शासन को एकीकृत करने के आर्थिक लाभ पर ध्यान केंद्रित किया गया। चर्चा सामूहिक रूप से आवश्यक कार्यों के साथ समाप्त हुई, जैसे मल्टी-एजेंट सिस्टम के लिए परीक्षण बेंचमार्क विकसित करना, साझा सीखने के लिए खुले प्लेटफार्मों को बढ़ावा देना, और भय को कम करने व व्यापक अपनाने को बढ़ावा देने के लिए AI शब्दावली को स्पष्ट करना।"
                    }
                }
            ],
            "onlyCodeFrame": true,
            "errorList": [],
            "codeFrameGenerateType": "CODE_FRAME_ONLY",
            "requestType": "CODEFRAME",
            "requestMode": "DEVELOPMENT",
            "sourceLanguage": "English",
            "targetLanguage": "Hindi",
            "response": {
                "status": "OK",
                "data": {
                    "status": "OK",
                    "PromptTokenCount": 10362,
                    "CandidatesTokenCount": 3838,
                    "TotalTokenCount": 15605,
                    "thoughtsTokenCount": 1405,
                    "finishReason": "STOP",
                    "data": [
                        {
                            "Net": "एजेंट AI और स्वायत्तता की अवधारणा",
                            "Subnets": [
                                {
                                    "Subnet": "एजेंट की परिभाषा और दायरा",
                                    "Categories": [
                                        {
                                            "Category": "स्वायत्तता का स्पेक्ट्रम",
                                            "Summary": "एजेंटों को बुनियादी चैटबॉट से लेकर पूरी तरह से स्वायत्त वाहनों तक स्वायत्तता के एक निरंतरता के रूप में देखना।",
                                            "code": 3
                                        },
                                        {
                                            "Category": "मल्टी-एजेंट सिस्टम",
                                            "Summary": "एजेंटों का एक-दूसरे के साथ बातचीत करना और सहयोग करना।",
                                            "code": 4
                                        },
                                        {
                                            "Category": "AI एजेंटों के प्रकार",
                                            "Summary": "चैटबॉट, रिसर्च एजेंट, एक्शन-टेकिंग एजेंट।",
                                            "code": 5
                                        }
                                    ],
                                    "Summary": "यह उप-विषय AI एजेंटों की मौलिक परिभाषाओं, उनके कार्य क्षेत्र और विभिन्न प्रकार के एजेंटों को स्पष्ट करता है, उन्हें स्वायत्तता के एक स्पेक्ट्रम पर रखता है।",
                                    "code": 2
                                },
                                {
                                    "Subnet": "एजेंट AI के साथ चुनौतियाँ",
                                    "Categories": [
                                        {
                                            "Category": "परिणामों की अनिश्चितता",
                                            "Summary": "यह न जानना कि एजेंट कैसे प्रतिक्रिया देंगे या उनके उद्देश्य कैसे बदलेंगे।",
                                            "code": 7
                                        },
                                        {
                                            "Category": "सुरक्षा और खतरे",
                                            "Summary": "एजेंटों के खतरनाक या अप्रत्याशित होने की आशंका।",
                                            "code": 8
                                        },
                                        {
                                            "Category": "नियंत्रण और स्वामित्व",
                                            "Summary": "यह तय करना कि सिस्टम का नियंत्रण किसके हाथ में होगा।",
                                            "code": 9
                                        }
                                    ],
                                    "Summary": "यह उप-विषय AI एजेंटों के उपयोग से जुड़ी आंतरिक चुनौतियों, विशेष रूप से उनके अप्रत्याशित व्यवहार और सुरक्षा जोखिमों पर प्रकाश डालता है।",
                                    "code": 6
                                },
                                {
                                    "Subnet": "एजेंट AI का विकास",
                                    "Categories": [
                                        {
                                            "Category": "इंटरनेट ऑफ AI एजेंट्स",
                                            "Summary": "AI एजेंटों के लिए बुनियादी ढांचे का निर्माण।",
                                            "code": 11
                                        },
                                        {
                                            "Category": "अगली पीढ़ी का AI",
                                            "Summary": "मल्टी-एजेंट सिस्टम को AI का अगला चरण मानना।",
                                            "code": 12
                                        }
                                    ],
                                    "Summary": "यह उप-विषय AI एजेंट प्रौद्योगिकी के वर्तमान और भविष्य के विकास को संबोधित करता है, जिसमें AI एजेंटों के लिए नए बुनियादी ढांचे और मल्टी-एजेंट सिस्टम की उभरती भूमिका पर जोर दिया गया है।",
                                    "code": 10
                                }
                            ],
                            "Summary": "यह विषय AI एजेंटों की पहचान, उनकी बुनियादी परिभाषाओं, कार्यक्षमताओं, और उनसे जुड़ी प्राथमिक चुनौतियों को समाहित करता है। यह AI के इस उभरते क्षेत्र के सैद्धांतिक और व्यावहारिक पहलुओं पर केंद्रित है।",
                            "code": 1
                        },
                        {
                            "Net": "एजेंट AI की शासन, सुरक्षा और विश्वास",
                            "Subnets": [
                                {
                                    "Subnet": "शासन का महत्व",
                                    "Categories": [
                                        {
                                            "Category": "रनटाइम शासन",
                                            "Summary": "एजेंट के कार्यों के बाद प्रतिक्रिया देने के बजाय वास्तविक समय में शासन।",
                                            "code": 15
                                        },
                                        {
                                            "Category": "आधारभूत सुरक्षा और ट्रस्ट",
                                            "Summary": "AI एजेंटों को तैनात करने से पहले सुरक्षा और भरोसे के मुद्दों को संबोधित करना।",
                                            "code": 16
                                        },
                                        {
                                            "Category": "जवाबदेही और पारदर्शिता",
                                            "Summary": "यह सुनिश्चित करना कि एजेंटों के कार्यों के लिए जवाबदेही हो और वे पारदर्शी रूप से कार्य करें।",
                                            "code": 17
                                        }
                                    ],
                                    "Summary": "यह उप-विषय AI एजेंटों के लिए प्रभावी शासन संरचनाओं की महत्वपूर्ण आवश्यकता पर केंद्रित है, जिसमें वास्तविक समय की निगरानी और मौलिक सुरक्षा व भरोसे के सिद्धांतों को एकीकृत करने पर जोर दिया गया है।",
                                    "code": 14
                                },
                                {
                                    "Subnet": "शासन स्टैक की अवधारणा",
                                    "Categories": [
                                        {
                                            "Category": "बिल्ड टाइम शासन",
                                            "Summary": "डेटा शासन और मॉडल संस्करण के आसपास कंपनी की वास्तुकला।",
                                            "code": 19
                                        },
                                        {
                                            "Category": "डिप्लॉय टाइम शासन",
                                            "Summary": "नीति ट्रैकिंग, अनुमतिकरण और रहस्य प्रबंधन पर विचार।",
                                            "code": 20
                                        },
                                        {
                                            "Category": "रनटाइम शासन (तकनीकी)",
                                            "Summary": "वास्तविक समय की अवलोकनीयता और किल स्विच सुनिश्चित करना।",
                                            "code": 21
                                        }
                                    ],
                                    "Summary": "यह उप-विषय AI सिस्टम के जीवनचक्र में शासन को एकीकृत करने के लिए एक संरचित दृष्टिकोण का विवरण देता है, जिसमें विकास से लेकर परिनियोजन तक के विभिन्न चरणों को शामिल किया गया है।",
                                    "code": 18
                                },
                                {
                                    "Subnet": "त्रुटियों और जोखिमों का प्रबंधन",
                                    "Categories": [
                                        {
                                            "Category": "त्रुटियों की पहचान",
                                            "Summary": "यह समझना कि एजेंट कब गलती कर सकते हैं और उनका पता कैसे लगाया जाए।",
                                            "code": 23
                                        },
                                        {
                                            "Category": "खराब अभिनेताओं को हटाना",
                                            "Summary": "बुरे उद्देश्यों के लिए एजेंटों का उपयोग करने वाले व्यक्तियों को हटाना।",
                                            "code": 24
                                        },
                                        {
                                            "Category": "घटना ट्रैकिंग और प्रतिक्रिया",
                                            "Summary": "त्रुटियों के लिए ऑडिट ट्रेल और घटना प्रतिक्रिया वास्तुकला का होना।",
                                            "code": 25
                                        }
                                    ],
                                    "Summary": "यह उप-विषय AI एजेंटों के साथ संभावित त्रुटियों और सुरक्षा उल्लंघनों की पहचान करने, ट्रैक करने और प्रतिक्रिया देने के लिए प्रक्रियाओं पर केंद्रित है, जिससे प्रभावी जोखिम प्रबंधन सुनिश्चित होता है।",
                                    "code": 22
                                },
                                {
                                    "Subnet": "कानूनी और नियामक विचार",
                                    "Categories": [
                                        {
                                            "Category": "नवाचार बनाम विनियमन",
                                            "Summary": "प्रौद्योगिकी नवाचार को बढ़ावा देने के लिए विनियमन और मानकीकरण के बीच संतुलन।",
                                            "code": 27
                                        },
                                        {
                                            "Category": "क्षेत्रीय नीतियां",
                                            "Summary": "कुछ क्षेत्रीय नीतियों का नवाचार पर संभावित प्रभाव।",
                                            "code": 28
                                        },
                                        {
                                            "Category": "बाजार स्वीकार्यता",
                                            "Summary": "विनियामक ढांचे के साथ बाजार स्वीकार्यता का संबंध।",
                                            "code": 29
                                        }
                                    ],
                                    "Summary": "यह उप-विषय AI एजेंटों के आसपास के कानूनी और नियामक परिदृश्य की पड़ताल करता है, जिसमें नवाचार को बाधित किए बिना प्रभावी नीतियों को तैयार करने की चुनौतियों और बाजार अपनाने पर उनके प्रभाव पर विचार किया गया है।",
                                    "code": 26
                                }
                            ],
                            "Summary": "यह विषय AI एजेंटों के संचालन में विश्वास स्थापित करने, जोखिमों को कम करने और नैतिक उपयोग सुनिश्चित करने के लिए शासन, सुरक्षा प्रोटोकॉल और विश्वसनीय सिस्टम की आवश्यकता को समाहित करता है। यह तकनीकी और नीतिगत दोनों पहलुओं को कवर करता है।",
                            "code": 13
                        },
                        {
                            "Net": "एजेंट AI के लिए मानक और प्रमाणन",
                            "Subnets": [
                                {
                                    "Subnet": "मानक और प्रमाणन की आवश्यकता",
                                    "Categories": [
                                        {
                                            "Category": "एजेंट पहचान",
                                            "Summary": "एजेंटों की उत्पत्ति, प्रशिक्षण और स्वामित्व को ट्रैक करना।",
                                            "code": 32
                                        },
                                        {
                                            "Category": "एजेंट प्रमाणन",
                                            "Summary": "AI एजेंटों के प्रदर्शन, सुरक्षा और विश्वसनीयता के लिए औपचारिक प्रमाणीकरण प्रक्रिया।",
                                            "code": 33
                                        },
                                        {
                                            "Category": "IEEE मानक",
                                            "Summary": "AI में डेटा पारदर्शिता, आयु-उपयुक्त डिज़ाइन और जवाबदेही के लिए मानक।",
                                            "code": 34
                                        }
                                    ],
                                    "Summary": "यह उप-विषय AI एजेंटों के लिए स्पष्ट पहचान, सत्यापन और गुणवत्ता आश्वासन के लिए मानकों और प्रमाणन की आवश्यकता पर जोर देता है, जिससे उनके सुरक्षित और विश्वसनीय संचालन को सुनिश्चित किया जा सके।",
                                    "code": 31
                                },
                                {
                                    "Subnet": "प्रमाणीकरण प्रक्रियाएं",
                                    "Categories": [
                                        {
                                            "Category": "शासन संरचना",
                                            "Summary": "प्रमाणीकरण के लिए एक मजबूत शासन संरचना का होना।",
                                            "code": 36
                                        },
                                        {
                                            "Category": "पुनरावृत्तीय दृष्टिकोण",
                                            "Summary": "निरंतर निगरानी और समायोजन के साथ एक पुनरावृत्तीय प्रमाणीकरण प्रक्रिया।",
                                            "code": 37
                                        },
                                        {
                                            "Category": "संदर्भित मूल्यांकन",
                                            "Summary": "विशेष व्यवहारों को समझने के लिए एजेंटों के संदर्भ में मूल्यांकन करना।",
                                            "code": 38
                                        }
                                    ],
                                    "Summary": "यह उप-विषय AI एजेंटों के लिए प्रमाणीकरण को सफल बनाने के लिए आवश्यक प्रक्रियाओं और दृष्टिकोणों को रेखांकित करता है, जिसमें निरंतर सुधार और अनुकूलनशीलता पर जोर दिया गया है।",
                                    "code": 35
                                },
                                {
                                    "Subnet": "सामुदायिक भागीदारी",
                                    "Categories": [
                                        {
                                            "Category": "बहु-विषयक समुदाय",
                                            "Summary": "कानूनविदों, डॉक्टरों, कलाकारों और इंजीनियरों सहित विविध समूहों की भागीदारी।",
                                            "code": 40
                                        },
                                        {
                                            "Category": "सामुदायिक ज्ञान",
                                            "Summary": "विशेषज्ञों के विविध समुदाय से अंतर्दृष्टि का लाभ उठाना।",
                                            "code": 41
                                        }
                                    ],
                                    "Summary": "यह उप-विषय मानकों और प्रमाणन प्रक्रियाओं के विकास में व्यापक सामुदायिक भागीदारी के महत्व पर प्रकाश डालता है, यह सुनिश्चित करता है कि विभिन्न दृष्टिकोणों को एकीकृत किया जाए।",
                                    "code": 39
                                }
                            ],
                            "Summary": "यह विषय AI एजेंटों की विश्वसनीयता, सुरक्षा और जवाबदेही सुनिश्चित करने के लिए मानकों, दिशानिर्देशों और प्रमाणन कार्यक्रमों के विकास और कार्यान्वयन की पड़ताल करता है। यह इस क्षेत्र में उद्योग निकायों और विशेषज्ञों की भूमिका पर प्रकाश डालता है।",
                            "code": 30
                        },
                        {
                            "Net": "तकनीकी कार्यान्वयन और उपकरण",
                            "Subnets": [
                                {
                                    "Subnet": "एजेंटों के लिए बुनियादी ढाँचा",
                                    "Categories": [
                                        {
                                            "Category": "न्यूनतम व्यवहार्य विश्वास स्टैक",
                                            "Summary": "किसी भी एजेंटिक AI उत्पाद के लिए एक बुनियादी विश्वास ढांचा।",
                                            "code": 44
                                        },
                                        {
                                            "Category": "ऑर्केस्ट्रेशन लेयर गार्ड्रेल",
                                            "Summary": "एजेंटों के बीच संचार और समन्वय के लिए सुरक्षित नियम।",
                                            "code": 45
                                        },
                                        {
                                            "Category": "वास्तविक समय अवलोकनीयता",
                                            "Summary": "एजेंटों के व्यवहार की वास्तविक समय में निगरानी करने की क्षमता।",
                                            "code": 46
                                        }
                                    ],
                                    "Summary": "यह उप-विषय AI एजेंटों के सुरक्षित और प्रभावी संचालन के लिए आवश्यक तकनीकी बुनियादी ढांचे और न्यूनतम आवश्यकताओं पर केंद्रित है, जिसमें विश्वास और नियंत्रण सुनिश्चित करने के लिए महत्वपूर्ण घटकों को रेखांकित किया गया है।",
                                    "code": 43
                                },
                                {
                                    "Subnet": "ओपन-सोर्स फ्रेमवर्क",
                                    "Categories": [
                                        {
                                            "Category": "न्यूरो AI मल्टी-एजेंट एक्सेलेरेटर",
                                            "Summary": "कोग्निजेंट द्वारा विकसित एक लो-कोड/नो-कोड मल्टी-एजेंट फ्रेमवर्क।",
                                            "code": 48
                                        },
                                        {
                                            "Category": "क्लाउड और LLM अज्ञेयवादी",
                                            "Summary": "विभिन्न क्लाउड वातावरण और बड़े भाषा मॉडल के साथ संगतता।",
                                            "code": 49
                                        },
                                        {
                                            "Category": "अंतर्निहित सुरक्षा और प्रोटोकॉल",
                                            "Summary": "सुरक्षा को एक मूलभूत विशेषता के रूप में और एजेंट-से-एजेंट संचार के लिए विभिन्न प्रोटोकॉल का समर्थन करना।",
                                            "code": 50
                                        }
                                    ],
                                    "Summary": "यह उप-विषय ओपन-सोर्स मल्टी-एजेंट फ्रेमवर्क के विकास और क्षमताओं का विवरण देता है, जो AI एजेंटों के परिनियोजन और प्रबंधन को गति देने के लिए डिज़ाइन किए गए हैं।",
                                    "code": 47
                                },
                                {
                                    "Subnet": "परीक्षण और मूल्यांकन",
                                    "Categories": [
                                        {
                                            "Category": "परीक्षण विधियों का विकास",
                                            "Summary": "मल्टी-एजेंट सिस्टम के लिए परीक्षण और मूल्यांकन बेंचमार्क का उन्नयन।",
                                            "code": 52
                                        },
                                        {
                                            "Category": "फ्रंटियर लैब रिपोर्टिंग",
                                            "Summary": "बड़े मॉडल की सीमाओं और क्षमताओं पर पारदर्शी रिपोर्टिंग।",
                                            "code": 53
                                        },
                                        {
                                            "Category": "वितरणात्मक AGI सुरक्षा",
                                            "Summary": "AGI के एक मोनोलिथिक सिस्टम के बजाय उप-एजेंटों के रूप में उभरने की संभावना।",
                                            "code": 54
                                        }
                                    ],
                                    "Summary": "यह उप-विषय AI एजेंटों, विशेष रूप से मल्टी-एजेंट सिस्टम के लिए प्रभावी परीक्षण और मूल्यांकन रणनीतियों की आवश्यकता पर जोर देता है, ताकि सुरक्षा और विश्वसनीयता सुनिश्चित की जा सके।",
                                    "code": 51
                                }
                            ],
                            "Summary": "यह विषय AI एजेंटों के विकास, परिनियोजन और प्रबंधन के लिए आवश्यक तकनीकी समाधानों, उपकरणों और फ्रेमवर्क पर केंद्रित है, जिसमें व्यावहारिक कार्यान्वयन और परीक्षण के तरीकों पर प्रकाश डाला गया है।",
                            "code": 42
                        },
                        {
                            "Net": "उद्योग तुलना और आर्थिक प्रभाव",
                            "Subnets": [
                                {
                                    "Subnet": "अन्य उद्योगों से सीख",
                                    "Categories": [
                                        {
                                            "Category": "विमानन और ड्रोन सुरक्षा",
                                            "Summary": "विमानन से सुरक्षा अवधारणाओं को AI एजेंटों पर लागू करना।",
                                            "code": 57
                                        },
                                        {
                                            "Category": "लूप में मानव से कमांड में मानव",
                                            "Summary": "सुरक्षा प्रणालियों में मानव भागीदारी के स्तर का विकास।",
                                            "code": 58
                                        },
                                        {
                                            "Category": "सुरक्षा-महत्वपूर्ण उद्योग",
                                            "Summary": "अन्य सुरक्षा-महत्वपूर्ण क्षेत्रों में सामना की गई समान चुनौतियाँ।",
                                            "code": 59
                                        }
                                    ],
                                    "Summary": "यह उप-विषय AI एजेंटों के लिए सुरक्षा और शासन रणनीतियों को सूचित करने के लिए विमानन जैसे अच्छी तरह से स्थापित, सुरक्षा-महत्वपूर्ण उद्योगों से प्राप्त सबक की पड़ताल करता है।",
                                    "code": 56
                                },
                                {
                                    "Subnet": "शासन का आर्थिक मूल्य",
                                    "Categories": [
                                        {
                                            "Category": "प्रतिस्पर्धात्मक लाभ",
                                            "Summary": "शासन को एक लागत बोझ के बजाय एक प्रतिस्पर्धी लाभ के रूप में देखना।",
                                            "code": 61
                                        },
                                        {
                                            "Category": "उद्यम स्वीकार्यता",
                                            "Summary": "उद्यमों में AI एजेंटों को अपनाने के लिए शासन की आवश्यकता।",
                                            "code": 62
                                        },
                                        {
                                            "Category": "स्टार्टअप चुनौतियाँ",
                                            "Summary": "सीमित संसाधनों वाले स्टार्टअप के लिए नियामक आवश्यकताओं का पालन करने की चुनौती।",
                                            "code": 63
                                        }
                                    ],
                                    "Summary": "यह उप-विषय AI एजेंटों के लिए शासन के आर्थिक निहितार्थों पर विचार करता है, जिसमें यह दर्शाया गया है कि यह बाजार में प्रवेश और उद्यम अपनाने के लिए एक महत्वपूर्ण कारक कैसे हो सकता है।",
                                    "code": 60
                                },
                                {
                                    "Subnet": "भारतीय डिजिटल अवसंरचना",
                                    "Categories": [
                                        {
                                            "Category": "आधार और UPI",
                                            "Summary": "भारत के डिजिटल पहचान और भुगतान प्रणालियों की सफलता की कहानी।",
                                            "code": 65
                                        },
                                        {
                                            "Category": "डेटा गोपनीयता और मानकीकरण",
                                            "Summary": "डेटा गोपनीयता और इंटरफेस के मानकीकरण में सरकार की भूमिका।",
                                            "code": 66
                                        },
                                        {
                                            "Category": "सरकार की भूमिका",
                                            "Summary": "AI एजेंट विकास को मानकीकृत करने और सुविधा प्रदान करने में सरकार का महत्वपूर्ण योगदान।",
                                            "code": 67
                                        }
                                    ],
                                    "Summary": "यह उप-विषय भारतीय डिजिटल अवसंरचना के विकास को एक उदाहरण के रूप में प्रस्तुत करता है, जिसमें AI एजेंटों के लिए मानकीकरण और सरकारी समर्थन के लिए सबक पर जोर दिया गया है।",
                                    "code": 64
                                }
                            ],
                            "Summary": "यह विषय AI एजेंटों के विकास और अपनाने के निहितार्थों को समझने के लिए अन्य उद्योगों से समानताएं खींचता है। यह शासन, सुरक्षा और नवाचार के बीच आर्थिक संतुलन की भी पड़ताल करता है।",
                            "code": 55
                        },
                        {
                            "Net": "भविष्य की दिशा और सामूहिक प्रयास",
                            "Subnets": [
                                {
                                    "Subnet": "सामूहिक रूप से आगे बढ़ना",
                                    "Categories": [
                                        {
                                            "Category": "खुला नवाचार",
                                            "Summary": "नवाचार को खुला रखना और विनियमन के साथ संतुलन बनाना।",
                                            "code": 70
                                        },
                                        {
                                            "Category": "सामान्य मार्गदर्शन",
                                            "Summary": "प्रयोग को प्रोत्साहित करने के लिए सामान्य मार्गदर्शन और रूपरेखा।",
                                            "code": 71
                                        },
                                        {
                                            "Category": "सामूहिक सीख",
                                            "Summary": "ऑडिटबिलिटी और ट्रेसबिलिटी पर खुले प्लेटफार्मों के माध्यम से सामूहिक सीख।",
                                            "code": 72
                                        }
                                    ],
                                    "Summary": "यह उप-विषय AI एजेंट प्रौद्योगिकी को आगे बढ़ाने के लिए एक सहयोगी दृष्टिकोण की वकालत करता है, जिसमें खुला नवाचार और साझा ज्ञान को बढ़ावा देने के लिए मार्गदर्शन और प्लेटफार्मों की स्थापना पर जोर दिया गया है।",
                                    "code": 69
                                },
                                {
                                    "Subnet": "जटिलता को दूर करना",
                                    "Categories": [
                                        {
                                            "Category": "शब्दावली को हटाना",
                                            "Summary": "AI एजेंटों से संबंधित तकनीकी शब्दावली को सरल बनाना।",
                                            "code": 74
                                        },
                                        {
                                            "Category": "मूल्य का प्रदर्शन",
                                            "Summary": "तकनीक के मूल्य को स्पष्ट रूप से संवाद करना।",
                                            "code": 75
                                        },
                                        {
                                            "Category": "भय को कम करना",
                                            "Summary": "गलतफहमी के कारण होने वाले अनावश्यक भय को कम करना।",
                                            "code": 76
                                        }
                                    ],
                                    "Summary": "यह उप-विषय AI एजेंटों के बारे में व्यापक समझ और स्वीकार्यता को बढ़ावा देने के लिए स्पष्ट संचार और जन जागरूकता के महत्व पर प्रकाश डालता है, जिससे अनावश्यक भय को दूर किया जा सके।",
                                    "code": 73
                                },
                                {
                                    "Subnet": "विश्वास के लिए बुनियादी ढाँचा",
                                    "Categories": [
                                        {
                                            "Category": "खंडित विश्वास",
                                            "Summary": "विभिन्न घटकों में खंडित विश्वास को एकीकृत करना।",
                                            "code": 78
                                        },
                                        {
                                            "Category": "बहु-आवाज भागीदारी",
                                            "Summary": "शिक्षाविदों, कंपनियों, नीति निर्माताओं और युवा पीढ़ी से सुनना।",
                                            "code": 79
                                        },
                                        {
                                            "Category": "विकेंद्रीकृत प्रणाली",
                                            "Summary": "परीक्षण मानकों के साथ विकेंद्रीकृत और एकीकृत प्रणाली की आवश्यकता।",
                                            "code": 80
                                        }
                                    ],
                                    "Summary": "यह उप-विषय AI एजेंटों के लिए एक मजबूत और एकीकृत विश्वास बुनियादी ढांचे के निर्माण की पड़ताल करता है, जिसमें विविध हितधारकों की भागीदारी और एक विकेंद्रीकृत दृष्टिकोण के माध्यम से जटिलता को दूर करने पर जोर दिया गया है।",
                                    "code": 77
                                }
                            ],
                            "Summary": "यह विषय AI एजेंटों के भविष्य को आकार देने के लिए आवश्यक सामूहिक प्रयासों, सहयोग और रणनीतियों पर केंद्रित है, जिसमें ज्ञान साझा करने, संचार और सामुदायिक भागीदारी के महत्व पर प्रकाश डाला गया है।",
                            "code": 68
                        }
                    ],
                    "Summary": "यह चर्चा AI एजेंटों और स्वायत्तता के विकसित होते परिदृश्य पर केंद्रित है। Google, Project Nanda, IEEE और Insight Partners के वक्ताओं ने स्वयं और अपने काम का परिचय दिया। प्रमुख विषयों में AI एजेंटों को स्वायत्तता के एक स्पेक्ट्रम के रूप में परिभाषित करना, मल्टी-एजेंट सिस्टम में वास्तविक समय के शासन और विश्वास की महत्वपूर्ण आवश्यकता, मौलिक सुरक्षा का महत्व, और जवाबदेही व पारदर्शिता सुनिश्चित करने में प्रमाणन व मानकों (जैसे IEEE) की भूमिका शामिल है। AI के लिए \"शासन स्टैक\" की अवधारणा पर भी प्रकाश डाला गया है, जिसमें अवलोकनीयता (observability), किल स्विच, ऑडिट ट्रेल और जवाबदेही पर जोर दिया गया है। एक ओपन-सोर्स मल्टी-एजेंट फ्रेमवर्क (Neuro AI मल्टी-एजेंट एक्सेलेरेटर) का प्रदर्शन किया गया, जिसमें एजेंट ऑर्केस्ट्रेशन, संचार और अंतर्निहित सुरक्षा की क्षमताओं को उजागर किया गया। विमानन जैसे सुरक्षा-महत्वपूर्ण उद्योगों के साथ समानताएं खींची गईं, जिसमें \"लूप में मानव\" से \"कमांड में मानव\" में बदलाव की वकालत की गई। नवाचार और विनियमन के बीच संतुलन पर बहस की गई, जिसमें स्टार्टअप के लिए प्रारंभिक चरण के \"न्यूनतम व्यवहार्य विश्वास स्टैक\" और शुरुआत से ही शासन को एकीकृत करने के आर्थिक लाभ पर ध्यान केंद्रित किया गया। चर्चा सामूहिक रूप से आवश्यक कार्यों के साथ समाप्त हुई, जैसे मल्टी-एजेंट सिस्टम के लिए परीक्षण बेंचमार्क विकसित करना, साझा सीखने के लिए खुले प्लेटफार्मों को बढ़ावा देना, और भय को कम करने व व्यापक अपनाने को बढ़ावा देने के लिए AI शब्दावली को स्पष्ट करना।"
                }
            },
            "summaryList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 10023,
                        "CandidatesTokenCount": 3631,
                        "TotalTokenCount": 15206,
                        "thoughtsTokenCount": 1552,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Summary": "<h2>एजेंटिक एआई और स्वायत्तता पर व्यापक संश्लेषण रिपोर्ट</h2><h3>परिचय और पैनलिस्ट अवलोकन</h3><p>इस सत्र में एजेंटिक एआई और स्वायत्त प्रणालियों के शासन की अवधारणा पर गहन चर्चा की गई, जो आज की तेजी से विकसित हो रही तकनीकी दुनिया में एक महत्वपूर्ण विषय है (स्रोत: इनपुट डेटा)। सत्र की शुरुआत पैनलिस्टों के परिचय के साथ हुई:</p><ul><li><b>एडी सहगल (गूगल):</b> एआई और उभरती हुई तकनीक, विशेष रूप से एजेंट एआई और रोबोटिक्स पर नीति प्रबंधक।</li><li><b>महेश (प्रोजेक्ट नंदा):</b> इंटरनेट ऑफ एआई एजेंट्स के लिए मूलभूत अवसंरचना के निर्माण में अग्रणी, विश्वास को एक महत्वपूर्ण नींव के मुद्दे के रूप में पहचानते हैं।</li><li><b>अल्पेश शाह (आईईईई स्टैंडर्ड्स एसोसिएशन):</b> आईईईई स्टैंडर्ड्स एसोसिएशन के प्रबंध निदेशक और आईईईई में प्रबंधन परिषद के सदस्य, जो प्रमाणन और मानकों पर ध्यान केंद्रित करते हैं।</li><li><b>अपूर्व गोयल (इनसाइट पार्टनर्स):</b> प्रिंसिपल, यूएस-भारत निवेश प्रयासों का नेतृत्व करते हैं, जो वैश्विक सॉफ्टवेयर कंपनियों में निवेश करने वाले 90 बिलियन डॉलर के परिसंपत्ति प्रबंधन वाले फंड का प्रतिनिधित्व करते हैं।</li><li><b>प्रवीण (कॉग्निजेंट एआई लैब):</b> सह-मॉडरेटर, ऑर्केस्ट्रेशन और मल्टी-एजेंट फ्रेमवर्क पर विशेषज्ञ।</li></ul><p>चर्चा ने इस बात पर प्रकाश डाला कि स्वायत्तता की अवधारणा कई लोगों को भयभीत करती है क्योंकि एजेंटों के निर्माण, उनकी प्रतिक्रियाओं और उद्देश्यों में संभावित परिवर्तनों के बारे में पूरी जानकारी नहीं है। शासन, जो प्रणालियों के परिणामों को नियंत्रित करने, प्रक्रियाओं, स्वामित्व और जोखिम प्रबंधन को नियंत्रित करने की एक व्यापक अवधारणा है, एजेंटों की दुनिया में केंद्रीय बन जाती है (स्रोत: इनपुट डेटा)। लक्ष्य स्वायत्तता को इस तरह से नियंत्रित करना है कि यह भरोसेमंद हो और मानव नियंत्रण से बाहर न हो।</p><h3>मल्टी-एजेंट सिस्टम बनाम पारंपरिक एआई</h3><p>एडी सहगल ने मल्टी-एजेंट प्रणालियों और पारंपरिक एआई के बीच अंतर को स्पष्ट किया, जिसमें 'एजेंसी' को एक स्पेक्ट्रम के रूप में देखने पर जोर दिया गया (स्रोत: इनपुट डेटा)। इसमें बुनियादी चैटबॉट (जैसे गूगल डीप रिसर्च) शामिल हैं जो जानकारी प्रस्तुत करते हैं लेकिन कार्रवाई नहीं करते, और पूर्ण स्वायत्त वाहन जो वास्तविक दुनिया में अंत-से-अंत तक कार्रवाई करते हैं। एजेंटों को मेमोरी, प्लानिंग (दीर्घकालिक और अल्पकालिक), निष्पादन और स्वायत्तता सहित विभिन्न आयामों पर स्वायत्तता की एक निरंतरता के रूप में देखा जाना चाहिए। एजेंटों को केवल 'कार्रवाई करने वाले' के रूप में परिभाषित करने के बजाय, उन्हें मूल स्वायत्तता से लेकर पूरी तरह से स्वायत्त प्रणालियों तक निरंतरता पर रखा जाता है।</p><h3>एजेंटों में त्रुटियों को समझना और उनका प्रबंधन करना</h3><p>महेश ने जोर दिया कि एजेंटों की 'निरंतरता' प्रकृति के कारण, पारंपरिक निगरानी मॉडल पर्याप्त नहीं हैं (स्रोत: इनपुट डेटा)। एजेंट नेटवर्क अभिनेता बन जाते हैं, जो हर चरण में दिखाई या अदृश्य रूप से कार्य करते हैं, जिससे रनटाइम गवर्नेंस अनिवार्य हो जाता है। यह एक पूर्व-सोची-समझी समाधान नहीं हो सकता है, बल्कि इसे एक पहले सिद्धांत के रूप में संबोधित किया जाना चाहिए। सुरक्षा, शासन, विश्वास और परिनियोजन क्षमता को मौलिक स्तर पर संबोधित किया जाना चाहिए, साथ ही निष्पादन की खुलेपन और जवाबदेही का भी त्याग किए बिना। चर्चा ने इस बात पर प्रकाश डाला कि चुनौती केवल बुरे अभिनेताओं को हटाने में नहीं है, बल्कि एजेंटों द्वारा स्वयं की जाने वाली त्रुटियों को प्रबंधित करने में है। 'एजेंटिक पहचान' की अवधारणा प्रस्तुत की गई थी - एजेंटों की उत्पत्ति, प्रशिक्षण डेटा, निर्माता, स्वामित्व और संचार भाषा को जानने का एक तरीका - जो एजेंटों को प्रमाणित करने की आवश्यकता की ओर ले जाती है।</p><h3>एजेंट प्रमाणन और शासन संरचनाएँ</h3><p>अल्पेश शाह ने एजेंट प्रमाणन और मजबूत शासन संरचनाओं की आवश्यकता पर प्रकाश डाला (स्रोत: इनपुट डेटा)। परिणामों को समझने में निहित कठिनाई के कारण, यह जानना महत्वपूर्ण है कि क्या एजेंट सही ढंग से काम कर रहे हैं। उनके अनुसार, यह आवश्यक है:</p><ul><li><b>मजबूत शासन संरचनाएँ:</b> इन्हें जगह पर होना महत्वपूर्ण है।</li><li><b>स्पष्ट प्रक्रियाएँ:</b> यह निर्धारित करने के लिए कि क्या तैनात किया जाता है या रनटाइम स्थिति में जाता है।</li><li><b>पुनरावृत्तीय दृष्टिकोण:</b> निगरानी एक बार का अभ्यास नहीं है, क्योंकि शिफ्ट अप्रत्याशित हो सकते हैं; निरंतर निगरानी आवश्यक है।</li><li><b>संदर्भित करना:</b> एजेंटों के व्यवहार को प्रभावित करने वाले अनदेखे ट्रिगर को समझना।</li><li><b>पारदर्शी और जवाबदेह लॉगिंग:</b> न केवल लॉगिंग, बल्कि प्रक्रिया में मानव भागीदारी भी।</li></ul><p>आईईईई के प्रमाणन कार्यक्रम का उद्देश्य यह सुनिश्चित करना है कि उत्पादन के साधनों का मूल्यांकन किया जा सके, भले ही 100% पूर्णता की गारंटी न दी जा सके। आईईईई मानकों में डेटा पारदर्शिता, आयु-उपयुक्त डिज़ाइन, जवाबदेही और डेटा गोपनीयता जैसे महत्वपूर्ण तत्व शामिल हैं। अल्पेश ने वकील, डॉक्टर, कलाकार और इंजीनियर सहित लोगों के विविध समुदाय की भागीदारी के महत्व पर भी जोर दिया, जिन्होंने प्रासंगिक समस्याओं को समझने में मदद की।</p><h3>शासन स्टैक: स्वायत्त दुनिया के लिए एक मॉडल</h3><p>अपूर्व गोयल ने एक 'शासन स्टैक' की कल्पना की, जो एक स्वायत्त दुनिया के लिए एक 5-स्तरीय मॉडल है (स्रोत: इनपुट डेटा):</p><ul><li><b>बिल्ड टाइम:</b> डेटा शासन और मॉडल संस्करण पर केंद्रित।</li><li><b>डिप्लॉय टाइम:</b> नीति, ट्रैकिंग, अनुमति और सीक्रेट्स प्रबंधन।</li><li><b>रनटाइम:</b> रियल-टाइम ऑब्जर्वेबिलिटी और 'किल स्विच' जैसे आपातकालीन तंत्र सुनिश्चित करना।</li><li><b>रेमेडिएशन:</b> पोस्ट-मॉर्टम और घटना प्रतिक्रिया के लिए ऑडिट ट्रेल्स।</li><li><b>जवाबदेही:</b> जवाबदेही, रिपोर्टिंग संरचनाओं और अनुपालन मैपिंग को परिभाषित करना।</li></ul><p>शासन को अब केवल एक अनुपालन मुद्दे के रूप में नहीं देखा जाता है, बल्कि एक प्रतिस्पर्धी लाभ और 'उत्पाद-बाजार फिट' का एक महत्वपूर्ण हिस्सा माना जाता है। कंपनियां, विशेष रूप से बड़े उद्यम, सही शासन सुनिश्चित करने के लिए लाखों डॉलर खर्च करने को तैयार हैं। स्टार्टअप के लिए, अपूर्व ने एक 'न्यूनतम व्यवहार्य विश्वास स्टैक' का प्रस्ताव दिया:</p><ul><li>यह स्पष्ट करना कि एजेंट को क्या करना चाहिए।</li><li>यह पुष्टि करना कि एजेंट वास्तव में वही कर रहा है जो उसे करना चाहिए।</li><li>यदि एजेंट भटकता है तो उसे वास्तविक समय में रोकने में सक्षम होना।</li></ul><p>इस स्टैक के लिए एक स्पष्ट रूप से परिभाषित एजेंट पहचान रजिस्ट्री, ऑर्केस्ट्रेशन लेयर पर गार्डरेल्स, एक रियल-टाइम ऑब्जर्वेबिलिटी आर्किटेक्चर और स्पष्ट परिभाषित निरीक्षण की आवश्यकता होती है।</p><h3>ऑर्केस्ट्रेशन और कोऑर्डिनेशन: कॉग्निजेंट का न्यूरो एआई</h3><p>प्रवीण ने मल्टी-एजेंट प्रणालियों को एआई का अगला चरण बताया और कॉग्निजेंट के ओपन-सोर्स फ्रेमवर्क, 'न्यूरो एआई मल्टी-एजेंट एक्सीलरेटर' का प्रदर्शन किया (स्रोत: इनपुट डेटा)। यह एक लो-कोड, नो-कोड फ्रेमवर्क है जिसे प्रोटोटाइप विकास से स्केलिंग तक तेजी लाने के लिए डिज़ाइन किया गया है। इसकी प्रमुख विशेषताएं:</p><ul><li><b>क्लाउड अज्ञेयवादी और एलएलएम अज्ञेयवादी:</b> उपयोगकर्ता को एलएलएम और परिनियोजन वातावरण पर नियंत्रण देता है।</li><li><b>अंतर-संचालनीयता:</b> एजेंट से एजेंट संचार के लिए कई प्रोटोकॉल का समर्थन करता है।</li><li><b>सुरक्षा अंतर्निहित:</b> कोर में निर्मित, बाद में सोचा जाने वाला नहीं।</li><li><b>सरल:</b> एआई/एमएल इंजीनियरों से लेकर सीईओ तक के लिए सुलभ।</li></ul><p>डेमो में दिखाया गया कि कैसे एक बहु-एजेंट नेटवर्क को प्रॉम्प्ट के माध्यम से बनाया जा सकता है, एजेंटों को परिभाषित किया जा सकता है, ऑर्केस्ट्रेटर से जोड़ा जा सकता है, और भूमिकाओं और जिम्मेदारियों के साथ आबाद किया जा सकता है। इसमें रियल-टाइम ऑब्जर्वेबिलिटी (आंतरिक चैट, लॉग) और ईएसडी मेट्रिक्स (ऊर्जा उपयोग, कार्बन पदचिह्न, लागत) शामिल हैं। प्रवीण ने यह भी बताया कि प्रत्येक एजेंट नेटवर्क को एक विशिष्ट एलएलएम तक पहुंच हो सकती है, और पत्ती नोड एजेंटों को उच्च सटीकता के लिए विशिष्ट एलएलएम के साथ सशक्त किया जा सकता है। यह फ्रेमवर्क Apache 2.0 लाइसेंस के तहत GitHub पर उपलब्ध है और इसने GitHub सुरक्षा के उच्चतम मानकों को पूरा किया है।</p><h3>अन्य उद्योगों से सीख: विमानन और परमाणु ऊर्जा</h3><p>एडी सहगल ने जोर दिया कि सुरक्षा की अवधारणा, विशेष रूप से सुरक्षा-महत्वपूर्ण उद्योगों में, नई नहीं है (स्रोत: इनपुट डेटा)। उन्होंने विमानन और ड्रोन का उदाहरण दिया। एफएए ने शुरू में 'दृश्य रेखा' (VLOS) नियमों को लागू किया, जिसमें पायलट को ड्रोन की दृश्य रेखा बनाए रखनी होती थी। हालांकि, एआई डिटेक्ट-एंड-अवॉइड सिस्टम की सुरक्षा में प्रगति के साथ, विनियमन 'दृश्य रेखा से परे' (BVLOS) में बदल रहा है। यह दर्शाता है कि एआई सिस्टम अब कुछ सुरक्षा कार्यों में मनुष्यों से बेहतर प्रदर्शन कर सकते हैं। एजेंटों से तुलना करते हुए, 'लूप में मानव' मॉडल (VLOS के समान) एजेंट की उपयोगिता को कम करता है। जैसे-जैसे एजेंट सुरक्षित और अधिक विश्वसनीय होते जाते हैं, 'लूप में मानव' से 'कमांड में मानव' (BVLOS के समान) की ओर बढ़ना होगा, जहाँ मनुष्य पर्यवेक्षण करता है लेकिन हर कदम को मंजूरी नहीं देता है।</p><h3>इंजीनियरिंग दिशानिर्देश बनाम विनियमन और मानक</h3><p>चर्चा इस बात पर केंद्रित थी कि इंजीनियरिंग दिशानिर्देशों और तकनीकी डिजाइन के बीच विनियमन, मानदंडों और मानकों के साथ संतुलन कैसे बनाया जाए (स्रोत: इनपुट डेटा)।</p><ul><li><b>महेश का दृष्टिकोण:</b> इंटरनेट के शुरुआती दिनों से एक सादृश्य, जहां नवाचार खुला रखा गया था। उनका सुझाव है कि नवाचार को एक डेटा प्लेन के रूप में खुला रखा जाना चाहिए, जबकि शासन को एक नियंत्रण प्लेन के रूप में समानांतर रूप से बनाया जाना चाहिए।</li><li><b>अल्पेश का दृष्टिकोण:</b> अत्यधिक शुरुआती विनियमन नवाचार को बाधित कर सकता है। उन्होंने 'फ्रेमिंग' के महत्व पर जोर दिया - यह जानने के लिए कि बहुत दूर क्या है - विनियमन के बजाय शुरुआती चरणों में। परिपक्वता पर विनियमन अधिक समझ में आता है। उन्होंने 'मिनी ट्रस्ट स्टैक' के आर्थिक पहलू पर भी ध्यान दिया, जिसमें स्टार्टअप के लिए प्रवेश बाधाओं को कम करने और सरकारों और नियामकों के लिए माप को आसान बनाने के लिए मानकों और नीतियों को 'गेम से पहले' ही शामिल करना शामिल है।</li><li><b>अपूर्व का दृष्टिकोण:</b> जिम्मेदार एआई को तैनात करना वाणिज्यिक परिनियोजन के साथ संरेखित होता है, न कि इसके साथ तनाव में। उद्यमों में बेचने के लिए, शासन मूलभूत है और एक प्रतिस्पर्धी लाभ है। एआई मूल टीमों के लिए, शासन एक साप्ताहिक परिचालन अनुक्रम का हिस्सा बन गया है, जिसमें मूल्यांकन, रेड टीमिंग और पोस्ट-मॉर्टम समीक्षाएं शामिल हैं। यह सुरक्षा के समान है, जो प्रारंभिक चरण के स्टार्टअप के लिए भी महत्वपूर्ण है जो उद्यम ग्राहकों को बेचते हैं।</li></ul><h3>विश्वास और भविष्य की दिशा के लिए बुनियादी ढांचा</h3><p>एकल सबसे महत्वपूर्ण चीज पर विचार करते हुए जो सामूहिक प्रगति को सक्षम करेगी:</p><ul><li><b>एडी सहगल:</b> मल्टी-एजेंट प्रणालियों के लिए परीक्षण विधियों और मूल्यांकन बेंचमार्क को विकसित करना। उन्होंने 'डिस्ट्रिब्यूशनल एजीआई सेफ्टी' पेपर का उल्लेख किया, जो एजीआई को कई विशेष एजेंटों के रूप में उभरने का सुझाव देता है जो एक साथ काम करते हैं। उन्होंने सीमांत प्रयोगशालाओं के लिए इन बेंचमार्क के खिलाफ परीक्षण करने और सीमाओं और क्षमताओं के बारे में पारदर्शी रूप से रिपोर्ट करने के महत्व पर जोर दिया।</li><li><b>महेश:</b> मल्टी-एजेंट प्रणालियों में विभिन्न खिलाड़ियों (ओपन-सोर्स और गैर-ओपन-सोर्स) द्वारा प्रदान किए गए मॉडलों का उपयोग करने की जटिलता पर प्रकाश डाला। उन्होंने सभी आवाजों - शिक्षाविदों, एलएलएम कंपनियों, सिस्टम इंटीग्रेटरों, हाइपरस्केलर्स, नीति निर्माताओं और युवा पीढ़ी को - चर्चा में शामिल करने के लिए एक विकेन्द्रीकृत और एकीकृत प्रणाली के महत्व पर जोर दिया, ताकि परीक्षण पैरामीटर सभी प्रणालियों में मान्य हों।</li><li><b>अपूर्व:</b> ऑडिटबिलिटी और पता लगाने की क्षमता के महत्व पर जोर दिया, ताकि जो हो रहा है उसकी अपरिवर्तनीय तरीके से सटीक रिपोर्टिंग हो सके। उन्होंने सामूहिक सीखने के लिए वैश्विक प्लेटफार्मों की कमी को नोट किया, जहां स्टार्टअप यह सीख सकते हैं कि सिस्टम किस पैमाने पर और किस बुनियादी ढांचे के स्तर पर विफल होते हैं, जो नवाचार को तेज करेगा।</li><li><b>अल्पेश:</b> दो प्रमुख बिंदु: पहला, प्रौद्योगिकी को एक 'कॉमन' या सार्वजनिक उपयोगिता के रूप में मानना, जैसा कि एआई कॉमन्स करता है, ताकि सभी को पहुंच मिल सके। दूसरा, शब्दजाल से छुटकारा पाने और 'एजेंटिक स्वायत्तता को नियंत्रित करना' जैसे विषयों को रहस्यमय बनाने पर बहुत अधिक समय खर्च करना, ताकि अधिक लोग मूल्य को समझ सकें और अनावश्यक भय को कम किया जा सके।</li></ul><h3>निष्कर्ष और कार्रवाई योग्य सिफारिशें</h3><h4>निष्कर्ष</h4><p>एजेंटिक एआई और स्वायत्त प्रणालियों का उदय तकनीकी नवाचार में एक महत्वपूर्ण बदलाव का प्रतिनिधित्व करता है, लेकिन यह शासन, विश्वास और सुरक्षा से संबंधित गंभीर चुनौतियां भी प्रस्तुत करता है। पारंपरिक एआई और मशीन-से-मशीन बातचीत के लिए विकसित मौजूदा शासन मॉडल मल्टी-एजेंट प्रणालियों की निरंतर, गतिशील और नेटवर्क प्रकृति के लिए अपर्याप्त हैं। स्वायत्तता को प्रभावी ढंग से प्रबंधित करने के लिए एक बहुआयामी दृष्टिकोण आवश्यक है, जिसमें मजबूत इंजीनियरिंग प्रथाएं, पुनरावृत्तीय प्रमाणन, नियामक ढांचा और सामूहिक हितधारकों की भागीदारी शामिल है (स्रोत: इनपुट डेटा)। शासन को अब केवल एक अनुपालन लागत के रूप में नहीं, बल्कि वाणिज्यिक सफलता और प्रतिस्पर्धी लाभ के चालक के रूप में देखा जाना चाहिए।</p><h4>कार्य योजनाएँ और हितधारक जिम्मेदारियाँ</h4><ul><li><b>उद्योग और स्टार्टअप:</b></li><ul><li>'न्यूनतम व्यवहार्य विश्वास स्टैक' को डिज़ाइन और परिनियोजन के शुरुआती चरणों में एकीकृत करें।</li><li>उत्पाद विकास में रियल-टाइम ऑब्जर्वेबिलिटी, किल स्विच और ऑडिट ट्रेल्स जैसी शासन सुविधाओं को एम्बेड करें।</li><li>सुरक्षा और जवाबदेही के लिए नियमित मूल्यांकन (evals), रेड टीमिंग और पोस्ट-मॉर्टम समीक्षा आयोजित करें।</li></ul><li><b>मानक संगठन (जैसे आईईईई):</b></li><ul><li>डेटा पारदर्शिता, आयु-उपयुक्त डिज़ाइन, जवाबदेही और डेटा गोपनीयता को कवर करने वाले मजबूत प्रमाणन कार्यक्रम और मानकों को विकसित और परिष्कृत करना जारी रखें।</li><li>विभिन्न हितधारकों (कानूनविदों, डॉक्टरों, कलाकारों, इंजीनियरों) को शामिल करके समुदाय-आधारित दृष्टिकोण को बढ़ावा दें ताकि शासन के लिए संदर्भ-समृद्ध दिशानिर्देश बनाए जा सकें।</li></ul><li><b>नियामक और नीति निर्माता:</b></li><ul><li>शुरुआती चरणों में 'फ्रेमिंग' के माध्यम से मार्गदर्शन प्रदान करें ताकि नवाचार को बाधित किए बिना सीमाओं को परिभाषित किया जा सके।</li><li>मानकों को 'कपड़े में' एकीकृत करने के लिए तंत्र का पता लगाएं ताकि स्टार्टअप के लिए अनुपालन को आसान बनाया जा सके और नियामकों के लिए मापने योग्य मीट्रिक प्रदान की जा सकें।</li><li>'डिजिटल पब्लिक इन्फ्रास्ट्रक्चर' मॉडल पर विचार करें, जैसे भारत के आधार और यूपीआई, एजेंटिक एआई के लिए मानकीकृत इंटरफेस प्रदान करने के लिए।</li></ul><li><b>शोधकर्ता और डेवलपर:</b></li><ul><li>मल्टी-एजेंट प्रणालियों के लिए नए परीक्षण विधियों और मूल्यांकन बेंचमार्क को सक्रिय रूप से विकसित करें।</li><li>क्षमताएं और सीमाएं दोनों पर पारदर्शी रिपोर्टिंग में संलग्न हों।</li><li>विभिन्न एलएलएम मॉडल और प्रदाताओं के साथ इंटरऑपरेटिंग मल्टी-एजेंट सिस्टम में विश्वास और सुरक्षा के मुद्दों को संबोधित करने पर ध्यान दें।</li></ul><li><b>सभी हितधारक:</b></li><ul><li>एजेंटिक एआई से जुड़े शब्दजाल को ध्वस्त करने और प्रौद्योगिकी के मूल्य को स्पष्ट करने के लिए सामूहिक प्रयास करें।</li><li>'कॉमन' या सार्वजनिक उपयोगिता के रूप में एजेंटिक एआई प्रौद्योगिकियों तक पहुंच की वकालत करें।</li><li>विभिन्न हितधारकों से सामूहिक सीखने और ज्ञान साझा करने के लिए वैश्विक प्लेटफार्मों को बढ़ावा दें।</li></ul></ul>"
                            }
                        ]
                    }
                }
            ]
        }
    ],
    "headerCols": {
        "col1": "Part Id",
        "col2": "FileName",
        "col3": "Content"
    },
    "fileName": "",
    "projectType": "DOCUMENT",
    "sourceLanguage": "English",
    "targetLanguage": "Hindi",
    "__v": 0
},
{
    "_id": "69944f6025c9f4e0d7e70997",
    "projectId": "413645",
    "projectName": "Governing Autonomy_ Trust in Agentic AI, Multi-Agent Systems and the Infrastructure of Trust - English",
    "projectDescription": "Governing Autonomy_ Trust in Agentic AI, Multi-Agent Systems and the Infrastructure of Trust",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T11:22:05.998Z",
    "projectUpdatedDate": "2026-02-17T11:22:05.998Z",
    "projectStatus": "Active",
    "companyId": "69453bf2e603a0a401b6f35b",
    "projectCreatedBy": "69453bf2e603a0a401b6f35c",
    "projectUpdatedBy": "69453bf2e603a0a401b6f35c",
    "projectLastAccessDate": "2026-02-17T11:22:05.998Z",
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "Content",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "Governing Autonomy_ Trust in Agentic AI, Multi-Agent Systems and the Infrastructure of Trust.mp3",
            "col3": " We'll say probably a few words about what they do. I'm sorry, but Eddie Sahai from Google. Go ahead, Eddie. Hi, everyone. I'm Eddie Sahai. I am policy manager working on AI and emerging tech, particularly agent AI and robotics. Hello, and good afternoon. This is Mahesh. Thank you all, and thanks to Amir and Pauline for having us here. And also, they are generous enough to put my younger version.  in the picture. And so I'm from Project Nanda. We are pioneering the building and foundational infrastructure for the internet of AI agents. It's a new kind of internet is born, or getting born, from last year onwards. And one of the most foundational issues, the way I say that, trust, that is me. That's my photo. The same thing, how we are gonna solve this problem using  in this internet of AI agents, and those primitives need to be solved at the foundational layer. It's not like after the fact or after the thought solutioning. It has to happen as a first principle. Yeah, thank you. Alpesh. Hello, I'm Alpesh Shah. I am the Managing Director of the IEEE Standards Association and a member of the Management Council at IEEE. Thank you. Thank you.  Hey, guys. Hi. My name is Apoorv Goyal. I'm a principal at Insight Partners. I help lead a lot of our US-India investing efforts. Just as context, we're one of the leading global technology investment funds. We manage close to $90 billion in AUM investing out of a $12.5 billion fund into leading global software companies around the world, including a bunch of companies based out of India. Wonderful. Thank you so much. I'll just be getting to Praveen. He's my co-moderator. Praveen is the head of AI Lab.  but before we start talking about it, with that, as a raise of hand, how many of you have built an agent or played with an agent? How many of you are scared of what's gonna happen with agents? One is probably like 5%, 10%. So it thinks that 90 or 80 or 90% of you don't feel that agents are gonna be dangerous or anything else. Have you followed what happened with Open Cloud recently and Notebook?  Do you feel this is cool or exciting or energizing or fun? Who's excited about what happened? It's a beautiful experiment, right? So we're learning about all these things and as we talk about agentic world, the concept of autonomy scares a lot of people. Why? Because we don't fully know how they're built. We don't fully know how they're going to react, if their objective is going to be changing, if they're going to be...  mixing with each other, including new babies, we don't know what's gonna happen. And then in that world of autonomy, a lot of decisions that enterprises take and societies functioning may not be able to be put in the hands of agents. So the concept of governance, which is a concept which is very encompassing about how we control the outcome of systems and procedures and ownership and risk management.  becomes central again in the world of agent, which was not the case about even a year ago. So today we're going to be talking about this concept of governing autonomy, how we can actually make sure that this world, this future world, we're going to be fully autonomous, not only is our hand to some extent, but we can somehow trust it, and we don't have to really extend. So to get started, maybe we can have a quick demo. What do you think, Praveen? Questions?  Okay, perfect, that's a good thing. So maybe I can start with Eli. Eli, in your vantage point, can you help us understand what's different between multi-agent environment and systems and traditional AI that we've been used to? How should we think about it? So with the raise of hand that I saw in this room, I think we are talking with a group of experts.  From my perspective, and many of you have already worked with ModBug and have seen OpenClaw and have built agents, so you know better than me that the way that we think about agents, it's not a specific line or a specific point in time that these group of things are agentic and this group of things are non-agentic, rather thinking about them as a spectrum of agency or autonomy going from  basic chatbots that can do a little bit of research and at the end present the result. For example, Google Deep Research does a research given your prompt and finds the information and presents that information. That has some agentic features, but doesn't take actions.  compared to let's say going all the way towards the end of the autonomy that is, for example, autonomous cars that are completely end to end and they take action in the real world. So I think seeing agents as a spectrum or a continuum of autonomy across different dimensions, including memory, including planning and long-term planning, short-term planning, planning horizon that they can plan.  execute things and also autonomy itself. So what I'm trying to say is that we don't want to say agents are the ones that take actions. We want to say agents live on a continuum from very basic autonomy all the way to things like autonomous vehicles that are fully end-to-end autonomous. Mahesh, you started talking about interesting  issues, but let's follow what Ellie said. How do we know agents could make an error? How can we actually even know? Yeah, so she used a very nice word. It's the continuum, right? So it's not like a single point responses that you can keep monitoring, you know, like after the agent took an action. And another thing that she mentioned, they become network actors. They will be acting at every single stage.  visibly or invisibly. And that makes it mandatory for us to make it, I would say that it has to be a runtime governance, rather than wait for, let it happen, and then we reverse. So the traditional governance models that we have or exist were very much evolved from our understanding of machine-to-machine interaction. Then it evolved into a SaaS model and a cloud model. Then we evolved to the microservices.  architecture and we kind of assumed that you know like somebody else is taking care of those governance after we reach at a scale you know like so when autonomy will reach at a scale then we cannot think of like okay we need to solve this problem now. Now we are thinking of like agent as like okay Chad GPT is one of the agents I'm just looking for the search answers.  rather than, and slowly we're saying that, oh, okay, it can just go and do some action, it can crawl some web pages. Okay, then we say that, okay, then there will be multi-agent systems, agent to agent interaction, collaboration, that will happen, and just we are adopting it without addressing the foundational issue that the safety and the governance has to be addressed, that's one, that, along with the trust, but also to make it, like, how it can become a deployable.  across the enterprises or the public services or the civil society without sacrificing, which is a very important factor, without sacrificing the openness of this execution and the accountability. So those are the two points I would like to mention. So you gave us more information than I asked for. Thank you.  incident that happens, of course, we're going to remove bad actors, right? People that are using agents for bad purposes. We're not talking about that, which is a different issue, security, cyber security, or control. We talk about agent themselves making an error or having a mistake. Some of you know that we're talking recently about agentic identity as a way to make sure that who are these agents? Where are they coming from? What's their training? What data have been used? Who created them?  who they belong, they have their passport, how they're gonna talk, what language they speak, can we talk with each other in some language and so forth, which brings now the concept of certifying agents. And I'm gonna ask to Alpesh to question, because Alpesh in the IEEE stands, and we're working a lot on many, many concepts ahead of his time on energetic, how do you think about certifying agents in the world of autonomy? How should we think about it? I think this works out.  I guess that's one of the most important points, right? Knowing when it's on and when it's not. I say that in jest, but this is also part of when we're thinking about these various models and how they interact with each other. Not having the ability to understand what is the end result or what good looks like creates a difficulty in understanding the parameters by which you can state this is working as it should. Which means that very much, and already we've  seeing this done quite successfully in a number of the use cases we've been involved in and prototypes and implementations around this that we've done is that First making sure you have a very strong governance structure in place is critical second is making sure that you have a very clear process by which you are determining what does and doesn't go out or is ready to go into the sandbox or  what you're ready to even put into that runtime state, as you were sharing earlier. It also requires ensuring that you have an iterative approach to it. These are not one-time exercises because you won't know necessarily when there might be a shift or where things may go in a different direction, right? And so this requires sort of a constant monitoring approach. Adding to this is also the need to contextualize a lot of what you're looking at.  some of the reasons why you may see the behaviors of the multi-agents may be because there was a trigger that you typically do not see or are unaware of, which means that the bounds that you currently had in mind may not be the right bounds. And therefore, it's important to also ensure that you have transparent, accountable, not only logging, but people in that process. Ultimately, at the end, what it is that you can evaluate, let's say,  at a much more level of confidence is the means by which it achieves the output. You won't know 100% if the thing that is out there is perfect. But over iterative series, you begin to clearly understand, right? This, out of the Monte Carlo simulation or whatever model you choose to apply here, at some point you have enough data to know.  what good should be, right? So our certified program allows for this. At the same time, we have a series of standards that Amir was instrumental in being involved, and a number of folks in the front row here as well, which are focused on data transparency, focused on age-appropriate design, focused on also the critical elements around what does it mean to be accountable, transparent.  What does data privacy mean in the context of systems that we don't know what they were to be, right? And what we found the greatest power in is people. It was the community of people that worked on this. They were heavily representative of a mix of lawyers, doctors, artists were even involved with, engineers, and technical people that really understood the contextual problem so well.  that it allowed for us to better understand and appreciate what some of those variances might have been. Thank you, Alpesh. Apurva, you invest in many, many startups and companies, and you are in the first line to evaluate them and see if the company is going to have solid technical background and the teams is right. But because Alpesh mentioned the terms of governance, we're talking about governance and governing AI, how do you imagine  because we talk about the technical stack, we're all familiar with the technical stack, but what would be a governance stack look like when everything is autonomous? I think the way we think about governance stack in this AI world is a 5K clear model. If I kind of go first is like the build time. Around the build time the idea is okay, how is the company architected around the idea of data governance, model versioning.  when you kind of come to the next step, which is basically deploy time. How do you think about the idea of policy, tracking, permissioning, secrets management? When we come to the third piece of it, we think about this whole idea of runtime. How are you ensuring that your architecture allows for real time observability? How do you ensure that there is an idea of kill switch? If things go bad, how do you ensure that you're able to cut it off at the right point of time? Then we think about the whole idea of  How do you drive remediation? So do you have the right audit trails to be able to do a post-mortem? Do you have the right incidence response architecture to be able to attack that? And then how can you drive a whole accountability layer, which is around the whole idea of who drives accountability? How do you have the right reporting structures? How can you conduct right post-mortem? Do you have the right set of tools to be able to do a compliance mapping there? So I think we think of  governance stack in today's world in a holistic 5K clear model and anybody who thinks about this holistically is where we feel comfortable about the fact that okay, this is the team that is truly thinking about governance in a holistic way and as you think about the best AI native teams, governance is becoming less of a compliance issue but more of like an issue which is new product in GTM because whoever is embedding governance into product is truly winning on the GTM side.  And that is becoming a competitive advantage, and so it's becoming a key part of how we evaluate a lot of the companies. Some people jokingly said a few years ago, 10, 20 years ago, we talk about intern insight when the PC era was that. Now we're talking about governance insight. Would you agree with that? 100%. I feel like with so much of uncertainty, I feel like every enterprise that is procuring today, like the upfront conversation before they take on any business through multiple layers of  evaluation is talking through a bunch of questions around what is your auditability, what's your traceability, what's your data handling practices, are the kills which is in place because the costs of things going wrong are so high that people are willing to spend millions of dollars in ensuring the right governance while even buying a half a million ACV contract. So I feel like it's becoming a core part of...  So how startup, sorry this is a question I didn't plan to ask but how startups that don't have all the funding and all the resources necessary could comply with this level of scrutiny or requirement? I think the way I think about it is when you're starting an agentic AI business in today's world, I call it like you need to have a minimal viable trust stack. You need to be at a high level, you need to be able to tell what is this agent supposed to do? Is it actually doing what it's supposed to do?  And if not, can you actually have someone stop doing what it's supposed to do on a real-time basis? And that becomes critical and for that you basically need to have what you call as a clearly defined agent identity registry. You need to have the right set of guardrails at the orchestration layer. You need to have a clear real-time observability architecture. And lastly, you need to have very clear set of defined oversight. I think if you don't have any of this,  I don't think you should be launching any Gentic ecosystem into production because you're bound to meet more failures than success and it's not that expensive to do these four basic things and of course as you scale you'll have a lot of compliance start building into your product but this is the minimal viable thing that any startup that's looking to start in the Gentic AI needs to do to be able to go into production. I think you should write an article about that because a lot of us would need that little four stack layer to get started.  To use the term orchestration, I would like to pass on to Praveen as we talk about orchestration and coordination. Because as we define multi-agent systems, they need to be coordinated in some ways to do certain things, but also to not do certain things. So Praveen, is you ready for a demo, quick demo? I think it's very tiny, so maybe you can explain. I truly believe that the next phase of AI is going to be  Or the agent system. Am I audible? Sorry about that. I thought this was working. So we at Cognizant and Cognizant AI lab truly believe that the next frontier is going to be multi-agent systems. And in that endeavor, sometime in May 2025, we sort of open-sourced our multi-agent framework, which is called Neuro AI Multi-Agent Accelerator, which you can see on the screen here. It's a low-code, no-code framework.  that's designed to accelerate the prototype development to scaling. It's a production-grade environment. Currently, it's available under Apache 2.0 license on GitHub. There's an active community that is sort of working on building and enhancing this framework as we speak. Some of the features and the capabilities of this accelerator is that it's a little bit agnostic, right?  It's cloud agnostic. So you as a user will have the control over which LLMs that you want the agent network to talk to. You have a control on which environment that you want to deploy on that. And it's interoperable. I mean, currently, we support multiple protocols from an agent to agent communication, our own OSR protocol, as well as A2N MCP server. And most importantly, as we heard about the governance, autonomy, and  the guardrails, security is built into it. So we build this from security at the core of it, not as an afterthought. So in the next one minute, and most importantly, we've made it so simple that even a seasoned AI ML engineer to a CXO who has minimal or no coding experience can start building their agent network in a matter of minutes. And that's all. The whole reason is that we want to make this as a de facto standard.  and we want everyone sort of adopt it and sort of, you know, not just experiment it, actually build those agent networks at scale and deploy them, right? So let me just do a quick demonstration. And today, what I'm going to do is we'll quickly wipe code an agent network in front of you guys. So this is a wiping environment. So let me just, anybody wants to give me a prompt? Anybody? Anything? Anything you want prompted? Yeah.  All right. Can you read, because it's not easy to read. Yeah. I'm saying create a multi-agent network. I'm saying create a multi-agent network for India AI Summit New Delhi, right? That's what they asked. OK. So what you can really see is just give it a few more seconds. What you will see is that there's an hour.  multi-agent architect and a designer which is working behind just to understand the intent. And the intent that I gave was just create a multi-agent network for India AI Summit. New Delhi, that's currently happening. First thing that you will see is that the large language model understood my requirement and it created a set of agents. It's defined the agents. Now what it's actually doing is that it's actually connecting those agents to an orchestrator. It's still building through it.  And then as a next step, what will happen really is that basically, now every agent needs to be defined. It needs to have its own roles, responsibilities, and what it should do and what it should not do. So currently, that's a step that we are actually going through it, where you can see that the matter of moment, all these individual agents will be populated with descriptors. You can see that, right? So those descriptions are nothing but its definitions, what it should be doing at the end of the day, right?  It's still finalizing the agent network. The two things that you need to see at this here is that there are agents which have been specifically created for that agent network. And then there are a few of them which are under the leaf nodes, orange in color. So basically what the designers understood that there are certain agents which are already  created and they were available in my environment, so it just connected them directly. So instead of rebuilding them, reinvesting in creating the agent network, it just connected them so you could use it. So it also allows as a designer of the agent network is that you have the control, auditability and traceability of it. At any point you think that that leaf node is not required in that agent, you can just delete it and save it, the server understands the changes and your new agent network is ready to be played around with. So now that the agent network is ready,  What I'm really going to do is launch it. You can see that the agent network is made available. Again, just let me blow it a little bit here. And before I fire a query, and again, this is a development environment. The reason we are showing a development environment is just to showcase a couple of capabilities. What you see on the right-hand side panel is basically your chat window. And then you have an internal chat.  which basically shows in real time the communication between the agents, what kind of information is being transmitted, what the nature of it is. So you as a designer have the control over it as well. So you can take a dump of it later on and you can sort of go through it, review it to see if any confidential data is being leaked through a large language model got access to it, right? And then you have the logs. The logs are pretty granular in nature in the sense that it gives you details of a level of  the token usage and the cost associated with each of that communication. And last but not the least, I mean, most of the folks here are pro ESD metrics, right? So you also get a score of energy usage, carbon footprint, and the cost associated with each of the prom that you are filing. So let me just, so I'm asking the agent, it goes, can you provide a schedule for the keynote speeches at the India AI Summit, right? So I think it's.  OK, let me see if I can get. So currently, it's not grounded. It's just talking to a large language model behind. So every agent network that we create here, by default, it talks to GPT-4. But you, as a user, have a control. As I said, it's LLM agnostic. So if you in your organization have a qualified large language model that you want this agent network to, we can. So globally, each and every agent network has access to one large language model. But the framework has the coolest capability where  Every leaf node agent also can be empowered with two specialized or general purpose agents, LLMs as well. So they act as a fallback. And if you have a specialized agent, like for auditing or governance or compliance, you want higher accuracy, you can sort of connect a specific large language model just to that leaf node agent as well. So that's sort of the, and today it's available on GitHub.  It's been there for the last five months now. We have an active community. The GitHub repository gives you access to the code base under the Apache 2.0 license. And this framework has met the highest standard of GitHub security as well today. So that's an example of a... You have no queries to run? Yeah. So that's an example of a coordination which basically takes into account all the considerations that was discussed. And you said there's open source, right?  Yes, it's open source. So here is the QR code. And if you need more, here is a QR code that you can sort of scan. It will directly take you to the GitHub repository that we have. We encourage you to fork it, clone it, and if you like it, give us a star. And if you don't like it, let us know why. And we are here to work with you on that. I just have a question. If you have multiple agents running at the same time,  All of them have different contexts and different memories. How do you manage that in a single platform? Good question. In the interest of time, let's come back to it offline. But we'll get into that. We'll get into it. Great question. So that was just a demo, too, basically. Because the audience is diverse, choosing what these things look like. But this has been certified ISO 4 as a pneumonia compared with this framework, as well. So there's a lot of that behind the scene. But I think what I was talking about has been taken into account in terms of observability, and monitoring, and so forth.  Going back to the issues of incident and tracking incident and danger that could happen, there are examples in other areas such as aviation and nuclear power, et cetera. Maybe Ellie can give us a view of how the other industries have dealt with these issues because this issue is, today we talk about agentic autonomy. We have these issues before in other sectors. How have we dealt with them? Absolutely.  The technology may be new, but the concept of safety, especially in safety critical industries, is not new. We have seen that before and we've solved for it. Let's take the example of aviations and drones, for example. So drones are a newer technology and a lot of regulatory bodies are looking at regulating drones. And the interesting thing about  looking at drones and comparing them with agents is that, when you look at, let's say, the regulations that, at least in the U.S., aviation industry is considering for drones, it was originally based on what FAA or Federal Aviation Administration calls VLOS, meaning visual line of sight, meaning that a pilot has all these  is in command and they have to keep visual line of sight with the drone. And that is how they are accountable for managing that drone. However, as the safety and as the system safety based on using AR technology develops, now the VLAS regulation is moving to beyond visual line of sight.  And that means that now the safety of an AI detect and avoid system has gone above what a human can do. And based on this advantage of detect and avoid based on AI system, now the regulation is moving from a pilot has to always keep visual line of sight to beyond visual line of sight. And using that as...  as a way to maintain safety. So drawing a comparison to agents, if you think of, when we talk about human in the loop, human in the loop, we are basically talking about the loss. The human has to keep approving every step of the way that agents are taking. However, that undermines the utility of the agent. So as agents become safer and more reliable, similar to what detect and avoid in the aviation industry,  and got better and better and better over time, we have to be moving from human in the loop to pilot in command in aviation industry. In this case, we'll be human in command. So human will be doing the supervision, but not in the loop, approving everything that agent is supposed to do. So drawing this kind of parallels with respect to other industries that are moving from safety of human.  making those decisions make a lot of sense to apply to agentic AI because similar to aviation, in this case we see that maybe keeping the human always in the loop is not the best thing. Similar to detect and avoid, there could be better safety systems that AI can provide that keeps agents even safer.  I'm not saying that data-wise today, I'm saying similar to other industries that moved from one side to the other, agentic AI would also move. Now we are human in the loop, and then we have to move towards human in command. Thank you very much. As you talked about the similarity between drones and agents and FAA as a regulation, I'm going back to the rest of the audience, the two of you, to ask a question. In your opinion, what's the balance between engineering  guidelines and technical design at the core versus regulations and norms and standards, right? Because both of them are needed, we're not discussing that. But today with what's going on with agents, in your opinion, what's the balance of that? Should it be 80% engineering design and good practices, like observability, measurement, et cetera, or should it be probably a little bit less controlled in engineering, let innovation go, but also provide a framework of  certification and standards. How do you balance these two? So I'm having an analogy with the early days of internet, like how internet even evolved at first place, being sure that it's not owned by individual player or anything. So it was all inclusive and very much focused on solving the foundational primitives like identity, discovery, trust.  and that was in a fragmented system. It was never a centralized system. So in this context, like when you said, okay, so compliance, it should be, heavy compliance will kind of sabotage the innovation, like how the tear-off we will do, and my opinion on that is we should keep the innovation open, and that remains as a data plane.  But the governance has to be the control plane. So they have to build in parallel. We cannot choose one versus other. So yeah, that's my thought. I'm curious to hear Apish's and Apuva's opinion on that as well. Yeah, I mean, I think, Amir, if we took some real world examples, there's some policies regionally that some have claimed were too early in the days  And that have led to unnecessary regulation. And the effects of that unnecessary regulation have compromised, perhaps, the innovation that could have emerged as a result. I'm not so sure whether or not I agree with that. I think I'll let time claim that one. But what I would say is that it really just depends what you're trying to do.  If you're in early days, you're just trying to understand what is possible and the bounds within your own sandbox, you should be able to test it out and understand. You need the chance to even observe it before you can jump to the conclusion that it's something that's very good to market. At the same time, when you get to the proper maturation point, as my colleagues have shared, this is when regulation starts to make a little bit more sense from a requirements  standpoint. Policy can also function, though, in an earlier stage to begin framing. And the framing is important to know just so you have a sense of where is too far, right? But framing is different than regulation. And at a certain point, as I'm sure Pruva knows as well, is that there comes a point where there's a market acceptance of what is normalized. And at that point in time,  That's when the market differentiation really starts to play a role. That's when you see the major uptick. Prior to that, you're not necessarily going to see everyone buy into paying $100,000 or $150,000 to get something certified. The costs become a barrier to entry. But this also means that there is perhaps a different way of approaching this. I go back to the comment Amir and Apurva were speaking to earlier of the mini trust stack.  You know, in many ways, we can also consider the economic problem of how you make it easier for startups to come into play. And one of those ways could be, let's say theoretically, but maybe some of us are working on this, is the integration of the required standards and policies baked into the fabric from the get-go. And then as you build and stage get beyond that, what you've already done is you've bought time.  for those startups. You've also given a level of trust for the governments as they're playing here. And you've given regulators at least something much clearer to measure against. It's very, very difficult to ask someone to measure something and hold you accountable when they don't know what they're supposed to hold you accountable against. And it's the same problem for companies. They don't want to waste time on something that makes no sense in all this market play.  So Amir, I hope that helps answer your question, but that's my take on it. Thank you. Can I ask a follow-up? Please go ahead. I may want to answer too. In your experience... Alpesh captured most of what I said, so I would save the time. Actually, I wanted to get your perspective on how you see founders actually balancing this. Again, I feel like the idea is everybody's working with the certain idea of what that framework looks like. And I feel like...  there is not a lot of mutual tension between deploying responsibly AI and what leads to high commercial deployment. It's like if you need to be a half a billion ARR business today, you need to be able to sell into enterprises in mid-market. And those mid-market solutions are selling into either government institutions, large-scale enterprises, and all of them today have a much larger business to protect.  And so if you're experimenting with a new technology that is evolving very, very fast, and if you're not able to provide for certain frameworks of those responsible guidelines, it's like, hey, I wanna make sure my data is being handled well. I wanna make sure my secrets are being managed in the right way. The agents have the right set of permissions and policies. Those guidelines are being, and I think today those frameworks are very individual.  are localized to the company that you're trying to serve. But if you kind of distill that, those will broadly come down to like 15, 20 large principles. And most of the innovation that is happening is happening within the bound of those 20, 25 frameworks. I think to Altesha's point, that is going to start solidifying a lot more as the technology matures. But I think people always try and frame  being responsible and commercial innovation as a trade-off, I feel like with the way the world is moving, it's not a tension. It's actually very, very aligned. And that's how even most of the board conversations happen is because if you can't do that, you would not be able to scale. So on that note, staying with you, Apurva, right? You talked about the trade-offs. You talked about the investment, right? If we switch gears into the economics of it, right?  So when do you see the governance actually becoming a competitive advantage and not a cost burden to the organization? To be very honest, in very early stages of the company, we have a lot of our portfolio companies at seed series A stage start selling into enterprises day one. So if you are selling into the cognizance of the world or if you're selling into the Googles of the world and all of that, a bunch of these companies today and just how innovative they are and how fast they're moving, like day one they sell into.  a half a million, a million ARR contract into a very large enterprise. And when you're doing that, you can't do that without governance. Like SOCTU used to be that version, SOCTU compliance for like your old age SAS framework there. So I feel like most of these conversations become pretty much very, very embedded in your seed series. The moment the company's about to sell, if governance is not an embedded conversation in your product,  it will not be a competitive edge for you in the GTN. And to be honest, I meet 100 AI companies a month. The best AI native teams today have actually embedded it so deeply that, like security, it's part of the weekly operating cadence. They'll talk about the idea around evals. How solid are your eval reviews? Are you doing it on a weekly basis? Are you making sure that...  meeting the standards that you want to espouse serving into an enterprise customer. They're thinking about red teaming in a very solid way. They're doing very high quality post-mortem reviews if things don't work the way they are supposed to work. So I feel like governance is like security today and they will start playing out pretty much in a series A conversation if you're selling into an enterprise. I feel like that conversation becomes later if you're a PLG led business.  that is serving individual creators, individual designers. I don't think if you are trying to buy a $50 per month contract as a designer and using that to create videos, you're thinking as much about, hey, is this meeting those governance standards that I need to? But anyone selling into a mid-market enterprise solution, this becomes a very early on conversation.  I have just a follow-up thought on that, right? So our India's digital infrastructure story, like when we built Aadhaar, we built UPI payment system, we solved identity problem, we solved the payment problem, then we were like, okay, what about the data privacy, then we have the DG Locker, then we have OCD, and then, now what is happening, that's very important success story, and very much aligned with this agentic AI, because,  the government took a stand saying that, hey, we are going to standardize the interfaces, okay? And what we really want to do, the rest of the people or the companies, they can do any application that they want on top of it. So that government plays a very important role in this agentic AI evolution, I would say. Thank you. Thank you. At the beginning, we talked about governing autonomy and we discussed about, autonomy talks about governance to some extent, I think.  What would be, in your opinion, well, there are two aspects to that question. One is if we count about trust, what is infrastructure for trust? Because it's so fragmented between different pieces. And then how should we think about the one single thing that would enable us to go to the right direction? Because we talked about, again, many things, but in your opinion, what would be the one single thing that we have to do today, given what you expect?  What you mentioned is just that we need to give innovation open, we need to experiment, we need to have some general guidance and framing, of course. But besides that, what would be one little thing that we need to really be pushing collectively, not as individuals but collectively, to make everybody progress, besides open sourcing it, of course? Eileen, do you want to start? Sure. Yeah, I can start.  Maybe one of the things that we can think about today is how can we evolve testing methods and evaluation benchmarks for multi-agent systems. And the reason I say that is that we have a paper called Distributional AGI Safety, meaning that the paper discusses the fact that AGI may not emerge as one monolithic, big, powerful system, but it may actually emerge as a lot of sub.  top AGI agents, specialized agents, working with each other in a multi-agent system. So we may be talking about multi-agent systems that reach AGI level capabilities. And for that to be successful, what should we be thinking about today when it arrives?  In that paper, we discussed four different categories of safety, including market design, including baseline safety for each agent itself, including inter-agent communication, et cetera, et cetera. Please feel free to refer to that paper. But I think what we need to do today is making sure that we build evaluation benchmarks that are relevant to multi-agent systems.  Frontier Labs that are developing these models that are core to the agents, it's important for them to, and by them I mean Google Live Mind being one of them, to report, test against these benchmarks, report transparently about the limitations, report transparently about the capabilities, and I think that is one of the goals that.  Google DeepMind has been pursuing, and a lot of that information is available on the website. Please feel free to read our frameworks on AGI safety as well. So I'll stop there. So I'll just, again, follow up on that, right? One of the important factors that we're missing in the multi-agent system, we are assuming that it's a single model agent, right? Across the multi-agent system, I'm using one LLM model.  But when we extend that thought, that what will happen when I'm using models provided by different players, open source or non-open source, you know, like all of those models will try to... So, single-matter responses are easy to verify. We can have the metrics or all the telemetry associated with it to verify whether the agent is claiming. But when it goes across those boundaries...  And it becomes like a multimodal system, multi-agentic system. And every single company is saying, hey, trust us. So one of the thought experiments, or the way we try to solve it, is get every voice heard. Meaning that let's hear from all the academia how they think about this problem. Get all the LLM or a big.  per companies getting involved in this discussion. The same thing with the system integrators and hyperscalers also participate in it. And not just them, but including the policy makers as well as the younger generation who is gonna be the agentic AI native. So how are we preparing for that? That's very important. So the problem doesn't...  get solved with the single model responses. So the only solution that we kind of lean towards to make it enough decentralized or enough integrated system so that we can have those test parameters across every single thing. Yeah, thank you. I think for me, largely, the thing which I think can solve a lot of it because that kind of distills enough information to people to start working from is  the idea around auditability and traceability to be able to capture and accurate reporting in an immutable way of a lot of what is happening. And as long as there are more open source platforms that allow for a lot of those, you know, anonymized way of auditability and traceability and the incidents that's coming to come up, I think it drives for more collective learning at a more global platform level. I don't know if there are...  And this is a conversation where when we meet a lot of startups, they're actually learning a lot from talking to each other. And they keep doing about the fact that there are not a lot of global platforms that just talk about just learning around auditable trails of what is happening, how is it happening, where do systems fail at what scale and level of infrastructure. And I think something that solves for just a collective platform to learn from.  will go a long way. It's a very innovative environment around the world and what's happening. So as long as you provide the right infrastructure for even knowing what's going wrong, you know, I think there's a lot that can actually accelerate the learning and development there. That's a cue for our session. Oh, I thought that was where you programmed the agenda. Yeah, so I think there's two.  Two key points I'd like to just leave everyone with. The first, Amir runs a very innovative organization called AI Commons that many of us have been a part of over the years. And really the critical nature of what they have been looking at, which very much aligned with a lot of what we were looking at as well, was the idea of having these technologies  treated as a commons, a public utility. Because if you don't do that, you can't do exactly what everyone here is speaking to. Only those that have access to that can do that. And the point on education is very critical. But the point, I would just read into this then, is the thing to do today, we have to spend a lot more time getting rid of the jargon. Even the title of today's session.  isn't something that's accessible readily to the same people we're asking to make use of these technologies, right? And so a lot of the effort needs to be spent on demystifying what we're talking about. Show the value of it. Because if we don't do that, then you're going to have a lot of fear. And a lot of the fear is unnecessary. And I think these would be the critical things. I would also love to use your app to turn off that alarm. So maybe you can teach us later about that. Thanks. Thank you very much.  Thank you. Thank you everyone for coming.",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771327320927.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:33",
                    "text": " We'll say probably a few words about what they do. I'm sorry, but Eddie Sahai from Google. Go ahead, Eddie. Hi, everyone. I'm Eddie Sahai. I am policy manager working on AI and emerging tech, particularly agent AI and robotics. Hello, and good afternoon. This is Mahesh. Thank you all, and thanks to Amir and Pauline for having us here. And also, they are generous enough to put my younger version."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:33",
                    "end": "00:01:03",
                    "text": " in the picture. And so I'm from Project Nanda. We are pioneering the building and foundational infrastructure for the internet of AI agents. It's a new kind of internet is born, or getting born, from last year onwards. And one of the most foundational issues, the way I say that, trust, that is me. That's my photo. The same thing, how we are gonna solve this problem using"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:03",
                    "end": "00:01:32",
                    "text": " in this internet of AI agents, and those primitives need to be solved at the foundational layer. It's not like after the fact or after the thought solutioning. It has to happen as a first principle. Yeah, thank you. Alpesh. Hello, I'm Alpesh Shah. I am the Managing Director of the IEEE Standards Association and a member of the Management Council at IEEE. Thank you. Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:33",
                    "end": "00:02:04",
                    "text": " Hey, guys. Hi. My name is Apoorv Goyal. I'm a principal at Insight Partners. I help lead a lot of our US-India investing efforts. Just as context, we're one of the leading global technology investment funds. We manage close to $90 billion in AUM investing out of a $12.5 billion fund into leading global software companies around the world, including a bunch of companies based out of India. Wonderful. Thank you so much. I'll just be getting to Praveen. He's my co-moderator. Praveen is the head of AI Lab."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:04",
                    "end": "00:02:36",
                    "text": " but before we start talking about it, with that, as a raise of hand, how many of you have built an agent or played with an agent? How many of you are scared of what's gonna happen with agents? One is probably like 5%, 10%. So it thinks that 90 or 80 or 90% of you don't feel that agents are gonna be dangerous or anything else. Have you followed what happened with Open Cloud recently and Notebook?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:36",
                    "end": "00:03:03",
                    "text": " Do you feel this is cool or exciting or energizing or fun? Who's excited about what happened? It's a beautiful experiment, right? So we're learning about all these things and as we talk about agentic world, the concept of autonomy scares a lot of people. Why? Because we don't fully know how they're built. We don't fully know how they're going to react, if their objective is going to be changing, if they're going to be..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:03",
                    "end": "00:03:28",
                    "text": " mixing with each other, including new babies, we don't know what's gonna happen. And then in that world of autonomy, a lot of decisions that enterprises take and societies functioning may not be able to be put in the hands of agents. So the concept of governance, which is a concept which is very encompassing about how we control the outcome of systems and procedures and ownership and risk management."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:28",
                    "end": "00:03:57",
                    "text": " becomes central again in the world of agent, which was not the case about even a year ago. So today we're going to be talking about this concept of governing autonomy, how we can actually make sure that this world, this future world, we're going to be fully autonomous, not only is our hand to some extent, but we can somehow trust it, and we don't have to really extend. So to get started, maybe we can have a quick demo. What do you think, Praveen? Questions?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:58",
                    "end": "00:04:27",
                    "text": " Okay, perfect, that's a good thing. So maybe I can start with Eli. Eli, in your vantage point, can you help us understand what's different between multi-agent environment and systems and traditional AI that we've been used to? How should we think about it? So with the raise of hand that I saw in this room, I think we are talking with a group of experts."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:27",
                    "end": "00:04:55",
                    "text": " From my perspective, and many of you have already worked with ModBug and have seen OpenClaw and have built agents, so you know better than me that the way that we think about agents, it's not a specific line or a specific point in time that these group of things are agentic and this group of things are non-agentic, rather thinking about them as a spectrum of agency or autonomy going from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:55",
                    "end": "00:05:15",
                    "text": " basic chatbots that can do a little bit of research and at the end present the result. For example, Google Deep Research does a research given your prompt and finds the information and presents that information. That has some agentic features, but doesn't take actions."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:15",
                    "end": "00:05:45",
                    "text": " compared to let's say going all the way towards the end of the autonomy that is, for example, autonomous cars that are completely end to end and they take action in the real world. So I think seeing agents as a spectrum or a continuum of autonomy across different dimensions, including memory, including planning and long-term planning, short-term planning, planning horizon that they can plan."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:45",
                    "end": "00:06:15",
                    "text": " execute things and also autonomy itself. So what I'm trying to say is that we don't want to say agents are the ones that take actions. We want to say agents live on a continuum from very basic autonomy all the way to things like autonomous vehicles that are fully end-to-end autonomous. Mahesh, you started talking about interesting"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:15",
                    "end": "00:06:44",
                    "text": " issues, but let's follow what Ellie said. How do we know agents could make an error? How can we actually even know? Yeah, so she used a very nice word. It's the continuum, right? So it's not like a single point responses that you can keep monitoring, you know, like after the agent took an action. And another thing that she mentioned, they become network actors. They will be acting at every single stage."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:44",
                    "end": "00:07:14",
                    "text": " visibly or invisibly. And that makes it mandatory for us to make it, I would say that it has to be a runtime governance, rather than wait for, let it happen, and then we reverse. So the traditional governance models that we have or exist were very much evolved from our understanding of machine-to-machine interaction. Then it evolved into a SaaS model and a cloud model. Then we evolved to the microservices."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:14",
                    "end": "00:07:38",
                    "text": " architecture and we kind of assumed that you know like somebody else is taking care of those governance after we reach at a scale you know like so when autonomy will reach at a scale then we cannot think of like okay we need to solve this problem now. Now we are thinking of like agent as like okay Chad GPT is one of the agents I'm just looking for the search answers."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:38",
                    "end": "00:08:07",
                    "text": " rather than, and slowly we're saying that, oh, okay, it can just go and do some action, it can crawl some web pages. Okay, then we say that, okay, then there will be multi-agent systems, agent to agent interaction, collaboration, that will happen, and just we are adopting it without addressing the foundational issue that the safety and the governance has to be addressed, that's one, that, along with the trust, but also to make it, like, how it can become a deployable."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:07",
                    "end": "00:08:30",
                    "text": " across the enterprises or the public services or the civil society without sacrificing, which is a very important factor, without sacrificing the openness of this execution and the accountability. So those are the two points I would like to mention. So you gave us more information than I asked for. Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:30",
                    "end": "00:08:57",
                    "text": " incident that happens, of course, we're going to remove bad actors, right? People that are using agents for bad purposes. We're not talking about that, which is a different issue, security, cyber security, or control. We talk about agent themselves making an error or having a mistake. Some of you know that we're talking recently about agentic identity as a way to make sure that who are these agents? Where are they coming from? What's their training? What data have been used? Who created them?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:57",
                    "end": "00:09:36",
                    "text": " who they belong, they have their passport, how they're gonna talk, what language they speak, can we talk with each other in some language and so forth, which brings now the concept of certifying agents. And I'm gonna ask to Alpesh to question, because Alpesh in the IEEE stands, and we're working a lot on many, many concepts ahead of his time on energetic, how do you think about certifying agents in the world of autonomy? How should we think about it? I think this works out."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:37",
                    "end": "00:10:11",
                    "text": " I guess that's one of the most important points, right? Knowing when it's on and when it's not. I say that in jest, but this is also part of when we're thinking about these various models and how they interact with each other. Not having the ability to understand what is the end result or what good looks like creates a difficulty in understanding the parameters by which you can state this is working as it should. Which means that very much, and already we've"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:11",
                    "end": "00:10:40",
                    "text": " seeing this done quite successfully in a number of the use cases we've been involved in and prototypes and implementations around this that we've done is that First making sure you have a very strong governance structure in place is critical second is making sure that you have a very clear process by which you are determining what does and doesn't go out or is ready to go into the sandbox or"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:40",
                    "end": "00:11:11",
                    "text": " what you're ready to even put into that runtime state, as you were sharing earlier. It also requires ensuring that you have an iterative approach to it. These are not one-time exercises because you won't know necessarily when there might be a shift or where things may go in a different direction, right? And so this requires sort of a constant monitoring approach. Adding to this is also the need to contextualize a lot of what you're looking at."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:11",
                    "end": "00:11:41",
                    "text": " some of the reasons why you may see the behaviors of the multi-agents may be because there was a trigger that you typically do not see or are unaware of, which means that the bounds that you currently had in mind may not be the right bounds. And therefore, it's important to also ensure that you have transparent, accountable, not only logging, but people in that process. Ultimately, at the end, what it is that you can evaluate, let's say,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:41",
                    "end": "00:12:11",
                    "text": " at a much more level of confidence is the means by which it achieves the output. You won't know 100% if the thing that is out there is perfect. But over iterative series, you begin to clearly understand, right? This, out of the Monte Carlo simulation or whatever model you choose to apply here, at some point you have enough data to know."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:11",
                    "end": "00:12:41",
                    "text": " what good should be, right? So our certified program allows for this. At the same time, we have a series of standards that Amir was instrumental in being involved, and a number of folks in the front row here as well, which are focused on data transparency, focused on age-appropriate design, focused on also the critical elements around what does it mean to be accountable, transparent."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:41",
                    "end": "00:13:10",
                    "text": " What does data privacy mean in the context of systems that we don't know what they were to be, right? And what we found the greatest power in is people. It was the community of people that worked on this. They were heavily representative of a mix of lawyers, doctors, artists were even involved with, engineers, and technical people that really understood the contextual problem so well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:10",
                    "end": "00:13:41",
                    "text": " that it allowed for us to better understand and appreciate what some of those variances might have been. Thank you, Alpesh. Apurva, you invest in many, many startups and companies, and you are in the first line to evaluate them and see if the company is going to have solid technical background and the teams is right. But because Alpesh mentioned the terms of governance, we're talking about governance and governing AI, how do you imagine"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:41",
                    "end": "00:14:08",
                    "text": " because we talk about the technical stack, we're all familiar with the technical stack, but what would be a governance stack look like when everything is autonomous? I think the way we think about governance stack in this AI world is a 5K clear model. If I kind of go first is like the build time. Around the build time the idea is okay, how is the company architected around the idea of data governance, model versioning."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:08",
                    "end": "00:14:38",
                    "text": " when you kind of come to the next step, which is basically deploy time. How do you think about the idea of policy, tracking, permissioning, secrets management? When we come to the third piece of it, we think about this whole idea of runtime. How are you ensuring that your architecture allows for real time observability? How do you ensure that there is an idea of kill switch? If things go bad, how do you ensure that you're able to cut it off at the right point of time? Then we think about the whole idea of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:38",
                    "end": "00:15:07",
                    "text": " How do you drive remediation? So do you have the right audit trails to be able to do a post-mortem? Do you have the right incidence response architecture to be able to attack that? And then how can you drive a whole accountability layer, which is around the whole idea of who drives accountability? How do you have the right reporting structures? How can you conduct right post-mortem? Do you have the right set of tools to be able to do a compliance mapping there? So I think we think of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:07",
                    "end": "00:15:35",
                    "text": " governance stack in today's world in a holistic 5K clear model and anybody who thinks about this holistically is where we feel comfortable about the fact that okay, this is the team that is truly thinking about governance in a holistic way and as you think about the best AI native teams, governance is becoming less of a compliance issue but more of like an issue which is new product in GTM because whoever is embedding governance into product is truly winning on the GTM side."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:35",
                    "end": "00:16:04",
                    "text": " And that is becoming a competitive advantage, and so it's becoming a key part of how we evaluate a lot of the companies. Some people jokingly said a few years ago, 10, 20 years ago, we talk about intern insight when the PC era was that. Now we're talking about governance insight. Would you agree with that? 100%. I feel like with so much of uncertainty, I feel like every enterprise that is procuring today, like the upfront conversation before they take on any business through multiple layers of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:04",
                    "end": "00:16:29",
                    "text": " evaluation is talking through a bunch of questions around what is your auditability, what's your traceability, what's your data handling practices, are the kills which is in place because the costs of things going wrong are so high that people are willing to spend millions of dollars in ensuring the right governance while even buying a half a million ACV contract. So I feel like it's becoming a core part of..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:29",
                    "end": "00:16:59",
                    "text": " So how startup, sorry this is a question I didn't plan to ask but how startups that don't have all the funding and all the resources necessary could comply with this level of scrutiny or requirement? I think the way I think about it is when you're starting an agentic AI business in today's world, I call it like you need to have a minimal viable trust stack. You need to be at a high level, you need to be able to tell what is this agent supposed to do? Is it actually doing what it's supposed to do?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:59",
                    "end": "00:17:28",
                    "text": " And if not, can you actually have someone stop doing what it's supposed to do on a real-time basis? And that becomes critical and for that you basically need to have what you call as a clearly defined agent identity registry. You need to have the right set of guardrails at the orchestration layer. You need to have a clear real-time observability architecture. And lastly, you need to have very clear set of defined oversight. I think if you don't have any of this,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:28",
                    "end": "00:17:58",
                    "text": " I don't think you should be launching any Gentic ecosystem into production because you're bound to meet more failures than success and it's not that expensive to do these four basic things and of course as you scale you'll have a lot of compliance start building into your product but this is the minimal viable thing that any startup that's looking to start in the Gentic AI needs to do to be able to go into production. I think you should write an article about that because a lot of us would need that little four stack layer to get started."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:58",
                    "end": "00:18:33",
                    "text": " To use the term orchestration, I would like to pass on to Praveen as we talk about orchestration and coordination. Because as we define multi-agent systems, they need to be coordinated in some ways to do certain things, but also to not do certain things. So Praveen, is you ready for a demo, quick demo? I think it's very tiny, so maybe you can explain. I truly believe that the next phase of AI is going to be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:33",
                    "end": "00:19:01",
                    "text": " Or the agent system. Am I audible? Sorry about that. I thought this was working. So we at Cognizant and Cognizant AI lab truly believe that the next frontier is going to be multi-agent systems. And in that endeavor, sometime in May 2025, we sort of open-sourced our multi-agent framework, which is called Neuro AI Multi-Agent Accelerator, which you can see on the screen here. It's a low-code, no-code framework."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:01",
                    "end": "00:19:24",
                    "text": " that's designed to accelerate the prototype development to scaling. It's a production-grade environment. Currently, it's available under Apache 2.0 license on GitHub. There's an active community that is sort of working on building and enhancing this framework as we speak. Some of the features and the capabilities of this accelerator is that it's a little bit agnostic, right?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:24",
                    "end": "00:19:52",
                    "text": " It's cloud agnostic. So you as a user will have the control over which LLMs that you want the agent network to talk to. You have a control on which environment that you want to deploy on that. And it's interoperable. I mean, currently, we support multiple protocols from an agent to agent communication, our own OSR protocol, as well as A2N MCP server. And most importantly, as we heard about the governance, autonomy, and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:52",
                    "end": "00:20:23",
                    "text": " the guardrails, security is built into it. So we build this from security at the core of it, not as an afterthought. So in the next one minute, and most importantly, we've made it so simple that even a seasoned AI ML engineer to a CXO who has minimal or no coding experience can start building their agent network in a matter of minutes. And that's all. The whole reason is that we want to make this as a de facto standard."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:23",
                    "end": "00:21:01",
                    "text": " and we want everyone sort of adopt it and sort of, you know, not just experiment it, actually build those agent networks at scale and deploy them, right? So let me just do a quick demonstration. And today, what I'm going to do is we'll quickly wipe code an agent network in front of you guys. So this is a wiping environment. So let me just, anybody wants to give me a prompt? Anybody? Anything? Anything you want prompted? Yeah."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:01",
                    "end": "00:21:43",
                    "text": " All right. Can you read, because it's not easy to read. Yeah. I'm saying create a multi-agent network. I'm saying create a multi-agent network for India AI Summit New Delhi, right? That's what they asked. OK. So what you can really see is just give it a few more seconds. What you will see is that there's an hour."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:43",
                    "end": "00:22:09",
                    "text": " multi-agent architect and a designer which is working behind just to understand the intent. And the intent that I gave was just create a multi-agent network for India AI Summit. New Delhi, that's currently happening. First thing that you will see is that the large language model understood my requirement and it created a set of agents. It's defined the agents. Now what it's actually doing is that it's actually connecting those agents to an orchestrator. It's still building through it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:11",
                    "end": "00:22:41",
                    "text": " And then as a next step, what will happen really is that basically, now every agent needs to be defined. It needs to have its own roles, responsibilities, and what it should do and what it should not do. So currently, that's a step that we are actually going through it, where you can see that the matter of moment, all these individual agents will be populated with descriptors. You can see that, right? So those descriptions are nothing but its definitions, what it should be doing at the end of the day, right?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:41",
                    "end": "00:22:58",
                    "text": " It's still finalizing the agent network. The two things that you need to see at this here is that there are agents which have been specifically created for that agent network. And then there are a few of them which are under the leaf nodes, orange in color. So basically what the designers understood that there are certain agents which are already"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:58",
                    "end": "00:23:27",
                    "text": " created and they were available in my environment, so it just connected them directly. So instead of rebuilding them, reinvesting in creating the agent network, it just connected them so you could use it. So it also allows as a designer of the agent network is that you have the control, auditability and traceability of it. At any point you think that that leaf node is not required in that agent, you can just delete it and save it, the server understands the changes and your new agent network is ready to be played around with. So now that the agent network is ready,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:27",
                    "end": "00:23:54",
                    "text": " What I'm really going to do is launch it. You can see that the agent network is made available. Again, just let me blow it a little bit here. And before I fire a query, and again, this is a development environment. The reason we are showing a development environment is just to showcase a couple of capabilities. What you see on the right-hand side panel is basically your chat window. And then you have an internal chat."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:54",
                    "end": "00:24:21",
                    "text": " which basically shows in real time the communication between the agents, what kind of information is being transmitted, what the nature of it is. So you as a designer have the control over it as well. So you can take a dump of it later on and you can sort of go through it, review it to see if any confidential data is being leaked through a large language model got access to it, right? And then you have the logs. The logs are pretty granular in nature in the sense that it gives you details of a level of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:21",
                    "end": "00:24:53",
                    "text": " the token usage and the cost associated with each of that communication. And last but not the least, I mean, most of the folks here are pro ESD metrics, right? So you also get a score of energy usage, carbon footprint, and the cost associated with each of the prom that you are filing. So let me just, so I'm asking the agent, it goes, can you provide a schedule for the keynote speeches at the India AI Summit, right? So I think it's."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:53",
                    "end": "00:25:23",
                    "text": " OK, let me see if I can get. So currently, it's not grounded. It's just talking to a large language model behind. So every agent network that we create here, by default, it talks to GPT-4. But you, as a user, have a control. As I said, it's LLM agnostic. So if you in your organization have a qualified large language model that you want this agent network to, we can. So globally, each and every agent network has access to one large language model. But the framework has the coolest capability where"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:23",
                    "end": "00:25:50",
                    "text": " Every leaf node agent also can be empowered with two specialized or general purpose agents, LLMs as well. So they act as a fallback. And if you have a specialized agent, like for auditing or governance or compliance, you want higher accuracy, you can sort of connect a specific large language model just to that leaf node agent as well. So that's sort of the, and today it's available on GitHub."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:50",
                    "end": "00:26:25",
                    "text": " It's been there for the last five months now. We have an active community. The GitHub repository gives you access to the code base under the Apache 2.0 license. And this framework has met the highest standard of GitHub security as well today. So that's an example of a... You have no queries to run? Yeah. So that's an example of a coordination which basically takes into account all the considerations that was discussed. And you said there's open source, right?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:25",
                    "end": "00:26:55",
                    "text": " Yes, it's open source. So here is the QR code. And if you need more, here is a QR code that you can sort of scan. It will directly take you to the GitHub repository that we have. We encourage you to fork it, clone it, and if you like it, give us a star. And if you don't like it, let us know why. And we are here to work with you on that. I just have a question. If you have multiple agents running at the same time,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:55",
                    "end": "00:27:24",
                    "text": " All of them have different contexts and different memories. How do you manage that in a single platform? Good question. In the interest of time, let's come back to it offline. But we'll get into that. We'll get into it. Great question. So that was just a demo, too, basically. Because the audience is diverse, choosing what these things look like. But this has been certified ISO 4 as a pneumonia compared with this framework, as well. So there's a lot of that behind the scene. But I think what I was talking about has been taken into account in terms of observability, and monitoring, and so forth."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:24",
                    "end": "00:27:52",
                    "text": " Going back to the issues of incident and tracking incident and danger that could happen, there are examples in other areas such as aviation and nuclear power, et cetera. Maybe Ellie can give us a view of how the other industries have dealt with these issues because this issue is, today we talk about agentic autonomy. We have these issues before in other sectors. How have we dealt with them? Absolutely."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:52",
                    "end": "00:28:19",
                    "text": " The technology may be new, but the concept of safety, especially in safety critical industries, is not new. We have seen that before and we've solved for it. Let's take the example of aviations and drones, for example. So drones are a newer technology and a lot of regulatory bodies are looking at regulating drones. And the interesting thing about"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:19",
                    "end": "00:28:48",
                    "text": " looking at drones and comparing them with agents is that, when you look at, let's say, the regulations that, at least in the U.S., aviation industry is considering for drones, it was originally based on what FAA or Federal Aviation Administration calls VLOS, meaning visual line of sight, meaning that a pilot has all these"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:48",
                    "end": "00:29:13",
                    "text": " is in command and they have to keep visual line of sight with the drone. And that is how they are accountable for managing that drone. However, as the safety and as the system safety based on using AR technology develops, now the VLAS regulation is moving to beyond visual line of sight."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:13",
                    "end": "00:29:41",
                    "text": " And that means that now the safety of an AI detect and avoid system has gone above what a human can do. And based on this advantage of detect and avoid based on AI system, now the regulation is moving from a pilot has to always keep visual line of sight to beyond visual line of sight. And using that as..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:42",
                    "end": "00:30:10",
                    "text": " as a way to maintain safety. So drawing a comparison to agents, if you think of, when we talk about human in the loop, human in the loop, we are basically talking about the loss. The human has to keep approving every step of the way that agents are taking. However, that undermines the utility of the agent. So as agents become safer and more reliable, similar to what detect and avoid in the aviation industry,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:10",
                    "end": "00:30:35",
                    "text": " and got better and better and better over time, we have to be moving from human in the loop to pilot in command in aviation industry. In this case, we'll be human in command. So human will be doing the supervision, but not in the loop, approving everything that agent is supposed to do. So drawing this kind of parallels with respect to other industries that are moving from safety of human."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:35",
                    "end": "00:30:59",
                    "text": " making those decisions make a lot of sense to apply to agentic AI because similar to aviation, in this case we see that maybe keeping the human always in the loop is not the best thing. Similar to detect and avoid, there could be better safety systems that AI can provide that keeps agents even safer."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:59",
                    "end": "00:31:29",
                    "text": " I'm not saying that data-wise today, I'm saying similar to other industries that moved from one side to the other, agentic AI would also move. Now we are human in the loop, and then we have to move towards human in command. Thank you very much. As you talked about the similarity between drones and agents and FAA as a regulation, I'm going back to the rest of the audience, the two of you, to ask a question. In your opinion, what's the balance between engineering"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:29",
                    "end": "00:31:59",
                    "text": " guidelines and technical design at the core versus regulations and norms and standards, right? Because both of them are needed, we're not discussing that. But today with what's going on with agents, in your opinion, what's the balance of that? Should it be 80% engineering design and good practices, like observability, measurement, et cetera, or should it be probably a little bit less controlled in engineering, let innovation go, but also provide a framework of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:59",
                    "end": "00:32:28",
                    "text": " certification and standards. How do you balance these two? So I'm having an analogy with the early days of internet, like how internet even evolved at first place, being sure that it's not owned by individual player or anything. So it was all inclusive and very much focused on solving the foundational primitives like identity, discovery, trust."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:28",
                    "end": "00:32:57",
                    "text": " and that was in a fragmented system. It was never a centralized system. So in this context, like when you said, okay, so compliance, it should be, heavy compliance will kind of sabotage the innovation, like how the tear-off we will do, and my opinion on that is we should keep the innovation open, and that remains as a data plane."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:57",
                    "end": "00:33:35",
                    "text": " But the governance has to be the control plane. So they have to build in parallel. We cannot choose one versus other. So yeah, that's my thought. I'm curious to hear Apish's and Apuva's opinion on that as well. Yeah, I mean, I think, Amir, if we took some real world examples, there's some policies regionally that some have claimed were too early in the days"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:35",
                    "end": "00:34:02",
                    "text": " And that have led to unnecessary regulation. And the effects of that unnecessary regulation have compromised, perhaps, the innovation that could have emerged as a result. I'm not so sure whether or not I agree with that. I think I'll let time claim that one. But what I would say is that it really just depends what you're trying to do."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:02",
                    "end": "00:34:32",
                    "text": " If you're in early days, you're just trying to understand what is possible and the bounds within your own sandbox, you should be able to test it out and understand. You need the chance to even observe it before you can jump to the conclusion that it's something that's very good to market. At the same time, when you get to the proper maturation point, as my colleagues have shared, this is when regulation starts to make a little bit more sense from a requirements"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:32",
                    "end": "00:35:05",
                    "text": " standpoint. Policy can also function, though, in an earlier stage to begin framing. And the framing is important to know just so you have a sense of where is too far, right? But framing is different than regulation. And at a certain point, as I'm sure Pruva knows as well, is that there comes a point where there's a market acceptance of what is normalized. And at that point in time,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:05",
                    "end": "00:35:36",
                    "text": " That's when the market differentiation really starts to play a role. That's when you see the major uptick. Prior to that, you're not necessarily going to see everyone buy into paying $100,000 or $150,000 to get something certified. The costs become a barrier to entry. But this also means that there is perhaps a different way of approaching this. I go back to the comment Amir and Apurva were speaking to earlier of the mini trust stack."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:36",
                    "end": "00:36:06",
                    "text": " You know, in many ways, we can also consider the economic problem of how you make it easier for startups to come into play. And one of those ways could be, let's say theoretically, but maybe some of us are working on this, is the integration of the required standards and policies baked into the fabric from the get-go. And then as you build and stage get beyond that, what you've already done is you've bought time."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:06",
                    "end": "00:36:33",
                    "text": " for those startups. You've also given a level of trust for the governments as they're playing here. And you've given regulators at least something much clearer to measure against. It's very, very difficult to ask someone to measure something and hold you accountable when they don't know what they're supposed to hold you accountable against. And it's the same problem for companies. They don't want to waste time on something that makes no sense in all this market play."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:33",
                    "end": "00:37:06",
                    "text": " So Amir, I hope that helps answer your question, but that's my take on it. Thank you. Can I ask a follow-up? Please go ahead. I may want to answer too. In your experience... Alpesh captured most of what I said, so I would save the time. Actually, I wanted to get your perspective on how you see founders actually balancing this. Again, I feel like the idea is everybody's working with the certain idea of what that framework looks like. And I feel like..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:06",
                    "end": "00:37:34",
                    "text": " there is not a lot of mutual tension between deploying responsibly AI and what leads to high commercial deployment. It's like if you need to be a half a billion ARR business today, you need to be able to sell into enterprises in mid-market. And those mid-market solutions are selling into either government institutions, large-scale enterprises, and all of them today have a much larger business to protect."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:34",
                    "end": "00:38:02",
                    "text": " And so if you're experimenting with a new technology that is evolving very, very fast, and if you're not able to provide for certain frameworks of those responsible guidelines, it's like, hey, I wanna make sure my data is being handled well. I wanna make sure my secrets are being managed in the right way. The agents have the right set of permissions and policies. Those guidelines are being, and I think today those frameworks are very individual."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:02",
                    "end": "00:38:29",
                    "text": " are localized to the company that you're trying to serve. But if you kind of distill that, those will broadly come down to like 15, 20 large principles. And most of the innovation that is happening is happening within the bound of those 20, 25 frameworks. I think to Altesha's point, that is going to start solidifying a lot more as the technology matures. But I think people always try and frame"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:29",
                    "end": "00:38:56",
                    "text": " being responsible and commercial innovation as a trade-off, I feel like with the way the world is moving, it's not a tension. It's actually very, very aligned. And that's how even most of the board conversations happen is because if you can't do that, you would not be able to scale. So on that note, staying with you, Apurva, right? You talked about the trade-offs. You talked about the investment, right? If we switch gears into the economics of it, right?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:56",
                    "end": "00:39:24",
                    "text": " So when do you see the governance actually becoming a competitive advantage and not a cost burden to the organization? To be very honest, in very early stages of the company, we have a lot of our portfolio companies at seed series A stage start selling into enterprises day one. So if you are selling into the cognizance of the world or if you're selling into the Googles of the world and all of that, a bunch of these companies today and just how innovative they are and how fast they're moving, like day one they sell into."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:24",
                    "end": "00:39:51",
                    "text": " a half a million, a million ARR contract into a very large enterprise. And when you're doing that, you can't do that without governance. Like SOCTU used to be that version, SOCTU compliance for like your old age SAS framework there. So I feel like most of these conversations become pretty much very, very embedded in your seed series. The moment the company's about to sell, if governance is not an embedded conversation in your product,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:51",
                    "end": "00:40:15",
                    "text": " it will not be a competitive edge for you in the GTN. And to be honest, I meet 100 AI companies a month. The best AI native teams today have actually embedded it so deeply that, like security, it's part of the weekly operating cadence. They'll talk about the idea around evals. How solid are your eval reviews? Are you doing it on a weekly basis? Are you making sure that..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:15",
                    "end": "00:40:41",
                    "text": " meeting the standards that you want to espouse serving into an enterprise customer. They're thinking about red teaming in a very solid way. They're doing very high quality post-mortem reviews if things don't work the way they are supposed to work. So I feel like governance is like security today and they will start playing out pretty much in a series A conversation if you're selling into an enterprise. I feel like that conversation becomes later if you're a PLG led business."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:41",
                    "end": "00:41:00",
                    "text": " that is serving individual creators, individual designers. I don't think if you are trying to buy a $50 per month contract as a designer and using that to create videos, you're thinking as much about, hey, is this meeting those governance standards that I need to? But anyone selling into a mid-market enterprise solution, this becomes a very early on conversation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:00",
                    "end": "00:41:29",
                    "text": " I have just a follow-up thought on that, right? So our India's digital infrastructure story, like when we built Aadhaar, we built UPI payment system, we solved identity problem, we solved the payment problem, then we were like, okay, what about the data privacy, then we have the DG Locker, then we have OCD, and then, now what is happening, that's very important success story, and very much aligned with this agentic AI, because,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:29",
                    "end": "00:42:01",
                    "text": " the government took a stand saying that, hey, we are going to standardize the interfaces, okay? And what we really want to do, the rest of the people or the companies, they can do any application that they want on top of it. So that government plays a very important role in this agentic AI evolution, I would say. Thank you. Thank you. At the beginning, we talked about governing autonomy and we discussed about, autonomy talks about governance to some extent, I think."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:01",
                    "end": "00:42:30",
                    "text": " What would be, in your opinion, well, there are two aspects to that question. One is if we count about trust, what is infrastructure for trust? Because it's so fragmented between different pieces. And then how should we think about the one single thing that would enable us to go to the right direction? Because we talked about, again, many things, but in your opinion, what would be the one single thing that we have to do today, given what you expect?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:30",
                    "end": "00:42:55",
                    "text": " What you mentioned is just that we need to give innovation open, we need to experiment, we need to have some general guidance and framing, of course. But besides that, what would be one little thing that we need to really be pushing collectively, not as individuals but collectively, to make everybody progress, besides open sourcing it, of course? Eileen, do you want to start? Sure. Yeah, I can start."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:55",
                    "end": "00:43:24",
                    "text": " Maybe one of the things that we can think about today is how can we evolve testing methods and evaluation benchmarks for multi-agent systems. And the reason I say that is that we have a paper called Distributional AGI Safety, meaning that the paper discusses the fact that AGI may not emerge as one monolithic, big, powerful system, but it may actually emerge as a lot of sub."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:24",
                    "end": "00:43:47",
                    "text": " top AGI agents, specialized agents, working with each other in a multi-agent system. So we may be talking about multi-agent systems that reach AGI level capabilities. And for that to be successful, what should we be thinking about today when it arrives?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:47",
                    "end": "00:44:16",
                    "text": " In that paper, we discussed four different categories of safety, including market design, including baseline safety for each agent itself, including inter-agent communication, et cetera, et cetera. Please feel free to refer to that paper. But I think what we need to do today is making sure that we build evaluation benchmarks that are relevant to multi-agent systems."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:16",
                    "end": "00:44:41",
                    "text": " Frontier Labs that are developing these models that are core to the agents, it's important for them to, and by them I mean Google Live Mind being one of them, to report, test against these benchmarks, report transparently about the limitations, report transparently about the capabilities, and I think that is one of the goals that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:41",
                    "end": "00:45:07",
                    "text": " Google DeepMind has been pursuing, and a lot of that information is available on the website. Please feel free to read our frameworks on AGI safety as well. So I'll stop there. So I'll just, again, follow up on that, right? One of the important factors that we're missing in the multi-agent system, we are assuming that it's a single model agent, right? Across the multi-agent system, I'm using one LLM model."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:07",
                    "end": "00:45:33",
                    "text": " But when we extend that thought, that what will happen when I'm using models provided by different players, open source or non-open source, you know, like all of those models will try to... So, single-matter responses are easy to verify. We can have the metrics or all the telemetry associated with it to verify whether the agent is claiming. But when it goes across those boundaries..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:33",
                    "end": "00:45:57",
                    "text": " And it becomes like a multimodal system, multi-agentic system. And every single company is saying, hey, trust us. So one of the thought experiments, or the way we try to solve it, is get every voice heard. Meaning that let's hear from all the academia how they think about this problem. Get all the LLM or a big."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:57",
                    "end": "00:46:23",
                    "text": " per companies getting involved in this discussion. The same thing with the system integrators and hyperscalers also participate in it. And not just them, but including the policy makers as well as the younger generation who is gonna be the agentic AI native. So how are we preparing for that? That's very important. So the problem doesn't..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:23",
                    "end": "00:46:56",
                    "text": " get solved with the single model responses. So the only solution that we kind of lean towards to make it enough decentralized or enough integrated system so that we can have those test parameters across every single thing. Yeah, thank you. I think for me, largely, the thing which I think can solve a lot of it because that kind of distills enough information to people to start working from is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:56",
                    "end": "00:47:21",
                    "text": " the idea around auditability and traceability to be able to capture and accurate reporting in an immutable way of a lot of what is happening. And as long as there are more open source platforms that allow for a lot of those, you know, anonymized way of auditability and traceability and the incidents that's coming to come up, I think it drives for more collective learning at a more global platform level. I don't know if there are..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:21",
                    "end": "00:47:46",
                    "text": " And this is a conversation where when we meet a lot of startups, they're actually learning a lot from talking to each other. And they keep doing about the fact that there are not a lot of global platforms that just talk about just learning around auditable trails of what is happening, how is it happening, where do systems fail at what scale and level of infrastructure. And I think something that solves for just a collective platform to learn from."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:46",
                    "end": "00:48:16",
                    "text": " will go a long way. It's a very innovative environment around the world and what's happening. So as long as you provide the right infrastructure for even knowing what's going wrong, you know, I think there's a lot that can actually accelerate the learning and development there. That's a cue for our session. Oh, I thought that was where you programmed the agenda. Yeah, so I think there's two."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:16",
                    "end": "00:48:46",
                    "text": " Two key points I'd like to just leave everyone with. The first, Amir runs a very innovative organization called AI Commons that many of us have been a part of over the years. And really the critical nature of what they have been looking at, which very much aligned with a lot of what we were looking at as well, was the idea of having these technologies"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:46",
                    "end": "00:49:16",
                    "text": " treated as a commons, a public utility. Because if you don't do that, you can't do exactly what everyone here is speaking to. Only those that have access to that can do that. And the point on education is very critical. But the point, I would just read into this then, is the thing to do today, we have to spend a lot more time getting rid of the jargon. Even the title of today's session."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:16",
                    "end": "00:49:48",
                    "text": " isn't something that's accessible readily to the same people we're asking to make use of these technologies, right? And so a lot of the effort needs to be spent on demystifying what we're talking about. Show the value of it. Because if we don't do that, then you're going to have a lot of fear. And a lot of the fear is unnecessary. And I think these would be the critical things. I would also love to use your app to turn off that alarm. So maybe you can teach us later about that. Thanks. Thank you very much."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:48",
                    "end": "00:49:51",
                    "text": " Thank you. Thank you everyone for coming."
                }
            ]
        }
    ],
    "projectDeleted": false,
    "codeFrameColumn": [
        {
            "col1": "col1",
            "col2": "col3",
            "researchObjective": "",
            "filterColumn": [],
            "netSetting": true,
            "subNetSetting": true,
            "categorySetting": true,
            "sentimentSetting": true,
            "thinkingBudgetSetting": 2000,
            "questionType": "OPENEND",
            "otherBrand": false,
            "existingCodeFrame": [],
            "isList": false,
            "codeFrameWithDataList": [],
            "codeFrameList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 10470,
                        "CandidatesTokenCount": 2443,
                        "TotalTokenCount": 14324,
                        "thoughtsTokenCount": 1411,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Net": "AI Agent Definition & Evolution",
                                "Subnets": [
                                    {
                                        "Subnet": "Understanding Agents",
                                        "Categories": [
                                            {
                                                "Category": "Basic Chatbot to Autonomous Vehicle Continuum",
                                                "Summary": "Agents exist on a spectrum from simple chatbots to fully autonomous vehicles."
                                            },
                                            {
                                                "Category": "Network Actors",
                                                "Summary": "Agents will act as visible or invisible actors within a network."
                                            }
                                        ],
                                        "Summary": "Exploring the fundamental nature and scope of AI agents."
                                    },
                                    {
                                        "Subnet": "Spectrum of Autonomy",
                                        "Categories": [
                                            {
                                                "Category": "Memory, Planning, Execution Dimensions",
                                                "Summary": "Autonomy is defined across dimensions like memory, short/long-term planning, and execution capability."
                                            },
                                            {
                                                "Category": "Agentic Features vs. Full Autonomy",
                                                "Summary": "Differentiating between systems with agentic features and those with full end-to-end autonomy."
                                            }
                                        ],
                                        "Summary": "Detailing the continuous range of capabilities and independence exhibited by AI agents."
                                    },
                                    {
                                        "Subnet": "Multi-Agent Systems",
                                        "Categories": [
                                            {
                                                "Category": "Agent-to-Agent Interaction",
                                                "Summary": "The collaboration and interaction between multiple autonomous agents."
                                            },
                                            {
                                                "Category": "Orchestration & Coordination",
                                                "Summary": "The process of managing and directing multiple agents to achieve specific outcomes."
                                            }
                                        ],
                                        "Summary": "Focusing on environments where multiple AI agents interact and collaborate."
                                    }
                                ],
                                "Summary": "This topic covers the definition of AI agents, their characteristics, the spectrum of autonomy they possess, and their evolution into multi-agent systems."
                            },
                            {
                                "Net": "Governance & Regulation of AI",
                                "Subnets": [
                                    {
                                        "Subnet": "Governance Frameworks",
                                        "Categories": [
                                            {
                                                "Category": "Runtime Governance",
                                                "Summary": "The necessity for continuous monitoring and control of agents during operation."
                                            },
                                            {
                                                "Category": "Iterative Approach",
                                                "Summary": "Governance as an ongoing process that adapts to changes and learning."
                                            },
                                            {
                                                "Category": "Contextualization",
                                                "Summary": "Considering the specific operational environment and triggers when defining governance bounds."
                                            }
                                        ],
                                        "Summary": "Establishing the structures and processes for overseeing AI agent behavior."
                                    },
                                    {
                                        "Subnet": "Certification & Standards",
                                        "Categories": [
                                            {
                                                "Category": "Agentic Identity (Passport)",
                                                "Summary": "Identifying agents' origins, training data, creators, and communication protocols."
                                            },
                                            {
                                                "Category": "IEEE Standards (Data Transparency, Accountability)",
                                                "Summary": "Standards focused on data transparency, age-appropriate design, accountability, and data privacy for AI systems."
                                            },
                                            {
                                                "Category": "Certified Programs",
                                                "Summary": "Programs aimed at validating and ensuring agents meet specific operational and safety criteria."
                                            }
                                        ],
                                        "Summary": "Developing methods and criteria for validating and regulating AI agents."
                                    },
                                    {
                                        "Subnet": "Regulatory Balance (Innovation vs. Control)",
                                        "Categories": [
                                            {
                                                "Category": "Data Plane vs. Control Plane",
                                                "Summary": "Maintaining open innovation (data plane) while building robust governance (control plane) in parallel."
                                            },
                                            {
                                                "Category": "Early-Stage Sandbox Testing",
                                                "Summary": "Allowing innovation in controlled environments before widespread deployment."
                                            },
                                            {
                                                "Category": "Framing vs. Regulation",
                                                "Summary": "The role of early policy in setting boundaries versus detailed, prescriptive regulation."
                                            }
                                        ],
                                        "Summary": "Addressing the challenge of balancing rapid technological advancement with necessary oversight."
                                    }
                                ],
                                "Summary": "This topic explores the various frameworks, standards, and regulatory considerations essential for governing AI agents, balancing control with innovation."
                            },
                            {
                                "Net": "Safety & Trust in AI Systems",
                                "Subnets": [
                                    {
                                        "Subnet": "Error & Risk Management",
                                        "Categories": [
                                            {
                                                "Category": "Identifying Agent Errors",
                                                "Summary": "Methods for detecting mistakes or unintended behaviors in AI agents."
                                            },
                                            {
                                                "Category": "Kill Switch/Cut-off Mechanism",
                                                "Summary": "The capability to halt an agent's operation if things go wrong."
                                            }
                                        ],
                                        "Summary": "Strategies and mechanisms to prevent, detect, and mitigate failures and risks associated with AI agents."
                                    },
                                    {
                                        "Subnet": "Building Trust",
                                        "Categories": [
                                            {
                                                "Category": "Foundational Layer Solutions",
                                                "Summary": "Addressing trust issues at the fundamental architectural level of AI agents."
                                            },
                                            {
                                                "Category": "Openness & Accountability",
                                                "Summary": "Ensuring transparency in agent operations and clear responsibility for outcomes."
                                            },
                                            {
                                                "Category": "Minimal Viable Trust Stack",
                                                "Summary": "The essential set of components needed for a startup to ensure basic trust and accountability in AI agents."
                                            }
                                        ],
                                        "Summary": "The process and infrastructure required to foster confidence in AI agent reliability and ethical operation."
                                    },
                                    {
                                        "Subnet": "Analogies from Other Industries",
                                        "Categories": [
                                            {
                                                "Category": "Aviation Safety (VLOS to BVLOS)",
                                                "Summary": "Drawing parallels from drone regulations moving from visual line of sight to beyond visual line of sight."
                                            },
                                            {
                                                "Category": "Human-in-the-Loop to Human-in-Command",
                                                "Summary": "Shifting from constant human approval to human supervision as agents become safer."
                                            }
                                        ],
                                        "Summary": "Learning from how other safety-critical sectors have managed and evolved their safety protocols."
                                    }
                                ],
                                "Summary": "This topic covers the critical aspects of ensuring AI agent safety, managing potential errors, and establishing trust through robust mechanisms and lessons from other industries."
                            },
                            {
                                "Net": "Development & Deployment Strategies",
                                "Subnets": [
                                    {
                                        "Subnet": "Technical Stacks for Governance",
                                        "Categories": [
                                            {
                                                "Category": "Build Time Governance (Data, Model Versioning)",
                                                "Summary": "Implementing governance during the development phase, including data and model management."
                                            },
                                            {
                                                "Category": "Deploy Time Governance (Policy, Permissions, Secrets)",
                                                "Summary": "Ensuring governance during deployment, covering policy tracking, access controls, and sensitive data management."
                                            },
                                            {
                                                "Category": "Run Time Governance (Observability, Kill Switch)",
                                                "Summary": "Governance during live operation, focusing on real-time monitoring and emergency controls."
                                            }
                                        ],
                                        "Summary": "The architectural layers and components required to embed governance directly into AI systems."
                                    },
                                    {
                                        "Subnet": "Open-Source Initiatives",
                                        "Categories": [
                                            {
                                                "Category": "Neuro AI Multi-Agent Accelerator",
                                                "Summary": "A low-code, no-code open-source framework for building and scaling multi-agent systems."
                                            },
                                            {
                                                "Category": "Community-Driven Development",
                                                "Summary": "The role of an active community in enhancing and standardizing open-source AI frameworks."
                                            }
                                        ],
                                        "Summary": "Leveraging open-source tools and communities to foster AI development and standardization."
                                    },
                                    {
                                        "Subnet": "Operational Best Practices",
                                        "Categories": [
                                            {
                                                "Category": "Real-time Observability",
                                                "Summary": "The ability to monitor agent communication, token usage, and environmental impact in real-time."
                                            },
                                            {
                                                "Category": "Auditability & Traceability",
                                                "Summary": "Maintaining clear records and paths to understand agent actions and outcomes."
                                            }
                                        ],
                                        "Summary": "Practical approaches and methods for effectively building, testing, and managing AI agent systems."
                                    }
                                ],
                                "Summary": "This topic focuses on the practical aspects of building and deploying AI agents, including the necessary technical architectures, open-source contributions, and operational best practices."
                            },
                            {
                                "Net": "Economic & Societal Impact",
                                "Subnets": [
                                    {
                                        "Subnet": "Investment & Market Acceptance",
                                        "Categories": [
                                            {
                                                "Category": "Startup Evaluation Criteria",
                                                "Summary": "How investors assess AI companies based on technical background and governance thinking."
                                            },
                                            {
                                                "Category": "Market Differentiation",
                                                "Summary": "How responsible AI practices become a unique selling point in the market."
                                            }
                                        ],
                                        "Summary": "The financial and market implications of AI agent development and adoption."
                                    },
                                    {
                                        "Subnet": "Governance as Competitive Advantage",
                                        "Categories": [
                                            {
                                                "Category": "Embedded Product Feature",
                                                "Summary": "Integrating governance as a core product feature rather than an afterthought."
                                            },
                                            {
                                                "Category": "Enterprise & Mid-Market Adoption",
                                                "Summary": "The necessity of robust governance for selling AI solutions to large organizations."
                                            },
                                            {
                                                "Category": "Weekly Operating Cadence",
                                                "Summary": "Governance, like security, becomes a routine, integral part of operations."
                                            }
                                        ],
                                        "Summary": "How strong governance practices can provide a strategic advantage in the AI market."
                                    },
                                    {
                                        "Subnet": "Public Utility & Education",
                                        "Categories": [
                                            {
                                                "Category": "AI as a Commons",
                                                "Summary": "Treating AI technologies as a shared public resource or utility."
                                            },
                                            {
                                                "Category": "Demystifying AI Jargon",
                                                "Summary": "Simplifying complex AI concepts to make them accessible and reduce fear."
                                            }
                                        ],
                                        "Summary": "The broader societal role of AI and the importance of public understanding and accessibility."
                                    }
                                ],
                                "Summary": "This topic delves into the economic motivations for adopting governance in AI, its impact on investment, market competitiveness, and the broader societal implications of AI agent development."
                            },
                            {
                                "Net": "Challenges & Future Outlook",
                                "Subnets": [
                                    {
                                        "Subnet": "Testing & Benchmarking",
                                        "Categories": [
                                            {
                                                "Category": "Evolving Testing Methods",
                                                "Summary": "The need for new evaluation techniques specifically for multi-agent systems."
                                            },
                                            {
                                                "Category": "Distributional AGI Safety",
                                                "Summary": "Considering AGI as emerging from interconnected specialized agents, requiring specific safety benchmarks."
                                            },
                                            {
                                                "Category": "Transparent Reporting",
                                                "Summary": "The importance of openly reporting on model limitations and capabilities by frontier labs."
                                            }
                                        ],
                                        "Summary": "Identifying current gaps and future needs in evaluating the performance and safety of AI agents."
                                    },
                                    {
                                        "Subnet": "Collective Learning & Auditability",
                                        "Categories": [
                                            {
                                                "Category": "Global Platform for Learning",
                                                "Summary": "A shared platform for insights on auditable trails, system failures, and infrastructure scaling."
                                            },
                                            {
                                                "Category": "Immutable Reporting",
                                                "Summary": "Capturing and accurately reporting what is happening in AI systems in an unchangeable way."
                                            }
                                        ],
                                        "Summary": "The importance of shared knowledge and transparent data to accelerate AI development and safety."
                                    },
                                    {
                                        "Subnet": "Demystifying AI",
                                        "Categories": [
                                            {
                                                "Category": "Reducing Fear",
                                                "Summary": "Addressing unnecessary public apprehension about AI through clear communication."
                                            },
                                            {
                                                "Category": "Accessible Language",
                                                "Summary": "Using simple, understandable language to explain AI concepts to a diverse audience."
                                            }
                                        ],
                                        "Summary": "Efforts to make AI concepts more understandable and relatable to a broader audience."
                                    }
                                ],
                                "Summary": "This topic examines the current obstacles in AI agent development and governance, and outlines key areas for future focus and collective progress."
                            },
                            {
                                "Net": "996 Non specific TOPIC",
                                "Subnets": [
                                    {
                                        "Subnet": "Irrelevant Responses",
                                        "Categories": [
                                            {
                                                "Category": "Responses not pertaining to question",
                                                "Summary": "User responses that are not relevant to the posed question."
                                            },
                                            {
                                                "Category": "Gibberish/Non-language characters",
                                                "Summary": "Responses containing meaningless text or non-alphanumeric characters."
                                            }
                                        ],
                                        "Summary": "Responses that are off-topic, nonsensical, or difficult to categorize within the main codeframe, potentially requiring re-evaluation or upcoding."
                                    }
                                ],
                                "Summary": "This topic captures responses that do not fit into the primary thematic structure and require special handling."
                            },
                            {
                                "Net": "997 Don’t Know Can’t Say",
                                "Subnets": [
                                    {
                                        "Subnet": "Non-committal/Vague Responses",
                                        "Categories": [
                                            {
                                                "Category": "Uncertainty expressed",
                                                "Summary": "Responses indicating the respondent is unsure or lacks sufficient information."
                                            },
                                            {
                                                "Category": "Vague or ambiguous statements",
                                                "Summary": "Responses that are not clear or specific enough to be categorized meaningfully."
                                            }
                                        ],
                                        "Summary": "Responses where respondents express a lack of knowledge, certainty, or provide non-committal answers."
                                    }
                                ],
                                "Summary": "This topic is for responses where respondents explicitly state they do not know or cannot provide a definitive answer."
                            },
                            {
                                "Net": "998 Not Applicable",
                                "Subnets": [
                                    {
                                        "Subnet": "Out of Scope Responses",
                                        "Categories": [
                                            {
                                                "Category": "Not relevant to project",
                                                "Summary": "Responses that, while potentially coherent, are not applicable to the specific project or survey context."
                                            },
                                            {
                                                "Category": "Question not understood",
                                                "Summary": "Responses indicating the respondent misunderstood the question and provided an irrelevant answer."
                                            }
                                        ],
                                        "Summary": "Responses that are irrelevant to the specific question asked or the overall project scope."
                                    }
                                ],
                                "Summary": "This topic covers responses that are deemed not applicable to the survey question or research objectives."
                            },
                            {
                                "Net": "999 None-Nothing",
                                "Subnets": [
                                    {
                                        "Subnet": "Empty or Null Responses",
                                        "Categories": [
                                            {
                                                "Category": "Blank/Empty submission",
                                                "Summary": "Responses where nothing was entered by the respondent."
                                            },
                                            {
                                                "Category": "Placeholder text",
                                                "Summary": "Responses containing only placeholder or default text."
                                            }
                                        ],
                                        "Summary": "Responses that are empty, null, or contain no meaningful content."
                                    }
                                ],
                                "Summary": "This topic is for responses that are completely blank, empty, or convey no information whatsoever."
                            }
                        ],
                        "Summary": "The discussion covers various perspectives from experts on AI and emerging technologies, focusing on the definition, capabilities, risks, and governance of AI agents and autonomous systems. It highlights the transition from traditional AI to multi-agent environments, emphasizing the need for robust governance frameworks, trust infrastructure, and safety measures from the foundational layer. Speakers discuss technical stacks for governance, certification processes, the balance between innovation and regulation, and the economic advantages of embedding governance into product development, using analogies from other safety-critical industries like aviation. The conversation also touches upon open-source initiatives and the importance of collective learning and demystifying AI concepts for broader adoption and responsible development."
                    }
                }
            ],
            "onlyCodeFrame": true,
            "errorList": [],
            "codeFrameGenerateType": "CODE_FRAME_ONLY",
            "requestType": "CODEFRAME",
            "requestMode": "DEVELOPMENT",
            "sourceLanguage": "English",
            "targetLanguage": "English",
            "response": {
                "status": "OK",
                "data": {
                    "status": "OK",
                    "PromptTokenCount": 10470,
                    "CandidatesTokenCount": 2443,
                    "TotalTokenCount": 14324,
                    "thoughtsTokenCount": 1411,
                    "finishReason": "STOP",
                    "data": [
                        {
                            "Net": "AI Agent Definition & Evolution",
                            "Subnets": [
                                {
                                    "Subnet": "Understanding Agents",
                                    "Categories": [
                                        {
                                            "Category": "Basic Chatbot to Autonomous Vehicle Continuum",
                                            "Summary": "Agents exist on a spectrum from simple chatbots to fully autonomous vehicles.",
                                            "code": 3
                                        },
                                        {
                                            "Category": "Network Actors",
                                            "Summary": "Agents will act as visible or invisible actors within a network.",
                                            "code": 4
                                        }
                                    ],
                                    "Summary": "Exploring the fundamental nature and scope of AI agents.",
                                    "code": 2
                                },
                                {
                                    "Subnet": "Spectrum of Autonomy",
                                    "Categories": [
                                        {
                                            "Category": "Memory, Planning, Execution Dimensions",
                                            "Summary": "Autonomy is defined across dimensions like memory, short/long-term planning, and execution capability.",
                                            "code": 6
                                        },
                                        {
                                            "Category": "Agentic Features vs. Full Autonomy",
                                            "Summary": "Differentiating between systems with agentic features and those with full end-to-end autonomy.",
                                            "code": 7
                                        }
                                    ],
                                    "Summary": "Detailing the continuous range of capabilities and independence exhibited by AI agents.",
                                    "code": 5
                                },
                                {
                                    "Subnet": "Multi-Agent Systems",
                                    "Categories": [
                                        {
                                            "Category": "Agent-to-Agent Interaction",
                                            "Summary": "The collaboration and interaction between multiple autonomous agents.",
                                            "code": 9
                                        },
                                        {
                                            "Category": "Orchestration & Coordination",
                                            "Summary": "The process of managing and directing multiple agents to achieve specific outcomes.",
                                            "code": 10
                                        }
                                    ],
                                    "Summary": "Focusing on environments where multiple AI agents interact and collaborate.",
                                    "code": 8
                                }
                            ],
                            "Summary": "This topic covers the definition of AI agents, their characteristics, the spectrum of autonomy they possess, and their evolution into multi-agent systems.",
                            "code": 1
                        },
                        {
                            "Net": "Governance & Regulation of AI",
                            "Subnets": [
                                {
                                    "Subnet": "Governance Frameworks",
                                    "Categories": [
                                        {
                                            "Category": "Runtime Governance",
                                            "Summary": "The necessity for continuous monitoring and control of agents during operation.",
                                            "code": 13
                                        },
                                        {
                                            "Category": "Iterative Approach",
                                            "Summary": "Governance as an ongoing process that adapts to changes and learning.",
                                            "code": 14
                                        },
                                        {
                                            "Category": "Contextualization",
                                            "Summary": "Considering the specific operational environment and triggers when defining governance bounds.",
                                            "code": 15
                                        }
                                    ],
                                    "Summary": "Establishing the structures and processes for overseeing AI agent behavior.",
                                    "code": 12
                                },
                                {
                                    "Subnet": "Certification & Standards",
                                    "Categories": [
                                        {
                                            "Category": "Agentic Identity (Passport)",
                                            "Summary": "Identifying agents' origins, training data, creators, and communication protocols.",
                                            "code": 17
                                        },
                                        {
                                            "Category": "IEEE Standards (Data Transparency, Accountability)",
                                            "Summary": "Standards focused on data transparency, age-appropriate design, accountability, and data privacy for AI systems.",
                                            "code": 18
                                        },
                                        {
                                            "Category": "Certified Programs",
                                            "Summary": "Programs aimed at validating and ensuring agents meet specific operational and safety criteria.",
                                            "code": 19
                                        }
                                    ],
                                    "Summary": "Developing methods and criteria for validating and regulating AI agents.",
                                    "code": 16
                                },
                                {
                                    "Subnet": "Regulatory Balance (Innovation vs. Control)",
                                    "Categories": [
                                        {
                                            "Category": "Data Plane vs. Control Plane",
                                            "Summary": "Maintaining open innovation (data plane) while building robust governance (control plane) in parallel.",
                                            "code": 21
                                        },
                                        {
                                            "Category": "Early-Stage Sandbox Testing",
                                            "Summary": "Allowing innovation in controlled environments before widespread deployment.",
                                            "code": 22
                                        },
                                        {
                                            "Category": "Framing vs. Regulation",
                                            "Summary": "The role of early policy in setting boundaries versus detailed, prescriptive regulation.",
                                            "code": 23
                                        }
                                    ],
                                    "Summary": "Addressing the challenge of balancing rapid technological advancement with necessary oversight.",
                                    "code": 20
                                }
                            ],
                            "Summary": "This topic explores the various frameworks, standards, and regulatory considerations essential for governing AI agents, balancing control with innovation.",
                            "code": 11
                        },
                        {
                            "Net": "Safety & Trust in AI Systems",
                            "Subnets": [
                                {
                                    "Subnet": "Error & Risk Management",
                                    "Categories": [
                                        {
                                            "Category": "Identifying Agent Errors",
                                            "Summary": "Methods for detecting mistakes or unintended behaviors in AI agents.",
                                            "code": 26
                                        },
                                        {
                                            "Category": "Kill Switch/Cut-off Mechanism",
                                            "Summary": "The capability to halt an agent's operation if things go wrong.",
                                            "code": 27
                                        }
                                    ],
                                    "Summary": "Strategies and mechanisms to prevent, detect, and mitigate failures and risks associated with AI agents.",
                                    "code": 25
                                },
                                {
                                    "Subnet": "Building Trust",
                                    "Categories": [
                                        {
                                            "Category": "Foundational Layer Solutions",
                                            "Summary": "Addressing trust issues at the fundamental architectural level of AI agents.",
                                            "code": 29
                                        },
                                        {
                                            "Category": "Openness & Accountability",
                                            "Summary": "Ensuring transparency in agent operations and clear responsibility for outcomes.",
                                            "code": 30
                                        },
                                        {
                                            "Category": "Minimal Viable Trust Stack",
                                            "Summary": "The essential set of components needed for a startup to ensure basic trust and accountability in AI agents.",
                                            "code": 31
                                        }
                                    ],
                                    "Summary": "The process and infrastructure required to foster confidence in AI agent reliability and ethical operation.",
                                    "code": 28
                                },
                                {
                                    "Subnet": "Analogies from Other Industries",
                                    "Categories": [
                                        {
                                            "Category": "Aviation Safety (VLOS to BVLOS)",
                                            "Summary": "Drawing parallels from drone regulations moving from visual line of sight to beyond visual line of sight.",
                                            "code": 33
                                        },
                                        {
                                            "Category": "Human-in-the-Loop to Human-in-Command",
                                            "Summary": "Shifting from constant human approval to human supervision as agents become safer.",
                                            "code": 34
                                        }
                                    ],
                                    "Summary": "Learning from how other safety-critical sectors have managed and evolved their safety protocols.",
                                    "code": 32
                                }
                            ],
                            "Summary": "This topic covers the critical aspects of ensuring AI agent safety, managing potential errors, and establishing trust through robust mechanisms and lessons from other industries.",
                            "code": 24
                        },
                        {
                            "Net": "Development & Deployment Strategies",
                            "Subnets": [
                                {
                                    "Subnet": "Technical Stacks for Governance",
                                    "Categories": [
                                        {
                                            "Category": "Build Time Governance (Data, Model Versioning)",
                                            "Summary": "Implementing governance during the development phase, including data and model management.",
                                            "code": 37
                                        },
                                        {
                                            "Category": "Deploy Time Governance (Policy, Permissions, Secrets)",
                                            "Summary": "Ensuring governance during deployment, covering policy tracking, access controls, and sensitive data management.",
                                            "code": 38
                                        },
                                        {
                                            "Category": "Run Time Governance (Observability, Kill Switch)",
                                            "Summary": "Governance during live operation, focusing on real-time monitoring and emergency controls.",
                                            "code": 39
                                        }
                                    ],
                                    "Summary": "The architectural layers and components required to embed governance directly into AI systems.",
                                    "code": 36
                                },
                                {
                                    "Subnet": "Open-Source Initiatives",
                                    "Categories": [
                                        {
                                            "Category": "Neuro AI Multi-Agent Accelerator",
                                            "Summary": "A low-code, no-code open-source framework for building and scaling multi-agent systems.",
                                            "code": 41
                                        },
                                        {
                                            "Category": "Community-Driven Development",
                                            "Summary": "The role of an active community in enhancing and standardizing open-source AI frameworks.",
                                            "code": 42
                                        }
                                    ],
                                    "Summary": "Leveraging open-source tools and communities to foster AI development and standardization.",
                                    "code": 40
                                },
                                {
                                    "Subnet": "Operational Best Practices",
                                    "Categories": [
                                        {
                                            "Category": "Real-time Observability",
                                            "Summary": "The ability to monitor agent communication, token usage, and environmental impact in real-time.",
                                            "code": 44
                                        },
                                        {
                                            "Category": "Auditability & Traceability",
                                            "Summary": "Maintaining clear records and paths to understand agent actions and outcomes.",
                                            "code": 45
                                        }
                                    ],
                                    "Summary": "Practical approaches and methods for effectively building, testing, and managing AI agent systems.",
                                    "code": 43
                                }
                            ],
                            "Summary": "This topic focuses on the practical aspects of building and deploying AI agents, including the necessary technical architectures, open-source contributions, and operational best practices.",
                            "code": 35
                        },
                        {
                            "Net": "Economic & Societal Impact",
                            "Subnets": [
                                {
                                    "Subnet": "Investment & Market Acceptance",
                                    "Categories": [
                                        {
                                            "Category": "Startup Evaluation Criteria",
                                            "Summary": "How investors assess AI companies based on technical background and governance thinking.",
                                            "code": 48
                                        },
                                        {
                                            "Category": "Market Differentiation",
                                            "Summary": "How responsible AI practices become a unique selling point in the market.",
                                            "code": 49
                                        }
                                    ],
                                    "Summary": "The financial and market implications of AI agent development and adoption.",
                                    "code": 47
                                },
                                {
                                    "Subnet": "Governance as Competitive Advantage",
                                    "Categories": [
                                        {
                                            "Category": "Embedded Product Feature",
                                            "Summary": "Integrating governance as a core product feature rather than an afterthought.",
                                            "code": 51
                                        },
                                        {
                                            "Category": "Enterprise & Mid-Market Adoption",
                                            "Summary": "The necessity of robust governance for selling AI solutions to large organizations.",
                                            "code": 52
                                        },
                                        {
                                            "Category": "Weekly Operating Cadence",
                                            "Summary": "Governance, like security, becomes a routine, integral part of operations.",
                                            "code": 53
                                        }
                                    ],
                                    "Summary": "How strong governance practices can provide a strategic advantage in the AI market.",
                                    "code": 50
                                },
                                {
                                    "Subnet": "Public Utility & Education",
                                    "Categories": [
                                        {
                                            "Category": "AI as a Commons",
                                            "Summary": "Treating AI technologies as a shared public resource or utility.",
                                            "code": 55
                                        },
                                        {
                                            "Category": "Demystifying AI Jargon",
                                            "Summary": "Simplifying complex AI concepts to make them accessible and reduce fear.",
                                            "code": 56
                                        }
                                    ],
                                    "Summary": "The broader societal role of AI and the importance of public understanding and accessibility.",
                                    "code": 54
                                }
                            ],
                            "Summary": "This topic delves into the economic motivations for adopting governance in AI, its impact on investment, market competitiveness, and the broader societal implications of AI agent development.",
                            "code": 46
                        },
                        {
                            "Net": "Challenges & Future Outlook",
                            "Subnets": [
                                {
                                    "Subnet": "Testing & Benchmarking",
                                    "Categories": [
                                        {
                                            "Category": "Evolving Testing Methods",
                                            "Summary": "The need for new evaluation techniques specifically for multi-agent systems.",
                                            "code": 59
                                        },
                                        {
                                            "Category": "Distributional AGI Safety",
                                            "Summary": "Considering AGI as emerging from interconnected specialized agents, requiring specific safety benchmarks.",
                                            "code": 60
                                        },
                                        {
                                            "Category": "Transparent Reporting",
                                            "Summary": "The importance of openly reporting on model limitations and capabilities by frontier labs.",
                                            "code": 61
                                        }
                                    ],
                                    "Summary": "Identifying current gaps and future needs in evaluating the performance and safety of AI agents.",
                                    "code": 58
                                },
                                {
                                    "Subnet": "Collective Learning & Auditability",
                                    "Categories": [
                                        {
                                            "Category": "Global Platform for Learning",
                                            "Summary": "A shared platform for insights on auditable trails, system failures, and infrastructure scaling.",
                                            "code": 63
                                        },
                                        {
                                            "Category": "Immutable Reporting",
                                            "Summary": "Capturing and accurately reporting what is happening in AI systems in an unchangeable way.",
                                            "code": 64
                                        }
                                    ],
                                    "Summary": "The importance of shared knowledge and transparent data to accelerate AI development and safety.",
                                    "code": 62
                                },
                                {
                                    "Subnet": "Demystifying AI",
                                    "Categories": [
                                        {
                                            "Category": "Reducing Fear",
                                            "Summary": "Addressing unnecessary public apprehension about AI through clear communication.",
                                            "code": 66
                                        },
                                        {
                                            "Category": "Accessible Language",
                                            "Summary": "Using simple, understandable language to explain AI concepts to a diverse audience.",
                                            "code": 67
                                        }
                                    ],
                                    "Summary": "Efforts to make AI concepts more understandable and relatable to a broader audience.",
                                    "code": 65
                                }
                            ],
                            "Summary": "This topic examines the current obstacles in AI agent development and governance, and outlines key areas for future focus and collective progress.",
                            "code": 57
                        }
                    ],
                    "Summary": "The discussion covers various perspectives from experts on AI and emerging technologies, focusing on the definition, capabilities, risks, and governance of AI agents and autonomous systems. It highlights the transition from traditional AI to multi-agent environments, emphasizing the need for robust governance frameworks, trust infrastructure, and safety measures from the foundational layer. Speakers discuss technical stacks for governance, certification processes, the balance between innovation and regulation, and the economic advantages of embedding governance into product development, using analogies from other safety-critical industries like aviation. The conversation also touches upon open-source initiatives and the importance of collective learning and demystifying AI concepts for broader adoption and responsible development."
                }
            },
            "summaryList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 10131,
                        "CandidatesTokenCount": 2235,
                        "TotalTokenCount": 13769,
                        "thoughtsTokenCount": 1403,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Summary": "<h2>Introduction to the Agentic AI Landscape</h2><h3>Panelist Introductions and Foundational Concepts</h3><p>The session commenced with introductions from key experts in the field of AI and emerging technologies. Eddie Sahai, a Policy Manager at Google, specializes in AI, particularly agent AI and robotics. Mahesh, representing Project Nanda, highlighted their pioneering efforts in building foundational infrastructure for the nascent 'internet of AI agents,' emphasizing trust as a critical, first-principle issue. Alpesh Shah, Managing Director of the IEEE Standards Association, brought a perspective on standardization and governance. Apoorv Goyal, a Principal at Insight Partners, offered insights from the investment side, focusing on technology companies. Praveen, Head of Cognizant AI Lab, served as a co-moderator, guiding the discussion on the practicalities and future of multi-agent systems.</p><h3>The Spectrum of Autonomy and Emerging Concerns</h3><p>The discussion quickly moved to the evolving nature of AI, particularly the concept of 'agentic AI.' It was established that agency exists on a spectrum, not as a binary state. Eli from Google illustrated this, explaining that systems range from basic chatbots that perform research (like Google Deep Research) to fully autonomous vehicles that take real-world actions. This spectrum is defined by dimensions such as memory, planning capabilities (long-term, short-term, planning horizon), and the ability to execute actions. A core challenge highlighted was the public's fear surrounding autonomy, stemming from a lack of understanding regarding agent construction, reaction patterns, objective changes, and inter-agent interactions. This fear makes 'governance' – encompassing control over system outcomes, procedures, ownership, and risk management – a central concern, a shift from even a year prior.</p><h2>Governance and Trust in the Age of Autonomous Agents</h2><h3>The Imperative for Runtime Governance</h3><p>Mahesh underscored that traditional governance models, which evolved from machine-to-machine, SaaS, cloud, and microservices interactions, are inadequate for the scale and complexity of autonomous agents. Given that agents function as 'network actors' operating visibly and invisibly at every stage, governance must be 'runtime' rather than an afterthought. Foundational issues of safety, trust, deployability, openness, and accountability must be designed into the systems from the outset, not retrofitted.</p><h3>Establishing Agent Identity and Certification</h3><p>The conversation introduced the concept of 'agentic identity' – defining an agent's origin, training data, creator, ownership, and communication protocols. This leads to the critical idea of 'certifying agents,' a challenge that Alpesh of IEEE addressed. He stressed that a strong governance structure and clear processes are paramount for deploying agents safely. Certification requires an iterative monitoring approach due to potential shifts in behavior and the need to contextualize unexpected multi-agent interactions. The IEEE's certification program focuses on evaluating the 'means' by which an agent achieves its output, rather than demanding 100% perfection from the start. Key standards include data transparency, age-appropriate design, accountability, and data privacy. Alpesh emphasized the value of diverse community involvement – lawyers, doctors, artists, engineers – in understanding and addressing complex problems.</p><h3>The 5-Key Governance Stack for Autonomous AI</h3><p>Apoorv Goyal outlined a comprehensive '5-key governance stack' model crucial for investors evaluating AI companies, envisioning governance as a competitive advantage rather than just a compliance cost:</p><ol><li><strong>Build Time:</strong> Focus on data governance and model versioning.</li><li><strong>Deploy Time:</strong> Addressing policy, tracking, permissioning, and secrets management.</li><li><strong>Runtime:</strong> Ensuring real-time observability and the presence of a 'kill switch' for immediate control.</li><li><strong>Remediation:</strong> Establishing audit trails for post-mortems and robust incident response architectures.</li><li><strong>Accountability:</strong> Defining reporting structures, conducting post-mortem reviews, and mapping compliance.</li></ol><p>For startups, Apoorv proposed a 'minimal viable trust stack' comprising a clearly defined agent identity registry, guardrails at the orchestration layer, real-time observability, and clear oversight. These four basic elements are essential for launching any agentic ecosystem into production safely and are not prohibitively expensive.</p><h2>Multi-Agent System Implementation and Real-World Parallels</h2><h3>Cognizant's Neuro AI Multi-Agent Accelerator</h3><p>Praveen introduced Cognizant's 'Neuro AI Multi-Agent Accelerator,' an open-source (Apache 2.0 license) low-code, no-code framework launched in May 2025. Designed for accelerating prototype development to scaled production, it offers LLM and cloud agnosticism, interoperability (supporting OSR and A2N MCP server protocols), and built-in security. Praveen demonstrated its ease of use, showing how it could rapidly create a multi-agent network from a simple prompt. The platform offers real-time monitoring of inter-agent communication, granular logs (token usage, cost), and ESG metrics (energy usage, carbon footprint) – all crucial for auditability and traceability. It also allows for the integration of specialized LLMs for specific agents, like those focused on auditing or compliance, further enhancing accuracy and control.</p><h3>Lessons from Other Safety-Critical Industries</h3><p>Eli drew parallels between agentic AI and other safety-critical industries like aviation. She explained the evolution of drone regulations from 'visual line of sight' (VLOS) where a human pilot maintains constant visual contact, to 'beyond visual line of sight' (BVLOS). This shift was driven by the development of AI-powered detect-and-avoid systems that proved more capable than human observation. This analogy suggests a similar trajectory for agentic AI: moving from a 'human in the loop' model (where humans approve every action) to a 'human in command' model (where humans supervise, but AI systems handle routine tasks with inherent safety). This transition acknowledges that, as agents become safer and more reliable, constant human intervention can actually undermine their utility and potentially introduce human error.</p><h2>Balancing Innovation, Regulation, and Collective Progress</h2><h3>Striking the Right Balance: Engineering vs. Regulation</h3><p>The panel discussed the delicate balance between fostering innovation and implementing regulations. Mahesh advocated for keeping innovation open (the 'data plane') while building governance in parallel (the 'control plane'), drawing an analogy to the early, decentralized evolution of the internet. Alpesh cautioned against premature, overly restrictive regulation that could stifle innovation. He emphasized that early stages require freedom for experimentation within a 'sandbox' to understand possibilities and bounds. Policy, however, can provide early 'framing' to define boundaries without imposing full regulation. As technology matures and market acceptance grows, standards and certification become more sensible. He also suggested integrating required standards and policies into the 'fabric from the get-go' to ease the burden on startups and provide clear metrics for regulators.</p><h3>Governance as a Competitive Advantage</h3><p>Apoorv reiterated that for companies selling into enterprises, robust governance is not a trade-off with commercial innovation but a direct enabler of scale. It's becoming a 'competitive advantage' and a 'new product in GTM.' Enterprises are willing to spend significant amounts on governance assurance due to the high costs of failure. For AI-native teams, governance, like security, is now an embedded part of their weekly operations, involving solid eval reviews, red teaming, and post-mortem analyses. This is particularly true for mid-market and enterprise solutions, where governance conversations happen from the seed/Series A stage.</p><h3>The Path Forward: Collective Action and Transparency</h3><p>The session concluded with a discussion on the single most impactful thing needed for collective progress. Eli proposed evolving testing methods and evaluation benchmarks for multi-agent systems, especially given the potential for Distributed AGI. She emphasized the importance of Frontier Labs (like Google DeepMind) transparently reporting on model limitations and capabilities against these benchmarks. Mahesh added that this problem is complex because multi-agent systems will likely involve models from diverse providers. He advocated for an inclusive, decentralized approach, involving academia, LLM providers, system integrators, hyperscalers, policymakers, and the younger generation. Apoorv emphasized 'auditability and traceability' – capturing and immutably reporting incidents to foster collective learning at a global platform level. Alpesh highlighted the importance of treating these technologies as a 'public utility' or 'commons' and, crucially, demystifying the jargon surrounding agentic AI to reduce unnecessary fear and demonstrate its value.</p><h2>Conclusions</h2><p>The advent of agentic AI marks a significant shift, demanding a re-evaluation of traditional AI governance. Autonomy, while powerful, introduces complexities and fears that necessitate a proactive, rather than reactive, approach to safety and control. The consensus is that governance must be an integral, foundational layer, evolving from runtime monitoring to comprehensive identity and certification frameworks. These frameworks, as demonstrated by the IEEE and detailed in the 5-key governance stack, transform governance from a mere compliance burden into a vital competitive advantage for market entry and scaling.</p><h2>Actionable Recommendations</h2><ol><li><strong>Integrate Runtime Governance:</strong> Organizations must implement runtime governance models, moving beyond traditional post-facto solutions to continuously monitor and control agent behaviors in real-time.</li><li><strong>Develop Agent Identity and Certification Standards:</strong> Industry bodies and research institutions should collaborate to establish clear standards for agent identity, provenance, training, and certification to build trust and accountability.</li><li><strong>Adopt a Comprehensive Governance Stack:</strong> Companies, particularly startups, should implement a 'minimal viable trust stack' covering build-time data governance, deploy-time policy enforcement, real-time observability with kill switches, robust remediation processes, and clear accountability structures.</li><li><strong>Invest in Multi-Agent System Benchmarking:</strong> Research and development efforts should prioritize the creation and adoption of advanced testing methods and evaluation benchmarks specifically designed for multi-agent systems to ensure their safe and reliable development.</li><li><strong>Foster Open Collaboration and Transparency:</strong> Encourage open-source initiatives and platforms for collective learning on audit trails, incident reporting, and best practices. Promote transparent reporting from Frontier Labs on model capabilities and limitations.</li><li><strong>Demystify Agentic AI:</strong> Stakeholders across academia, industry, and policy must work collectively to simplify the language and concepts surrounding agentic AI, demonstrating its value and addressing public fears to facilitate broader adoption and understanding.</li><li><strong>Learn from Safety-Critical Industries:</strong> Apply lessons from sectors like aviation regarding the evolution of safety protocols, moving from constant human intervention ('human in the loop') to supervisory roles ('human in command') as AI systems demonstrate enhanced safety and reliability.</li></ol>"
                            }
                        ]
                    }
                }
            ]
        }
    ],
    "headerCols": {
        "col1": "Part Id",
        "col2": "FileName",
        "col3": "Content",
        "col4": "Status"
    },
    "fileName": "",
    "projectType": "VIDEO",
    "sourceLanguage": "English",
    "targetLanguage": "English",
    "__v": 0
},
{
    "_id": "69943bf225c9f4e0d7e70552",
    "projectId": "866079",
    "projectName": "AI Diplomacy Asia Leads -Governing Artificial Intelligence in Fragmented ",
    "projectDescription": "AI Diplomacy Asia Leads -Governing Artificial Intelligence in Fragmented ",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T09:59:13.737Z",
    "projectUpdatedDate": "2026-02-17T09:59:13.737Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-17T10:52:16.835Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "AI Diplomacy Asia Leads -Governing Artificial Intelligence in Fragmented World.mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771322340328.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:28",
                    "text": " Safety Asia, if I might take a bit of time, works across the region and globally to bring together policymakers, diplomats, researchers, and industry to think through the governance and crisis implications of advanced AI systems, particularly from Asian and Global South perspectives. Today's session is part of a broader set of conversations we've been convening on what we're calling AI crisis diplomacy, how governments coordinate across borders."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:28",
                    "end": "00:00:57",
                    "text": " when AI-related incidents unfold at speed and with uncertain attribution. So this discussion is intended to help continue that conversation and move us toward more concrete frameworks for AI crisis diplomacy and engagement across borders. So let me begin with a simple observation. In diplomacy, we are used to thinking in months and years. We negotiate, we deliberate, we build consensus."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:57",
                    "end": "00:01:26",
                    "text": " AI enabled crises do not move at diplomatic speed. Financial systems react in seconds, synthetic media spreads in minutes, autonomous systems can act before governments even know something has gone wrong. So the central question we are exploring today is how do governments coordinate diplomatically in response to AI related crises that unfold across borders at machine speed."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:26",
                    "end": "00:01:55",
                    "text": " We circulated a short fire starter paper ahead of this session with our panelists describing an AI crisis as one that crosses borders, unfolds faster than any single authority can manage, involves attribution uncertainty. These are not hypothetical challenges and yet while we have built deeply interconnected technological systems, we have not yet built the corresponding cross-border"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:55",
                    "end": "00:02:19",
                    "text": " crisis coordination infrastructure. This is what we want to explore together today, this morning. So, if I might, before we move into my just presenting some scenarios, let me briefly introduce our speakers. We're joined this morning by Professor Stuart Russell, Director of the Center for Human Compatible AI at the University of California, Berkeley, and co-host of this session. Thank you, Professor Russell."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:19",
                    "end": "00:02:47",
                    "text": " Wan Sze Lee, Director of the Infocomm Media Development Authority, IMGA of Singapore. Harayama Yuko, former Executive Director of the Global Partnership on AI and a leading voice in international AI governments from Japan. And Aji John Azimi, apologies if I mispronounce, but founding chair of the Artificial Intelligence Council of the Republic of Tajikistan."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:47",
                    "end": "00:03:17",
                    "text": " We will also be hearing a short video intervention from Audrey Tang, the Cyber Ambassador of Taiwan and former digital minister. And we are delighted at some point, she might be in the room already, but she said she might be delayed, Ambassador Anne-Marie Engtoft-Meldgaard, Denmark's tech ambassador, who may be joining us shortly given her scheduling. So please join me in welcoming our speakers."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:17",
                    "end": "00:03:45",
                    "text": " Let me just quickly frame some scenarios. These are illustrative. They are simply to help us think concretely about coordination challenges with regard to AI. So scenario one, a financial cascade. An AI-driven model misinterprets data and generates a false policy signal affecting another country's currency."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:45",
                    "end": "00:04:13",
                    "text": " Automated trading systems react instantly across jurisdictions. Markets halt, financial links strain, officials call each other. One side asks, why did you take this action? The other says, we didn't. Both may be telling the truth. The core problem is not only malfunction, it is the absence of a rapid cross-border verification channel to establish what happened before trust collapses."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:13",
                    "end": "00:04:43",
                    "text": " Here's another scenario, scenario two, deep fake driven instability. A highly realistic synthetic video appears shortly before an election or peace negotiation. It spreads rapidly across platforms hosted in multiple jurisdictions. Verification takes longer than political timeline itself. By the time the video is debunked, the damage is already done. The issue here is not detection."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:43",
                    "end": "00:05:13",
                    "text": " It is the absence of a shared cross-border response mechanism at speed. Third scenario, autonomous infrastructure incident. An AI system managing interconnected infrastructure makes a decision that affects another country. The system may be hosted in one jurisdiction, operated in another, and affecting a third. Government suddenly asks, who has authority? Who is responsible? And who do you call? Who can act?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:13",
                    "end": "00:05:43",
                    "text": " Our current diplomatic and legal frameworks assume human decision makers. They are not designed for agentic systems acting across borders. Across all three scenarios, all three scenarios that I've outlined, the challenge is not only technological, it is coordination under uncertainty. So how do we establish shared reality quickly enough to prevent escalation? How do we clarify intent? How do we coordinate response?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:43",
                    "end": "00:06:15",
                    "text": " Now, to frame the broader implications, let me turn now to Professor Stuart Russell. Stuart, may I invite you to offer some reflections on why increasingly autonomous and agentic systems create new challenges for governance and diplomacy? Please. Thank you very much for hosting this session and for the opportunity to speak. So, AI has been around for about 80 years."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:15",
                    "end": "00:06:44",
                    "text": " I wrote my first AI program almost 50 years ago. And we've gone through a number of boom and bust cycles. Obviously we're in a boom, as one can see, right now. And that is the result of mainly large language model technology. And that technology is increasingly being augmented with so-called agentic capabilities, which is sort of ironic because my..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:44",
                    "end": "00:07:09",
                    "text": " My textbook, which came out in 1994, has on the front cover the intelligent agent book. And so the idea of an agent is central to how we think about AI systems. They're systems that perceive and act, and generally they will act in the pursuit of their objectives. So you asked about the challenges to governance."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:09",
                    "end": "00:07:37",
                    "text": " from this fast moving and technology and the fast unfolding scenarios. And in fact, this is one myth, I think, that is often trotted out by people who are opposed to regulation. They say, oh, you can't regulate something that is changing because by the time you've regulated, your regulation will be out of date. This is utter and complete nonsense."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:37",
                    "end": "00:08:09",
                    "text": " Think about how we regulate medicines. Medicines have to be shown to be safe and shown to be effective before they can be sold to the public. That regulation is going to remain in place for the next 500,000 years until we all become immortal and we don't have to take medicine at all. It has nothing to do with the technology, it has to do with the risk. We set the level..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:09",
                    "end": "00:08:40",
                    "text": " of risk that we are willing to accept from the technology, and we leave the technologist to figure out how to make the technology that satisfies that level of risk and delivers as much benefit as possible. And so, you know, another permanent and future-proof form of regulation is liability. If you harm someone, you are legally responsible and you have to pay."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:40",
                    "end": "00:09:11",
                    "text": " that's gonna be true for the next 500 billion years. And again, it has nothing to do with the technology. But unfortunately, pretty much all tech companies disclaim liability. So if you ever have a few hours to spare and you really want to read your Microsoft end user license agreement, you will find that Microsoft disclaims all liability and in no case can you ever claim more than $5 in compensation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:11",
                    "end": "00:09:37",
                    "text": " That's what it says. So that's a regulatory tool that's been around for thousands of years that the tech industry has managed to avoid. And I think it would be very sensible if we were to restore that tool, which works very well for fast-moving technologies, for things that unfold very quickly and so on."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:37",
                    "end": "00:10:07",
                    "text": " The only problem with liability is that it's not much use for extinction risk. In fact, that's a mathematical theorem. It doesn't deter people from imposing extinction risks on the rest of us at all, because of course, if the risk comes to pass, they won't pay. Myth number two. AI is a general purpose technology, and you can't regulate general purpose technologies. This is a strange axiom."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:07",
                    "end": "00:10:34",
                    "text": " And you might say, okay, why? And then they say, oh, but look at electricity. We don't regulate electricity. It's a general purpose technology. But that's nonsense, right? Electricity as a general technology is highly regulated. The voltage and currents have to be regulated. Otherwise you would set fire to every city on earth. The wiring is regulated."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:34",
                    "end": "00:11:01",
                    "text": " all employees who work with electricity are regulated and trained. The plugs, the appliances are regulated. So these myths are trotted out by people who are professionals, who have successfully in the past opposed regulation of cigarettes, of carbon dioxide, you name it. It's the same playbook that's being used here to prevent the regulation of technology."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:02",
                    "end": "00:11:32",
                    "text": " So as I mentioned, effective regulation is about setting the level of risk that we're willing to accept and then requiring demonstrable proof or a high degree of evidence that the technology meets that level of risk. And this is how, here's some water. Why can I drink this? Why can I drink it? Because it's regulated."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:32",
                    "end": "00:12:01",
                    "text": " I know that I can drink it because the companies are subject to regulation on the quality and safety of this product. Same with food, same with this building. How can I sit in this building without being worried that it's gonna collapse on my head? Because it's regulated, because it has to meet building codes before anyone can come into it. Most of you flew here on an airplane or took a train. Those are highly regulated. If you go upstairs in the elevator, that's highly regulated."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:01",
                    "end": "00:12:30",
                    "text": " Nuclear power is often cited as an example of how the risk is quantified very precisely. We accept a risk of nuclear meltdown of between one in a million per year and one in 10 million per year. And that risk has been set by regulators because that's what we consider to be a reasonable trade-off to get the benefits of nuclear power."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:30",
                    "end": "00:12:59",
                    "text": " It's very difficult if you set the risk to zero to prove that your nuclear power station will never have an accident. But we have been able to prove, and those proofs take the form of millions of pages of probability calculations that the level of risk for nuclear power meets that standard. So what can we do with AI? You know, I think with planes, for example,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:59",
                    "end": "00:13:29",
                    "text": " the notion of safety is pretty clear. It's not supposed to touch the ground unless you mean to touch the ground, right? That's pretty much it. With AI, because it's so general, it's difficult to say what is the line between safe and unsafe behaviors by AI. So this approach of what we call behavioral red lines is gaining some currency. And behavioral red lines are specific"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:29",
                    "end": "00:13:56",
                    "text": " obviously unacceptable behaviors by AI systems. And the idea would be that companies need to demonstrate that their systems will not cross those red lines with some reliability that's appropriate for the level of risk that we're talking about. And so if a red line says AI systems should not explain to terrorists how to build a new biological weapon,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:56",
                    "end": "00:14:27",
                    "text": " That's a fairly high risk thing, so you'd want a fairly low probability of that happening. Another red line might be an AI system should not pretend to be a human being, whether a general human being or a specific human being. That's maybe a bit less of a risk, so maybe you might have a slightly higher threshold for that. Now when it comes to extinction, which many, or in fact all of the leading CEOs of AI companies have,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:27",
                    "end": "00:14:55",
                    "text": " have said is a pretty likely outcome of their effort to create AGI, what's an acceptable level of risk? Should it be zero? Well, probably not because there is already a background level of extinction risk from other things like asteroid collisions and so on. And so we might say, well, obviously it should be safer than nuclear power station, right?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:55",
                    "end": "00:15:25",
                    "text": " Extinction is much worse than a nuclear meltdown. So let's say for the sake of argument, one in a hundred million per year. And I think that's a bit generous, but we'll go with that. So what do the companies say the risk is? The CEOs have given numbers for the risk of extinction from the technology that they are trying to build between 10 and 50%. So put another way..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:25",
                    "end": "00:15:54",
                    "text": " we need the technology to be 10 million times safer than it is. Also, those numbers that they're giving out are, I would say, seat of the pants. They're not even back at the envelope, right? They're not based on any actual calculation. It's just a gut feel that they have about how easy it will be to lose control as we make AI systems more and more capable. So we need regulations."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:54",
                    "end": "00:16:24",
                    "text": " that will require a demonstration that the probability of loss of control leading to human extinction is less than one in a hundred million. And that applies even to systems now, right? It should be easy to come up with that evidence now. As the systems get more capable, it's only gonna get harder. So the company's view is, well, we don't know how to show any level of risk."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:26",
                    "end": "00:16:55",
                    "text": " But if they can't do it now, how are they going to do it next year and the year after when the systems are in fact much more dangerous and much more capable? So this argument that technology companies come up with often is We don't know how to comply with any required level of risk and therefore you can't require it Let me say that again you the human race are not allowed to protect yourselves"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:55",
                    "end": "00:17:29",
                    "text": " from the technology that we are building that may well make you go extinct. So that's the argument that they're making. I think I'll stop there. Thank you. Great. Thank you very much, Professor Stewart, Professor Russell. Now if I might turn to the panelists. One scene, if I can start with you. There's much that you can react to. The scenarios that I posed, the comments by Professor Russell,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:29",
                    "end": "00:17:58",
                    "text": " I'm thinking Singapore sits at the intersection of finance, infrastructure, and digital connectivity in Southeast Asia. So if one of the scenarios I mentioned unfolded in the region today, how would coordination actually begin in practice? Which channels would move first? And where do you see the most dangerous friction, verification, authority, speed? Up to you on what you want to comment on."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:58",
                    "end": "00:18:23",
                    "text": " Thanks, Alejandro. First of all, thanks for having me here on this panel. I think when we were looking at the notes that you sent us, when I was looking at the notes that you sent us, I don't think I can speak on behalf of..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:23",
                    "end": "00:18:50",
                    "text": " I thought it was quite interesting because I don't think we have had this conversation before in terms of thinking about how do we coordinate amongst governments when a crisis happens in AI. I have to say this is a somewhat theoretical and hypothetical discussion right now because it's not something that we have set up specifically for AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:50",
                    "end": "00:19:18",
                    "text": " But we have done it for other types of crisis and I define it very specifically on crisis because I think Stuart talked about governance and regulations of AI and I think here we are perhaps gonna, I'm gonna maybe focus on your earlier kind of provocations, which is how do we cooperate when it comes to something that has happened in the crisis, meaning some bad thing has happened and what we can do about it, right?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:18",
                    "end": "00:19:45",
                    "text": " But we have had crisis in the past separate from AI whether it's the pandemic and cyber security and so on so there are ways in which Governments are working together in concert to deal with crisis. The question is what has changed in the age of AI, right? And I think We should be able to what has changed in terms of perhaps in terms of all the the scenarios that you've highlighted doesn't sound like it's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:45",
                    "end": "00:20:14",
                    "text": " Well, I mean, it has changed in terms of the speed, the skill, and our ability to assess the situation. But it is still applied to these scenarios in which we find ourselves in the past, cyber security, healthcare, and so on, pandemics. And I think in that case, we should continue to rely on the kind of institutions that we have set up amongst ourselves to address these issues."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:14",
                    "end": "00:20:44",
                    "text": " But maybe what's additional is to your earlier provocation on how do we identify and how do we verify, right? I work with a few other institutions that, some are called AI safety institutions, but some are not anymore. And there are other names as well. And one of the things we do a lot of is joint testing efforts, evaluations and so on. And it's not so much"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:44",
                    "end": "00:21:14",
                    "text": " the outcome of the joint testing efforts is important. I mean, the results and all that for us is less critical. Building capacity for evaluating is important. But what was really important for us is the ability to start building trust. I think for governments to work together, you have trust with one another at various levels. At the diplomatic level, our foreign ministries, they have channels to engage and discuss."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:14",
                    "end": "00:21:41",
                    "text": " sectoral level, the healthcare agencies have a way to engage and discuss, the cyber security agencies have a way to engage and discuss. How then do the people working in AI have a way to engage and discuss? And I'd like to suggest that this is one initial way in which we're starting to build that trust among people who are within government, dealing with AI, learning to, well, working together on even small projects is a way to start doing that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:41",
                    "end": "00:22:11",
                    "text": " and doing it on a regular basis, exchanging information that allows us to build trust. And when you have trust, then you can pick up a phone and call someone and say, you know, my colleague in Japan is Akiko, right? And pick up a phone and call Akiko and say, hey, you know, something has happened. Can we just figure out, is this what you're observing today as well or not? And then coming back then to Stuart's earlier sort of discussions around then how then do we regulate and think about bigger..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:11",
                    "end": "00:22:40",
                    "text": " think about the issues around requirements on labs to demonstrate that they're not putting us at risk in very severe scenarios. This is also one way in which we can then start to do that together. Governments have to, one, build the capacity to also technically evaluate and assess. We can't rely on the labs themselves to show us."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:40",
                    "end": "00:23:05",
                    "text": " So if they give us a number, how do we validate that? And we can't do it on our own because collectively, I think we even have less resources than a single lab in doing some of this work. So definitely we have to do it together. So these are ways in which we can try to address what you call in the title, diplomatic crisis."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:05",
                    "end": "00:23:33",
                    "text": " I'll just summarize, I know this is a very long kind of intervention, but I'll summarize in three points. One, we already have existing ways in which we are dealing with crisis, whether it's in healthcare, in cyber security and so on, we should do that. I think on the AI front, we can start building a lot more trust amongst, well I don't want to use allies, amongst colleagues who are working in this space, especially around technical evaluation. And then three,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:33",
                    "end": "00:24:02",
                    "text": " as we start to build this technical evaluation to not just deal with crisis, but we can then start dealing with issues around governance. And I have more to say on that, but maybe I'll just have my fellow panelists provide some provocations as well. Thanks. If you could give us the perspective from Japan, but also from the global governance standpoint, which you're quite familiar with."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:02",
                    "end": "00:24:38",
                    "text": " If one of the scenarios again that I mentioned occurred tomorrow or any other scenario, which institutions would be expected to coordinate? Where do you see the biggest institutional gaps for blind spots, if you will? Thank you for having me. And the topics you are already focusing on, diplomatic crisis, AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:38",
                    "end": "00:25:11",
                    "text": " It's really something not evident, but we have to tackle all together. I agree with this. And I don't think we have solution, internationalization, something like that. We should be innovative in the way to tackle these issues. Because first of all, AI is technology. As such, we know that we have to tackle... I would say..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:11",
                    "end": "00:25:41",
                    "text": " We are facing many crisis today more frequently, more in depth and handling all these issues we have experienced by the past. When you have cross-border crisis, diplomatic way of handling these issues. And this has been, we have accumulation past experiences and doing well and something"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:42",
                    "end": "00:26:10",
                    "text": " it works, something else, sometimes it doesn't work. But we have some kind of consolidated way of tackling this issue. But when you have AI, which is amplifier of existing crisis, the difference is the speed and we are less equipped human to make decision quickly."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:10",
                    "end": "00:26:41",
                    "text": " and also to have a way to have overview quickly gross water issues. So we are less equipped to address all these issues. So how to do that? We are not alone as a human being. And also I would say the differences with the existing crisis and air-driven or air-implified crisis that usually when we human are tackling."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:41",
                    "end": "00:27:10",
                    "text": " this crisis. We have kind of social consensus but it is absent in case of AI. So the difference is there and I think its phenomena is becoming more critical because we are seeing, observing that AI is gaining in agency more than more. So how to do that? So my view"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:10",
                    "end": "00:27:38",
                    "text": " on this issue is that we need to consolidate what we have accumulated by the past. But not only because we see the limits because of the capacity, the human capacity. So what is needed is to have kind of new practice, new framework to address these cross-border classes."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:38",
                    "end": "00:28:11",
                    "text": " power generated by air. There's no one solution, but fundamentally what we needed is that more to more human touch within. That means human to human, nation to nation, institution to institution, channels at the basis. And it should be consolidated by mutual trust. And what we needed?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:11",
                    "end": "00:28:40",
                    "text": " is not just addressing crisis you have in your front, but everyday life. You need to consolidate all these channels with employing human capacity, and that's the key to foresee innovative way of tackling crisis generated by as of here. Great. Thank you very much, Yuka-san."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:40",
                    "end": "00:29:07",
                    "text": " Now if I might turn to Azizan Azemi. Now I'm sure I mispronounced the name. Azizan, if you could then talk to us a bit about the Central Asian perspective. What kinds of AI related crises feel most plausible in your regional context? And how included do countries in the region feel?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:07",
                    "end": "00:29:24",
                    "text": " in terms of global coordination discussions and what would make a crisis coordination mechanisms are genuinely usable from your point of view from where you're sitting."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:24",
                    "end": "00:29:54",
                    "text": " I think it's working out. Perfect. Thank you so much. First of all, I want to say Namaste, India. Thank you so much for hosting all of us. It's been a phenomenal time in Delhi for everyone who has come over. Let me say a few words in the beginning about what's been interesting in Central Asia that I think could be a model for other regions of the world. Last year, I was involved in a process together with our colleagues from the government of Tajikistan and the government of Central Asia in drafting a United Nations"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:54",
                    "end": "00:30:21",
                    "text": " which was passed over the summer. It was passed on July 25th of last year. And that resolution is quite unique in several aspects. First and foremost, it built a common Central Asian approach to artificial intelligence, where we laid out that eventually our intent with AI adoption is to achieve sustainable development goal implementation. And as part of that, there were two features that were very interesting for our discussion. Number one."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:21",
                    "end": "00:30:50",
                    "text": " The UN resolution called for the establishment of a regional center for AI. So it's the first time when, at a regional level, there's a center for AI in a particular part of a region that will help at least coordinate between the different AI bodies that are responsible for it. Our friends from the Agency for Innovation and Digital Technologies of Tajikistan are here today. The deputy director is here, and they're working very hard on realizing this regional center for AI initiative now, together with other governments of Central Asia. Number two, the resolution recognized..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:50",
                    "end": "00:31:19",
                    "text": " mechanisms of AI self-regulation that have come out of Central Asia. And Tajikistan has been really at the frontier of that discussion. Now let me be a little bit libertarian here and give you the perspective that we've had at least in our country. So there was mentioning on this panel from other esteemed speakers about, for example, electricity as a technology, cars, aviation. What is the main cause for plane crashes? About 80% of plane crashes are because of human error."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:19",
                    "end": "00:31:48",
                    "text": " It's because of pilot decisioning that was faulty. What is the main reason for car crashes? Almost all of the time it's human decisioning, right? It's because of different factors that are tied to the human drivers who are making faulty decisions. You mentioned in your finance case, and I'll tie this point back, the case of how in trading, for example, financial trading, most of the trading at hedge funds is done by AI models. Why is that the case? Why have hedge funds moved away from human decisioning?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:48",
                    "end": "00:32:11",
                    "text": " because in any backtest that you can run, an AI model that's as basic as a classification model running on a very simple boosting algorithm outperforms human judgment. And the point I'm trying to make is that these technologies, including artificial intelligence technologies, are permeating into decision making not because of human perfection, but because they perform superior."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:11",
                    "end": "00:32:41",
                    "text": " to human imperfection. So when we talk about regulation of AI, the question that puzzles me as someone who is a founder of a startup that is now helping our government with AI policy, is that we always talk about AI as an object of regulation, right? And that's not really the case because who's eventually making the decisions? It's actually humans that are using AI as an instrument to make decisions. So why do we hold AI to a standard that is much tougher and stricter than"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:41",
                    "end": "00:33:10",
                    "text": " human decision making. I find that to be very puzzling. And the perspective of Tajikistan and Central Asia, at least from our conversations with our peers in the region, is that think about it from our perspective. Tajikistan is a developing economy. We're just permitting our AI policies. We've put AI into 100 schools across the country. There's a national LLM. We've developed locally. We export AI to 20 countries. A lot of this work is done by my good colleagues who are in the room today. And the question we raise is, if we were to step in,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:10",
                    "end": "00:33:38",
                    "text": " to regulate AI at this point. What does that mean about our competitive advantage on the global scale where countries are, look at China for example, countries are developing and deploying AI models at scale never seen in human history before. Regulation slows down our ability to innovate fast. But it's not a question of not regulating. Of course we all agree that AI should be regulated somehow. We all agree on that point. It's a question of do we hold AI to perfect standards?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:38",
                    "end": "00:34:05",
                    "text": " while human decisioning isn't perfect, that's where I'll push back. And one comment I wanna make is, Zippo, the company that I founded, as a founder and CEO, I chaired the AI Council under our Ministry of Industry and New Technologies in Tajikistan, we work on using synthetic data in order to automate decisioning for risk models in banks. So whenever we go to a bank, let me explain to you the journey of what we go through as start-upers. We go to a bank, we tell them, folks,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:05",
                    "end": "00:34:33",
                    "text": " Don't trust our words. Let's run a backtest. We can help you reduce your non-performing loan rate by using our proprietary synthetic data models. You know what the bankers tell us? Fantastic. Let's run a test, but let's make it X condition, Y condition, Z condition. Six months pass, we've done 10 backtests, and then they come back and say, let's run one more test of hypothetical scenario of this happening. We go through seven rings of hell before deploying one AI model into production."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:33",
                    "end": "00:35:03",
                    "text": " when we're serving enterprises. And banking is a very highly regulated sector. And if someone tells me that a government bureaucrat can regulate me better than my customer, I'm sorry, but this is not the case. And we service 60 banks across 20 countries. The world's largest Islamic bank is our customer. We just onboard them. To onboard them as a customer, we have to go through 12 months of piloting and testing, right? And my customer is the ultimate judge of how capable is my AI model and whether it was..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:03",
                    "end": "00:35:26",
                    "text": " developed in a properly regulated environment. So my call is to think about it a bit broader where I understand that for advanced economies, strongly regulated policies make sense. But let's not go back into colonial thinking of the whole world should follow that model. It's not the case. We're living in a different world. And frontier economies like Tajikistan have a voice of their own to make."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:26",
                    "end": "00:35:55",
                    "text": " So this is my call to have a bit of kind of counter-intuitive thinking and to at least give a chance to developing countries to showcase that we can have our own independent sovereign policies that are innovation first and that will help us to have our own seat on the table not to be on the menu at the table. Thank you. Thank you very much. I see one seat wants to respond a bit. Before we go to Audrey Tang's video, please. Thanks for that. I mean I absolutely agree that I think there are many things that we should allow"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:55",
                    "end": "00:36:25",
                    "text": " economies to be able to do and be able to use AI effectively in Singapore, we do that as well. I mean, AI is going to be a big part of our economy and our society. We've launched a lot of recent, well, made a lot of recent announcements to actually be able to use AI effectively. But I thought maybe I wanted to break down the regulation conversation a little bit better because I think it's a little bit more complex than that, I feel. So I work in AI governance and safety policy."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:25",
                    "end": "00:36:51",
                    "text": " We talk about AI regulation in Singapore, so that's really what I will be focusing on, what I'll be working on if we have AI regulations in Singapore. I think we look at one, regulations in terms of outcomes, not technology. I think that's important and that actually is reflected in how that comes out in the sectors, in financial services, in healthcare, in legal, in HR and so on. You focus on the outcomes."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:51",
                    "end": "00:37:18",
                    "text": " and not the technology. And those regulations already exist today. But of course, then you start thinking, but in some cases, it might be difficult in the existing regulations to be able to address the new impact that AI might have in these specific use cases. So what do you do? You either clarify your laws or you introduce new ones. So for example, in the elections, we introduced a law relating to deepfakes a couple of years ago. So that it's very specific, it's very surgical."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:18",
                    "end": "00:37:48",
                    "text": " Only during elections, you cannot be using deepfakes. And we have mechanisms to, one, decide if it's a deepfake. And a deepfake doesn't have to come from AI. It can come from Photoshop. It can come from somebody just putting somebody's voice over a video, right? That doesn't require AI. It just scales it up, makes it easier, and so on. And then we have means to say then who then tells the platforms, and we're talking about social media platforms, WhatsApp, and so on, who then tells them to remove this. So you must have a mechanism to do all of that as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:48",
                    "end": "00:38:18",
                    "text": " So these are laws that you can put in place to address specific issues that you have to deal with. And I think for your citizens, they will expect some level of protection from harms. Not everything, but some level of protection from harms. And actually you've already done that in many ways in your sectors. Now let's come to AI in a different way. What Stu was talking about was how do we deal with significant risks from AI at the broad level across the world, right? And that I think requires a form of..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:18",
                    "end": "00:38:47",
                    "text": " collaboration and working across governments, right? We're talking about companies with a lot of influence, which is already alluded to by Stuart. Then how then do we work together on that? And that's slightly different from how we want to use AI domestically within our countries, right? But even then, when we think about something like chat GBT from open AI or any of the models that are dealing with our citizens on a very regular daily basis, and you worry about the harms that can come from that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:47",
                    "end": "00:39:16",
                    "text": " And that is a domestic issue that we have to deal with. What do we say as unacceptable harms? Whether it's CSM, non-consensual sexual images and so on. What is unacceptable and that's something that we can actually address. We have online safety laws in Singapore and we can use that to be able to deal with these things. So that's the point I'm trying to make. Regulation is a complex topic. You have different parts of it that you want to address. There are things that we should do internationally together and that takes time and effort. And maybe diplomacy is an important piece of that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:16",
                    "end": "00:39:45",
                    "text": " Thank you very much. I see that all our panelists want to respond and I'll have you respond. Can you just keep it brief because I would like to show the video of Dr. Audrey Tang and we'll make that at the end. So please, Stuart, your comment. Yeah, so with respect to the last remarks from our Tajikistan colleague, regulation is not a single uniform thing."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:45",
                    "end": "00:40:14",
                    "text": " It always needs to be calibrated to the risk that's being posed. And your bank customers, because you are not liable as an AI company for losses incurred by the banks trading with your software, they have a very strong incentive to impose lots of testing, which is effectively regulation but happening in the commercial arena. But when a company like OpenAI is..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:14",
                    "end": "00:40:44",
                    "text": " selling it directly to the user. The user is not aware. There's a huge information asymmetry. I may not know that this software is capable of convincing my children to commit suicide. And so what kinds of regulation depend a lot on the information status of the participants, their economic resources and so on. And regulation does not have to be an enormous burden."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:44",
                    "end": "00:41:14",
                    "text": " My friend Emily runs Food Standards in the UK and she says that there's far more regulations on sandwiches than there are on AI systems. So a little sandwich shop can manage to comply with a heavy regulatory burden and we open across the world more than a million new innovative restaurants and sandwich shops every year. So somehow regulation and innovation are going hand in hand."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:14",
                    "end": "00:41:44",
                    "text": " And as I mentioned, vast parts of our economy, water, food, buildings, transportation, these are all regulated and yet they function very well. So I think that's really important to understand. I just want to come back to something that Wan Sze and Yukio said about cooperation among regulators. I think this is a really crucial thing. And I think cooperating around a specific"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:44",
                    "end": "00:42:13",
                    "text": " regulatory projects that are not moonshots, that are not trying to regulate everything in one giant omnibus global treaty. For example, we should have a right to know, every human being should have a right to know if they're interacting with a human or a machine. It's a very simple proposition. It's already in the European Union AI Act. I would love it if regulators around the world could cooperate to figure out"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:13",
                    "end": "00:42:36",
                    "text": " how to put that into their national laws and how to operationalize it, for example, so that social media platforms screen out bots so that you know that the message you're getting is not from a bot but from a real human being. That would be wonderful. Now, just one minute each, please. One minute each because I know the time. Just one minute."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:36",
                    "end": "00:43:05",
                    "text": " Regarding regulation, we have to move from this traditional classical principle agent model, high authority controlling companies access into more co-responsible model in the way that everyone on the same boat and including those who are at the government, regulatory issues, but private sectors and users and we have to take"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:05",
                    "end": "00:43:31",
                    "text": " responsibility to each of us, and then changing the way of reading. So just as usual. Thank you so much. We'll keep it super short. Yes. I think the reason that there are more regulations on sandwiches than artificial intelligence is because AI is a bit more complex than making sandwiches. So we need to take time on how to actually think through these policies. One question I just want to, rather common I want to add is that I absolutely agree with my colleague from Singapore, because..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:31",
                    "end": "00:43:51",
                    "text": " It's not a question of we are working on the AI regulation policy in Tajikistan. My colleagues from the government are here as well. And we understand how it needs to be risk-based. But our startups in the country, and we're one of the few developing countries that is exporting AI software today at scale in production, are not entering the European Union markets because EU AI Act is too complex."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:51",
                    "end": "00:44:15",
                    "text": " we can debate theoretically what is right. Is the risk-based approach framework great or not? But de facto, in the real world of practice, companies are not entering the EU market because we simply don't understand what is in that thousand page code that I think even GPT will have a hard time contemplating upon. So the question I want to pose for everyone just to think about is that our approach is AI self-regulation where..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:15",
                    "end": "00:44:45",
                    "text": " Exactly as my colleague from Singapore said, it needs to be outcome based. We have seven principles that we've enshrined in that resolution that passed. AI has to be explainable. The models have to go through validation, out of sample testing, out of time testing before they go into deployment. We understand that because we make those models. But it's a question of should it be up to a government agency to deploy these into practice or should it be the market to self-regulate up until we understand a little bit better what is AGI, how does it feel like, and whether we should eat sandwiches with it or not."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:45",
                    "end": "00:45:25",
                    "text": " Thank you. Okay, now I'm sorry, Russian into the local time. I'm Audrey Tong, Taiwan's cyber ambassador, first digital minister and 2025 right livelihood laureate. My heartfelt thanks to AI Safety Asia for staging this important conversation at the India AI Impact Summit. In diplomacy, we think in years, it takes time to draft texts to build consensus."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:25",
                    "end": "00:45:54",
                    "text": " and ratify commitments. In the AI world, crises unfold in milliseconds. What I wish to discuss today is this incompatibility between diplomatic and algorithmic time. These crises are not on the horizon they're unfolding now. Deep fake videos featuring public figures, synthetic voice scams spreading across borders, automated systems amplifying harm,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:54",
                    "end": "00:46:25",
                    "text": " before regulators can respond. And we have seen what algorithmic time does to markets. The 2010 flash crash, where US markets plunged and recovered in minutes. We can see what it does to trust. Europol has warned that organized crime is using AI-driven impersonation to scale fraud and evade detection across jurisdictions. And here is the shift."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:25",
                    "end": "00:46:53",
                    "text": " that changes everything. AI is no longer just a tool, it is a participant. NIST's most recent guidance described AI agent systems as capable of planning and taking autonomous actions that affect real-world systems, as OpenClaw recently demonstrated. Once incidents become agentic, response cannot depend on heroic improvisation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:53",
                    "end": "00:47:24",
                    "text": " It requires institutionalized cross-border mechanisms. So what does AI crisis diplomacy look like? Three building blocks and one regional proposal. First, for trust, build a white list for public integrity. Taiwan's 111 government SMS is a dedicated short code for official messages so citizens can instantly verify what is real."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:24",
                    "end": "00:47:55",
                    "text": " a blue check mark for public communication. Every message shows the agency's name and the last three digits of your phone number, proof that the sender knows who you are and the network guarantees who they are. When people trust the channel, phishing and impersonation lose steam. Every country needs its own version, a low-friction verifiable trust channel that works even in crisis. Second, for consensus."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:55",
                    "end": "00:48:25",
                    "text": " Use AI to listen at scale and reward bridging, not outrage. Tools like POLIS simplify interaction to agree or disagree, removing reply threads that amplify emotions. They surface bridging statements, ideas that people's opposing views still find reasonable and make them visible. The VTaiwan process combines this with in-person dialogue."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:25",
                    "end": "00:48:55",
                    "text": " transforming polarized issues into workable policy. Tools like Talk to the City push the skill of listening further with auditability at the core. Every theme traces back to original participant statements so society can verify whether the summary is feasible. In crisis diplomacy, legitimacy comes from speed and verifiability. Third, for safety."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:55",
                    "end": "00:49:25",
                    "text": " treat AI incidents as civil defense. Deepfakes do not stop at borders. Market cascades do not either. Work on defining and monitoring AI incidents is emerging, but we still lack the operational connected tissue for cross-border coordination. So here is my proposal. Establish a regional AI crisis liaison network, a technical hotline for the algorithmic age."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:26",
                    "end": "00:49:55",
                    "text": " Now we do not need to start from scratch. In cybersecurity, FIRST provides a global network for incident response teams. APCERT offers a trusted contact framework for the Asia-Pacific. What we need is to extend that capacity to cover AI-specific incidents, whether by deepening existing mandates, embedding AI expertise within the structures, or establishing."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:55",
                    "end": "00:50:26",
                    "text": " complementary liaison points that plug into the networks we already have. The goal is to ensure that when the millisecond scale crisis hits, cross-border cooperation is not improvised. It is ready to roll. This does not require a political alignment. It requires technical trust. Finally, a window of opportunity. Let Asia be not just a rule taker."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:26",
                    "end": "00:50:48",
                    "text": " by the supplier of safety infrastructure. This summit is the first event of this scale hosted by the Global South, possible through growing recognition of the institutional power of the India Digital Public Infrastructure. ATAR and UPI are increasingly seen as leaders"
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6994516d25c9f4e0d7e70a4a",
    "projectId": "487282",
    "projectName": "Scaling Trusted AI Global Practices, Local Impact - French",
    "projectDescription": "Scaling Trusted AI Global Practices, Local Impact - French",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T11:30:57.772Z",
    "projectUpdatedDate": "2026-02-17T11:30:57.772Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T11:31:47.747Z",
    "projectDeleted": false,
    "projectDataLength": 3,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "Scaling Trusted AI Global Practices, Local Impact.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "Scaling Trusted AI Global Practices, Local Impact.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "French"
},
{
    "_id": "699450c025c9f4e0d7e70a04",
    "projectId": "632829",
    "projectName": "Scaling Trusted AI Global Practices, Local Impact - Hindi",
    "projectDescription": "Scaling Trusted AI Global Practices, Local Impact - Hindi",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T11:28:04.956Z",
    "projectUpdatedDate": "2026-02-17T11:28:04.956Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T11:28:57.357Z",
    "projectDeleted": false,
    "projectDataLength": 3,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "Scaling Trusted AI Global Practices, Local Impact.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "Scaling Trusted AI Global Practices, Local Impact.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Hindi"
},
{
    "_id": "69944e4f25c9f4e0d7e7093f",
    "projectId": "444788",
    "projectName": "Scaling Trusted AI Global Practices, Local Impact - English",
    "projectDescription": "Scaling Trusted AI Global Practices, Local Impact - English",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T11:17:40.629Z",
    "projectUpdatedDate": "2026-02-17T11:17:40.629Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T11:23:24.121Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "Scaling Trusted AI Global Practices, Local Impact.mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771327047632.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:28",
                    "text": " projected to add 15.7 trillion to the global economy by 2030. Here in India, that number could reach as large as 500 billion in the GDP contribution. 47% of Indian enterprises already have multiple AI use cases, not in pilots, but in production. And 74% just over the last two years have accelerated their AI rollout."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:28",
                    "end": "00:00:45",
                    "text": " From precision agriculture to fraud detection in banking to AI power diagnostics in rural healthcare, the deployment in AI is real and it is accelerating. So I truly believe that it is an amazing time to build."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:45",
                    "end": "00:01:07",
                    "text": " But I think what is even more exciting is it is time to build responsibly with trust and confidence because the question that will define the next wave of AI is not who builds the fastest or the best performing model in the world, but it is who builds with measurable trust. Who gives people enterprises?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:07",
                    "end": "00:01:36",
                    "text": " institutions and nations, genuine confidence that this AI technology will deliver prosperity and progress for everybody. The next wave of AI will not be defined by capability alone, and this is a core belief we've had since founding the company. It will be defined by trust that you can define, measure, and prove. And that is why the framing of the Impact AI India Summit is so critical."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:36",
                    "end": "00:02:03",
                    "text": " Congratulations to the Indian government for convening the summit around the imperative of impact. India is not just asking the question, should we build AI? I think the exciting thing is India is asking the question, which is a hard one. How do we build AI that works for 1.4 billion people? And how do we prove it, that it can be trusted? And how do we make sure that it is impactful?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:03",
                    "end": "00:02:32",
                    "text": " And there's a reason that this question is being asked here. You know, India built Aadhaar, enrolling 1.3 billion people into digital identity system that now underpins everything from banking to healthcare access. India also built UPI, processing over 13 billion transactions a month, making peer-to-peer payments seamless for hundreds and millions of Indians built on the India AI stack."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:32",
                    "end": "00:03:00",
                    "text": " Now, these are not just technology achievements. I think these are really trust achievements. These work because people trust them. Institution, that institutional memory building systems that billion people can rely on is rare. And that is what this moment in AI really demands. Now, with the release of the AI governance guidelines, India has done some really important work in establishing the foundation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:00",
                    "end": "00:03:17",
                    "text": " which is really underpinning the AI ecosystem in India. Now the guidelines represent the framework around the seven sutras that translate responsible AI principle into actionable pathways for builders, enterprises and institutions."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:17",
                    "end": "00:03:41",
                    "text": " But most importantly, as this technology and sector specific need evolves across India, the framework can be built upon and augmented. And this is something that we've been actively working with the Indian government on over the past couple of months. Now, this is emblematic of what good effective governance by design looks like. And we really commend the Indian government for their leadership here."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:41",
                    "end": "00:04:06",
                    "text": " Last month I had the opportunity to visit with professor sued who is leading the office of the principal scientific advisor to Prime Minister Modi and they've recently published the techno legal framework for AI governance. Now for those unfamiliar with this term which is techno legal it just means that governance is not just written in policies but it is embedded throughout the AI life cycles."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:06",
                    "end": "00:04:24",
                    "text": " It is measured scientifically, and you are constantly making sure that the trust can be proven. So it is compliance by design, governance by design, not as an afterthought. And the framework basically maps governance across the full AI lifecycle, from how data is collected to how models are trained."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:24",
                    "end": "00:04:46",
                    "text": " to how AI systems are making decisions, to how autonomous agents are going to be operating with full agency. And critically, it is built for India. Purpose-built institutions calibrated to India's sectoral diversity and digital infrastructure. Now, this is the kind of thinking that excites us and that really matters in this age of AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:46",
                    "end": "00:05:14",
                    "text": " The next step is really the hard part, making it real. And that means translating this policy into operational standards that enterprises can actually follow. It means investing in AI governance as a professional discipline, not just as AI literacy, but the specific skills needed to assess AI risks, verify AI's trustworthiness, and manage AI systems continuously."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:14",
                    "end": "00:05:42",
                    "text": " It also means that venture capital flowing into trustworthy startups and governance tooling. It also means incentivizing all the builders in this room with measurable trust from day one. Defining the framework we believe is necessary, but operationalizing it is the bigger challenge, and it will require a lot of deep work in investment, in scaling, in incentives, and in policy to determine what it changes next."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:42",
                    "end": "00:06:08",
                    "text": " Now Credo AI has spent past five years building software that helps organizations worldwide to operationalize AI governance. And what that means is we take trustworthy AI principles and we turn them into daily workflows, measurable outcomes, measurable checks, and organizational alignment. We work with Fortune 500 companies, many of those that you're going to hear from today on the panel, and there are a few things that we've learned."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:08",
                    "end": "00:06:34",
                    "text": " The first thing that we've learned is context is so important because the same AI system that can be helpful or harmful depending upon where and how it is used. A simple example, if you have a large language model that works well for answering employee questions inside a company becomes a very serious risk when it is giving financial advice to a retail customer who may act on that advice."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:34",
                    "end": "00:06:60",
                    "text": " a facial recognition system that is accurate in one country might fail in another just because of demographic disparities. Now this is why you can't govern AI in abstract and that's why context really matters. Second is governance really needs to be across the full AI life cycle. Now most I would say organizations treat AI governance as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:60",
                    "end": "00:07:25",
                    "text": " airport security. You know, it's like the worst thing ever, but it's a single checkpoint before deployment. And we've seen those companies are actually not successful. And the reason for that is you have to really think from your designing to your implementation to putting it in production. How do you actually in in that system define that measurable trust and make sure that those systems are holistically governed?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:25",
                    "end": "00:07:41",
                    "text": " The third thing which I do want to underscore, because this was a very interesting set of conversations yesterday at the summit, is many assume governance is all about regulation. And I actually think that is a framework or mindset we need to shift."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:41",
                    "end": "00:07:56",
                    "text": " AI governance actually is a competitive advantage. This is one of the biggest opportunities knowing the risk that you are having in your AI systems that you're either building or buying, aligning your teams and deploying that with confidence."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:56",
                    "end": "00:08:21",
                    "text": " So I think as we increasingly think about governance, governance is becoming a market access requirement. Enterprises want to know, can I trust the AI? Governments want to know, does it meet our standards? Trading partners want to know, can I rely on this AI across borders? And the organizations that can answer yes with evidence are the ones that are winning contracts."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:21",
                    "end": "00:08:31",
                    "text": " And so we really see governance as a steering wheel and not a brake that is going to guide AI to be useful across not only India but also global south."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:31",
                    "end": "00:09:01",
                    "text": " And you know, it's something that we've been working very extensively with across the globe with standard setting bodies, with regulators, with policymakers is really thinking about how trust needs to be embedded across standards, institutions and procurement. Trust is not just a product feature. It has to be woven into the entire ecosystem. Standard bodies need interoperable AI governance benchmarks so that an AI system can be assessed as trustworthy in India is recognized."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:01",
                    "end": "00:09:26",
                    "text": " in Europe is recognized in Middle East, in Africa, institutions, banks, hospitals, government agencies really need clear definition of what trust in AI means and not weak principles but specific measurable requirements. And this is the work that we've been doing over the past five years, making sure that for every application we can define what that trust means and can we measure it. And so trust is..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:26",
                    "end": "00:09:55",
                    "text": " how we standardize and how we institutionalize how we buy and sell AI. So I also want to quickly address before we bring on the panel a couple of few things that there is this big conversation happening globally around AI taking jobs. And I would like to offer a very different lens. AI governance actually is creating a whole new category of professional jobs that didn't even exist like"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:55",
                    "end": "00:10:10",
                    "text": " one year ago or two years ago. And these roles require distinctly human skills, judgment, contextual reasoning, ethical analysis, stakeholder communication that AI in itself cannot replace."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:10",
                    "end": "00:10:38",
                    "text": " AI evaluators, professionals who test and stress test AI systems for safety, fairness and accuracy before and after deployment. AI governors, people who design and manage the policy frameworks and the workflows that keep AI aligned with organizational and regulatory requirements. Managers of AI agents. I mean, you know, if claw bot was any indication over the past 10 days, we know that we need humans to really think through what does"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:38",
                    "end": "00:11:03",
                    "text": " automated governance look like, but at the same time be able to manage this massive army of agents. And so trust and verification specialists are also becoming a core role. It's a professional who audits AI systems, produces the evidence that enterprises and regulators need to have confidence. So these are not speculative roles. These are happening right now, right here across the globe."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:03",
                    "end": "00:11:30",
                    "text": " For anyone in the room who's early in their career or even starting a startup, I would certainly suggest that AI governance is going to be that next frontier of job creation that humans are going to be critical for. The future of work in AI is not just about building AI. It is about verifying, governing, and earning trust in AI. And that's where some of the most important careers for the next decades will be."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:30",
                    "end": "00:11:59",
                    "text": " So everything that I've shared just now is coming from practice. It's not coming from theory. As I said, for the past five years, we've been working extensively with governments and enterprises to really define what trust looks like and to embed that in their software development, in their AI development processes. And so today we are so excited to one, launch the global compendium of contextual AI use cases with some of our customers and partners."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:59",
                    "end": "00:12:27",
                    "text": " Some of those will be on the panel in just a bit and you'll get to hear a little bit more about them. But I think this is really important because as we are thinking about what does context mean in terms of AI governance, you can't have a peanut butter approach to AI governance. It really needs to be grounded in impact that each and every AI application is going to have, whether it's in healthcare, whether it is in lending, whether it is in facial recognition systems, whether it is in defense. And so certainly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:27",
                    "end": "00:12:51",
                    "text": " If you have your phones out, please check out the compendium. I am so grateful for our partners and customers, MasterCard, Autodesk, G42, IAPP, and Cisco for contributing some really exciting use cases so that as you go on into your organizations, you can bring these use cases to bear."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:51",
                    "end": "00:13:13",
                    "text": " We are all here because we all believe in potential of AI or we are afraid of what is going to create. But for whatever reason, you're all here. Let's be honest where things stand. Um, you know, AI governance feels super overwhelming. You know, everyone I speak with around AI governance doesn't really know how to dissect this complex challenge, a child, uh, you know, a topic."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:13",
                    "end": "00:13:32",
                    "text": " New rules every week. We have 72 countries with over 1000 policy initiatives. The EU AI Act, the AI governance guidelines in India, South Korea's basic AI Act, sector specific frameworks from RBI to SEC. Risks are expanding fast and so are the opportunities."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:32",
                    "end": "00:14:01",
                    "text": " So whether you're a startup founder building your first product or an enterprise leader scaling across market, the question really is, where do I begin? And five years of building AI governance software has taught us that there is no single credible source of truth in AI governance. Everyone means different things when they talk about AI governance. And it's actually as a company that has created this category, it's painful to watch. So today I am really excited to be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:01",
                    "end": "00:14:30",
                    "text": " solving that challenge because today we are also launching the first global AI governance insights hub and this is a really monumental thing and I'll share a little bit more about what this hub is hopes to do but it is a comprehensive continuously updated single source of truth single source of truth for AI policies for AI risk controls models"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:30",
                    "end": "00:14:60",
                    "text": " and also vendors. As you're buying a lot of third party AI systems, how do you know which ones to trust? How do you know which ones to buy? How do you unpack the very complex AI supply chain? Our hope is that Credo AI's AI governance hub is the answer to that solution. It is not just augmented by our governance AI agents, but also by Credo AI's human intelligence through our team at Credo AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:60",
                    "end": "00:15:28",
                    "text": " The exciting thing is this is not a static document. It is actually tailored to your need. As I mentioned, context is so important in governing these AI systems and your input, your industry, your regions, your specific risk profiles, your use cases in plain language. This hub takes in that information and with Credo AI's governance intelligence gives you an AI powered snapshot of your governance posture and a plan how you can get started."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:28",
                    "end": "00:15:53",
                    "text": " I also want to just quickly focus on the three pillars that this hub really tries to address right now and then where we are augmenting it. So the first pillar is the policy tracker. Again, as I mentioned, regulation is just one teeny tiny component of governance, but we just want to make sure that Credo AI can provide you the single system of record for what truth looks like in."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:53",
                    "end": "00:16:16",
                    "text": " even regulations. So right now, the regulatory landscape, as you know, is exploding. We have aggregated all the major regulations globally and translated them into plain language so that you all can understand what this means for your businesses. But this is also where we actually work with the policymakers and regulators to make sure all the implementation that is shared here actually is validated by them."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:16",
                    "end": "00:16:42",
                    "text": " The second is the risk taxonomy. Again, one of the key things when we talk about AI risk is what is the harm to your brand, to your revenue, to the people that it serves. So this risk taxonomy provides 16 distinct risk types mapped from policy violations to bias, to security vulnerabilities, hallucinations, so that you can see exactly where you're exposed."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:42",
                    "end": "00:17:00",
                    "text": " And then the third thing is risk and policies are kind of futile without the right controls in place. And this is we believe is the bridge between the knowing and the doing. And a specific mitigation strategy for every risk, not just the diagnosis, but from human in the loop oversight to technical stress tests."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:00",
                    "end": "00:17:30",
                    "text": " And what is really exciting is obviously, you know, we do have a platform, but we are making this available for free to everyone in Global South, because what is really important to us is that it improves with your contributions. So I'll share the QR code in just a bit, but this hub lets you, lets the community vote on risk control mappings and the rate of impact of policies, the severity of these risks. And this is a collective effort, governance intelligence as a shared responsibility."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:30",
                    "end": "00:17:56",
                    "text": " So I really hope that all of you would join us. I'll leave this on for just a second so that you can, and we'll be sharing this after the session as well. Your contributions are going to be really critical to make sure that Global South continues to lead in trusted AI, because without that, we are just going to be beholden to models that might not work in context that is required for Global South."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:57",
                    "end": "00:18:17",
                    "text": " And so I do want to now switch to going back to something I started with, which is it is a really exciting time to build with AI. And today we are so excited to bring some extraordinary folks to the panel. These are the builders."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:17",
                    "end": "00:18:36",
                    "text": " who are not only defining what the next era of AI is going to be, but these are the builders who are defining what responsible and trustworthy AI looks like. They are the ones who understand that trust is truly the unlock across technology, across standards, across institutions, across procurement, across borders."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:36",
                    "end": "00:19:05",
                    "text": " And what really gets built with trust doesn't just serve India. We believe that it's going to serve the entire global South. And these builders that I'm going to be inviting are actually making that happen across enterprises, from financial sector to retail to frontier model to healthcare. So with that, Magesh Bhagwati, who is the Senior Vice President and Global Head of Data and Analytics and AI at PepsiCo, please welcome to the stage."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:08",
                    "end": "00:19:39",
                    "text": " Caroline LeVu, who is the Chief Privacy, AI and Data Responsibility Officer at MasterCrag. Fabrice CS, who is the Group Head of Responsible AI at G42. G42 is leading some really exciting work at the frontier of AI, building their own models for the Global South. So really excited to have you here, Fabrice. And lastly, Rajiv Gupta, who is the President at PB FinTech. Rajiv, welcome."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:39",
                    "end": "00:20:13",
                    "text": " So we are now going to kick into a very exciting panel conversation, because everything that we've shared today, these builders are operationalizing that at scale across their businesses. Hello, how's everyone doing? Wonderful, thank you. All of you have a very important responsibility in your organizations, not only to lead in AI, but to build responsible AI. So maybe Fabrice, I'll start with you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:13",
                    "end": "00:20:45",
                    "text": " What does that role mean and why is it important to G42? Is that working? In G42, we are a global, as you say, technology company based in Abu Dhabi. We bring the full AI infrastructure. We build data centers at scale. We build the biggest data center capability and capacity outside of the US. We bring the digital infrastructure, which is safe, sovereign, to create tokens of intelligence."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:45",
                    "end": "00:20:56",
                    "text": " develop AI models across a lot of industry in healthcare, in geospatial activities, in smart cities. We bring frontier model to bear."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:56",
                    "end": "00:21:16",
                    "text": " frontier model we develop in Indian English, in Arabic and English, in Kazakh and English. So for us, we need to bring responsible AI at national scale and regional scale. And to do that, responsible AI, as you say Navrina, is a no-brainer, has to be part of the strategy because what we develop has to be really secure."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:16",
                    "end": "00:21:37",
                    "text": " Data has to be protected, the handling of data has to be sovereign, but we need trust to develop AI at government level and really at scale across the regions. So as you say, responsible AI is not an airport security check. It's really, really embedded as a strategic imperative in everything we do. Thank you Fabrice."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:37",
                    "end": "00:22:04",
                    "text": " You know, Carolyn, you and I have worked for a very, very long time together. MasterCard was our second customer when we were just getting started. And you've made this a core part of your strategy, but you've implemented this across the entire financial inclusion for all the financial products. So tell me a little bit more about where is MasterCard today? How are you making responsible AI real, especially across some of the agents that you've developed recently as well?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:04",
                    "end": "00:22:31",
                    "text": " Thank you so much, Navrina. So happy to be here with all of you today. So let me first say a quick word about MasterCard and AI because it may not be obvious to all of you. So AI is nothing new for MasterCard. We have been leveraging AI for decades to make our network safer and more secure. So for example, we use AI to make sure that you are who you say you are when you make a payment or to detect legitimate fraud and spending behaviors."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:31",
                    "end": "00:23:01",
                    "text": " And then more recently, we are leveraging Gen.AI techniques that help us increase the speed and the accuracy of our fraud detection tools by up to 300%. So it's a real game changer. And as you said, Navrina, for us, the only AI is responsible AI. It is all about trust, way beyond regulations, to your point. Because for innovation to scale, people have to trust it, right? So we have been investing in AI governance for way before it was a hot topic."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:01",
                    "end": "00:23:29",
                    "text": " Now, to be honest, our customers have also been very explicit. That's what they expect. And they're very clear that one of the key reasons why they select us is because they trust us more. As simple as that. And we've learned early on that actually responsible AI is not a friction. It actually provides a business with clarity, with confidence, with guardrails that it can embed."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:29",
                    "end": "00:23:51",
                    "text": " into the design of the innovations from day one throughout their life cycle. And so if done well, it can really accelerate innovation. I'm going to keep that. No, I think you just hit on something which is so important. Like the trust is how your customers really believe in MasterCard. And I think there's such a great validation for Magesh for you guys, PepsiCo, where"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:51",
                    "end": "00:24:15",
                    "text": " Brand is so critical to making sure that any AI that is used is actually bolstering that customer trust, that PepsiCo's reputation. And you play a very important role leading the entire AI portfolio and data strategy at PepsiCo. So tell us a little bit more about concretely, maybe an example of where responsible AI has sort of helped shape a lot of your PepsiCo's work."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:15",
                    "end": "00:24:45",
                    "text": " Well, thank you. Thanks for having me. It's good to see everyone here We've got a full crowd here. So thank you all for coming By the way, I was actually 30 years back. I was just in your backyard I was working in narrow place before I ended up here. So Well, thank you But yes local makes a difference With regards to Pepsi Co on a daily basis, we serve around 1.4"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:45",
                    "end": "00:25:14",
                    "text": " billion consumers. So that's sips and bites every day. So for us, it's really about serving the community. It's about ensuring that truly the Prime Minister's vision of Vasudeva Kudumbam really comes to life. And what that really means is really it's welfare for all, it's happiness for all as well. And so when we talk about that, when we talk about AI, we almost switch very easily switch to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:14",
                    "end": "00:25:44",
                    "text": " Okay, there is benefits for me, but there could also be harm for me or for us or for the collective overall planet itself. So it's something that we've got to really walk the fine line because we're focused on two things. One is we want to all innovate at speed, but we also want to ensure that we are not taking undue risk as well as we're innovating at speed. So that's a fine line that we're walking. So I'll give you an example."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:44",
                    "end": "00:26:11",
                    "text": " we were actually doing a campaign for Mountain Dew. The marketing team was doing a campaign for Mountain Dew. And the responsible AI team got involved in the campaign. And very quickly, one of the things we noticed was taking pictures of mountains. Well, certain mountains could be copyright infringement. So very quickly, the team got together, worked with the marketing team, and we ensured that we had the right marketing campaign."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:11",
                    "end": "00:26:36",
                    "text": " go out and also make changes to the applications as well, the app itself that was doing. That made sure that we are not only launching the capability with the right mechanics, but we're also launching the right capability for the right consumers with the right guardrails in place. So I think responsible AI is not a back office function like you mentioned, but it's really a front office function because it truly is welfare for all."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:36",
                    "end": "00:27:10",
                    "text": " Thank you so much my gosh, and I think you know as a big fan of pepsico product the sips and bites really resonates with me and that goes to health care because Rajiv you and I were just talking backstage about PB fintech is one of the largest provider of Insurance lending software, but also they're foraying into health care So maybe would love to understand like why is trust important to you? Rajiv and how are you embedding that across your AI lifecycle?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:10",
                    "end": "00:27:42",
                    "text": " Hello everyone Good afternoon to all of you and thank you for having me here As we spoke, you know a little while before the session PB fintech represents largely the bfsi segment and banking and financial services without trust Is not innovation right? It is liability. We are accountable to the regulators We are accountable to the customers. We are accountable, you know for long-term brand equity as well When PB fintech was launched we started policy bizarre first"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:42",
                    "end": "00:28:11",
                    "text": " And to answer your question, we deal with about close to 1 million customers every day. And over 200 million customers a year. I mean, leaving aside some of the holidays, so over 250 million customers. Many years later, we launched Pesa Bazaar, which is a lending platform. So we are today one of the largest lenders. So who gets an insurance policy? Who gets the loan sanctioned? Who, at the end of the day, is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:11",
                    "end": "00:28:40",
                    "text": " trying to do a fraud with us. Since we connect with so many people every day, how do we summarize the conversations we've had during the day? It's all thanks to AI. And governance for us is upfront, because everything we do is regulated. You know, RBI is one of the largest regulators here in the country. We have IRDA, which regulates the insurance sector. On the health side, obviously, there's no regulator, but there are responsibilities we have. We've just launched PB Health, where we're setting up"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:40",
                    "end": "00:29:10",
                    "text": " hospitals across the country. We launched this company this year. For the first five hospitals we have chosen Delhi and CR as a pilot. The whole objective is how do we deal with India as a country, you know, keeping in mind the whole uniqueness of this country, right? We have 140 crore people living in this country of which 30-40 crore people live below the poverty line, generally get covered by social security schemes unlike the US and Europe where social security scheme is available for everybody."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:10",
                    "end": "00:29:36",
                    "text": " We also have the top 20 crore 30 crore people who manage things on their own because they are either you know earning enough to be able to manage things But the largest challenge we have in this country is those 60 70 crore people Who live in middle class and lower middle class for them affordability is a big challenge right and In this country the two big areas we will need to obviously work very closely is how do we improve financial literacy?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:36",
                    "end": "00:30:05",
                    "text": " and how do we build that trust and AI is the biggest platform for us to use that and that's what we have done we adapted AI way back in 2019 and today thanks to AI we are able to handle volumes and our ability to reach to every pin code of the country you know and educate people on why understanding finance is important leave alone buying insurance leave alone you know buying and secondly you know how do this people living in the middle class decide"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:05",
                    "end": "00:30:33",
                    "text": " what is important for them and their families. If they buy a health insurance, can health be affordable for them? If I go to a hospital, can I afford that cost if I don't have a health insurance? And if I have a health insurance, is it enough for me to cover my challenges? So I think AI for us will do everything in a hospital scenario where the customer enters as a patient. Till the time he gets discharged, if he gets hospitalized, everything is going to be managed through AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:33",
                    "end": "00:31:02",
                    "text": " Absolutely Rajiv and you know this is something we were just discussing yesterday with Meti and others is the culturally sensitive profile in India the large population base also the need for inclusivity and making sure these AI systems work for all is so critical. So Fabrice would love to hear from you because something that has been really exciting to watch is how G42 is not just building frontier models but they're actually building models that work for Global South."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:02",
                    "end": "00:31:28",
                    "text": " a little bit more about obviously all the innovation on frontier models, but why is Global South important to you? It's very important. I give you an example. We are very proud. We are launching a pilot for digital schools in Azerbaijan. And the intention is to bring basically the knowledge in all the schools and also accelerate what teachers can do so we can be more efficient and share the knowledge more effectively."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:28",
                    "end": "00:31:34",
                    "text": " To do that, we develop language models that are based on the local language."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:34",
                    "end": "00:32:03",
                    "text": " due to scarcity of data, of resource or skills. It took a lot of effort to be very, very contextual in the way we developed the model, to do a lot of testing in the model. And when you develop something at national level and for schools, where you have to build a lot of security and you have to build a lot of trust. And for us, it's very, very important to bring all the skills we have developed internally in the Global South to benefit that. And we do the same for in the English languages, Arabic."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:03",
                    "end": "00:32:29",
                    "text": " large language models as well we are developing. So for us the global South is part of the construct of all what we do. Thank you Fabrice. Maybe Carolyn I'll just go to you. You're based in Europe. Obviously MasterCard is multinational. So tell me a little bit more about when you're building AI products how do you take into account the context especially for India and weave that into products so that it actually is inclusive and working for all?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:29",
                    "end": "00:32:58",
                    "text": " Yes, absolutely. I think it's important to understand that the principles remain the same. So I think it's really important for companies to be driven by a set of global principles that act as your no-star and that guide everything that you do with AI. But then to your point, you may need to localize the controls and the guardrails so that they fit for purpose to take into account the local regulations, the digital maturity, the cultural differences, societal expectations, and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:58",
                    "end": "00:33:25",
                    "text": " Of course, Global South is something that has to be top of mind. So for example, transparency. How do you make sure that your AI innovations are very clear and understandable by everyone? And I think for India, it's particularly relevant given the scale, the diversity, the multiple languages that are at play here. Same for fairness. Are you sure that your training data reflects the diversity of the population so that the outcomes are fair?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:25",
                    "end": "00:33:52",
                    "text": " Now the good news is that at MasterCard we have our AI garage that is based here in India, in Pune. And we have 6,000 technologies that are really working day in day out to develop the next generation of our innovations with the global South in mind. And they conduct also all our research and developments. And so that is totally top of mind for us. And then on top of that, we are also working with academic institutions."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:52",
                    "end": "00:34:20",
                    "text": " To bring that outside in perspective, which is really, really important so that you don't actually work in eco chamber, you get expertise from outside. And then given the pace of change of the technology, I think this is absolutely critical to keep everyone in mind so that AI works for everyone, including here. So Carolyn, I'm just very curious, is agent pay now available in India or is it only in India? It is and we just announced it today."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:20",
                    "end": "00:34:44",
                    "text": " So, fresh hot of the press. I think when you think about AI agents and everything that's coming in the future, maybe just very quickly on agentic AI governance, how are you thinking about it? Yes, agentic commerce brings a whole lot of exciting opportunities, but also, you know, it brings some anxieties to some of us."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:44",
                    "end": "00:35:09",
                    "text": " And look, we're going to be driven again by the same principles. I think our mission at MasterCAT is to make sure that agent-led transactions are equally safe and secure, and to bring everyone along the journey to set the standards so that it works at scale in a trusted manner. And so we're going to bring our capabilities in terms of fraud protection, cybersecurity, building standards. But that's very exciting."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:09",
                    "end": "00:35:38",
                    "text": " Certainly is and, you know, we look forward to sharing more about agentic AI governance frameworks we are thinking about for Global South and potentially an opportunity to collaborate there as well. Happy to partner in every now. Mahgesh, I'm very curious. I won't go into retail because like, you know that in and out, but are there things from other sectors that have either surprised you in responsible AI or excited you that you want to bring into retail so that your AI is more trusted, whether you are buying third party?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:38",
                    "end": "00:36:07",
                    "text": " AI or developing your own? I mean it's a great question because we're still in the early innings of AI right so we're all collectively learning so one of the things I truly believe is AI is an equal opportunity leveler right organizations are going to be flattened industries are going to be also having a lot more seamless interconnectedness unlike kind of the walls we actually have today so what"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:07",
                    "end": "00:36:25",
                    "text": " And when you think about PepsiCo, we're not just limited to certain components of maybe a brand that you think about. We're through the entire food chain or value chain, all the way from farm to manufacturing, to supply chain, to go to market, to shelves."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:25",
                    "end": "00:36:52",
                    "text": " to eventually to the consumers and the consumer's homes and the palettes, right? And brands as well and an associated marketing. So we threw the full value chain and they're part of the value chain. We're connected with almost every aspect of most industries. And let me share an example of kind of what we've leveraged from the tech world is we're actually, I mean, some of the favorite brands here, at least that actually started in India is Kukri."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:52",
                    "end": "00:37:21",
                    "text": " Right and but at the same time there's other brands as well like lays or Doritos and These brands we a lot of the innovation for these brands actually come because we've actually got local ingredients in them So we've got local farmers Embedded engaged with us as it actually as we're actually building these brands the freshness of the food the high quality of the food That's one so there's also another piece as well that we're actually collaborating right now is with meta"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:21",
                    "end": "00:37:50",
                    "text": " in terms of really WhatsApp ordering. So you can actually go into a Kinara store, and we actually have customers in Kinara stores that can actually now start ordering via voice, via image, via video, or just native ordering. And so it integrates the full chain with, and you talked about earlier, you've got Aadhaar, you've got UPI, you've got Mastercard. So we're integrating across the full chain with all the best practices that we have, but with still affordability."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:50",
                    "end": "00:38:18",
                    "text": " of the market, Rajiv, like you talked about, because that's a very key component for this market here. And, Magesh, as you are using facial recognition, voice recognition, I mean, governance becomes so central to making sure that you're protecting every citizen, their biometrics. So I'm very curious, could you talk a little bit more about how you're thinking about maybe doing more for India in responsible AI and governance so that these systems not only work, but are going to be trusted?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:18",
                    "end": "00:38:44",
                    "text": " See, this is where it's the coalition of the willing and the coalition of the partnership. I mean, I'll give a plug for Navrina and her team, right? Credo.ai is a very critical part of how we run the entire governance lifecycle. So as we're actually building these systems, we're actually using the software development lifecycle. Now we're calling it AI development lifecycle. And through the full lifecycle,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:44",
                    "end": "00:39:12",
                    "text": " It's not only responsibly I typically gets invoked or involved or cyber teams towards the end of the process, but through the full lifecycle. And in that process, we're actually embedding. How do we think about consumer protection? Legals and war way in the beginning rather than later in the processes. And the point is, as more and more consumer ecosystems really comes into our consumer data platform, we've got to be that much more responsible."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:12",
                    "end": "00:39:41",
                    "text": " as we execute our game going forward. Absolutely. And we are excited to continue in collaborating with you and making responsible AI more operational across PepsiCo. Rajiv, you mentioned something which really I've been thinking deeply about is, you know, it's not just about tooling and just building trust and magically it happens. It's a lot about education and AI literacy. So I would love for you to like maybe talk a little bit more about within PB FinTech."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:41",
                    "end": "00:40:09",
                    "text": " How are you making your employees not only understand AI, but also responsible AI? Because again, AI skilling is one thing. It's really understanding how do you put governance guardrails in place. So talk to us a little bit more about your AI literacy initiatives. So I mean, straight away to answer your question, first of all, we have now started monitoring AI usage across levels of employees. So we are encouraging that if you don't adopt AI now,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:09",
                    "end": "00:40:38",
                    "text": " you will have a problem in future. So if I have to be 3x or 4x from where I am today, adopting AI is very, very crucial. First of all, getting the right skilled people today who understand AI continues to be a big challenge in a country like ours. So while we have a lot of people, we are training people to use AI very effectively. And AI to us is all about fairness. Today we live in a country where we have other big challenges. Today we are not an online"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:38",
                    "end": "00:41:04",
                    "text": " country. Technically, we have done many things. We've created UPI, Aadhaar. Those are obviously facilitators for people to start coming online. But if you look at this country, a very significant portion today still believes that people should come and meet them face to face and do transactions. To move from that scenario, to ask people to trust online requires a lot of responsibility, requires"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:04",
                    "end": "00:41:33",
                    "text": " to ensure that we remain accountable to everything we do online. We also want to ensure that our accountability towards the regulators is very high because we are under very strict regulations. Any challenges we face there will completely close our business down. And so to us, not only an option of AI, using AI effectively,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:33",
                    "end": "00:41:59",
                    "text": " What can AI do for us? For example, I'll just tell you one small story on which we are working today. We believe while India is moving towards Vixit Bharat in 2047, today there is a population who are 60 plus and that's about 10 crore people. Now if you study, this number is going to be 60 crore people in 2047, which means a significant rise in senior citizen population."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:59",
                    "end": "00:42:28",
                    "text": " Which also means today's 40 year olds are tomorrow citizens senior citizens. So how do we identify? To customize products for today's 40 year old so that they safeguard their interest and they grow old They don't have an income plan. There will be a problem because with longevity increasing how will they survive? So AI is going to help us to identify such people Customize products for them and at the same time we reach out to every pin code as I mentioned to you in the country keeping in mind diversity"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:28",
                    "end": "00:42:57",
                    "text": " multiple languages that we have in this country because people will not understand English or Hindi. We'll have to also adopt, you know, multi-language training programs to be able to use AI effectively. So, when in 2047, when we have 60 crore people, can they live a life with respect? Because 80% of those people today don't have an income plan, don't have a health insurance plan. Imagine what impact it will have on the country which doesn't have a social security scheme if this is not addressed today. And that is where AI is helping us."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:57",
                    "end": "00:43:26",
                    "text": " Absolutely, Rajiv. We're going to go to the last question. I think this is a really important question because we all are gathered here to really think about what do we want from this India AI Impact Summit. So maybe, you know, quickly, maybe, Mahesh, we can start from you. What are your hopes and what does success look like for you and for PepsiCo coming out of this India Summit, like one actionable thing that you would like to see? And we'll make sure that the answers are limited to less than a minute so we can round it up."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:26",
                    "end": "00:43:51",
                    "text": " One of the things I'm super excited about what India is doing is really associated with sovereign AI and building the necessary infrastructure at a governmental level, at a policy level, at educational institutions, for enterprises and for industries, and really also trying to make sure it flows all the way down to its citizenry."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:51",
                    "end": "00:44:21",
                    "text": " For us as serving 1.4 billion consumers every day, leveraging this kind of capability is a huge advantage for us because we can piggyback off all of the digital capabilities, the AI capabilities that a country like India is offering and thereby ensuring that we are able to have the necessary fairness, equity, equality, but most importantly, welfare and happiness for all. Thank you so much, Mahesh. Carolyn?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:21",
                    "end": "00:44:50",
                    "text": " So quickly, one thing that we haven't really spoken about is, of course, we speak a lot about the huge opportunities of AI, but let's be clear. AI also lowers the barrier to entry for cybercrime. Think of how easy it has become now to generate deepfakes. Anyone can do that in their garage, right, or in the bedroom. And then how easy it has become to trick people into sending money or data to fraudsters. And the threat is only amplified now with agentic and new developments in AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:50",
                    "end": "00:45:20",
                    "text": " So if you want to stay one step ahead of cyber criminals, we must join forces, we must act together, and we must be able to share data intelligence across borders and across the public and private sector. That is absolutely mission critical. And so if you want AI to continue to be safe, we must combat fragmentation of standards and data silos because it only benefits cyber criminals. And so that's what I hope to achieve. Well, thank you so much. Rajiv and then Fabrice, bring us home."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:20",
                    "end": "00:45:44",
                    "text": " So as I mentioned to you in FinTech and InsureTech, for us, AI risks are amplified because decisions affect financial futures. Outcomes are obviously asymmetric. Errors compound socially. So for us, ensuring that responsible AI with highest governance standards is very crucial."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:44",
                    "end": "00:46:07",
                    "text": " It is early days for India, but India is growing fast as far as AI is concerned. Today, we use a mix of open AI and local AI to be able to build what we are trying to build across our businesses. To conclude, I would say financial services taught us to be cautious. AI first companies taught us to be systematic. Thank you so much, Rajeev. Fabrice? Yeah, for me, that is."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:07",
                    "end": "00:46:31",
                    "text": " Blueprint adoption and leadership. So if we could define a blueprint for responsible AI in the global South, taking into account all the contextual challenges that we need to address, making sure it's safe, making sure it's trusted, but making sure we respect basically all the local culture and requirements is very important to define this blueprint."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:31",
                    "end": "00:46:54",
                    "text": " adoption because if we don't adopt AI we don't get enough data points to improve AI and we don't make AI safer and more trusted so that's very very important and leadership on the global stage where we see a need for leadership in responsible AI so if we could wish that this place and India could be a leading place for the global science this would be great."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:54",
                    "end": "00:47:23",
                    "text": " Wow, Fabrice, that was exactly what I was going to say, but I'm just kidding. But I think that's a really good summary, like blueprint, what is trustworthy AI in India, but Global South really mean. I think this summit has such a great opportunity to make that happen. What does adoption at scale look like? Because it will need all of us to make sure responsible AI is embedded. And I think leadership also is responsibility. So how are we making sure, to Carolyn's point, we have the right standards and right guard?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:23",
                    "end": "00:47:47",
                    "text": " becomes really critical to make sure that our builders are not just building with AI, but they are doing so with trust and responsibility central to it. So with that, please give a big round of applause to all the panelists. Thank you so much."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:47",
                    "end": "00:48:15",
                    "text": " Thank you so much to all of our panelists and Navrina for such an insightful discussion. It was great to hear about how global practices are being employed here for local impact in India and across the global south. We heard some really great examples of AI use across the value chain and we hope that we've left this audience with insights that you can take with you as you drive the AI transformation journey in your enterprises. Now we're going to do a quick ceremony."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:16",
                    "end": "00:49:36",
                    "text": " of mementos for each of our speakers. So if we could just one by one have each of you. So Fabrice, we can start with you. Caroline, Magesh, and Rajiv. Thank you all so much for joining."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
    {
    "_id": "6994490225c9f4e0d7e7083f",
    "projectId": "303248",
    "projectName": "AI for Secure India Combating AI Enabled Cybercrime Deepfakes Darkweb Threats and Data - Spanish",
    "projectDescription": "AI for Secure India Combating AI Enabled Cybercrime Deepfakes Darkweb Threats and Data - Spanish",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T10:55:02.672Z",
    "projectUpdatedDate": "2026-02-17T10:55:02.672Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T10:55:02.672Z",
    "projectDeleted": false,
    "projectDataLength": 3,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "AI for Secure India Combating AI Enabled Cybercrime Deepfakes Darkweb Threats and Data.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "AI for Secure India Combating AI Enabled Cybercrime Deepfakes Darkweb Threats and Data.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Spanish"
},
{
    "_id": "6994487f25c9f4e0d7e7082b",
    "projectId": "518481",
    "projectName": "AI for Secure India Combating AI Enabled Cybercrime Deepfakes Darkweb Threats and Data - Hindi",
    "projectDescription": "AI for Secure India Combating AI Enabled Cybercrime Deepfakes Darkweb Threats and Data - Hindi",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T10:52:51.328Z",
    "projectUpdatedDate": "2026-02-17T10:52:51.328Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T10:52:51.328Z",
    "projectDeleted": false,
    "projectDataLength": 3,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "AI for Secure India Combating AI Enabled Cybercrime Deepfakes Darkweb Threats and Data.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "AI for Secure India Combating AI Enabled Cybercrime Deepfakes Darkweb Threats and Data.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Hindi"
},
{
    "_id": "6994441225c9f4e0d7e706fe",
    "projectId": "905827",
    "projectName": "AI for Secure India Combating AI Enabled Cybercrime Deepfakes Darkweb Threats and Data - English",
    "projectDescription": "AI for Secure India Combating AI Enabled Cybercrime Deepfakes Darkweb Threats and Data - English",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T10:33:58.538Z",
    "projectUpdatedDate": "2026-02-17T10:33:58.538Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T10:33:58.538Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "AI for Secure India Combating AI Enabled Cybercrime Deepfakes Darkweb Threats and Data .mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771324391832.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:27",
                    "text": " If you want to be part of the music, A play very significant role. So which of the domain just you take, you know, A play very vibrant and significant role from, you know, the end to end. On the current trends, if you see this close to one trillion from the economic standpoint, which is contributed by the IT, global economy, which is coming over there actually. From the estimated from the India standpoint, close to 20 percentage."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:27",
                    "end": "00:00:53",
                    "text": " A, going to contribute in the bigger way. It's a 20%. It's bigger, right? I mean, in terms of the GDP growth for the country. And if you take us as an organization, from the organization standpoint, from the layman standpoint, then just 10 years before, there is a demonetization, right? Where we are crunch on the money. The people are looking on the UPI. The UPI payment, like you know that, right? It reached the tender coconut side."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:53",
                    "end": "00:01:22",
                    "text": " who is selling the tender coconut at roadside he wants the technology enabled and then what are the money transactions even fraction of you know the you know PISA's where we are able to transfer and get things so there is adoption change but however it's a transformation is it gone into the bigger wing instead of the credit card debit card and then online banking or the UPI transaction is bigger and then wider now this wave it may not be the wave right it's a transformation I can say the wave will fall down right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:22",
                    "end": "00:01:41",
                    "text": " But the AI may not. It's absolutely the transformations where we are looking, where we are going to add value in the bigger ring. And if you see this IT industry standpoint, like always, when the AI came, of course it was at more than 20 years, but it's when come to the layman, that's what it is a bigger impact it is coming to, right?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:41",
                    "end": "00:02:07",
                    "text": " opening a chat cpd coming to the pictures everybody start to look like what exactly I want to do he and Then from that just I can you know get things as a extract as output I may not require to put my brains down actually on the specific area so from the in a cyber standpoint like Close to 40 to 50 percent of the organizations already be opted how we can detect and prevent by using the a component"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:07",
                    "end": "00:02:28",
                    "text": " But at the same time, the hacker also used the same A component, how they can, in terms of do the complete harvesting of what I think just really wanted to get things done, right? That's like other side, it's going on the biggest step. So we have a great experience. I can say it's a legend. I worked with all the leaders."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:28",
                    "end": "00:02:56",
                    "text": " and really we want to hear from you know their perspective like what exactly you know aim mean for a technology as a layman and then as organization and what are the crime just where we have handled from you know the end to end so just would like to you know start from you you know specifically the people are thinking they is it's for the technologist that's for the business you know not for the you know layman"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:56",
                    "end": "00:03:26",
                    "text": " and as well as when A way you know came into the pictures just where we are thinking this is an angel but of course from cyber side we have a devil also inside right so really love to hear from you there is for whom I mean it's for the layman or maybe it's for the technologists or for the business like how this weapon is going to be used on that well I think that when a new technology comes cyber criminals are the you know the negative people"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:26",
                    "end": "00:03:55",
                    "text": " they are the first to exploit any technology. And we always say that cyber criminals, at least in their domain, they are a step ahead than the police or law enforcement. It's a fact, you know. So for example, AI came, and they are the first one to exploit this AI technology. How they are exploiting this AI technology, they are creating defect video, they are creating synthetic video, synthetic image. And through that, they are, you know, committing digital arrest and sextortionist scam."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:55",
                    "end": "00:04:18",
                    "text": " Every scam is being operated now. It's IVR enable You know they have integrated AI in such a way that if call is coming from that is automated call But the call is coming from it. It appears to be that it's a genuine call. It's coming from bank It's coming from any insurance service provider. The fact is that is completely automated. They are using AI in such a way and that's why"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:18",
                    "end": "00:04:47",
                    "text": " It's just one month back, the figure was announced in parliament that 22,000 crore rupees in 2024, in one year, 22,000 crore rupees was lost by general public in India. And more than 26 lakh cases reported. And earlier the cyber criminals used to be localized, you know."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:47",
                    "end": "00:05:15",
                    "text": " then they organize at the district level, then at the state level, and then at the national level. Now they are internationally connected. They are operating from a huge cyber scam at the border of Thailand, Myanmar, and India, you know. Huge scam compounds. I mean, multiple stories, and they are operating from there. They are targeting Americans, Canadians, Indians, you know, and they have recruited people from the social media."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:15",
                    "end": "00:05:44",
                    "text": " based on their profile, based on their native. If they have to loot someone from India, they will give this task to Indian. If they have to operate something inside the USA, then they are recruiting from there and they are putting them in a scam compound, they are seizing their passport and then they are getting them worked as cyber criminals. Earlier they were recruited for some good job, you know."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:44",
                    "end": "00:06:05",
                    "text": " some IT work, some coder, some cyber security guys. But once they land up there, they are pushed inside the scam compound. And they are compelled to commit the crime. So this is the way the cyber criminals are operating. And now it's very worrisome that internationally they are connected."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:05",
                    "end": "00:06:25",
                    "text": " They are creating their own land somewhere. They are operating. A victim is somewhere else. Money is being withdrawn somewhere else. A negotiation is happening with someone who is in other parts of the world. So now it's very difficult for law enforcement agency and police to operate and successfully crack."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:25",
                    "end": "00:06:51",
                    "text": " Police, they don't have expertise. They don't have tools. They don't have a resource. They don't have international cooperation. It's not only technical. It's not only crime related. It's technically and legally very complex to operate. So now the police is facing the biggest challenge with the introduction of AI. The biggest challenge for law enforcement agency of police is that how to track the criminal."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:51",
                    "end": "00:07:15",
                    "text": " No, not necessarily sir, like specifically the last note what you have left with us, how do we track this, you know, criminals? Because earlier if you take a robbery, physically the thief will be present on the bank, around the bank. Or maybe if you wanted to, you know, physical attack, that's like always weak, there is a physical visibility of the, you know, the human being, right? Now it's like all digitally connected."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:15",
                    "end": "00:07:42",
                    "text": " more internet and then more threads. Now it's like more A coming to the pictures. It's easy to morph it like I'm from India. Another five minutes I may access from a different country. So it's not so easy to connect and take it forward. So with this note, really, I wanted to hear from you. You had shared this in terms of the real-time experience of what exactly the challenge is, so where we have it. But just the other side, really, we want because just you crafted a lot of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:42",
                    "end": "00:08:09",
                    "text": " Policies and then enforcement, you know, we love to hear The practical stuff like if you see the last year to this year that I foresee is announced like a close to 200 percentage of you know The cyber fraud has you know, the increased even now just it's going up in terms of the a site, right? You've grafted the policy and process and then guidelines for the country level like how you see in terms of the enforcement at ground level If that isn't fulfilling know some extent we would have avoided"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:09",
                    "end": "00:08:41",
                    "text": " Or maybe still there is a gap between the drafting policy and then enforcements on the ground level. So first of all, I'm Rakesh Maheshwari. I was heading Ministry of Electronics and IT, the cyber law, cybersecurity, and data governance side. See, the role of government for any policy initiative is to see that one, like AI for all, you could ever reach good usage, but simultaneously to also ensure"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:41",
                    "end": "00:09:11",
                    "text": " the law and order that the harms to the individual, harm to the society, harm to the nation does not happen and therefore the policies are written in such a manner that we actually require or want that the digital economy and digital platforms continue to comply and in a normal scenario when they are complying ideally government should not intervene at all but when things go wrong"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:11",
                    "end": "00:09:40",
                    "text": " once again at any level the law should be such that one it creates deterrence it creates possibilities of detection it creates maybe through the platforms possibilities of educating the users creating accountability of the platforms and finally when things go wrong as to be sort of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:40",
                    "end": "00:10:09",
                    "text": " governance function comes into picture and as to how exactly it can adjudicate or maybe do something about whatever wrong has happened. Particularly in the context of AI, while lots of use cases, good use cases are there, in fact, including in crime prevention, fraud detection, the fact remains the ease with which things can be faked, defects can be created and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:09",
                    "end": "00:10:36",
                    "text": " because of the very nature of the internet that the things go viral and hence the extent of damage that can happen is immense commensurate actions therefore have been initiated by the government and maybe subsequently when maybe as and when i am required to talk i'll talk a bit about the it rules 2021 which has been very recently just about 10 days back has been"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:36",
                    "end": "00:11:04",
                    "text": " Really I wanted to you know come back on the specific those area like tier 1 to tier 3 of course subsequent discussions we can have it so not just as Saar had mentioned in terms of the policy creations and the reinforcements at the ground level really just I want to hear from awareness standpoint right I mean we cannot measure the sample from tier 1 cities what exactly they have taken over there right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:04",
                    "end": "00:11:20",
                    "text": " It has to go and reach on the bottom level of people. Of course, it is going and reaching But just I want to know like where we are now I mean in terms of the awareness level specific to the a enforcement how the weapon can be used, right? So allow to hear from you know the your side"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:20",
                    "end": "00:11:49",
                    "text": " Okay, thank you very much. First of all, I sincerely thanks the organizer and Cybercrime Research Foundation was given me opportunity, the Research Foundation for sharing this stage and especially with the dignity. See, I'm a Dr. Sapna Bansal, Professor Shriram College of Commerce and Director Institutional Outreach. As a point of education, I feel there is a difference between theoretical and practical."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:49",
                    "end": "00:12:18",
                    "text": " When we talk about cybercrime laws, rules and regulation, the picture is in a different image. But when it comes to the practical aspects, it's altogether different because the police station, they are not equipped with the machineries and technology. And at the same time, we don't have an ecosystem. We have UPI. We have a lot of things. Right now, we are doing digital payment. We have dizzy lockers. We have Aadhaar card. But yes,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:18",
                    "end": "00:12:45",
                    "text": " Once again, this ecosystem creating a lot of problem as well because cyber attackers, they know how to leak it. And you know, it was a time when we say a picture can define a thousand of words and that can be easily, you can recognize, but right now you can say a fake video can change the liar voice. So things are changing day by day."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:45",
                    "end": "00:12:57",
                    "text": " and as a point of view of adjugator I can say that there has to be some coordination between government, adjugator and yes of course police"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:57",
                    "end": "00:13:26",
                    "text": " systems, rules and regulations. When the rules and regulation will match with the education sector and they will create awareness, maybe this problem can be solved. You better know a man in a Kanpur recently lost a one lakh rupees, a fake AI video from someone pretending to be a relative. And at the same time, the second case recently happened, you know, the senior citizen in Delhi paid money after listening the voice of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:26",
                    "end": "00:13:55",
                    "text": " child, okay, he stepped somewhere. And at the same time, few more cases, a tech worker in Bangalore almost lost 1.5 lakh rupees in, you know, recently this kind of a scam. So, so many things are taking place like hacking tools, fake identity, malware, stolen data, phishing emails and fake loans. Every day we are getting a lot of, lot of calls and it's very, very difficult."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:55",
                    "end": "00:14:21",
                    "text": " defects, creating a lot of problem even current scenario. So we should understand the things is changing only awareness, alertness and third very important part is that educate them in a very, very specific ground level. So that can solve the problem as the best of my knowledge. Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:21",
                    "end": "00:14:48",
                    "text": " Tarun just, you know, following up on Sapna, I really just want to know, she has touched upon the deep fake, you know, which has created a lot of scam. I mean, of course, it's not, you know, the illiterate, right? It's like a well-educated people just they are losing in terms of the money from, you know, the in-to-win. So really, we want to know, this is one is on the AI, yes sir, glad to have you, sir."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:55",
                    "end": "00:15:18",
                    "text": " So we want to know like, one is with the A, how we can prevent in terms of the incident. Second one is how I can detect the incident. And third one is, once the incident has happened, really just I want to know like, it's like a real or the fake, or how it is, and then from where it is. And just we want to know the route of this incident so that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:18",
                    "end": "00:15:47",
                    "text": " Furthermore, any incident come just can be avoided with the right set of the mitigation step, right? So allowed to hear from you like in terms of the forensic side, like what are the key steps, how the A is playing significant role in terms of detection as well as the preventions. Thanks. Thanks, Mr. Nathan. So let me touch upon the problem in brief before I talk about the solution. See, if we talk about a state like Delhi, we have a population of 1.52 crore people. We have a police force of 1 lakh 4,000 people."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:47",
                    "end": "00:16:17",
                    "text": " There is absolutely no way you can monitor 1.52 crore people with a force of 1 lakh 4 thousand I am pretty sure that 50% of the people in this room would have faced an incident That you would have lost your phone or maybe your wallet when you went to the police to complain it complain about it No FIR you would get a lost and found complaint and be done with it You may get a message on your phone saying that some inspector has been allocated for this case four days later You will get a message that the case has been closed"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:17",
                    "end": "00:16:44",
                    "text": " And the reason for that is that the entire department is overloaded with cases. The third problem as Dr. Sapna also mentioned, Triveni sir also mentioned, the guy who has joined the police force, you know, if he really wanted to become a technologist, he probably would not have joined the force, right? I mean, I'm not saying that he cannot become a technologist, but that is not his motto. That is not his career. It does not give him that zest for life, right?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:44",
                    "end": "00:17:10",
                    "text": " He wants to go to the field, he wants to catch criminals. He doesn't want to spend time learning a new technology and the technology is changing at such a rapid pace that he will have to learn a new technology after every 1-2 months and start using that technology. So if you look at the problems surrounding our police forces, they seem to be unsurmountable, right? Now the solution to this, what is the basic premise behind a crime?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:10",
                    "end": "00:17:40",
                    "text": " If you look at it, every crime, if it is not a white wolf attack, it's a network crime. You talk about somebody stealing your phone. At the end of the day, he has to sell that phone and make money. You talk about somebody stealing your wallet. You know, it's a network. Somebody, you know, you have all these money laundering is a network. You have narcotic, which is a network. These are all network crimes. How do you identify a network? How do you start building the patterns to identify a network? And that is where AI can help. See, I'll take a few examples."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:40",
                    "end": "00:18:09",
                    "text": " Bharatiya Nyaya Sahita has come out with a guideline that today if a person is arrested, you need to do the forensic image of his digital devices, laptop and mobile phone. Because literally everything that we are, our life is defined by our digital devices. If I get access to your phone or laptop for maybe one hour, I would be able to figure out 90% of what your life is and vice versa. So Bharatiya Nyaya Sahita came out with a guideline that when you arrest someone, you have to do an image."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:09",
                    "end": "00:18:33",
                    "text": " clone his mobile phone and laptop and keep that data. Now, that is huge volumes of data. Just to give you an example, with income tax investigation units that we work with, when they go on a simple rate on a single company, they get data in tune of 80 terabytes of data. How can you analyze 80 terabytes of data? And that is where AI plays a huge role in forensics analysis."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:33",
                    "end": "00:19:02",
                    "text": " So you dump all the data onto an AI engine and let the AI start trying to find patterns within that data. Imagine the kind of pattern that you can identify. I'll give you a few examples, some of the use cases that we have done ourselves with revenue intelligence, GST intelligence. Just using pattern recognition, a very small AI model, on the entire GST data of the country, we were, as per their own estimates, able to identify possible fraudulent GST transactions worth 5,577 crores."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:02",
                    "end": "00:19:21",
                    "text": " And this was just a three to four months effort. So that is the role that AI can play in identifying patterns in huge volumes of data and identifying these networks. So I think now, Nathan, I think I answered your question. Yeah, absolutely. Thanks a lot. Thank you so much. See, fundamentally,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:21",
                    "end": "00:19:50",
                    "text": " you know doing the same activity it's like a one go if you have the same pattern if the a will help you to detect and then give you this accuracy in terms of what went wrong and then how we can detect and then you know the preventive order actually so thank you so thank you so much for joining see from the your rich experience you know you you would have handled a lot of you know cyber cases and really just we love to hear from you in terms of the how the a can help you to hand"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:50",
                    "end": "00:20:22",
                    "text": " in terms of you know the fraudsters or maybe in terms of the threats because now this other side what people are using is AEs can be used to do the fraud right and just I'm looking on the both aspects like how you have utilized or how you have a plan of in terms of using a component to detect or to hand the respective you know the fraudsters. First of all I will want to reframe this question"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:22",
                    "end": "00:20:52",
                    "text": " because, you know, being a counsel, I am not an investigator. So I am not here to detect. There are times I am, in fact, mostly I am representing the accused, the condemned man, you know, because with all the presumption of innocence, proof beyond reasonable doubt and all of that. So I would want to, you know, speak of an ideal legal regime that I foresee with AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:52",
                    "end": "00:21:20",
                    "text": " technology, the policemen are there, the judges are there, all that is happening, the accused is there, the victim is there, the complainant is there. Now I, the few points that I come to my mind, first of all, privacy is very important. Right to privacy. It's a fundamental right, recognized by the Supreme Court in Puttaswami's case, flows out of Article 21. So privacy cannot be compromised, first of all."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:20",
                    "end": "00:21:50",
                    "text": " Now as an exception to the fundamental right to privacy one of the exceptions is Crime investigation crime prevention terror prevention terror investigation and prosecution. So that's an exception So first of all, we must respect the rights of people the privacy of the people as an exception We must have enough technology law enforcement"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:50",
                    "end": "00:22:20",
                    "text": " to prevent criminality, cyber criminality. Preventive. Prevention is better than cure. And as my colleague said on the dais, it's next to impossible to investigate all crime that is happening. So many people come to me and they say, ye ho gaya, aise ho gaya, waise ho gaya. You know, when a fraud is less than say five lakhs or ten lakhs, I tell them, just forget it. Nothing will happen."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:20",
                    "end": "00:22:51",
                    "text": " Why why why should good money chase something which you will not get back? So one is crime prevention at the level of crime prevention I would say that people themselves have to be cautious first of all be careful in your dealings. That's one Now an incident has happened Then I would say from a victims perspective. What does the victim want? First of all, he wants a redressal. He wants his money back"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:51",
                    "end": "00:23:18",
                    "text": " That's one. And secondly, he wants the damn guy to be prosecuted. He wants that guy who has committed the crime on him and made his life miserable, he wants that man to go to jail. So effective prosecution. Effective prosecution is where the police comes in, law enforcement comes in. Now you know, I hear or I have heard people saying that for example in the UK,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:19",
                    "end": "00:23:50",
                    "text": " Small time thefts up to say 10,000 pounds. No cognizance is taken. That means we can do more less than 10,000 crimes. So what do you do? What do you do? So then therefore the police has to investigate the crime properly, prosecute the offender. That's the another objective. Of course, AI plays a role over there. AI plays a double edged weapon. AI will be used as a cyber crime tool also."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:50",
                    "end": "00:24:18",
                    "text": " as a weapon also and to counter it also you can use AI, both ways. So therefore I will want a legal regime where the correct accused is found and prosecuted. Now I am saying correct because the internet has layers of service providers. It's very, very difficult to pinpoint the originator of a particular fraud. So I will want, you know, the man who has committed the crime to be booked."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:18",
                    "end": "00:24:44",
                    "text": " and then last but not the least where I come in the innocent man should not be implicated you know we have we still I must say after 33 years at the bar and as a defense counsel I must say that we still have a very colonial police force in the country I'm not saying it the Supreme Court has said it time and again we still you know the whole police system"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:44",
                    "end": "00:25:15",
                    "text": " Triveni Singhji is perhaps smiling and with a smirk that the whole police system, they start working from a confession and they work backwards. So we need a fairer system from the perspective of the accused also. Fortunately, we are a country where there is rule of law. Fortunately, we have an independent judiciary. Fortunately, we have the rights of the man who is arrested."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:15",
                    "end": "00:25:44",
                    "text": " We have the victim also who is now being looked at with some sympathy in the criminal justice system We are moving in the correct direction But of course systemic problems are there the number of policemen to the population number of judges to the population All that is there now. You can't have you know a policeman? Protecting every Indian citizens so some criminality will obviously go unpunished, but I hope"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:44",
                    "end": "00:26:11",
                    "text": " that most of the criminality should not go unpunished and innocence should not be implicated. And if they are, then I am there to look after the details. Thank you. Absolutely amazing, sir. Very impactful input. And that's what we are bringing here, sir. I mean, you may not have to have a proportional of populations and police. I mean, we can deduce that, actually. And then from there, we can add values. Amazing. Thank you so much for the great input."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:11",
                    "end": "00:26:40",
                    "text": " so just coming I mean from lawyer to police right so just I wanted to hear couple of stuff right when we debate and the cyber security you know side we see so that's like a national threats I mean what are the a coming over there of course the valuation is there but however we we feel that is like a national threats since so you know I have seen you like from the long time you have handled in terms of the crime"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:40",
                    "end": "00:27:08",
                    "text": " where the people to the smuggling and then, you know, other cybercrime just on the physically. Now you are into the digital area also, like in terms of the handling all the component and then how needs to be into that from, you know, the end to end standpoint. So just really just I wanted to hear from you like a both of aspects from police to the, you know, in terms of handling this a like a one of the prevention stuff where we can take it forward to prevent all the crimes. Love to hear your input, sir."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:08",
                    "end": "00:27:38",
                    "text": " So when we were selected for the police, you know, the focus was completely on swimming, horse riding, rifle, AK-47, you know, run 10 kilometers in the morning. So when we used to go to class at 10 o'clock, 90% of the girls used to sleep."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:38",
                    "end": "00:28:07",
                    "text": " Because from morning to evening, it's all physical, physical, physical."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:07",
                    "end": "00:28:35",
                    "text": " But this cybercrime is completely a different domain. Cyberforency is a completely different domain. So what we think, by giving two, three days training, we can make them smart, cyber smart. No. It's complete science. You will have to give at least one year, two years, three years rigorous training in cyber security tools, cyber technology, cyber forensic tools. And then this all big data platform, it's all technology."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:35",
                    "end": "00:28:56",
                    "text": " So for training police personnel, which is not happening right now, you need at least six months, a rigorous course. And only then you can train them, unless and until police understand technology behind the crime. For example, suppose somebody is committing crime through UPI. So unless and until you know the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:56",
                    "end": "00:29:26",
                    "text": " the structure of UPI, how NP, National Payment Corporation works, what is the infra behind RBI, NEFT and RTGC, you cannot, you know, crack the case. If some crime is happening through website, if you don't know how the domain is purchased, how the website is created, how the payment gateway is integrated at the back end, and how the criminals are using mule accounts and mule sim, you cannot crack the case."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:26",
                    "end": "00:29:55",
                    "text": " so that's why what is happening 99.99 percent police they are not well trained in cyber domain they are not well and not only cyber technology they have to be the domain export as well because if you want to investigate a case you should know the technology behind banking you know how this atm network and everything absolutely very technical so and that's why you always say that there has to be a clear cut a different"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:55",
                    "end": "00:30:20",
                    "text": " IT catered in the police department absolutely okay and only then we can solve otherwise it's very difficult for the police to crack this of course the force is ready sir no no wonderful see the current trend nope if you want to hack or if you want to bring down this way infrastructures I may not require to get into your laptop right if I pull down the power of your house"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:20",
                    "end": "00:30:46",
                    "text": " If you pull down the power of your internet, it's more than enough. So ultimately, I want to disturb you, right? I want you to stop what you're supposed to do there. I can stop your water. Like, what are the flow? It comes to your house. You cannot take a bath. You cannot come to office. So there's a supply chain that is coming in the bigger way, right? As well as now, if you wanted to bring those ecosystems, like I think most of the people you are aware, right? The entire power grid has been hacked."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:46",
                    "end": "00:31:11",
                    "text": " People are talking about internet, the power grid has been hacked and then it's like a completely shut down for this entire regions. So that aspect, now I'm finding all the mobile what is being used, everyone hand knows that, that is much more critical infrastructure for the country. Because it's like a weapon, right? As I correctly mentioned, right to be forgotten, like what they can tweet. Once it is done, just you cannot take it back. That's done."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:11",
                    "end": "00:31:39",
                    "text": " It's like, you know, within your friends, within your parents, and within other neighbors, you can squirrel, you can make a fight, but after a few days, it will go off. But the social media, one comment you leave it, that's it forever. It will be there, actually. So with that aspect, I know that just he was a, you know, the foundation, you know, the member of, in terms of drafting DBDB and data privacy and all things and all. So love to hear from you, like how, because A, I need more data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:39",
                    "end": "00:32:07",
                    "text": " data privacy where we should not send the data so allow to hear from your perspective sir yeah so when we talk maybe privacy is a fundamental right therefore when the laws for the country particularly the DP DP act the digital personal data protection act which is which is which has already been enacted which is getting into operationalization as of today the whole aim is collect"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:07",
                    "end": "00:32:38",
                    "text": " only what is really required don't over collect be transparent store it only for the minimum period that is required for the business if you are sharing it with anybody there must be a reason and there must be a control but we say everything for yes yes sir because we won't access the app right so so so in fact as of today probably it's only i agree to the terms and conditions and that's it but tomorrow you will find"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:38",
                    "end": "00:33:08",
                    "text": " you will get to know as to what data is being collected, number one. Number two, in case maybe more is being collected or has been collected, you will have the rights to ask and confirm as to what data has actually been collected, with whom it has been shared. And in case you are actually wanting that this data should no longer be processed further, you will have the right to even get it erased."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:08",
                    "end": "00:33:37",
                    "text": " Lastly, for some reason or another, because it's a cyber security issue, things can always go wrong, breaches can always happen intentionally, unintentionally, that it will be the liability of the organization who has collected, processing this data to report a breach, not only to the data protection board, which is coming up, but to also to the individuals whose data has been impacted. So that means transparency, trust, accountability,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:37",
                    "end": "00:34:05",
                    "text": " and possibility of financial fines which is all there in the system."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:05",
                    "end": "00:34:31",
                    "text": " You know, use, make use of this weapon very responsibly. Of course, there's a lot of framework which is available, but love to hear from you. Yeah, very good question. And yes, everybody wants to know key as a layman, how we can take a prevention because there's a very, very important part. We all literate people actually, you know, come under this trap because we don't know. We always share the OTP like we use the public by five."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:31",
                    "end": "00:34:52",
                    "text": " And sometimes we never know how to use the AI ethically. And AI ethically, it's a very, very important part. We are teaching cybercrime in the colleges and schools, but you know, ethical AI is still lacking. I feel like that the diploma degree and more awareness program can help a lot. There was a one share which was the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:52",
                    "end": "00:35:15",
                    "text": " So very important, the rules and regulation. The rules and regulation and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:15",
                    "end": "00:35:28",
                    "text": " Ethically, we have to make such pathways so that the Jainji can follow. Because if we are not providing the framework, definitely people will suffer. And we are talking about Viksit Bharat."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:28",
                    "end": "00:35:53",
                    "text": " Right? And Vixit Bharat in five trillion dollar economy can be factored if we are not taking a step ahead because it is Italy, we are in power. We are actually super power. But ethically, yes, most of the crimes taking place in India only. You find twenty two thousand crore rupees we have lost because of this crime and every minute five hundred five crimes every minute."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:53",
                    "end": "00:36:21",
                    "text": " every minute, you know, 505 crimes taking place. So it's a very serious matter. And this point of time, government have to make certain rules and regulation. Implementation should be very strict. And the best part is that awareness and yes, awareness, education and at the same time, make the guideline for everyone. And there has to be some two way verification."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:21",
                    "end": "00:36:50",
                    "text": " and ethical aspects. That's it. That's it. I want to say. Awesome. Thanks, Lala. Thank you. See, from where she left out in terms of the awareness play very significant role, right? But even though just as a layman, how I can make awareness is a one. But however, if you take this deep fake and then in terms of the voice cloning, even though you are a SME, you cannot make out like it's a real or it's a fake, right? And what are the techniques? Maybe as a layman, just I can..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:50",
                    "end": "00:37:18",
                    "text": " You know give it stop in terms of just this other video. These are the glimpses it comes There's a fake or this is a real right and because the voices already It's everywhere right if this voice is there with that they can clone whatever they want to do it And if you have my photo and then my mannerism just all things also can be cloned and that can be give it order right? So that aspect as a layman really just I want to understand like what are the key stuff? You know, they should look in terms of the deep fake as well as a wise cloning"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:19",
                    "end": "00:37:48",
                    "text": " So see deep fake is going to become a national security issue very soon, maybe as soon as the 2029 elections Imagine it's so easy to create a video. I mean frankly if you are in hostile country It's so easy to create a video of maybe a deep fake video of a cow getting slaughtered and the Head of the cow in front of a temple or a pig getting slaughtered and the head of a pig in front of a mosque It will lead to diets in specific states in the country. It will lead to diets"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:48",
                    "end": "00:38:07",
                    "text": " That has an impact on your economy. And with social media, it's very easy. It actually cost as little as, and I've seen it live, it actually cost as little as two and a half lakh rupees to make a Twitter trend, the sun rises in the west, go viral. So there was a time where the top..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:07",
                    "end": "00:38:35",
                    "text": " Three hashtags in Twitter. It was the sun rises in the West, right? So you can create a deep fake video, make it go viral on social media. Half of our generation is on social media, more than half. They will look at it and they will believe that this is the truth, right? So this is how big a problem deep fake has become. And another challenge with that is, as of now, and I've evaluated multiple AI models, there is not a single AI model which can detect a really good deep fake generated video."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:35",
                    "end": "00:39:03",
                    "text": " and tell you that it's a fake video. So the only thing that we can as responsible citizens do when you see something on social media or when somebody gives you a call, you have to do the verification of that through multiple other sources. You know, if you see something on social media, if it is actually genuine, it would have been covered by some newspaper media vernacular or mainstream in the country, right? So check it out. Check out search engines. If you receive a call from someone who you think is your brother, sister or whoever,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:03",
                    "end": "00:39:33",
                    "text": " You know, do a cross check. Call them back and figure out whether they are the same people or not. These are basic checks that you can do. Technology, as of now, to detect deepfake is lagging behind."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:33",
                    "end": "00:40:01",
                    "text": " And they can go and execute, report it, it's done, right? So that aspect, so now this is ramping up. You may not require it to be part of any crime, but you can hire the professionals. They come and do the work, execute, they go away. So it's a cyber crime as a service, which came into the pictures. You can pay and know where you get involved, and they also know where connected to the counterpart whom you wanted to make an attack on them."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:01",
                    "end": "00:40:31",
                    "text": " So from the legal standpoint really just I wanted to see us up in a cyber crime or services What are the preventions or have you come across any other cases like that has been you know? Executed from in to win and who is liable whether it is who has executed or who has you know? Hired the people so loud to hear from you Okay We have we have the law in place in fact the law the entire law of criminal conspiracy"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:31",
                    "end": "00:41:01",
                    "text": " has been there since Lord Macaulay's Indian Penal Code of 1860. That if people come together and they agree upon to commit a crime tacitly, explicitly, implicitly, they are liable. So the liability mechanism is there in place. And we have a lot of law that has developed. The law has gone so far. In fact, if you see the Evidence Act,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:01",
                    "end": "00:41:28",
                    "text": " There's a very interesting illustration that if a place was to be bombed and there was a conspiracy between A, B, C, D and E. So it is not necessary for law to prove the agreement of criminality between all the five. So if A is organizing the fund somewhere in the world, B is organizing the weapon or the bomb."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:28",
                    "end": "00:41:38",
                    "text": " or the making. So all by their actions you can prove a criminal conspiracy. So the law is there in place."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:38",
                    "end": "00:42:06",
                    "text": " If we apply it to AI, if I apply it to the internet, the problem that arises, the experience that we have seen, of course, we derive the benefit of all that, all the confusion from a defense perspective. The problem is how to attribute certain acts to individuals, how to identify them. For example, if you have an AI, if you create something, why have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:06",
                    "end": "00:42:35",
                    "text": " Why have five people? Why hire five people? Why not create a drone or create something that is not attributable to me? I have to only create it and put it into the internet. And that drone or that mechanism itself is committing a crime. So the attribution of an act to an individual is getting more and more complex. In the physical world, you could identify people."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:35",
                    "end": "00:43:01",
                    "text": " Whereas in the world of the internet, AI, cyber world, you can't identify people. And then we have problems of jurisdiction, collection of evidence. But the law is in place. Now the law enforcement agencies have to be smart enough to identify these hiring of cyber criminals. Because the best way to catch them is through the money trail."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:01",
                    "end": "00:43:23",
                    "text": " Money trail has become a very effective means of tracking cyber criminality laundering etc etc So there are see law also catches up with crime. It's the law is not you know not law is not so weak It catches up with criminality. Yeah criminals is helping to develop a law right one extent"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:24",
                    "end": "00:43:49",
                    "text": " In fact, if they are not there, I mean, we don't develop on that aspect, right? Just maybe one, you know, 10 seconds follow-up question, sir. Like, since all young chaps are there, sometime we think, like, if you do the crime from India, it's like a straightaway comes and hit on that, right? But if you do VPN to some other country, from there, if you launch, the people will think it may not come to our responsibility. So, love to hear your, you know, 30 seconds of input, sir."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:49",
                    "end": "00:44:13",
                    "text": " Of course this proxy criminals and the Interpol also come into the pictures They do come into the picture, but but you know it's very difficult to Prosecute offenders at that level you what we what we hear what we read in the papers That so and so person is going to be extradited to India. That's only a tip of the iceberg"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:13",
                    "end": "00:44:33",
                    "text": " you know somebody some young lawyer very interestingly he narrated to me that you know why are these you know Nirav Modi's and why are these few people being targeted you know because they have they have defrauded the bank what about those people who have defrauded private individuals private lenders"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:33",
                    "end": "00:45:01",
                    "text": " and they have run away from the country. So these are, you know, problems. Extradition is a problem. Collection of evidence is a problem. Sovereignty of a nation is a problem. So I guess, you know, mutual treaties amongst nations where you cooperate with one another. Interpol is there, but Interpol cannot really take care of all these problems. So we need to keep evolving."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:01",
                    "end": "00:45:31",
                    "text": " You know, from a law enforcement perspective. Absolutely. Thank you, sir. Thank you so much. And it's maybe I may take one more minute, sir. Just 10 seconds of input for takeaway for all the people, sir. Just each one. Just I want to hear from 10 seconds. Just one takeaway for them, like what they have to do and don't in terms of cybercrime as well as with the AI. For this, I think government will have to create massive awareness campaign among the public so that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:31",
                    "end": "00:45:59",
                    "text": " they could know how the criminals are operating, what are their modus operandi. And secondly, government will have to put much more emphasis on capacity building, not only in police, but in judiciary and prosecutors as well. And they will have to provide resources as well. I mean, cyber lab, advanced tools, technology, you know, and they will have to put much more emphasis in R&D as well, R&D, regarding Europe. For example, the current threat is deepfake."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:59",
                    "end": "00:46:24",
                    "text": " So we must focus on defect detection technology. So do we have? No. As Tarun told, so no. Right now, no. The best of the tools available, that can give you only 80% accuracy, not 100%. So it cannot be admissible in court of law. And that's why I think these are the things, if government is focusing, then definitely we will be able to control the cybercrime sequence."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:26",
                    "end": "00:46:55",
                    "text": " Yeah, I'll just like to mention it's not just about cyber frauds. It's also about so many illegalities which do happen in in the cyber world and will also like to say Prevention is the best way Understand greed this is a normal human. I mean behavior so Something which is giving you too much of money or or maybe exploiting your greed exploiting your fears exploiting your emotions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:55",
                    "end": "00:47:25",
                    "text": " Which is what we have to be I mean, which is what we need to overcome in the age of AI Seeing is not believing listening or maybe hearing from your known sources may also not be truth. So So maybe act responsibly before what we say is a zero trust architecture Only one line is stay alert. Once again, I'm repeating it stay alert"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:25",
                    "end": "00:47:52",
                    "text": " Stay aware, stay secure. These are the three fundamental aspects you should keep in mind. Definitely, you can save yourself and your family members. Thank you. I think I have a recommendation for the government, not for the citizens. I think the government of our country, we have to realize that intelligence has to become privatized. It has already become so in US. In US, you do not have..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:52",
                    "end": "00:48:22",
                    "text": " You know, the intelligence is connected by private individuals working in conjunction with the government and the operations are in the hands of the law enforcement agency. India will have to realize, like the income tax departments of the country, they have already done that. They are working with private entities to gather intelligence and they are only involved in the operations. I think all aspects of the government will have to follow suit. Yeah, we all work together. We'll make that. We need a deterrent law."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:22",
                    "end": "00:48:51",
                    "text": " to check cyber criminality. Secondly, no crime should go unpunished. And thirdly, we should not be obsessed with catching hold of the criminal. We must protect the rights of the innocent man, the accused also. The system has to be balanced. It cannot be a one-way street. Awesome. Amazing. And it's a wonderful conversation. We can give a great applause to the entire leadership."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:51",
                    "end": "00:49:07",
                    "text": " And of course, A is a double-edged sword, right? How you handle is a matter, actually. So that's way we can handle with a more responsible and more accountability to ensure that things are taken care with the positive note. Thanks a lot. Thank you so much, and have a wonderful day ahead. Thanks a lot, gentlemen."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
    {
    "_id": "69944a9f25c9f4e0d7e70860",
    "projectId": "625147",
    "projectName": "Deep Tech and AI Investing in India  Current and Future States - English",
    "projectDescription": "Deep Tech and AI Investing in India  Current and Future States - English",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T11:01:56.445Z",
    "projectUpdatedDate": "2026-02-17T11:01:56.445Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T11:01:56.445Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "Deep Tech and AI Investing in India  Current and Future States.mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771326099433.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:28",
                    "text": " I'm Gopal Jain. I am one of the co-founders of Gaja Capital. Gaja Capital is a homegrown firm with 27 years of history of investing in Indian startups and growth capital and growth stage firms. I went to IIT Delhi for an education at an early age. I grew an interest in one of the founding members of IDDA, which is the Indian Deep Tech Investors Association. And we, as a group..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:28",
                    "end": "00:00:51",
                    "text": " are also committed to fostering deep tech investing in India. And we, along with the others, have committed. We have made some deep tech investments. We've committed to make more deep tech investments. What are the other ways in which we connect with deep tech? It's through our giving program. We have a giving program called Gaja Gives. And through that, we've supported, we'll talk about it to universities. And I'm very excited to be here. Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:51",
                    "end": "00:01:16",
                    "text": " Thank you, Gopal. My name is Siddharth Pai. I'm the founding partner, CFO, and ESG officer of 314 Capital. We're a homegrown VC fund that invested in technology and technology-enabled companies with a special focus on deep tech as well. In addition to that, I do a fair amount of work when it comes to liaising with the regulators and the government when it comes to public policy. I'm an executive committee member of the Indian venture capitalization. I'm also the co-chair of regulatory affairs and sit on the apex of..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:16",
                    "end": "00:01:45",
                    "text": " the policy formation body by SEBI when it comes to alternative investments in the country. So I think deep tech is one of those areas that's truly been the alternative of alternatives. We're very fantastic to see it get mainstream as of now. I'd love to actually speak more about what that means, what the policy scaffolding and regulatory architecture should look like and how we can end up taking the handbrakes on Indian innovation and truly allow entrepreneurs' spirits to end up actually being unleashed. Thank you, Sudhart. We just have Nishat joined us. Nishat, please. Yeah, thank you so much. Sorry for..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:45",
                    "end": "00:02:13",
                    "text": " The chaos looks like we are celebrating here. Thank you so much for being present, so many numbers here. My name is Nishid Desai. I happen to be founder of Nishid Desai Association, India-centric global law firm. So we focus on futuristic technologies, look 10 to 20 years from now and try to see what new things will be happening and what would be the future strategy, legal, tax, ethical issues, so more on that side."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:13",
                    "end": "00:02:41",
                    "text": " that we try to shape the future of law. Thank you so much. Thank you, Nishat Bhai. We also have, as Gopal mentioned, this program was organized along with METI by IDTA, Indian Deep Tech Association. We have Sriram and Arun here from IDTA and Celesta. Thank you, Sriram and Arun. I want to say a few words. And Shiv, sorry, and Shiv, chief architect."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:44",
                    "end": "00:03:10",
                    "text": " It's great to see all of you here and this is really interesting to have all our IDTA members and many of their executive committee members coming together on this panel. I want to thank Chris for lending his time and for all his encouragement to what you're all trying to accomplish. Thank you. Thank you. My name is Sudhir Sethi. I'm from Chirathi. We've been investing in India for actually the 20th year and apart from..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:10",
                    "end": "00:03:37",
                    "text": " normal investments, we do deep tech investments also. We've done about $300 million worth of deep tech investments in the last 10 years and very proud to be here intermingling with yourself. Before I start the session, there's a lot of people at the back. Now because deep tech is very young and we have a lot of young founders here and young investors and ecosystem players, maybe, and I overheard Gopal say this,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:37",
                    "end": "00:04:16",
                    "text": " the younger people maybe want to sit down in the front. It's not a bad idea, right? Though it means, but that's normal in a startup, right? So if you're okay, some of the younger members, maybe 20 of you, can come here so that we can interact more at the front. This will be the only session in the whole of AI Summit where people were sitting on the floor. Yeah. Thank you very much. Excellent."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:17",
                    "end": "00:04:51",
                    "text": " Yeah, if the doors can close that would be helpful. Okay while people settle down Again, so let me start we are talking about deep tech. We'll be talking about the three areas the capital reality check the AI and beyond thesis I think in India deep AI is a very big theme right now the conference But we are talking about beyond thesis and robotics in space and other places also policy framework and Last but not the least the role of academic institutions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:51",
                    "end": "00:05:20",
                    "text": " as innovations engines in the country. As a prelude to our discussion, let me give you some numbers. IDTA has tied up with Venture Intelligence and we have somebody from Venture Intelligence here. Yes. We are releasing India's first deep tech report at 3 p.m. which is in collaboration with Venture Intelligence, NDA, as such."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:20",
                    "end": "00:05:50",
                    "text": " And you'll find that's very, very interesting what's coming out of that. But let me give you some highlights not to steal away the thunder as such. Deep tech investments in India since $28 billion worth of investments. It's not that small for India. And if we look at just AI, and again, some of these definitions because they were in the past years may vary a little bit from current definitions, but leaving aside those corrections which have to be made, 12 billion have been in AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:50",
                    "end": "00:06:19",
                    "text": " This includes native AI, this includes all the six, seven layers of AI, and also the enablement of AI in perhaps, as RDF defines it in companies which are in agriculture and so on and so forth. And only in 2025, DPEG investments were about five billion, and in AI was half that, about two and a half. So it's not small for India, it's actually pretty big overall, and by stage, interestingly,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:19",
                    "end": "00:06:46",
                    "text": " 70% was in growth and late stage. So I think there are many initiatives which have been taken in the country and we will discuss some of those initiatives today as how early stage can also become a much, much bigger allocation overall. So with that, let me start off, and as you know I have to change from paper to digital, the conference."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:46",
                    "end": "00:07:17",
                    "text": " The first session is on capital reality check and I'm going to address this question to Siddharth. Siddharth Gopal and Vardhan, how do you think, Siddharth, about capital formation in India as it exists today and specifically if you could comment on the RDIF intervention and the impact of that?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:17",
                    "end": "00:07:46",
                    "text": " And then I'll go to the next panelist. Deep tech has always been one of the, it's always been a focus sector for a number of Indian investors, but I don't think it's been the primary bulwark or the nucleus of their entire portfolio. Because what happens, a number of deep tech companies, as you know, very long gestation periods, higher risk, much higher risk that ends up actually entailing. There's a single product risk or a single feature risk that's actually inherent in the business model itself. And even when it comes to the acquisition framework, I don't think we have as many active acquirers of IP in those particular areas of what existed compared to what we're seeing as of now."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:46",
                    "end": "00:08:07",
                    "text": " So what we have actually seen, especially of recent times, is three broad changes. One is the LP appetite, which has now actually become a lot stronger for longer gestation products, for fund managers, domain actually spending 15, 20 years. Because the biggest issue that people had with Indian venture and Indian VC was the fact that exists the mistranslation of paper gains to actual hard cash."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:07",
                    "end": "00:08:28",
                    "text": " I think what we've seen in the IPO market as of now has actually shown that Indian assets can find buyers in the public market as well. They can perform exceptionally well over there. Therefore, if they've actually seen this happen with the initial thrust of technology-led companies, there's a lot more appetite to end up actually going deeper into the entire stack as of now. That's number one. The second thing we've also seen, and this is something that the RDR fund has actually done."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:28",
                    "end": "00:08:52",
                    "text": " The bridge between research to actual development and commercialization, I think, has actually become a lot stronger as of now. Previously, a large amount of intellectual research would actually be done over here, but developed and actually researched out in the West. We will talk about the university systems, the endowments, and what that entire ecosystem ends up actually looking like also over there. That research part has now actually become a lot stronger because funds like the RDI and initiatives by departments like DSD and METI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:52",
                    "end": "00:09:17",
                    "text": " and actually allow these researchers to start commercialization of their structures and to actually give them the support that they need. The third is actually the maturity of the entire Indian startup ecosystem. Previously, the founder was a jack of all trades where they would have to end up actually being the chief scientist, the product, the CEO, the product engineer, the chief compliance officer, the chief accountant, the chief hiring officer, the chief marketing guy. Everything needs to end up actually happening."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:17",
                    "end": "00:09:38",
                    "text": " But now we've seen a lot more bend strength where quality talent can end up actually going far deeper into what they have to, and the support system actually exists. And even the VCs as well are no longer just the providers of capital. Each of them have as a platform team. They give a large amount of support. They help you to end up actually solving those problems. So those are the three changes we've seen from the last 10 years till now. And the RDI has now become a nucleus."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:38",
                    "end": "00:10:07",
                    "text": " around which a large amount of capital formation can happen by being up to 50% of the funds by giving grants over to research, to development across the various TRL stages that actually exist. We've actually seen a government program that is acting far more, far more like the private sector than the private sector in terms of actually envisioning. And I think that is something that we're grateful for. But as usual, because this is deep tech, it's a long gestation period. I think everyone's just waiting for the one lakh crore to get spent in like a year and a half is not going to happen. It will take five, six, seven years to happen and the results will take another decade after that to come."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:07",
                    "end": "00:10:35",
                    "text": " But I think this initial enthusiasm translating to actual investments and then translate to exits and outcomes, that is something that I'm actually looking forward to the most. Thank you. Maybe I can ask a different question to Gopal. Gopal, two questions to you. One, what do you see the role of domestic capital, non-RDIF, and how do you think that is propelling itself in the country? And second is, RDIF is..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:35",
                    "end": "00:11:03",
                    "text": " which in today's currency is about $11 billion. And as I think Sadat mentioned, we are talking about 50% coming from private capital. To supplement that, that's an essential key regulatory requirement, which basically means that whether it is a 10-year cycle for disbursement or a 15-year cycle, and deep tech requires that long cycle, private capital needs to be raised by fund managers and companies about $22 billion."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:03",
                    "end": "00:11:32",
                    "text": " How do you see that happening in the country? Over to Gopal. Just to give some perspective, I mean, about, say, 25 years ago, Indian venture capital and private equity was about a billion dollars a year. Everybody still talks in billions, so I'm going to stick to that. And today, the industry has grown to about $40 billion. So for the entrepreneurs in this room, there's $40 billion annually that investors invest in, say, the startup ecosystem."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:32",
                    "end": "00:11:58",
                    "text": " India's deep tech investing is what, roughly about $2 billion a year now, right? So we are where, it's a very good number. We are where the entire venture capital and private equity ecosystem was saved 20 years ago. We have a long way to go. China's at about order of magnitude, say 100 billion, 80 to be exact. The US is at 150 billion. So how do we get to those numbers?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:58",
                    "end": "00:12:15",
                    "text": " and the framework is private sector, public sector. Public sector and policy have a huge role to play, especially in the initial stages of any sector when there is market failure. So right now the market isn't there. To support deep tech investing, that is the long and short of the answer, it has to be developed."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:15",
                    "end": "00:12:43",
                    "text": " but some very useful initiatives, very important, path-breaking initiatives. These remind me of what has happened in other countries like the U.S. and China. And also what happened in India when, you know, in the early years of the republic, these initiatives like RDIF, National Semiconductor Mission, the funding for, you know, the Anusandhan Foundation, the Quantum Mission, all of these are very important initiatives which are looking to address market failure."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:43",
                    "end": "00:13:12",
                    "text": " They are looking to provide a source of capital, both profit-seeking and grant-based source of capital to fund deep tech investing either directly or through managers. You have to make a start. We now have the catalyst in place. Governmental capital is not a substitute. It's meant to be a substitute. I think it is upon us now to raise this money from the Indian privates. Unlike China, we don't have the..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:12",
                    "end": "00:13:35",
                    "text": " I mean, the global image. It's very difficult to sell India. It's even tougher to sell Indian innovation and deep tech investing to global investors. Coming to Indian investors, again, we have our task cut out, but these initiatives are a starting point. We'll have to work harder. This is not capital that distributors can find for us."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:35",
                    "end": "00:14:04",
                    "text": " We'll have to come up with structures. We'll have to, first of all, success stories to start the virtuous cycle because nothing succeeds greed. Greed is good, right? So nothing succeeds like greed and then we have to find those pools of capital in the private sector. We have to find them from the public sector. There is a lot of headroom there. A lot of public sector entities haven't used up their headroom for alternative investing but it's going to take a few years."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:04",
                    "end": "00:14:34",
                    "text": " for us to be able to say that, listen, if RDIF, actually what I'd like to say is if RDIF is doing, say, 20,000 crores a year, then ideally the private sector should be doing at least 80,000 crores a year, right? For us to get to that level, it's going to take a few years for us to find that capital. It's going to be a lot of hard work. Thank you, Gopal. I'm going to ask the next two questions. Chris, if I may ask because you've been a mover and shaker in deploying Indian capital."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:34",
                    "end": "00:14:58",
                    "text": " and then I'm going to go to Sneh and then come to Ravi. Is that okay? Chris, any thoughts on Indian capital because that's very key so far and how do you supplement, how do you get Indian capital to come in at a bigger scale? So the good thing is the money is there, the wealth is there."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:58",
                    "end": "00:15:23",
                    "text": " It's just that it is not coming into the startup ecosystem. But over the last four or five years, I have seen a lot more family offices, a lot more high net worth individuals starting to invest in the startup ecosystem. I think the push was too hard because they all set up their own small, small AIFs. What we want now is actually larger AIFs which can do bigger checks."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:23",
                    "end": "00:15:47",
                    "text": " later stage etc hopefully that will happen and the RDIF so you know the the backstory of RDIF is we were all many of us you see here we're trying to create a billion dollar VC fund in India we don't have a billion dollar Indian capital VC fund and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:47",
                    "end": "00:16:15",
                    "text": " Then we approached the government to set up a billion-dollar VC fund. They surprised us by saying, why stop at a billion? Here's a $12 billion VC fund. So the government actually took a big step in setting up the RDI fund. They didn't start small. They started with a bang. Now the challenge is, as both Gopal and Sudhir has said, and Sid is also part of the, you know,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:15",
                    "end": "00:16:45",
                    "text": " different groups which are working to streamline this absorption capacity. 20,000 crores of RDF has to invest in a year because this fund has to be deployed over six years. One year is over, so only five years is left. That means 20,000 every year has to be deployed. There has to be an equivalent, at least an equivalent private capital coming in, otherwise that can't be deployed. So absorption capacity and that's why..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:45",
                    "end": "00:17:14",
                    "text": " We need larger funds, India-based, and it may not be 100 percent India-based. The only thing is the AF has to be registered in India. The source of funding can be anywhere, and then we need to find good deep tech companies. This only can be invested in deep tech companies, deep tech companies to actually absorb this investment. So I hope that it's possible."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:14",
                    "end": "00:17:43",
                    "text": " I'm optimistic because I've seen in the last three months, tremendous enthusiasm, momentum. Government is moving very fast. July 2025 was when it was announced. They formulated all the rules, regulations regarding that. They called for startup funds to apply, already opened and closed."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:43",
                    "end": "00:18:09",
                    "text": " This is just the first close. There will be more rounds of that. And the call for startups to apply is already open. And we are hoping that within a period of three months, the money will be dispersed once the application is accepted. And I'm hoping that this will probably catalyze a deep tech revolution in India. Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:09",
                    "end": "00:18:39",
                    "text": " Gopal. I forgot to mention that perhaps the most concrete evidence of capital formation in the private sector has been the initiative by IDTA and IDTA members have collectively committed $2 billion which is not a small amount to deep tech investing from their existing funds. In a three year time frame. In a three year time frame and you have to start somewhere. So there is some capital formation starting to happen even in the private sector that is heartening. No, thank you and you know if you look at the U.S."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:39",
                    "end": "00:19:06",
                    "text": " Deep technology investing was in an environment of capital abundance. China was capital abundance. India so far had been capital paucity. I think the message which is coming up is there's a lot of money which is available even today not going to be available. That's a big change which is happening as such. I just want to take the experience before I come to Ravi of SNE. Now we are all investors and we are talking about capital."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:06",
                    "end": "00:19:32",
                    "text": " started a company along with his co-founders called Meco. What has been your experience both raising domestic and Indian capital? What are the ease of stuff which you think or ease of raising capital now do you see compared to the last five years? Thank you so much. As I said, we started almost 10 years ago. We started in 2015. And we've seen multiple weathers of this."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:32",
                    "end": "00:20:02",
                    "text": " funding cycle so 2015 I think I heard no from almost 500 investors and every VC would say Once you get your first revenue Then you then we talk and we were building a deep technology product We built our own AI models over the past 10 years when AI was not heard of at that point of time And so we spent two and a half years building product with zero revenue like the first 25 percent of our journey was like this So most VC said no"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:02",
                    "end": "00:20:25",
                    "text": " We said we'll build a companion for a friend, which will teach a child. He said, build it, sell it, then you come. So those days were there. Then we saw a whole wave of massive funding up to 2021 with all investors funding front, left, and center, consumer internet funding boom. And then, of course, the deep tech startups were beneficiaries, not direct beneficiaries. But OK, we are putting so much in A. Why not let's?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:25",
                    "end": "00:20:45",
                    "text": " diversify and put some in BV. We were not the prime targets of most institutional venture money at that point of time. Then when the US started collapsing a bit, then after that growth capital started drying up after 2022 as we all know. But then the good thing that has happened in the past two, three years is Indian"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:45",
                    "end": "00:21:14",
                    "text": " family office capital has stepped up in a very big way. On our cap table, you'll see half of it coming in from institutions, half of it coming in from family offices. And for me, personally, more than the capital, what I really enjoy is I interact with every promoter of that family, getting experience. Indian families have built their businesses over 50 years with a lot of learnings, how to navigate India, how to navigate the world. And someone who has a wealth of experience of running a business for 20 years, when he's spending time..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:14",
                    "end": "00:21:41",
                    "text": " with you and he knows that there's a genuine direction that you're trying to take a quantum leap, which is unheard of. We have 95% of our businesses outside of India. We sell in the US. This is not made in India alone. This is made in India intelligence selling globally. So when they connect with you at a very good level, decisions are faster. The investment cycles, they are willing to support with follow-on capital. We've had family offices come for..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:41",
                    "end": "00:22:10",
                    "text": " unheard of before. Family Office had come forward and said, we will price the round and we are there to back you for the long term. And my 2015, 16, 17, 18 investors who came in as families, they say, no, no, we don't want to exit. We'll take it in IPO and we'll be with the company for the next 10, 20 years. So the Family Office ecosystem has come up in a very big manner in India, which has supported, I know, because I'm part of a lot of deep tech forums and I'm speaking on behalf of a lot of founders, this is when..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:10",
                    "end": "00:22:30",
                    "text": " Institutional capital is not available at growth stage. All of them have got very good access to family office capital in the past two, three years, which is just excellent. Indian companies raising money in the US, the shifts have changed significantly and the tone has changed significantly in the past one year. Since the news of public."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:30",
                    "end": "00:22:60",
                    "text": " market exits. If you see the kind of exits that a US company would do they say they have to scale to 350 million and that's a much longer cycle to be qualified at a decent IPO. Here at 100 million dollars of revenue companies are taking exit at 10x revenue multiples. Equivalent companies valuation multiples are much better. So people are preferring Indian businesses from the US. US investors are opening up to invest. This is our we've looked at the data. We've seen the sentiments of interactions and now"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:60",
                    "end": "00:23:23",
                    "text": " equivalent investors coming in from US investors coming into Indian capital is opened up because of public market boom which is happening. So just a question or not a statement out of your 60 million raised the first 50 was from Indian investors. In fact I would say out of 80 million raised"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:23",
                    "end": "00:23:52",
                    "text": " 70 was from India. OK, so that's a very interesting thing because I know we as investors invest in deep tech. 75% of the companies were supported in seed and series A round by Indian investors, including institution funds and family offices. That's a very big thing. And now, with that statement, I come to Ravi. Ravi, Premji is an investor in India and outside India, right? How do you view, and most of your deep tech investing is outside the country?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:52",
                    "end": "00:24:17",
                    "text": " How do you view investing in India, difference from US, and how do you think we can get global investors to come into India into this deep technology and do you think that will become a tsunami over a period of time because we still have to raise another $11 billion in the next six to seven years to do the 50% of RDIF? Thanks, Sudhir."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:17",
                    "end": "00:24:44",
                    "text": " Let me just kind of introduce Premji Invest. We are a family office. So what people are saying or talking about family offices, we are a large Indian family office-driven investor. You must have heard about our promoter Mr. Azeem Premji. So yes, family offices are now starting to look actively in the deep tech space. But before I get specifically into Sudhir's question, I see a lot of young faces here. I want to just introduce"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:44",
                    "end": "00:25:13",
                    "text": " The thought framework as I was coming here for deep tech. Why is deep tech important? And how do I think about deep tech when I am talking to regulators, government, policy makers? And the first thought that came to me was that it is like infrastructure. It is like a horizontal positive or a negative externality that you will have in every sector of your economy going forward. What does that mean? That means that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:13",
                    "end": "00:25:40",
                    "text": " how your deep tech sector is, it will determine the cars you drive, the phones you use, the laptop that you have, how fast is it, what sort of buffering speeds you have. All of those things are related to the health of your deep tech sector. Now, with that as the framework, take a step back and think, how has the government gone about building infrastructure?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:40",
                    "end": "00:26:11",
                    "text": " without getting into technicalities, I can tell you that a lot of push came from the government, from the lending side, from the equity side to relax and infuse capital into infrastructure, build roads, build dams, you know, give all sorts of incentives. Now, in the context of RDIF, what has the government done? Government has said that, okay, if something is ready for out of lab, but is having a funding gap from getting out of lab,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:11",
                    "end": "00:26:38",
                    "text": " to the market. That is where the government wants to infuse capital. And what does RDIF say? RDIF says that 50% will be by the government and 50% is by the LP. So that is the Indian story that the government is doing now. Right? Now, how is the US different from India? It's a very long sort of an answer. I'll just sort of give you two, three tidbits, right, to think about. In the US, of course, $30 trillion economy, very mature."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:38",
                    "end": "00:27:01",
                    "text": " deep ecosystem where the academia is interlinked with the industry, with the government. Meaning what? We are talking about a jurisdiction which invented the internet, which came up with GPS, right? And all of this came up through either defense government funding or some other means of academia, right? So very different ecosystem. When you come into India, and so the startups are very different."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:01",
                    "end": "00:27:31",
                    "text": " Right? So the kind of innovation that happens there is very different, which is what we see. So it is not only about the capital or it is not only about the talent. It is an ecosystem where in an informal setting, expertise is exchanged. Sudhir, if I... That would be the... I would say what catalyzes US investments. I was reading 40% of US family offices invest in deep tech or in related..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:31",
                    "end": "00:28:01",
                    "text": " If they deploy $100, 40% would be within these sectors. India is different. Like my colleague said, where is your ARR? If you go market, first Indians talk about is there a revenue stream? So that, I think, is the difference. But RDIF and various of our VCs and family offices are now stepping into this space because we see an opportunity in India because talent is there, capital is coming."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:01",
                    "end": "00:28:28",
                    "text": " And we see that this now, of course, we'll discuss the other questions later, how policy incentives and other things can develop that architecture and that ecosystem. That's the way I would put it. Just one more thing I'll just add. CalPERS is the California pension fund. Again, I was seeing about 15% to 20% of CalPERS money goes into private equity and venture capital."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:28",
                    "end": "00:28:56",
                    "text": " India and this is one of one of his topics and I've seen him argue this very passionately before the government that imagine if your provident fund or you know your LIC type of money starts coming into this sector it's huge it's huge issue and it's hopefully about to happen yeah so that's that that's where I would draw the sort of when we look at both India US this what we see but we see a lot of opportunity in India because of the talent so do we get it that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:56",
                    "end": "00:29:25",
                    "text": " Now Premji will start investing in deep tech in India. Yes, absolutely. And that's why we are also founding members of IDTA. Gopal, you wanted to make a point. I just wanted to say that, listen, coming back to India, all of you are here. I just want India's innovation moment is here. Innovation requires a long time. Like one of the companies in our portfolio started 25 years ago out of IIM Ahmedabad."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:25",
                    "end": "00:29:54",
                    "text": " And so innovation requires a long time to achieve. It also therefore requires patient capital, long-term capital to support it. Many things are happening here which will completely change the availability of capital for innovation. In this room are potential innovators. And what I'm saying is that, yes, there are many parts of the ecosystem, how universities behave, how investors behave."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:54",
                    "end": "00:30:25",
                    "text": " And we are at the moment of innovation reminds me, I was there in IIT in 91 when India changed. And we took the initial steps in this country to open up entrepreneurship to every Indian. And I think what's happening around us right now reminds me of that moment. That we are opening up innovation to every Indian. It's now no longer around workarounds, which is jugard. It is about boring through to create fresh technologies. And the longest distance between two dots is a straight line."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:25",
                    "end": "00:30:51",
                    "text": " you know, workarounds are always faster. So this is the time for boring through and finding innovation. There is a sea change in the amount of capital available. We can always say it is less compared to where it is in China and the US, but compared to where it was two or three years ago, it's a groundswell. Yeah, absolutely. If I just add one thing to that, just very quickly. I think during the entire summit, a large number of people have heard, you know, the US is doing this, China is doing this. Yeah, it doesn't matter. Et cetera. I think..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:51",
                    "end": "00:31:18",
                    "text": " Here's the most important part. The US is the largest developed nation, so the different league. China has spent 30 years learning from the best, doing a whole bunch of policy changes, building up that kind of asset base and that show base to end up actually doing that. I think, going back to the Cold War, you had these two large elephants that end up actually fighting. India actually truly represents a non-aligned movement. We are actually the middle path, and as Indians have always proclaimed that, we are the middle path that the rest of the world will end up actually following."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:18",
                    "end": "00:31:47",
                    "text": " The chaos that's actually inherent in India, and the frugality that's actually there, breeds innovation. I think the moment you have a large amount of surplus, you never look at optimization. You never look at how things can actually be better. It's just like, let's just throw cash at it. Something will end up actually managing. It doesn't happen that way. The greatest deep tech innovations across the world have always happened from a place of frugality, when they had to end up actually making something better. And that's actually inherent to what a large number of people up here end up actually doing. And some of the larger problems that people faced 10, 15, 25 years ago have actually been solved."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:47",
                    "end": "00:32:14",
                    "text": " capital. We do have a government that's not looking at deep tech and trying to figure out what. There's a joke that some of the ministers have actually cracked when they practically publicly say that the reason why the IT revolution happened in Bangalore is Bangalore is so far from Delhi. Delhi had no idea what was actually happening in Bangalore. That's how it could actually get incubated and that's how it ended up actually building up. But now in deep tech, they're no longer looking at a handbrake to end up actually stopping it. So we can't speak about capital, we can speak of policy. But I think the most important part of that is actually the assets."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:14",
                    "end": "00:32:44",
                    "text": " Unless we actually see people get off the fence, end up actually starting up, we're never going to end up seeing it. And I think as we as people, we look at failure and are like, failure is like failing an exam in school or something. It's something bad. It's something we're ashamed about. It's not. The whole process of entrepreneurship is iterative. But every way you find that doesn't work, you've actually learned, you can end up actually applying that knowledge to something else. So I think the biggest takeaway for anyone up here is it is a lot more fun to end up actually building when there's chaos than to go to someplace where everything works and just being another cog in that machine."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:44",
                    "end": "00:33:14",
                    "text": " Become a builder. You don't have to end up actually becoming a coach. Chris, and then Nishat Bhai. So I have been working with the research ecosystem and the startup ecosystem over the last 10 years. And I can tell you with confidence that what is required is to think big. What is required is to actually nudge these initiatives a little bit. And we have examples of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:15",
                    "end": "00:33:44",
                    "text": " technology from India going global. I will give you another example of technology from India that's gone global. Of course, all of you have heard of UIDI, but how many of you have heard of Mossip? Mossip, one hand partly raised. That's Model Open Source Identity Program. It's running in 29 countries across the world."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:44",
                    "end": "00:34:12",
                    "text": " 300 million people have got IDs with that. And pretty soon, a billion people will get ID from that. And on top of that, citizen services are being delivered. Our Diksha, which is an education system that, again, developed through philanthropy, running in many schools in India. It's actually running in many schools in Africa. And many of these are now moving to developed countries."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:12",
                    "end": "00:34:39",
                    "text": " So technology from India, from a developing country going across the world, you will see now more and more examples of that. Coming down the line, I can tell you 2D materials, which are the next promising materials for semiconductors. Because we need to reduce the power required for AI. We need green AI. So 2D materials, analog computing."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:39",
                    "end": "00:35:08",
                    "text": " developing the entire software stack on analog computing. Quantum of course, there are many startups which are already now getting commercial in India. You may have seen Parijan's presentation today morning at 11.30. But what you didn't realize or what you may not have fully realized is, today when you actually use voice translation using AI, your language is translated and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:08",
                    "end": "00:35:38",
                    "text": " The LLM is in English. It's not native in our language. So there is an opportunity to develop multi-modal, multi-linguistic AI, which is what Barajan is doing now. These are all, I would say, the next round of innovation, which will all now go from India to the world. Because very few countries are just monolingual. And even US is now multilingual."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:38",
                    "end": "00:36:06",
                    "text": " they will require a new set of LLMs if they want to do native AI. Thank you, Chris. One statement and then I'll come to Anishat. I think fundamentally one message which is perhaps underlying in this discussion so far is deep tech requires longer periods of gestation. Investors have to ask the right questions, not just after one year, what's your revenue model and profit model."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:06",
                    "end": "00:36:34",
                    "text": " And I think investors have to change themselves because investors in India have been asking the same question because of obviously, you know, return related issues. And I think the RDIF effort along with the private effort is going to change that because investors will not be able to survive if they don't ask the right question. So with that statement, there are many policy frameworks. And Nishat Bhai, this question is to you. There are many policy frameworks"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:34",
                    "end": "00:37:04",
                    "text": " which are undergoing change in the country for enabling deep tech investing. Could you highlight the top two which you think are important to be done in the next six months to 12 months? Thank you very much. What I see is that let me just give my work is significantly in the US as well. I think time has come to do look at innovation in all sectors."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:04",
                    "end": "00:37:33",
                    "text": " One side is technology side, but second thing is policy side. A lot of innovations are needed in policy side. And I think a lot of benchmarks are also available around the world. So the way in which we look at is that whenever new technology comes and new product comes, one side the industry wants to promote and grow, government's concern is to contain the bad actors."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:33",
                    "end": "00:38:01",
                    "text": " So balancing the two is a very strategic, important task. This is one part of the game. Second, that even in terms of focus, there are two stacks. One is the foundational technologies, whether it is AI, whether it's robotics, whether it's quantum, whether it's other ones, you know, photonics and so on and so forth. A lot of time and investment is required in the foundational technology."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:01",
                    "end": "00:38:28",
                    "text": " But then the application side is where we have great opportunity. For example, you converge AI with CRISPR and robotics. That would lead you to cyborgs, cyborgs, right? Ambience intelligence, digital twins. There are so many things that you can do beyond normal AI. Second thing I find here, slight difference."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:28",
                    "end": "00:38:57",
                    "text": " When in China or in the US, I see entrepreneurs, when they think of a market, they think global market. How big is the global market? In India, most of them, I feel, oh, India is a big market. Okay, immediate low-hanging fruit. India is a great laboratory, I see. It's a great laboratory because the cost of experimentation is low, scalability is high. So you have great opportunity. Trump has helped us to some extent to retain talent here."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:57",
                    "end": "00:39:28",
                    "text": " So, you know, you've got talent set. All that we need to do is how do you tap the capital? Today, local capitalists come, which is a great thing, but I believe that we really need significantly large capital. If you see the foundational technology otherwise, and in fact, we have to learn how to play musical chair. Why I'm telling this is that first of all, if you look at OpenAI, $100 billion, right?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:28",
                    "end": "00:39:57",
                    "text": " We are talking at a very different level today. And the needs are very large. So when you think of global markets and global products, that is where really, and secondly, that when you get investment, typically what would happen, people, investment also is a function of capital, is a function of exits. There is long term, which we are all agree that. But at the same time, we can find mechanisms, and you know what mechanism, so that,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:57",
                    "end": "00:40:25",
                    "text": " After four, five, six years, there is somebody else comes in. Then there is somebody else comes in. So startup, early stage funding, to growth equity, to private equity, to IPO maybe 10, 15 years from now. Who knows? But for 10 years, IPO market was in dumps, right? So suddenly, at the moment it is going on, everybody wants to ride on that. But you don't know how long it lasts. I'm not saying it should not last. Please don't get me wrong."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:25",
                    "end": "00:40:55",
                    "text": " But at the same time, we have to have a mechanism, both in terms of policy framework. For example, today our Companies Act, 2013, right? It's still old wine in new bottle, with respect, okay? A good attempt made. Income Tax Act, new act is coming. It's still somewhat similar, okay? So there are major innovations required, because when I go around the world, you can have so many different instruments, and you can structure."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:55",
                    "end": "00:41:24",
                    "text": " because instruments are nothing else but risk management tools. So you have different sets of risk. You know, third thing I find in policy, particularly grants are very important. Now, again, tax is a big problem, but the grant is an income or not. You must have experienced that, correct? And it's a big, lawyers take up half the grants. Accountants take remaining 50%. Entrepreneur is very happy."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:24",
                    "end": "00:41:54",
                    "text": " because we are also entrepreneurs, so please. But I'm saying that if you treat grant as a taxable income, you have to go around with this kind of thing so much. So there is a lot of need for innovation. One point which my friend made, the grid is good. He talked in a constructive sense. Please don't take it negatively. It's constructive sense."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:54",
                    "end": "00:42:23",
                    "text": " because greed could be bad. Okay, Michael Douglas taught us that lesson earlier, but it has been taken to a different level now. I would suggest that most important thing after building great companies, governance, okay, and ethics, organizational behavior, these are the subjects you cannot miss out. While you're building a business, be very careful. One of the other things, I'm sorry, I'm going a little long, is that when early stage..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:23",
                    "end": "00:42:48",
                    "text": " start working on that, when they start doing plans, because they want money, so there is a lot of temptation to fudge numbers, estimate numbers, okay? And after it starts doing good business, good funding, the next thing is you fudge accounts and financial, okay? I tell you, this is what is the biggest killer. Just be very careful, be patient."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:48",
                    "end": "00:43:19",
                    "text": " And those who are patient, that kind of money you want it. You don't want people to pressurize you to just get revenue and especially in deep tech. So I'm sorry I went on too long. But as a lawyer, what else can I do? I don't make money the way you guys make it. But if he would have come to me, I would have done one cent. No, thank you, Rishi. This is very interesting. I have just two more questions. I know we are running short of time. So if two questions, short answers, then I'd like to for five minutes open."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:19",
                    "end": "00:43:45",
                    "text": " to the audience, okay? Arun, exit, Nishad, in fact, alluded to that. Exits are very important. History of exits in India, secondaries, IPOs, M&A. M&A is very, very weak in India. Do you think M&A will become very important in technology divesting? If so, what do you think is happening today and what do you think is happening in the future? And then I'll come to the question for Chris. I think on M&A..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:45",
                    "end": "00:44:16",
                    "text": " as we see more deep tech companies coming out of India. Certainly we will see global M&A's from India. We've already seen that in smaller ways, but I would expect to see more. Second, I think the Indian companies too will start looking at that. Today again we're seeing them in small ways, but I would expect that M&A will be a major source of exit going forward. Okay, thank you. Chris, last question. Deep tech and AI, everybody is talking about how"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:16",
                    "end": "00:44:44",
                    "text": " Global markets are opening up. But there are challenges in India. How do you improve farm productivity on a half acre average land size? How do you remove grain storage losses? There are many Indian problems to be solved. Do you think that is, are you seeing some green shoots or a tsunami of those deep tech entrepreneurs who are addressing Indian challenges at scale? Okay."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:44",
                    "end": "00:45:13",
                    "text": " Do you think that's going to be very important, not just a global solution? In fact, I would say that yes, we're all talking about LLMs and voice response systems and things like that. The opportunity is actually application of AI in agriculture, in healthcare, in education, in automotive systems. All the cars will have now AI enabled."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:13",
                    "end": "00:45:42",
                    "text": " even GPS systems and things like that. So if the opportunity in the foundational layer, which is the deep tech layer, is about, let's say, 100 million, the opportunity in the application layer is probably 1 billion, 10 times that. So that's the opportunity. And then from India to the world, I don't remember now who said this."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:42",
                    "end": "00:46:08",
                    "text": " where the cost of doing innovation in India is one fifth the cost. So you have one fifth the cost and the market is the entire world. That opportunity is significant in India. And I'm seeing now, I met, every day I meet entrepreneurs, startup founders, et cetera, and the kind of problems there."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:08",
                    "end": "00:46:36",
                    "text": " actually attempting solve is very exciting and very inspiring, actually. So with this money, and that's the reason why I'm extremely glad all these people are sitting in front of you and saying that we're going to fund all of you, right? So it's a very exciting period. And I'm also glad because then RDI will get deployed. RDI fund will get deployed. So I'm hopeful that the next 10 years, India will become the deep tech capital of the world."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:36",
                    "end": "00:47:05",
                    "text": " Yeah, I believe. Thank you. Thank you. Very nice. So I know there are a lot of questions. We have actually five minutes to the end of this. And as a sutradhar, I may have to. I just wanted to say that we have Anand Kalamath from Applied Materials who has joined, was a big supporter of the IDTA. And I thought I want to welcome him. OK, excellent. Thank you. Thank you, Arun, for inviting Applied Materials to be part of IDTA. I think two observations here right from the discussion."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:05",
                    "end": "00:47:31",
                    "text": " I think it'll be useful to think less about what the stock markets are, I mean, my view of it is you build great companies, exits will take care of themselves, right? And the question is, given the talent here in the room and outside the room, right? How do you build the next NVIDIA? How do you build the next SpaceX? How do you build the next Grok or ILabs and so on, right? Because I don't see that discussion happening, right? Everybody's like, oh, the stock market is X, I'm going to do Y, and I'm going to make Z money."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:31",
                    "end": "00:47:59",
                    "text": " If you see, we have been investing in deep tech for 20 years, right? And I don't see that discussion happening in the companies in Israel, for example. Nobody is building a company saying that I'll sell it, I'll do it, right? They're starting and saying, can I build a big company because I'm solving a big problem? I would encourage that level of discussion. And the second thing on the deep tech, like somebody mentioned, they've been investing in the company for 25 years. Also figure out how you can reduce the risk, right? Because we have seen the cycles. Like in cleantech, people came and went out."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:59",
                    "end": "00:48:27",
                    "text": " Internet people came in, came out. So I'm just worried like two years from now, we shouldn't have tourists leaving part and there's going to be a death of the deep tech sector, right? So just observation is like one is I think the boundaries of what to think about is not a billion exit. Like you know, companies are now exiting at a trillion plus, right? How do you build that trillion plus company because you're solving a trillion plus value thing, right? That's what I think, right? But great, great team and look forward to be part of it. Thank you. I'm going to open this to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:27",
                    "end": "00:49:00",
                    "text": " I think we have time for three questions. Please name yourself. Please address the question to one of us here. And then we'll answer. And then we go to second and third. Yes, please. Question only, not comment. Yeah. Big question or a long question? Long question. I'll make it in 30 seconds. You've already taken 30 seconds. So rather than asking these many questions, straightforward, I'm asking. Yeah."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:00",
                    "end": "00:49:05",
                    "text": " Right now, just like rapid fire who are ready for a 10 CR investment?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:05",
                    "end": "00:49:32",
                    "text": " without seeing the pitch deck. Because there are many startups who cannot express themselves if you say about Bharat Edge and all. I would like to compare it with Google and Meta. But when it comes to same for us startups, they are going to compare it with Sarvamaya and Bhasne and say that, how are you different? And we are not going to invest. And as Sir said, they are going to, of course, reject for the first 500 times. Don't mistake investing for gambling. I think that's number one. And as Nishibhai also said, investors have to do their diligence."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:32",
                    "end": "00:50:00",
                    "text": " Third, there are enough LLMs that are very articulate. I'm fairly certain you can put in what you want and make something articulate. The hygiene is crucial. There is a way to prepare for funding without a deck. Your caliber and your credentials must be outstanding. In our world, that happens all the time. We meet people. We commit funding without seeing a business plan. This happens in the US. This happens to some extent in India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:00",
                    "end": "00:50:28",
                    "text": " but please understand for that you have to be like a five, listen to me, it has to be a two, you have to be three standard or four standard deviations better than the normal. So if you are three or four standard deviations exceptional, you don't need a business plan, you just need to announce to the world I want to be an innovator, they'll be capital. So I think the answer to that is very simple. You would like any investor including this room to listen to you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:28",
                    "end": "00:50:56",
                    "text": " I think without hesitation, I can speak on behalf of all investors. They will listen to you right to the investors. Somebody will listen to you in each firm. There's a lady here up front. One minute, please. There's a lady here up front. Hello, panel. I'm Pankhari. I'm building an AI as well. My question right now is actually to Sid Pai, as everyone calls you. You mentioned frugality in investment, right?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:56",
                    "end": "00:51:25",
                    "text": " I absolutely understand. It actually helps. It promotes cost-effective solutions. However, when this similar kind of ideas, at least like at the early stage, in India receive, say, a 1, 2 CR check, right, and add a certain dilution. And in the US, they do receive a huge amount, which is 4x, 5 times, right, the kind of amount that we're talking about. Do you think Indian startups will be able to..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:25",
                    "end": "00:51:44",
                    "text": " scale at global or will be able to build for globe we will because I think see the important part is investors end up funding it basis what the entrepreneur wants Indian entrepreneurs come and say I will do this in 50 lakhs I will do this in a crore an American entrepreneur says I need I need 15 million ends up actually raising seven I think increase your ambitions I think that most part"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:44",
                    "end": "00:52:11",
                    "text": " Frugality doesn't mean that you end up putting handcuffs on yourself. Frugality means you have a large pool that you don't have to dip into whenever you need to, but you end up optimizing at the edge. I think that optimization is what's more important, not the fact that you end up putting handcuffs on yourself and then cry about the fact that you can't compete with the rest of the world. Yeah, I think the answer to that is very simple. Ask for more. But prove that or talk about the fact that this is deserved from a point of view of where you will spend that money. Many of the investments we've made"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:11",
                    "end": "00:52:44",
                    "text": " are in the range of 7 to 14 crores at the day one. Third question, please. Yes. I'm pointing at him, please. There is a sequence in my head right now. Yes, please. Just a quick question. In 2026 or 2027, if you have to evaluate a startup, would you look at their team size?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:44",
                    "end": "00:53:14",
                    "text": " if a startup makes use of only AI and AI agents to solve a problem, that would cut the bill. Thank you. I think size really doesn't end up mattering, I think in a large number of these things. I think what's actually important is the competence of the entrepreneur and the team they end up actually bringing in. Whether you are using AI, whether you're using a large number of people sitting over there for that matter, those are choices of yours. There are pros and cons that end up actually coming in, but..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:14",
                    "end": "00:53:41",
                    "text": " VC investing can't be boiled down to a set of five rubrics and hacks, unlike what the LLMs end up actually generating. So it depends on what you're actually building, how much cash you need to end up building it, what your chances of success are. What are you bringing to the table? If you're primarily an integrator of other people's ideas and intermediary, there's really not much value that ends up getting added. So look at value added, not the ingredients that go into it. We have one more question. This gentleman here. We just need four good people. One more. Sorry?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:41",
                    "end": "00:54:18",
                    "text": " I just wanted to say that, listen, four is a magical number. At a startup stage, most startups, you need four good people. So you can enhance yourself with AI and all. This is not about 40 people. As you are looking to build a team, it's not about tens of people. It's about two, three, four good people. We have one more question. Can you please write to all of them? Yeah, OK. That's it. Write to all of us."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:18",
                    "end": "00:54:46",
                    "text": " Write to all of us, find out offline. One of them will say yes or no. I'm going to halt all the questions right now. I'm going to halt all the questions right now because all of us will be available outside after this session. I know we're ending at this. Again, thank you to Meti and IGTA for organizing this function. And thank you all for coming here."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
    {
    "_id": "69944bd325c9f4e0d7e708a8",
    "projectId": "577939",
    "projectName": "Deep Tech and AI Investing in India  Current and Future States - Hindi",
    "projectDescription": "Deep Tech and AI Investing in India  Current and Future States - Hindi",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T11:07:03.895Z",
    "projectUpdatedDate": "2026-02-17T11:07:03.895Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T11:12:21.380Z",
    "projectDeleted": false,
    "projectDataLength": 3,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "Deep Tech and AI Investing in India  Current and Future States.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "Deep Tech and AI Investing in India  Current and Future States.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Hindi"
},
    {
    "_id": "69944d5525c9f4e0d7e7090c",
    "projectId": "865410",
    "projectName": "Deep Tech and AI Investing in India  Current and Future States - French",
    "projectDescription": "Deep Tech and AI Investing in India  Current and Future States - French",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T11:13:29.718Z",
    "projectUpdatedDate": "2026-02-17T11:13:29.718Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T11:14:17.314Z",
    "projectDeleted": false,
    "projectDataLength": 3,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "Deep Tech and AI Investing in India  Current and Future States.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "Deep Tech and AI Investing in India  Current and Future States.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "French"
},
    {
    "_id": "6994343825c9f4e0d7e703d2",
    "projectId": "593820",
    "projectName": "AI masterclass in Financial Services - English ",
    "projectCategory": "AI-impact-summit-2026",
    "projectDescription": "AI masterclass in Financial Services | AI Summit India 2026",
    "projectCreatedDate": "2026-02-17T09:26:12.785Z",
    "projectUpdatedDate": "2026-02-17T09:26:12.785Z",
    "projectStatus": "Active",
    "companyId": "69453bf2e603a0a401b6f35b",
    "projectCreatedBy": "69453bf2e603a0a401b6f35c",
    "projectUpdatedBy": "69453bf2e603a0a401b6f35c",
    "projectLastAccessDate": "2026-02-17T09:26:12.785Z",
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "Content",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "_AI Transforming BFSI_ Practical Strategies _ Safe Scaling _ India AI Impact Summit 2026_(MP3_128K).mp3",
            "col3": " Quick context setting on what we see from our vantage point as well in financial services in terms of the adoption. What's happening on the ground a lot being talked about it, but it's also fact about making it work in an enterprise in the form. So that's on the first part that will do will then move to the next one, which is about you wanted all of you to experience a bit of especially from a wipe session standpoint more in the area of taking an idea through a proof of value.  More of a vibe design and a vibe build session. So you have a lot of laptops out there. There are a series of instructions. It will be both assisted as well as if some of you wanted to do on your own, more than welcome to do it as well. But we'll assist in the process flow. Get a feel for it. How does it work through, especially the design and the initial prototype build part of it. And then of course we'll wrap it up as a final thing. Let me give a little bit of context setting from our vantage point where we see us.  I think it's all well recognized that AI has really got to a hyper acceleration phase. Maybe two or three key call out that I would do out here. I'm sure most of you are familiar with it, but I'll just make a call out. One is the fact that there is a genuine movement from the traditional chat bot kind of an interface to more of agents in action. It's really happening out there in the market, in the industry as well, and BFSI is definitely in leading that as well. There are multiple...  I would call it as S waves that are happening right now. Multiple interventions happening at the same point. So that kind of giving you a bit of challenge as well as an opportunity. Exciting, but at the same time, it's also giving a lot of challenge in terms of the adoption standpoint. We see a significant focus on multimodal adoption. And believe that's the future in which it will shape up. You would judiciously use multiple models to accomplish a particular task or a process flow. And the most important thing is that  End of the day, it's all about context engineering. You go to build the context before you engage with the model. So three callouts, because there's a lot of flurry of activity happening. As you could see, a series of models that are coming into the market, competing ones, quality is improving. For instance, think about these two. The 5.3 Codex and the Sonnet 4.6 were launched in a space of 20 minutes. And you could expect what would happen down the line as well.  So with this flurry of activity, it's important that we keep those three callouts in mind because we're going to judicially use which models would factor for and what context in an enterprise as well. While that happens, this flurry of activity happening, what we believe is that there is also convergence happening in this space. There is a vertical convergence happening. Right from so far, the focus has not been on models and applications. It's all about hardware and infrastructure as well.  could see a lot of investments being called out in this space globally within India and from our organization standpoint, also in the group, we have announced the launch of an AI data center, a one gigawatt data center over the period of next five years as well. But the point is that there is a vertical alignment that's happening right through infrastructure to the models, both specialized and foundation models, and to the applications. It's very important to think so that it has got off to a  inflection point that cannot be seen in isolation. It has to be seen in terms of the entire vertical stack to make it more impactful. We did a survey among customers both globally and in India. It was 170 of our customers we did a survey and see what's adoption and how is it shaping up. It was a very recent survey that we did. The multiple things that came out, one good news is all about that. Close to 90 odd percentage of the customers are in some form or shape. And this cut across banking,  capital markets and insurance customers, as I said, both India and globally, are in some form or shape adopting it with almost 60% fairly advanced state of the adoption standpoint as well. I just want to make one specific call-out here. It'll be interesting to many of you as well. 50% of the customer base feels that a lot of focus needs to be given on enterprise-specific models and some fine-tuned models as well. This is something which is very important given the regulated nature of the financial service industry.  how much of secure and how much of accurate models are also needed from an enterprise standpoint. Then of course, a majority, overwhelming majority also felt there is a good need to measure the outcome. Metrics are needed. It's one to talk about POC and pilots, but another to talk about the business outcome. In India, I think it's also interesting to see India, the opportunity and challenges, the nuances around the language. India, I would call it, is a very voice-led  India is a land of many voices, spoken in different languages,  across diverse geographies and, most importantly, with unique financial needs. At Tata Capital, financial inclusion means more than just access to finance. It means being present, being understood and being reachable to everyone, everywhere. Today, we are proud to showcase a multi-modal voice AI solution supporting our customers across every key moment of their financial journey. Truly,  end to end.  Okay, so first of all, are you salaried or self-employed? I don't understand, madam. Excuse me, I mean, do you work somewhere or do you have a business of your own? I have my own shop. Okay, so you have your own shop. Can you tell me what your business is about? Madam, I have a rental shop. Okay, so you have a grocery shop. So instead of your shop, do you have your own shop or do you have it on rent? My rental shop is on rent. Okay, so your shop is on rent?  I have sent you a personal information form on WhatsApp which requires some basic details. Fill it in. If you have any doubts, you can ask me. Yes, I have got the form. I will fill it in, madam. It won't take much time, right? No, it won't take much time. Fill it in slowly. I am here. If you have any questions, you can ask me. Madam, what do I have to put in the mother's maiden name? Okay, you are asking about the mother's maiden name. This means the name of your mother before marriage.  Do you have a call? If you need anything, let me know. I will help you. Yes, I am filling the form. Madam, I am in the middle. I am filling the form. Okay, no problem. You can fill the form. I am here. Yes, Madam. I have filled the form. Very good. Thank you for filling the form. So, let me tell you again that you will get a loan of 5 lakhs for 3 years at an interest rate of 12.5%. The EMI of which will be 16,727.  What you just saw. Well, that is the future. The future of conversations, of customer engagement and of AI led transformation. This isn't just a concept. It is already powering millions of customer conversations. This is how Tata Capital is reimagining engagement. At scale.  with responsibility, with empathy, and with real impact. Because when trust finally finds a voice, Bharat engages. And when Bharat engages, Bharat progresses.  India-based model based on Sarvam. So I think a lot of linguistic variation that is addressed, and I think this is important because on one side we have the global models available, but it's also important to address the local nuances out there. And the second thing is it's also now allowing from a multi-channel standpoint. So I think the ability to come together more from a customer servicing, I think that has taken on a different league altogether. While that happens, let me call you out some key pointers from.  What we have been experiencing over the last, I would say, nine to 12 months, some very grounded reality has been popped up here. From our vantage, we were looking at the adoption in terms of AI for business and AI for IT. IT is pretty straightforward in terms of, it's mostly around a lot of focus on coding assistance and plus the fact that you do a lot using modernization. Huge focus in many forms in terms of technology modernization, the ability of AI to accelerate it. These are not POCs and pilots really getting into production.  Then the business side of it, if you look at it, it's got a broad stick happening out there. Right from customer servicing to horizontal functions, corporate functions, and also into wealth and investment advisory as well. So there is a significant progress happening in this space, but just also wanted to call out some of the ground reality that are popping up. ROI challenges still persist. There is still challenges in terms of the accuracy, the.  compliance aspect of it, that's still an important thing to be addressed and factored in building an AI solution. The other part is that building an AI solution is one thing. Making it work in an enterprise in terms of integration is so critical. A lot of ROs have been built around building solution, but not necessarily the effort that you need to make it work. Connecting to your enterprise systems, connecting to CRM, your core banking, your learning application, that's a critical aspect of this whole process flow.  I think it's fairly recognized that individual use case are not giving you the benefit, the real benefit come by process flows and by workflows. And that you're going to look at workflows differently, not to the traditional left to right, but more from right to left in terms of an outcome standpoint. We see and we believe fine tuning is something that financial services would pick up in a significant way. If you're already seeing that, that would really do an uptake in the next six to nine months.  Definitely see that picking up in a big way. Why? Because of the need for much more accurate, much more precise models that need to happen. To say that it will be a hybrid scenario, you would have fine-tuned models on one hand, you'll also have large language models, you would judiciously use it. There is definitely a shift towards Hintiq. Having said that, these are not Hintiq as in high level of autonomy, but a slow.  a pedigree of autonomy is being enabled in it, but agentic workflows are happening. The bottom point, I think we refer to that the vertical stack that's evolving, very important to look at an infrastructure in the platform-based thought process. Again, the whole point is that I can do a great POC pilot shiny object, but how do I scale it across and you need a platform engineered thought process to make it happen. So these are some of the experiences that we're going to picked up, thought I can relate that all of you as well. Just wanted to.  give a quick perspective around the fine tuning that we talked about. This is something which piggyback on the fact that it will be in hybrid AI solutions and models that you would end up with. You would have both the scenarios. You will continue to leverage large language model for broad, measurable enterprise task and activities with proper grounding. I think it's very important I just double down on the fact that it has to be engineered.  How well do you orchestrate your workflows? How well do you build your evals around it? It's very critical. But I think this is an equally important. How well do you build your domain-adapted models in compliance area, in advisory, in credit research? I think it's important to see how well they build appropriate and high accuracy. So our vantage point, we believe, it's going to be convergence between the two, not one over the other because each has its own space. It's very important to traditionally use this. I'll just give you a couple of quick.  and adults around it, these are real life examples. Something that we have done for litigation prediction in a claims value chain. If you look at it, what a base model provided versus what a fine tune model provides. And this is not something that happens straight away, you need to iteratively build it, but it's very important to continue, make sure that right from base model, the different techniques of fine tuning to make it more domain adapted in that context. And you could see the high percentage of  accuracy as you build that incrementally. And this is where business gets a lot glued to it saying that, hey, now it's come to a point. It's not just technical. It's also important to make it work. Same thing in compliance. And we have used two different models just to give a variation. There are many models out there. But these are models that you could enable it. And whether it's Lama or Mistral Family, there are many other models. For instance, as I said, for several models, we've been working on it in terms of voice, AI standpoint. But the key message I want to leave out here is that.  fine-tuning or making it proprietary, it's not just technology. I think it's all about building this domain curation. How well do you build the instruction set for a fine-tuning? That need a lot of domain understanding. Once you build that, the initial set, then I think it's a lot about applying the right techniques. It's also equally important to look at the right models. Your slew of models out there,  and there is a method to the madness in terms of which model would fit into what context. So all the three are very important. Building the right instruction sets for fine-tuning. For all of you who I'm sure fine-tuning is something that all of you are aware of it, but essentially what you're trying to do is that making sure a broader model has been fine-tuned towards specific domain and adaptation so that it get much more relevant in that context and accuracy of the decisions it takes us to higher level.  a series of business scenarios that are evolving in this area. We see this in India, we see this globally. So this is something from our vantage, we see a strong uptick that's happening. So just wanted to give a quick flavor of that before we get into the white thought process. I think it's important to look at it from the context engineering, very important. Enterprise adoption is not just about building use cases, but also making it work in the broad end.  it's very critical to make sure that you judicious use of large language models and domain adapted models. With that, let me just show you one specific scenario. Good afternoon. Thank you for calling. Do I have the pleasure of speaking with you? Hi, Catherine. This is Daniel Thompson. I need to report an accident that happened earlier today morning at 4.30 AM.  I am sorry to hear. Could I please have your policy number? Yes, it is ALC 76-12-39-05. Thanks, Daniel. I thought that's your policy. Everything looks active. Could you describe what happened? I was already on the highway. I was driving southbound on the M1 Pacific motorway past the Gosford exit in New South Wales, maintaining my lane and speed when a silver SUV suddenly emerged from the left.  onto the motorway and site wiped my vehicle so the impact was on the left side yes quite severely it hit around the front left door and fender area i had no chance to react the other driver didn't yield or check for traffic understood um with a police call to the scene  Yes, our patrol unit came by. They've noted the details. I have the event number. I also took several photos of the damage. Great. If you can send us the event number, police notes and any photos that will help us proceed. We'll also need the repair estimate or tow docket if you've received one. I will email all the documents shortly. Perfect.  Drive safe and let us know if you need anything else. All right. As you could relate, I think the accent of this is for a UK-based customer. And the context is a little different from the first one that you saw from a voice here, which was a very specific scenario where, in the Tata Capital's story, it was AI at the thick and center of everything in terms of engaging in a customer service. But not actions are not getting taken. It's more of a conversational flow.  This one is scenario in which a multi-model with a fine-tune model being brought into the picture. So if you could see the focus of the customer service rep is more in terms of looking at the information that can be gleaned around it. So this is an augmented AI. AI was listening to the whole conversation, providing from an augmented sample what additional information. Normally, this whole process called FN oil, the first notice of loss from the insurance background would know.  When you meet with an accident, the first notice of loss that you make a call to the back office to inform. A lot of information you gather at that point. It's very important to capture that information. Today, it's a very back and forth happens, and your claims get delayed because of a simple reason, you haven't captured enough information. In this scenario, behind the scenes, so AI was not in the front-end center, behind the scenes capturing the information, and there was a litigation validator, which I was telling the fine-tune model.  looking at it, is there any fraud in this process so that there is a subsequent litigation possible down the line? That information, I don't know how many of you have watched that, that information was gleaned through in between for what I say, this looks like a good call, let's proceed. So I think that the modality is a little different out there, but the whole point is that it's very important to look from both the scenarios. So to put it in a nutshell, I think it's important from an enterprise form, I think proper calibration is very critical.  great technology, great capabilities evolving, but it's also important, how do you enable that in a context? Where do you not apply AI also? I think it's important to also look where you don't apply AI. Because at times, we have the thought process of applying AI all over the place. I think it's very important to break it and then make sure that the ones, and that's what our philosophy is that, don't look at from a traditional process flow from left to right, which you break it down to series of task and say where I can apply it.  Look at from a right to left and see outcome. What is outcome and how do I reapply back? So what are you going to do now for the next 60 minutes or so? What I thought is that, I believe that this need to be broken into three broad areas. You got to have a quick proof of value. You need to build that proof of value because there are certain things you will just move forward then. You say, okay, this is not working. Let me try the next one. You have the ability right now to do a quick prototyping, a quick vibe design and prototyping.  Then you go into a rapid build over a period of time, five to six weeks, and then you really scale it. This is a model that we would advocate as enterprises start picking up more and more. Because, as I said, it's very important to look at outcome backwards. It's very important to look for what is the business assurance that I bring in. So we have constructed the next 60 minutes or so in that format. We have certain challenge cards on your table. You have laptops out there. We're going to do this in two formats. We'll assist you in the process.  in how do I go through the process, and second is also in terms of those who are more wanting to do on their own, they feel free to try it out. Instructions are also given out there. Let me invite Kavita now to run you through the vibe coding design and prototyping session. Thank you. To continue on the topic of scaling, in financial services, AI doesn't scale when the model is ready. It's when legal compliance...  regulatory, all of the body comes together. When they are ready, it's when AI scales. So I think that's one of the challenges that we face when we look at a scaling AI. This is just one of them. And my name is Kavita, and I lead the AI labs for financial services. And I help customers on their journey to adopt AI and scale AI. And one of the other challenges that really come across  when we start discussing about AI use cases is, how do I get the business teams on board? Because if you look at the kind of implementations we have seen so far in AI, you've got knowledge assistance, you've got chatbots, you've got good document AI capability. And if you look at these implementations, they have actually filled the gap where the traditional AI, traditional IT has failed in the sense...  inability to access unstructured data. There was a gap which AI solved. So most of the use cases have been around these lines, and that's primarily because AI has not been, we have not let the business teams actually drive AI. So what does it take to help enable the business teams to understand the power of AI and look at more transformative use cases?  So that's the construct of what we want to do today. We want you to experience that how a business user or a business team can be enabled with AI to accelerate that whole journey of identifying the right opportunity, doing a working prototype, and actually even testing it out and seeing the proof of value. How do we shift left?  in the sense that how do we give business more control and working with the engineering team, build out a quick proof of value that you can validate and then scale to production. Because in one of the earlier conversations I had with one of you here was that scaling is a challenge. But sometimes you don't get to scale because you've not identified the right opportunity. You try to apply AI everywhere. So how do you use AI to help you in that journey?  And what we want to achieve through this is let business lead the whole AI redesign. So how do we do that? And that's something that we will cover today. And what we're going to demonstrate to you is not a tool, but how we can bring all the AI capability together for a business audience.  Yeah, switching on to that. So what we'll see today, disclaimer before I start, this tool is not a single tool. It's a workbench that we have created that brings together different AI capabilities. And then what we do is that we start off with something called a business challenge, because that's the starting point for a business team. Where do I begin my journey? What is the challenge that I need to address? And for the sake of this demonstration, I'll pick up one use case, something that's  A challenge that has come up in one of our customer conversations was that you have a lot of portfolio managers who need to, at the end of the quarter, generate a summary of how their portfolio has performed. And along with collecting a lot of data, they need to actually  check out the sentiment of the stock on the public domain. So there's a lot of data gathering, a lot of analysis that comes in, and as a result, there's a lot of time that goes into that activity, and that delays the overall report, and sometimes this gets delegated to a junior analyst, the quality of the report is affected. So how do I bring all of that in a way that AI can help the portfolio managers solve the problem? So that was the challenge that we picked up, and I'll show you a quick demonstration of  how the challenge works. The challenge was just creating an assistant to help the AI agent, to help the portfolio manager actually look up what is the right, what is the data that's available out there on the public domain, or even look up Yahoo Finance to see how the stock has performed. So all of that was being done by AI in the background. And this is a solution that...  or when we piloted with our customers, they saw a huge advantage, because today, they were actually outsourcing a lot of this activity to a third party organization, and they were able to bring that back, and bring down the overall time to generate the report, and even the quality of the report started coming out much better. I thought I'd give you a view of the final outcome, and take you back to where we started. And that's the platform that we were trying to show you. So switching on.  to that platform. So what we do as a first step is actually pick up the challenge and how do we use AI to ideate and come up with something called the opportunity canvas. So many of our users have, you know, if you see business users are all equipped with co-pilots because it's already available in the enterprise. But what we've observed is that just not giving them AI tools or training them on prompts is really going to help. It's about  helping them to understand that end-to-end process and letting them have complete control. So that's what we've done over here. So we looked at what could be a set of questions that a business user needs to look at when they see a challenge and convert it to an opportunity. So we've curated a set of prompts which can go through the data and come up with what we call the opportunity canvas. So we'll just walk you through those different steps. So I think one thing that we always need to keep in mind is that  When we give AI to the business users, the research is primarily on, you can use a lot of research tools available, but the organization context is more important. How do we know what the user is? What are the different users going through? What are their challenges and pain points? So loading the context is the first step.  And the reason I'm walking you through this is because once I finish the demo, I want you also to try and just see whether as a business leader or a practitioner how the whole experience will be and how can we really build that capability for business teams to use. So we'll go through this flow. So we have built the context. And then after that, you can even go back and do a deep research. Because sometimes your context is not enough. You want to know how your peers are performing.  pull out a deep research from Gemini or ChachiPT. So put that context also in, and then you have an overall knowledge base that your business team can now use to start ideating. So what we have done is created an ideation framework here. Now, as you go through the ideation process, you'll have a lot of questions. Have I captured all the pain points right? Because you want to be sure that you have actually  solving a problem. So there's pain points that you capture. And there are quite a lot of steps. I won't go through all of them. But if you quickly go through, you will have pain points. You'll want to assess what the business impact is. What's the possible solution direction that you want to take? And as we go through these steps, you will see that it's not enough just to analyze the data that's available.  but to actually start seeing AI as a thinking partner. That's what we want the business to do, be able to bounce ideas, validate them, feed the input, and finally, at the end of it, create what we call an opportunity canvas. So a lot of thought has gone in. So while these demonstrations, while he brings up the opportunity canvas, something that we have observed is that just giving a tool may not be the only solution.  What we do in TC, something that's worked for us, is that we conduct an AI day for business teams, where we try to create that immersive experience where they understand what are the right AI tools and also how to use them and how to apply them in the context. So I'll, if you can just go through the, yeah. So there's a lot of questions, so we don't, we, and it will take a minute to go through, but if our demo is ready, you can show that as well.  Yeah, so this is what the final product looked like. So you give a company name, and it went through the organization's research. It also went to the public domain and started picking up all the stock information and brought that up into an overall summary that can be then used by the portfolio manager to make a decision on how his stock has performed, how do I report that?  to his customers? How do I report that into the performance reporting? And if you see, it was instant. And it's gone through, I think, 15 to 20 sites. We've configured all the financial sites to go through and capture the sentiment, capture key financial metrics, all of that instantly on a very high-speed internet. Now, I think we would be done with the Opportunity Canvas. Last one step.  So it's important that we go through these steps. So that's why we have consciously made these steps so that you don't pick an idea which is not the right one. So identifying the opportunity, what is the possible solution direction, what's the impact, and what your roadmap will be. And there you have an opportunity canvas, which the user has co-created along with AI. Now you go on to the next step. And this is important.  All this is fine. The business team could be comfortable in Opportunity Canvas. But can now the business team, with the help of AI, drive the overall design process? So what you'll see is that you can brainstorm with AI and actually give a direction of what that design should be. And this is what AI has generated. It's kind of broken down that whole opportunity into a set of  different agents that can go in. Because as we covered in the previous session, a lot of attention focuses on moving to agentic. So can we look at building what that agentic solution would look like is what you're going to see here. It's not done the coding yet, but what it has done is broken up that entire opportunity into a series of agents which can address this need. And we'll just.  You can try that. You will get a chance to try this, so I won't go into the detail. But here, it's come up with four or five agents. It's organized. And as a business team, they can actually go and say, I want these kind of checks and balances. I want to do it this way. And now you see the difference, right? Earlier, it was all left to the engineering team to actually build the solution. But now, we are passing the power back to the business to drive all of this. Now, going forward,  What is the next important step is to validate what this opportunity looks like from an experience. So you use a simple wipe coding tool. We have here used both Lovable and Google AI Studio. This tool generates the prompt. You copy the prompt. And then you actually can see it through a visual experience.  simple as putting it and if you see the prompt it's all English and you know when you get into your wipe session you'll see that you can actually edit some of it and see what that experience looks like so this will take how much time maybe a minute or so do you have something do you have something that you've already pre-built yeah so this is something that that came out of the tool a very very nice experience which shows what the portfolio summary should look like what are the stocks etc so  kind of understood the opportunity and translated that. And you can play around with it and tweak it to the experience. This step becomes very important because what earlier took two to three weeks of building a UX prototype as the end of a design phase actually got converted to minutes. And this is something you take to these stakeholders, validate it, and you're ready with what is that real challenge you want to address, what's the real opportunity comes out so quickly.  We can't stop with just a UX prototype, because that just gives you an experience. You need to actually engineer some of the back-end capability. And there are low-core platforms that are available. So in today's live session, we will show you that as well. And one of the examples that we have shown here is a solution called Ne10. It's a low-core platform building. It's not that we are saying this is the right tool. As per your enterprise decision, you could have different options.  But this really helps, in our view, to ideate and understand or discuss with your engineering teams how your design should be. You can easily visualize what are the models being used here. Have I thought through all the guardrails? Have I built my business evaluations? All of this can be verified in one shot. And then from this, the code can get generated, and you'll have a working end-to-end solution.  So this is the process we followed for quickly prototyping. And the demo that you saw was an example of something that was built using this approach. So what we'd like to do now is shift back to you and see if you'd like to try this. I mean, some of you could. We have laptops configured with.  with user IDs, we have all the tools available. It's an opportunity for you to try out and we have mentors. We have mentors available for each table. A lot of our teams here to support you and we would really encourage you to try and we can guide you here as you go along. Explore, yeah. Okay, so let me just walk you through what you would do. So all of your laptops will have a handbook already preloaded.  which talks about the use case that you would be picking up. Okay, so, and that is in green, it's tagged as challenge. So that's the use case that you will start with. And you have different phases and you have resources for each of the phase. So feel free to go through this handbook, take it step by step, and we'll help you with it. You are workspace five.  So log into your workspace based on the number that's on your table. So when you start, it will be the discovery and design. Yeah, you can start the solution with the resource present there. Okay, so just a quick time check. We just have five minutes. So if you want some help on any particular section, just raise your hand. We'll come and help. But I'm so happy to see many of you have generated very nice UX prototypes.  And you can also move into that build phase. That's an optional one, but try it out. We'll have mentors to support you.  uh...  So I think we'll just conclude.  So, so far we spoke about, I only covered, I'll hand it over to him, but what you did essentially comes as step one. You actually did, and I'm so happy that many of you got challenge completed, and you saw your code getting converted, and I see a lot of happy faces because it was, I heard words like, amazing, I can do the same thing with my tools, so very happy to hear all that. So what we did,  was the first step, proof of value. We've tried to establish this is something, a problem worth solving for my organization and you may have an answer then, you have an answer at the end of the stage. But the journey to take it to production involves something called rapid building. What you've built is a foundation which you can add on more functionality and then create it into a full-fledged solution. And that's where we call the rapid build.  which is about engineering the full-fledged solution before you go to production. And then finally, the scale. And when you look at scale, and someone asked me that question, you have to build a lot of AI foundation capability, the control plane, a lot of things need to go in. And that's important if you want to bring that scale. I'll just hand it over to Onni to close. I think thank you. Thank you for being a real sport out there. I think all of you joining in, being through the wipe session as well. Great to see all of you continue out there.  Once again, thanks for being participating. I think this is one important thing that we believe AI would make a difference in the ability to quickly translate an idea to prototype. Not to say that that prototype is what you would eventually scale to production, but that gives you the feel of what can be done. And then you naturally scale it into production in a very calibrated way. But again, thanks for your time. And then being part of it, feel free to reach out. Some of these I'm assuming.  would have been captured in your email, et cetera. So it's available for you to do some prompting and wiping back home as well if you're adventurous and wanting to extend the focus. Thank you once again and have a good rest of the day.",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771320345334.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:01",
                    "end": "00:00:29",
                    "text": " Quick context setting on what we see from our vantage point as well in financial services in terms of the adoption. What's happening on the ground a lot being talked about it, but it's also fact about making it work in an enterprise in the form. So that's on the first part that will do will then move to the next one, which is about you wanted all of you to experience a bit of especially from a wipe session standpoint more in the area of taking an idea through a proof of value."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:29",
                    "end": "00:00:59",
                    "text": " More of a vibe design and a vibe build session. So you have a lot of laptops out there. There are a series of instructions. It will be both assisted as well as if some of you wanted to do on your own, more than welcome to do it as well. But we'll assist in the process flow. Get a feel for it. How does it work through, especially the design and the initial prototype build part of it. And then of course we'll wrap it up as a final thing. Let me give a little bit of context setting from our vantage point where we see us."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:59",
                    "end": "00:01:28",
                    "text": " I think it's all well recognized that AI has really got to a hyper acceleration phase. Maybe two or three key call out that I would do out here. I'm sure most of you are familiar with it, but I'll just make a call out. One is the fact that there is a genuine movement from the traditional chat bot kind of an interface to more of agents in action. It's really happening out there in the market, in the industry as well, and BFSI is definitely in leading that as well. There are multiple..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:29",
                    "end": "00:01:59",
                    "text": " I would call it as S waves that are happening right now. Multiple interventions happening at the same point. So that kind of giving you a bit of challenge as well as an opportunity. Exciting, but at the same time, it's also giving a lot of challenge in terms of the adoption standpoint. We see a significant focus on multimodal adoption. And believe that's the future in which it will shape up. You would judiciously use multiple models to accomplish a particular task or a process flow. And the most important thing is that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:59",
                    "end": "00:02:26",
                    "text": " End of the day, it's all about context engineering. You go to build the context before you engage with the model. So three callouts, because there's a lot of flurry of activity happening. As you could see, a series of models that are coming into the market, competing ones, quality is improving. For instance, think about these two. The 5.3 Codex and the Sonnet 4.6 were launched in a space of 20 minutes. And you could expect what would happen down the line as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:26",
                    "end": "00:02:55",
                    "text": " So with this flurry of activity, it's important that we keep those three callouts in mind because we're going to judicially use which models would factor for and what context in an enterprise as well. While that happens, this flurry of activity happening, what we believe is that there is also convergence happening in this space. There is a vertical convergence happening. Right from so far, the focus has not been on models and applications. It's all about hardware and infrastructure as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:55",
                    "end": "00:03:25",
                    "text": " could see a lot of investments being called out in this space globally within India and from our organization standpoint, also in the group, we have announced the launch of an AI data center, a one gigawatt data center over the period of next five years as well. But the point is that there is a vertical alignment that's happening right through infrastructure to the models, both specialized and foundation models, and to the applications. It's very important to think so that it has got off to a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:25",
                    "end": "00:03:57",
                    "text": " inflection point that cannot be seen in isolation. It has to be seen in terms of the entire vertical stack to make it more impactful. We did a survey among customers both globally and in India. It was 170 of our customers we did a survey and see what's adoption and how is it shaping up. It was a very recent survey that we did. The multiple things that came out, one good news is all about that. Close to 90 odd percentage of the customers are in some form or shape. And this cut across banking,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:57",
                    "end": "00:04:27",
                    "text": " capital markets and insurance customers, as I said, both India and globally, are in some form or shape adopting it with almost 60% fairly advanced state of the adoption standpoint as well. I just want to make one specific call-out here. It'll be interesting to many of you as well. 50% of the customer base feels that a lot of focus needs to be given on enterprise-specific models and some fine-tuned models as well. This is something which is very important given the regulated nature of the financial service industry."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:27",
                    "end": "00:04:57",
                    "text": " how much of secure and how much of accurate models are also needed from an enterprise standpoint. Then of course, a majority, overwhelming majority also felt there is a good need to measure the outcome. Metrics are needed. It's one to talk about POC and pilots, but another to talk about the business outcome. In India, I think it's also interesting to see India, the opportunity and challenges, the nuances around the language. India, I would call it, is a very voice-led"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:57",
                    "end": "00:05:26",
                    "text": " India is a land of many voices, spoken in different languages,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:26",
                    "end": "00:05:56",
                    "text": " across diverse geographies and, most importantly, with unique financial needs. At Tata Capital, financial inclusion means more than just access to finance. It means being present, being understood and being reachable to everyone, everywhere. Today, we are proud to showcase a multi-modal voice AI solution supporting our customers across every key moment of their financial journey. Truly,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:56",
                    "end": "00:06:26",
                    "text": " end to end."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:26",
                    "end": "00:06:55",
                    "text": " Okay, so first of all, are you salaried or self-employed? I don't understand, madam. Excuse me, I mean, do you work somewhere or do you have a business of your own? I have my own shop. Okay, so you have your own shop. Can you tell me what your business is about? Madam, I have a rental shop. Okay, so you have a grocery shop. So instead of your shop, do you have your own shop or do you have it on rent? My rental shop is on rent. Okay, so your shop is on rent?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:55",
                    "end": "00:07:26",
                    "text": " I have sent you a personal information form on WhatsApp which requires some basic details. Fill it in. If you have any doubts, you can ask me. Yes, I have got the form. I will fill it in, madam. It won't take much time, right? No, it won't take much time. Fill it in slowly. I am here. If you have any questions, you can ask me. Madam, what do I have to put in the mother's maiden name? Okay, you are asking about the mother's maiden name. This means the name of your mother before marriage."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:32",
                    "end": "00:08:06",
                    "text": " Do you have a call? If you need anything, let me know. I will help you. Yes, I am filling the form. Madam, I am in the middle. I am filling the form. Okay, no problem. You can fill the form. I am here. Yes, Madam. I have filled the form. Very good. Thank you for filling the form. So, let me tell you again that you will get a loan of 5 lakhs for 3 years at an interest rate of 12.5%. The EMI of which will be 16,727."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:06",
                    "end": "00:08:35",
                    "text": " What you just saw. Well, that is the future. The future of conversations, of customer engagement and of AI led transformation. This isn't just a concept. It is already powering millions of customer conversations. This is how Tata Capital is reimagining engagement. At scale."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:35",
                    "end": "00:09:05",
                    "text": " with responsibility, with empathy, and with real impact. Because when trust finally finds a voice, Bharat engages. And when Bharat engages, Bharat progresses."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:05",
                    "end": "00:09:33",
                    "text": " India-based model based on Sarvam. So I think a lot of linguistic variation that is addressed, and I think this is important because on one side we have the global models available, but it's also important to address the local nuances out there. And the second thing is it's also now allowing from a multi-channel standpoint. So I think the ability to come together more from a customer servicing, I think that has taken on a different league altogether. While that happens, let me call you out some key pointers from."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:33",
                    "end": "00:10:03",
                    "text": " What we have been experiencing over the last, I would say, nine to 12 months, some very grounded reality has been popped up here. From our vantage, we were looking at the adoption in terms of AI for business and AI for IT. IT is pretty straightforward in terms of, it's mostly around a lot of focus on coding assistance and plus the fact that you do a lot using modernization. Huge focus in many forms in terms of technology modernization, the ability of AI to accelerate it. These are not POCs and pilots really getting into production."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:03",
                    "end": "00:10:28",
                    "text": " Then the business side of it, if you look at it, it's got a broad stick happening out there. Right from customer servicing to horizontal functions, corporate functions, and also into wealth and investment advisory as well. So there is a significant progress happening in this space, but just also wanted to call out some of the ground reality that are popping up. ROI challenges still persist. There is still challenges in terms of the accuracy, the."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:28",
                    "end": "00:10:56",
                    "text": " compliance aspect of it, that's still an important thing to be addressed and factored in building an AI solution. The other part is that building an AI solution is one thing. Making it work in an enterprise in terms of integration is so critical. A lot of ROs have been built around building solution, but not necessarily the effort that you need to make it work. Connecting to your enterprise systems, connecting to CRM, your core banking, your learning application, that's a critical aspect of this whole process flow."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:56",
                    "end": "00:11:25",
                    "text": " I think it's fairly recognized that individual use case are not giving you the benefit, the real benefit come by process flows and by workflows. And that you're going to look at workflows differently, not to the traditional left to right, but more from right to left in terms of an outcome standpoint. We see and we believe fine tuning is something that financial services would pick up in a significant way. If you're already seeing that, that would really do an uptake in the next six to nine months."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:25",
                    "end": "00:11:51",
                    "text": " Definitely see that picking up in a big way. Why? Because of the need for much more accurate, much more precise models that need to happen. To say that it will be a hybrid scenario, you would have fine-tuned models on one hand, you'll also have large language models, you would judiciously use it. There is definitely a shift towards Hintiq. Having said that, these are not Hintiq as in high level of autonomy, but a slow."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:51",
                    "end": "00:12:21",
                    "text": " a pedigree of autonomy is being enabled in it, but agentic workflows are happening. The bottom point, I think we refer to that the vertical stack that's evolving, very important to look at an infrastructure in the platform-based thought process. Again, the whole point is that I can do a great POC pilot shiny object, but how do I scale it across and you need a platform engineered thought process to make it happen. So these are some of the experiences that we're going to picked up, thought I can relate that all of you as well. Just wanted to."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:21",
                    "end": "00:12:48",
                    "text": " give a quick perspective around the fine tuning that we talked about. This is something which piggyback on the fact that it will be in hybrid AI solutions and models that you would end up with. You would have both the scenarios. You will continue to leverage large language model for broad, measurable enterprise task and activities with proper grounding. I think it's very important I just double down on the fact that it has to be engineered."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:48",
                    "end": "00:13:18",
                    "text": " How well do you orchestrate your workflows? How well do you build your evals around it? It's very critical. But I think this is an equally important. How well do you build your domain-adapted models in compliance area, in advisory, in credit research? I think it's important to see how well they build appropriate and high accuracy. So our vantage point, we believe, it's going to be convergence between the two, not one over the other because each has its own space. It's very important to traditionally use this. I'll just give you a couple of quick."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:18",
                    "end": "00:13:48",
                    "text": " and adults around it, these are real life examples. Something that we have done for litigation prediction in a claims value chain. If you look at it, what a base model provided versus what a fine tune model provides. And this is not something that happens straight away, you need to iteratively build it, but it's very important to continue, make sure that right from base model, the different techniques of fine tuning to make it more domain adapted in that context. And you could see the high percentage of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:48",
                    "end": "00:14:17",
                    "text": " accuracy as you build that incrementally. And this is where business gets a lot glued to it saying that, hey, now it's come to a point. It's not just technical. It's also important to make it work. Same thing in compliance. And we have used two different models just to give a variation. There are many models out there. But these are models that you could enable it. And whether it's Lama or Mistral Family, there are many other models. For instance, as I said, for several models, we've been working on it in terms of voice, AI standpoint. But the key message I want to leave out here is that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:18",
                    "end": "00:14:44",
                    "text": " fine-tuning or making it proprietary, it's not just technology. I think it's all about building this domain curation. How well do you build the instruction set for a fine-tuning? That need a lot of domain understanding. Once you build that, the initial set, then I think it's a lot about applying the right techniques. It's also equally important to look at the right models. Your slew of models out there,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:44",
                    "end": "00:15:14",
                    "text": " and there is a method to the madness in terms of which model would fit into what context. So all the three are very important. Building the right instruction sets for fine-tuning. For all of you who I'm sure fine-tuning is something that all of you are aware of it, but essentially what you're trying to do is that making sure a broader model has been fine-tuned towards specific domain and adaptation so that it get much more relevant in that context and accuracy of the decisions it takes us to higher level."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:14",
                    "end": "00:15:41",
                    "text": " a series of business scenarios that are evolving in this area. We see this in India, we see this globally. So this is something from our vantage, we see a strong uptick that's happening. So just wanted to give a quick flavor of that before we get into the white thought process. I think it's important to look at it from the context engineering, very important. Enterprise adoption is not just about building use cases, but also making it work in the broad end."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:41",
                    "end": "00:16:09",
                    "text": " it's very critical to make sure that you judicious use of large language models and domain adapted models. With that, let me just show you one specific scenario. Good afternoon. Thank you for calling. Do I have the pleasure of speaking with you? Hi, Catherine. This is Daniel Thompson. I need to report an accident that happened earlier today morning at 4.30 AM."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:09",
                    "end": "00:16:38",
                    "text": " I am sorry to hear. Could I please have your policy number? Yes, it is ALC 76-12-39-05. Thanks, Daniel. I thought that's your policy. Everything looks active. Could you describe what happened? I was already on the highway. I was driving southbound on the M1 Pacific motorway past the Gosford exit in New South Wales, maintaining my lane and speed when a silver SUV suddenly emerged from the left."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:38",
                    "end": "00:17:01",
                    "text": " onto the motorway and site wiped my vehicle so the impact was on the left side yes quite severely it hit around the front left door and fender area i had no chance to react the other driver didn't yield or check for traffic understood um with a police call to the scene"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:01",
                    "end": "00:17:29",
                    "text": " Yes, our patrol unit came by. They've noted the details. I have the event number. I also took several photos of the damage. Great. If you can send us the event number, police notes and any photos that will help us proceed. We'll also need the repair estimate or tow docket if you've received one. I will email all the documents shortly. Perfect."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:29",
                    "end": "00:18:07",
                    "text": " Drive safe and let us know if you need anything else. All right. As you could relate, I think the accent of this is for a UK-based customer. And the context is a little different from the first one that you saw from a voice here, which was a very specific scenario where, in the Tata Capital's story, it was AI at the thick and center of everything in terms of engaging in a customer service. But not actions are not getting taken. It's more of a conversational flow."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:07",
                    "end": "00:18:32",
                    "text": " This one is scenario in which a multi-model with a fine-tune model being brought into the picture. So if you could see the focus of the customer service rep is more in terms of looking at the information that can be gleaned around it. So this is an augmented AI. AI was listening to the whole conversation, providing from an augmented sample what additional information. Normally, this whole process called FN oil, the first notice of loss from the insurance background would know."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:32",
                    "end": "00:19:00",
                    "text": " When you meet with an accident, the first notice of loss that you make a call to the back office to inform. A lot of information you gather at that point. It's very important to capture that information. Today, it's a very back and forth happens, and your claims get delayed because of a simple reason, you haven't captured enough information. In this scenario, behind the scenes, so AI was not in the front-end center, behind the scenes capturing the information, and there was a litigation validator, which I was telling the fine-tune model."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:00",
                    "end": "00:19:28",
                    "text": " looking at it, is there any fraud in this process so that there is a subsequent litigation possible down the line? That information, I don't know how many of you have watched that, that information was gleaned through in between for what I say, this looks like a good call, let's proceed. So I think that the modality is a little different out there, but the whole point is that it's very important to look from both the scenarios. So to put it in a nutshell, I think it's important from an enterprise form, I think proper calibration is very critical."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:28",
                    "end": "00:19:56",
                    "text": " great technology, great capabilities evolving, but it's also important, how do you enable that in a context? Where do you not apply AI also? I think it's important to also look where you don't apply AI. Because at times, we have the thought process of applying AI all over the place. I think it's very important to break it and then make sure that the ones, and that's what our philosophy is that, don't look at from a traditional process flow from left to right, which you break it down to series of task and say where I can apply it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:56",
                    "end": "00:20:26",
                    "text": " Look at from a right to left and see outcome. What is outcome and how do I reapply back? So what are you going to do now for the next 60 minutes or so? What I thought is that, I believe that this need to be broken into three broad areas. You got to have a quick proof of value. You need to build that proof of value because there are certain things you will just move forward then. You say, okay, this is not working. Let me try the next one. You have the ability right now to do a quick prototyping, a quick vibe design and prototyping."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:26",
                    "end": "00:20:55",
                    "text": " Then you go into a rapid build over a period of time, five to six weeks, and then you really scale it. This is a model that we would advocate as enterprises start picking up more and more. Because, as I said, it's very important to look at outcome backwards. It's very important to look for what is the business assurance that I bring in. So we have constructed the next 60 minutes or so in that format. We have certain challenge cards on your table. You have laptops out there. We're going to do this in two formats. We'll assist you in the process."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:55",
                    "end": "00:21:24",
                    "text": " in how do I go through the process, and second is also in terms of those who are more wanting to do on their own, they feel free to try it out. Instructions are also given out there. Let me invite Kavita now to run you through the vibe coding design and prototyping session. Thank you. To continue on the topic of scaling, in financial services, AI doesn't scale when the model is ready. It's when legal compliance..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:24",
                    "end": "00:21:53",
                    "text": " regulatory, all of the body comes together. When they are ready, it's when AI scales. So I think that's one of the challenges that we face when we look at a scaling AI. This is just one of them. And my name is Kavita, and I lead the AI labs for financial services. And I help customers on their journey to adopt AI and scale AI. And one of the other challenges that really come across"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:53",
                    "end": "00:22:20",
                    "text": " when we start discussing about AI use cases is, how do I get the business teams on board? Because if you look at the kind of implementations we have seen so far in AI, you've got knowledge assistance, you've got chatbots, you've got good document AI capability. And if you look at these implementations, they have actually filled the gap where the traditional AI, traditional IT has failed in the sense..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:20",
                    "end": "00:22:48",
                    "text": " inability to access unstructured data. There was a gap which AI solved. So most of the use cases have been around these lines, and that's primarily because AI has not been, we have not let the business teams actually drive AI. So what does it take to help enable the business teams to understand the power of AI and look at more transformative use cases?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:48",
                    "end": "00:23:15",
                    "text": " So that's the construct of what we want to do today. We want you to experience that how a business user or a business team can be enabled with AI to accelerate that whole journey of identifying the right opportunity, doing a working prototype, and actually even testing it out and seeing the proof of value. How do we shift left?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:15",
                    "end": "00:23:43",
                    "text": " in the sense that how do we give business more control and working with the engineering team, build out a quick proof of value that you can validate and then scale to production. Because in one of the earlier conversations I had with one of you here was that scaling is a challenge. But sometimes you don't get to scale because you've not identified the right opportunity. You try to apply AI everywhere. So how do you use AI to help you in that journey?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:43",
                    "end": "00:24:09",
                    "text": " And what we want to achieve through this is let business lead the whole AI redesign. So how do we do that? And that's something that we will cover today. And what we're going to demonstrate to you is not a tool, but how we can bring all the AI capability together for a business audience."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:11",
                    "end": "00:24:41",
                    "text": " Yeah, switching on to that. So what we'll see today, disclaimer before I start, this tool is not a single tool. It's a workbench that we have created that brings together different AI capabilities. And then what we do is that we start off with something called a business challenge, because that's the starting point for a business team. Where do I begin my journey? What is the challenge that I need to address? And for the sake of this demonstration, I'll pick up one use case, something that's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:42",
                    "end": "00:25:00",
                    "text": " A challenge that has come up in one of our customer conversations was that you have a lot of portfolio managers who need to, at the end of the quarter, generate a summary of how their portfolio has performed. And along with collecting a lot of data, they need to actually"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:00",
                    "end": "00:25:30",
                    "text": " check out the sentiment of the stock on the public domain. So there's a lot of data gathering, a lot of analysis that comes in, and as a result, there's a lot of time that goes into that activity, and that delays the overall report, and sometimes this gets delegated to a junior analyst, the quality of the report is affected. So how do I bring all of that in a way that AI can help the portfolio managers solve the problem? So that was the challenge that we picked up, and I'll show you a quick demonstration of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:30",
                    "end": "00:25:56",
                    "text": " how the challenge works. The challenge was just creating an assistant to help the AI agent, to help the portfolio manager actually look up what is the right, what is the data that's available out there on the public domain, or even look up Yahoo Finance to see how the stock has performed. So all of that was being done by AI in the background. And this is a solution that..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:56",
                    "end": "00:26:25",
                    "text": " or when we piloted with our customers, they saw a huge advantage, because today, they were actually outsourcing a lot of this activity to a third party organization, and they were able to bring that back, and bring down the overall time to generate the report, and even the quality of the report started coming out much better. I thought I'd give you a view of the final outcome, and take you back to where we started. And that's the platform that we were trying to show you. So switching on."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:25",
                    "end": "00:26:54",
                    "text": " to that platform. So what we do as a first step is actually pick up the challenge and how do we use AI to ideate and come up with something called the opportunity canvas. So many of our users have, you know, if you see business users are all equipped with co-pilots because it's already available in the enterprise. But what we've observed is that just not giving them AI tools or training them on prompts is really going to help. It's about"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:54",
                    "end": "00:27:25",
                    "text": " helping them to understand that end-to-end process and letting them have complete control. So that's what we've done over here. So we looked at what could be a set of questions that a business user needs to look at when they see a challenge and convert it to an opportunity. So we've curated a set of prompts which can go through the data and come up with what we call the opportunity canvas. So we'll just walk you through those different steps. So I think one thing that we always need to keep in mind is that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:25",
                    "end": "00:27:45",
                    "text": " When we give AI to the business users, the research is primarily on, you can use a lot of research tools available, but the organization context is more important. How do we know what the user is? What are the different users going through? What are their challenges and pain points? So loading the context is the first step."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:45",
                    "end": "00:28:15",
                    "text": " And the reason I'm walking you through this is because once I finish the demo, I want you also to try and just see whether as a business leader or a practitioner how the whole experience will be and how can we really build that capability for business teams to use. So we'll go through this flow. So we have built the context. And then after that, you can even go back and do a deep research. Because sometimes your context is not enough. You want to know how your peers are performing."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:15",
                    "end": "00:28:43",
                    "text": " pull out a deep research from Gemini or ChachiPT. So put that context also in, and then you have an overall knowledge base that your business team can now use to start ideating. So what we have done is created an ideation framework here. Now, as you go through the ideation process, you'll have a lot of questions. Have I captured all the pain points right? Because you want to be sure that you have actually"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:43",
                    "end": "00:29:08",
                    "text": " solving a problem. So there's pain points that you capture. And there are quite a lot of steps. I won't go through all of them. But if you quickly go through, you will have pain points. You'll want to assess what the business impact is. What's the possible solution direction that you want to take? And as we go through these steps, you will see that it's not enough just to analyze the data that's available."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:08",
                    "end": "00:29:36",
                    "text": " but to actually start seeing AI as a thinking partner. That's what we want the business to do, be able to bounce ideas, validate them, feed the input, and finally, at the end of it, create what we call an opportunity canvas. So a lot of thought has gone in. So while these demonstrations, while he brings up the opportunity canvas, something that we have observed is that just giving a tool may not be the only solution."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:36",
                    "end": "00:30:03",
                    "text": " What we do in TC, something that's worked for us, is that we conduct an AI day for business teams, where we try to create that immersive experience where they understand what are the right AI tools and also how to use them and how to apply them in the context. So I'll, if you can just go through the, yeah. So there's a lot of questions, so we don't, we, and it will take a minute to go through, but if our demo is ready, you can show that as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:08",
                    "end": "00:30:37",
                    "text": " Yeah, so this is what the final product looked like. So you give a company name, and it went through the organization's research. It also went to the public domain and started picking up all the stock information and brought that up into an overall summary that can be then used by the portfolio manager to make a decision on how his stock has performed, how do I report that?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:37",
                    "end": "00:31:10",
                    "text": " to his customers? How do I report that into the performance reporting? And if you see, it was instant. And it's gone through, I think, 15 to 20 sites. We've configured all the financial sites to go through and capture the sentiment, capture key financial metrics, all of that instantly on a very high-speed internet. Now, I think we would be done with the Opportunity Canvas. Last one step."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:10",
                    "end": "00:31:39",
                    "text": " So it's important that we go through these steps. So that's why we have consciously made these steps so that you don't pick an idea which is not the right one. So identifying the opportunity, what is the possible solution direction, what's the impact, and what your roadmap will be. And there you have an opportunity canvas, which the user has co-created along with AI. Now you go on to the next step. And this is important."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:39",
                    "end": "00:32:08",
                    "text": " All this is fine. The business team could be comfortable in Opportunity Canvas. But can now the business team, with the help of AI, drive the overall design process? So what you'll see is that you can brainstorm with AI and actually give a direction of what that design should be. And this is what AI has generated. It's kind of broken down that whole opportunity into a set of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:08",
                    "end": "00:32:33",
                    "text": " different agents that can go in. Because as we covered in the previous session, a lot of attention focuses on moving to agentic. So can we look at building what that agentic solution would look like is what you're going to see here. It's not done the coding yet, but what it has done is broken up that entire opportunity into a series of agents which can address this need. And we'll just."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:33",
                    "end": "00:33:01",
                    "text": " You can try that. You will get a chance to try this, so I won't go into the detail. But here, it's come up with four or five agents. It's organized. And as a business team, they can actually go and say, I want these kind of checks and balances. I want to do it this way. And now you see the difference, right? Earlier, it was all left to the engineering team to actually build the solution. But now, we are passing the power back to the business to drive all of this. Now, going forward,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:01",
                    "end": "00:33:33",
                    "text": " What is the next important step is to validate what this opportunity looks like from an experience. So you use a simple wipe coding tool. We have here used both Lovable and Google AI Studio. This tool generates the prompt. You copy the prompt. And then you actually can see it through a visual experience."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:33",
                    "end": "00:34:03",
                    "text": " simple as putting it and if you see the prompt it's all English and you know when you get into your wipe session you'll see that you can actually edit some of it and see what that experience looks like so this will take how much time maybe a minute or so do you have something do you have something that you've already pre-built yeah so this is something that that came out of the tool a very very nice experience which shows what the portfolio summary should look like what are the stocks etc so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:03",
                    "end": "00:34:32",
                    "text": " kind of understood the opportunity and translated that. And you can play around with it and tweak it to the experience. This step becomes very important because what earlier took two to three weeks of building a UX prototype as the end of a design phase actually got converted to minutes. And this is something you take to these stakeholders, validate it, and you're ready with what is that real challenge you want to address, what's the real opportunity comes out so quickly."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:32",
                    "end": "00:35:02",
                    "text": " We can't stop with just a UX prototype, because that just gives you an experience. You need to actually engineer some of the back-end capability. And there are low-core platforms that are available. So in today's live session, we will show you that as well. And one of the examples that we have shown here is a solution called Ne10. It's a low-core platform building. It's not that we are saying this is the right tool. As per your enterprise decision, you could have different options."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:02",
                    "end": "00:35:30",
                    "text": " But this really helps, in our view, to ideate and understand or discuss with your engineering teams how your design should be. You can easily visualize what are the models being used here. Have I thought through all the guardrails? Have I built my business evaluations? All of this can be verified in one shot. And then from this, the code can get generated, and you'll have a working end-to-end solution."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:30",
                    "end": "00:35:52",
                    "text": " So this is the process we followed for quickly prototyping. And the demo that you saw was an example of something that was built using this approach. So what we'd like to do now is shift back to you and see if you'd like to try this. I mean, some of you could. We have laptops configured with."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:52",
                    "end": "00:36:46",
                    "text": " with user IDs, we have all the tools available. It's an opportunity for you to try out and we have mentors. We have mentors available for each table. A lot of our teams here to support you and we would really encourage you to try and we can guide you here as you go along. Explore, yeah. Okay, so let me just walk you through what you would do. So all of your laptops will have a handbook already preloaded."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:46",
                    "end": "00:37:15",
                    "text": " which talks about the use case that you would be picking up. Okay, so, and that is in green, it's tagged as challenge. So that's the use case that you will start with. And you have different phases and you have resources for each of the phase. So feel free to go through this handbook, take it step by step, and we'll help you with it. You are workspace five."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:16",
                    "end": "01:07:22",
                    "text": " So log into your workspace based on the number that's on your table. So when you start, it will be the discovery and design. Yeah, you can start the solution with the resource present there. Okay, so just a quick time check. We just have five minutes. So if you want some help on any particular section, just raise your hand. We'll come and help. But I'm so happy to see many of you have generated very nice UX prototypes."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:07:22",
                    "end": "01:09:27",
                    "text": " And you can also move into that build phase. That's an optional one, but try it out. We'll have mentors to support you."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:09:27",
                    "end": "01:10:13",
                    "text": " uh..."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:10:13",
                    "end": "01:11:40",
                    "text": " So I think we'll just conclude."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:11:45",
                    "end": "01:12:14",
                    "text": " So, so far we spoke about, I only covered, I'll hand it over to him, but what you did essentially comes as step one. You actually did, and I'm so happy that many of you got challenge completed, and you saw your code getting converted, and I see a lot of happy faces because it was, I heard words like, amazing, I can do the same thing with my tools, so very happy to hear all that. So what we did,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:12:14",
                    "end": "01:12:43",
                    "text": " was the first step, proof of value. We've tried to establish this is something, a problem worth solving for my organization and you may have an answer then, you have an answer at the end of the stage. But the journey to take it to production involves something called rapid building. What you've built is a foundation which you can add on more functionality and then create it into a full-fledged solution. And that's where we call the rapid build."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:12:43",
                    "end": "01:13:14",
                    "text": " which is about engineering the full-fledged solution before you go to production. And then finally, the scale. And when you look at scale, and someone asked me that question, you have to build a lot of AI foundation capability, the control plane, a lot of things need to go in. And that's important if you want to bring that scale. I'll just hand it over to Onni to close. I think thank you. Thank you for being a real sport out there. I think all of you joining in, being through the wipe session as well. Great to see all of you continue out there."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:13:14",
                    "end": "01:13:41",
                    "text": " Once again, thanks for being participating. I think this is one important thing that we believe AI would make a difference in the ability to quickly translate an idea to prototype. Not to say that that prototype is what you would eventually scale to production, but that gives you the feel of what can be done. And then you naturally scale it into production in a very calibrated way. But again, thanks for your time. And then being part of it, feel free to reach out. Some of these I'm assuming."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:13:41",
                    "end": "01:13:54",
                    "text": " would have been captured in your email, et cetera. So it's available for you to do some prompting and wiping back home as well if you're adventurous and wanting to extend the focus. Thank you once again and have a good rest of the day."
                }
            ]
        }
    ],
    "projectDeleted": false,
    "codeFrameColumn": [
        {
            "col1": "col1",
            "col2": "col3",
            "researchObjective": "",
            "filterColumn": [],
            "netSetting": true,
            "subNetSetting": true,
            "categorySetting": true,
            "sentimentSetting": true,
            "thinkingBudgetSetting": 2000,
            "questionType": "OPENEND",
            "otherBrand": false,
            "existingCodeFrame": [],
            "isList": false,
            "codeFrameWithDataList": [],
            "codeFrameList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 8881,
                        "CandidatesTokenCount": 3053,
                        "TotalTokenCount": 13345,
                        "thoughtsTokenCount": 1411,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Net": "AI Adoption & Trends in Financial Services",
                                "Subnets": [
                                    {
                                        "Subnet": "Current State & Acceleration",
                                        "Categories": [
                                            {
                                                "Category": "Hyper Acceleration Phase",
                                                "Summary": "AI has entered a rapid development and adoption stage with widespread recognition."
                                            },
                                            {
                                                "Category": "High Adoption Rate",
                                                "Summary": "Approximately 90% of financial services customers are adopting AI, with 60% in advanced stages."
                                            },
                                            {
                                                "Category": "Shift to Agents in Action",
                                                "Summary": "Movement from traditional chatbots to more proactive, capable AI agents."
                                            }
                                        ],
                                        "Summary": "Highlights the rapid advancement of AI, its broad recognition, and significant adoption across financial services, indicating a mature state of deployment."
                                    },
                                    {
                                        "Subnet": "Key Trends & Focus Areas",
                                        "Categories": [
                                            {
                                                "Category": "Multimodal Adoption",
                                                "Summary": "Significant focus on using multiple AI models to accomplish specific tasks or process flows."
                                            },
                                            {
                                                "Category": "Context Engineering",
                                                "Summary": "Emphasizes the critical need to build relevant context before engaging with AI models for effective outcomes."
                                            },
                                            {
                                                "Category": "Enterprise-Specific Models",
                                                "Summary": "Strong demand for secure, accurate, and fine-tuned models tailored for regulated financial industries."
                                            }
                                        ],
                                        "Summary": "Explores emerging trends like multimodal adoption, the crucial role of context engineering, and the increasing importance of specialized enterprise AI models."
                                    },
                                    {
                                        "Subnet": "India-Specific Nuances",
                                        "Categories": [
                                            {
                                                "Category": "Voice-Led Market",
                                                "Summary": "India is identified as a predominantly voice-led country, requiring voice-centric AI solutions."
                                            },
                                            {
                                                "Category": "Linguistic Variation",
                                                "Summary": "The need to address diverse languages and dialects to ensure broader reach and financial inclusion across India."
                                            },
                                            {
                                                "Category": "Financial Inclusion",
                                                "Summary": "AI's role in making financial services accessible, understood, and reachable to everyone, everywhere in India."
                                            }
                                        ],
                                        "Summary": "Addresses the unique opportunities and challenges of AI adoption in India, particularly regarding linguistic diversity and its impact on financial inclusion."
                                    }
                                ],
                                "Summary": "Discusses the current hyper-acceleration phase of AI, key shifts in financial services adoption, and survey findings regarding customer readiness and focus areas globally and in India."
                            },
                            {
                                "Net": "AI Technology & Implementation",
                                "Subnets": [
                                    {
                                        "Subnet": "Evolution of AI Models",
                                        "Categories": [
                                            {
                                                "Category": "Rapid Model Releases",
                                                "Summary": "New AI models are being launched frequently, improving quality and driving competition."
                                            },
                                            {
                                                "Category": "Hybrid AI Solutions",
                                                "Summary": "Strategic use of both large language models (LLMs) for broad tasks and fine-tuned/domain-adapted models for specific needs."
                                            },
                                            {
                                                "Category": "Model Selection & Context",
                                                "Summary": "Importance of judiciously selecting the appropriate AI model for specific enterprise contexts to maximize effectiveness."
                                            }
                                        ],
                                        "Summary": "Covers the rapid pace of model development and the strategic imperative of utilizing different model types effectively within an enterprise."
                                    },
                                    {
                                        "Subnet": "Fine-tuning & Domain Adaptation",
                                        "Categories": [
                                            {
                                                "Category": "Improved Accuracy & Precision",
                                                "Summary": "Fine-tuning significantly enhances model accuracy and precision for specialized, domain-specific tasks."
                                            },
                                            {
                                                "Category": "Domain Curation",
                                                "Summary": "Building robust instruction sets for fine-tuning requires deep domain understanding to ensure relevance and performance."
                                            },
                                            {
                                                "Category": "Hybrid Scenarios for Accuracy",
                                                "Summary": "Combining LLMs with fine-tuned models to achieve high accuracy for critical functions like compliance and advisory."
                                            }
                                        ],
                                        "Summary": "Emphasizes the critical role of fine-tuning for achieving accuracy and relevance in specific financial domains, driven by domain expertise."
                                    },
                                    {
                                        "Subnet": "Vertical Convergence & Infrastructure",
                                        "Categories": [
                                            {
                                                "Category": "Infrastructure & Hardware Focus",
                                                "Summary": "Growing global and domestic investments in AI data centers and underlying hardware infrastructure."
                                            },
                                            {
                                                "Category": "Full Vertical Stack Integration",
                                                "Summary": "The necessity to view AI implementation as an integrated process across the entire vertical stack, from infrastructure to applications."
                                            },
                                            {
                                                "Category": "Platform-Based Approach",
                                                "Summary": "The need for an engineered platform thought process to effectively scale AI solutions across the enterprise beyond individual POCs."
                                            }
                                        ],
                                        "Summary": "Highlights the trend of integrating AI across the entire technology stack, from hardware to applications, for more impactful deployment."
                                    }
                                ],
                                "Summary": "Covers the technical evolution of AI, including model types, the importance of context and fine-tuning, and the vertical integration of AI infrastructure for scaled deployment."
                            },
                            {
                                "Net": "Challenges in AI Deployment",
                                "Subnets": [
                                    {
                                        "Subnet": "Business & Technical Hurdles",
                                        "Categories": [
                                            {
                                                "Category": "ROI Challenges",
                                                "Summary": "Persistent difficulties in demonstrating clear return on investment for AI initiatives."
                                            },
                                            {
                                                "Category": "Accuracy & Compliance Concerns",
                                                "Summary": "Ongoing challenges related to ensuring AI model accuracy and adherence to regulatory compliance standards."
                                            },
                                            {
                                                "Category": "Integration with Enterprise Systems",
                                                "Summary": "The critical effort required to integrate AI solutions with existing enterprise systems (CRM, core banking, etc.) for real-world functionality."
                                            }
                                        ],
                                        "Summary": "Addresses the significant business and technical hurdles encountered during the practical deployment and scaling of AI solutions in enterprises."
                                    },
                                    {
                                        "Subnet": "Scaling & Operationalization",
                                        "Categories": [
                                            {
                                                "Category": "Difficulty in Scaling POCs",
                                                "Summary": "Challenges in moving AI proof-of-concepts (POCs) and pilots into full-scale production."
                                            },
                                            {
                                                "Category": "Lack of AI Foundation Capability",
                                                "Summary": "The absence of robust AI foundation capabilities, control planes, and governance mechanisms hindering enterprise-wide scaling."
                                            },
                                            {
                                                "Category": "Regulatory & Legal Readiness",
                                                "Summary": "AI scaling is dependent on the readiness of legal, compliance, and regulatory bodies, not just model readiness."
                                            }
                                        ],
                                        "Summary": "Focuses on the complexities of scaling AI solutions from pilot projects to full production, including operational and governance aspects."
                                    },
                                    {
                                        "Subnet": "Strategic Application & Outcome Measurement",
                                        "Categories": [
                                            {
                                                "Category": "Identifying Right Opportunities",
                                                "Summary": "Challenge of determining where AI can provide the most transformative value versus applying it everywhere indiscriminately."
                                            },
                                            {
                                                "Category": "Measuring Business Outcome",
                                                "Summary": "The critical need to measure the tangible business outcomes and metrics, beyond just POCs and pilots."
                                            },
                                            {
                                                "Category": "Strategic Application of AI",
                                                "Summary": "Emphasizes the importance of knowing where and where not to apply AI strategically based on desired outcomes."
                                            }
                                        ],
                                        "Summary": "Discusses the challenges related to strategically applying AI, identifying the right use cases, and effectively measuring the business impact and outcomes."
                                    }
                                ],
                                "Summary": "Addresses the multi-faceted challenges faced during the practical implementation and scaling of AI in an enterprise environment, from ROI to regulatory compliance."
                            },
                            {
                                "Net": "AI Application & Use Cases",
                                "Subnets": [
                                    {
                                        "Subnet": "Customer Engagement & Service",
                                        "Categories": [
                                            {
                                                "Category": "Multi-modal Voice AI",
                                                "Summary": "Demonstration of a multi-modal voice AI solution supporting customers across their financial journey, e.g., Tata Capital."
                                            },
                                            {
                                                "Category": "Conversational AI Flows",
                                                "Summary": "AI-led transformation of customer conversations and engagement, enabling empathy and impact at scale."
                                            },
                                            {
                                                "Category": "Augmented AI for Service Reps",
                                                "Summary": "AI assisting customer service representatives by capturing information and providing insights during conversations, e.g., insurance claims."
                                            }
                                        ],
                                        "Summary": "Highlights AI's transformative role in enhancing customer engagement and service through advanced conversational and multi-modal solutions."
                                    },
                                    {
                                        "Subnet": "Financial Operations & Advisory",
                                        "Categories": [
                                            {
                                                "Category": "Portfolio Summary Generation",
                                                "Summary": "AI assisting portfolio managers in generating quarterly performance summaries and checking stock sentiment from public domains."
                                            },
                                            {
                                                "Category": "Litigation Prediction",
                                                "Summary": "Using fine-tuned AI models for predicting litigation in claims value chains, enhancing fraud detection and risk assessment."
                                            },
                                            {
                                                "Category": "Compliance & Advisory Support",
                                                "Summary": "AI's application in compliance, credit research, and investment advisory for improved decision-making and accuracy."
                                            }
                                        ],
                                        "Summary": "Showcases AI applications in financial operations, including portfolio management, litigation prediction, and compliance advisory roles."
                                    },
                                    {
                                        "Subnet": "IT & Business Function Modernization",
                                        "Categories": [
                                            {
                                                "Category": "Coding Assistance",
                                                "Summary": "AI's role in assisting with coding and technology modernization efforts within IT departments."
                                            },
                                            {
                                                "Category": "Process Flow Acceleration",
                                                "Summary": "Leveraging AI to accelerate various business and corporate functions beyond just customer service."
                                            },
                                            {
                                                "Category": "Production-Ready AI",
                                                "Summary": "Moving AI from POCs and pilots to real production environments for significant modernization impact."
                                            }
                                        ],
                                        "Summary": "Illustrates how AI is being applied to modernize both IT operations (coding, technology) and various horizontal corporate functions."
                                    }
                                ],
                                "Summary": "Highlights practical examples and scenarios where AI is being applied across various functions, from customer service to IT modernization and financial operations."
                            },
                            {
                                "Net": "Business-Led AI Development & Prototyping",
                                "Subnets": [
                                    {
                                        "Subnet": "Empowering Business Teams",
                                        "Categories": [
                                            {
                                                "Category": "Business-Driven AI Redesign",
                                                "Summary": "Shifting control to business teams to drive the AI solution design process."
                                            },
                                            {
                                                "Category": "AI as a Thinking Partner",
                                                "Summary": "Enabling business users to leverage AI for ideation, validating ideas, and creating opportunity canvases."
                                            },
                                            {
                                                "Category": "AI Day for Business Teams",
                                                "Summary": "Conducting immersive workshops to educate business teams on AI tools and their application within organizational context."
                                            }
                                        ],
                                        "Summary": "Focuses on strategies and methodologies for enabling business teams to actively lead and drive AI initiatives and design processes."
                                    },
                                    {
                                        "Subnet": "Rapid Proof-of-Value Process",
                                        "Categories": [
                                            {
                                                "Category": "Quick Prototyping & Vibe Design",
                                                "Summary": "The ability to rapidly build quick prototypes and design solutions to test ideas efficiently."
                                            },
                                            {
                                                "Category": "Opportunity Canvas Creation",
                                                "Summary": "AI assisting in curating prompts to help business users convert challenges into structured opportunity canvases."
                                            },
                                            {
                                                "Category": "UX Prototype Generation",
                                                "Summary": "Converting design ideas into visual UX prototypes in minutes using AI tools, accelerating stakeholder validation."
                                            }
                                        ],
                                        "Summary": "Describes a streamlined process for rapidly developing proof-of-value, from ideation and design to quick prototyping and validation."
                                    },
                                    {
                                        "Subnet": "Low-Code Build & Validation",
                                        "Categories": [
                                            {
                                                "Category": "Low-Code Platform Utilization",
                                                "Summary": "Using low-code platforms (e.g., Ne10) to engineer backend capabilities and visualize designs."
                                            },
                                            {
                                                "Category": "Guardrails & Business Evals",
                                                "Summary": "Verifying guardrails and building business evaluations within the prototyping phase to ensure responsible AI."
                                            },
                                            {
                                                "Category": "Working End-to-End Solution",
                                                "Summary": "Generating functional code from validated designs to create a working end-to-end solution rapidly."
                                            }
                                        ],
                                        "Summary": "Explains the process of leveraging low-code platforms for rapid backend engineering, validating designs, and generating working solutions."
                                    }
                                ],
                                "Summary": "Describes the methodology for involving business users in AI ideation, design, and rapid proof-of-value creation, emphasizing business control and quick iteration."
                            },
                            {
                                "Net": "Future Vision & Strategic Approach",
                                "Subnets": [
                                    {
                                        "Subnet": "Outcome-Driven Development",
                                        "Categories": [
                                            {
                                                "Category": "Right-to-Left Approach",
                                                "Summary": "Focusing on desired business outcomes first and then working backward to define AI applications."
                                            },
                                            {
                                                "Category": "Business Assurance",
                                                "Summary": "Prioritizing the business assurance that AI brings, not just technical feasibility."
                                            },
                                            {
                                                "Category": "Calibrated Scaling",
                                                "Summary": "Gradually and strategically scaling prototypes into full production with careful calibration."
                                            }
                                        ],
                                        "Summary": "Emphasizes an approach where AI development is guided by clear business outcomes rather than solely focusing on individual tasks or technologies."
                                    },
                                    {
                                        "Subnet": "Strategic AI Application",
                                        "Categories": [
                                            {
                                                "Category": "Judicious Use of Models",
                                                "Summary": "Careful consideration of when and where to apply large language models versus domain-adapted models."
                                            },
                                            {
                                                "Category": "Breaking Down Process Flows",
                                                "Summary": "Deconstructing traditional process flows into tasks to identify optimal points for AI application."
                                            },
                                            {
                                                "Category": "Avoiding AI Over-Application",
                                                "Summary": "Recognizing scenarios where AI may not be the optimal solution, avoiding indiscriminate application."
                                            }
                                        ],
                                        "Summary": "Discusses the strategic considerations for applying AI, including judicious model selection and understanding when and where AI delivers the most value."
                                    },
                                    {
                                        "Subnet": "Building for Scale",
                                        "Categories": [
                                            {
                                                "Category": "Rapid Build Phase",
                                                "Summary": "A phase dedicated to engineering a full-fledged solution with added functionality before production deployment."
                                            },
                                            {
                                                "Category": "AI Foundation Capability",
                                                "Summary": "The necessity of building robust AI foundational capabilities, control planes, and governance for enterprise-wide scaling."
                                            },
                                            {
                                                "Category": "Infrastructure & Platform Focus",
                                                "Summary": "Emphasizing an infrastructure and platform-based thought process for scalable AI deployment, rather than isolated POCs."
                                            }
                                        ],
                                        "Summary": "Outlines the long-term perspective and architectural considerations necessary to build, sustain, and scale AI solutions effectively across an enterprise."
                                    }
                                ],
                                "Summary": "Encompasses the long-term perspective, platform thinking, outcome-driven development, and balancing different AI models for sustainable and impactful deployment."
                            },
                            {
                                "Net": "996 Non specific TOPIC",
                                "Subnets": [
                                    {
                                        "Subnet": "Non specific SUBTOPIC",
                                        "Categories": [
                                            {
                                                "Category": "Non specific CATEGORY",
                                                "Summary": "Content that is irrelevant or cannot be easily coded into defined themes, potentially requiring further analysis."
                                            }
                                        ],
                                        "Summary": "Responses that are irrelevant but do not fit clearly into other specific non-codable categories and may require further upcoding."
                                    }
                                ],
                                "Summary": "This topic will contain all those responses which are irrelevant but could not be coded in the themes within the codeframe easily. These may need to be upcoded."
                            },
                            {
                                "Net": "997 Don't Know Can't Say",
                                "Subnets": [
                                    {
                                        "Subnet": "Non-committal/Vague Response",
                                        "Categories": [
                                            {
                                                "Category": "Unable to Provide Opinion",
                                                "Summary": "Responses indicating a lack of knowledge, uncertainty, or an inability to answer the question."
                                            }
                                        ],
                                        "Summary": "Responses that are non-committal, vague, or express uncertainty, suggesting the respondent cannot provide a clear answer."
                                    }
                                ],
                                "Summary": "This topic will contain all those responses which are non-committal and vague."
                            },
                            {
                                "Net": "998 Not Applicable",
                                "Subnets": [
                                    {
                                        "Subnet": "Irrelevant to Question",
                                        "Categories": [
                                            {
                                                "Category": "Out of Scope Response",
                                                "Summary": "Responses that are completely irrelevant to the question asked or the overall project scope."
                                            }
                                        ],
                                        "Summary": "Responses that are irrelevant and not applicable to the specific question asked or the project at hand."
                                    }
                                ],
                                "Summary": "This topic will contain all those responses which are irrelevant, not applicable to the question."
                            },
                            {
                                "Net": "999 None-Nothing",
                                "Subnets": [
                                    {
                                        "Subnet": "Gibberish/Meaningless",
                                        "Categories": [
                                            {
                                                "Category": "Unintelligible Content",
                                                "Summary": "Responses consisting of plain gibberish, non-language characters, or content that holds no discernible meaning."
                                            }
                                        ],
                                        "Summary": "Responses that are plain gibberish, contain non-language characters, or convey no meaningful information whatsoever."
                                    }
                                ],
                                "Summary": "This topic will contain all those responses which are plain gibberish non-language characters or mean nothing."
                            }
                        ],
                        "Summary": "The provided text discusses the accelerated adoption of AI in financial services, highlighting key trends, challenges, and practical implementation strategies. It covers the shift from chatbots to AI agents, the importance of multimodal adoption, context engineering, and fine-tuning for enterprise-specific needs, particularly in a diverse market like India. The speaker emphasizes the vertical convergence of AI, from infrastructure to applications, and the need for a platform-based approach to scale solutions. A significant portion details a 'vibe session' workshop designed to empower business users to lead AI ideation, design, and rapid prototyping, demonstrating how AI can aid in creating proof-of-value for complex business challenges like portfolio performance reporting and insurance claims processing. Key challenges such as ROI, accuracy, compliance, integration, and scaling are addressed, advocating for an outcome-driven development process where AI is applied strategically."
                    }
                }
            ],
            "onlyCodeFrame": true,
            "errorList": [],
            "codeFrameGenerateType": "CODE_FRAME_ONLY",
            "requestType": "CODEFRAME",
            "requestMode": "DEVELOPMENT",
            "sourceLanguage": "English",
            "targetLanguage": "English",
            "response": {
                "status": "OK",
                "data": {
                    "status": "OK",
                    "PromptTokenCount": 8881,
                    "CandidatesTokenCount": 3053,
                    "TotalTokenCount": 13345,
                    "thoughtsTokenCount": 1411,
                    "finishReason": "STOP",
                    "data": [
                        {
                            "Net": "AI Adoption & Trends in Financial Services",
                            "Subnets": [
                                {
                                    "Subnet": "Current State & Acceleration",
                                    "Categories": [
                                        {
                                            "Category": "Hyper Acceleration Phase",
                                            "Summary": "AI has entered a rapid development and adoption stage with widespread recognition.",
                                            "code": 3
                                        },
                                        {
                                            "Category": "High Adoption Rate",
                                            "Summary": "Approximately 90% of financial services customers are adopting AI, with 60% in advanced stages.",
                                            "code": 4
                                        },
                                        {
                                            "Category": "Shift to Agents in Action",
                                            "Summary": "Movement from traditional chatbots to more proactive, capable AI agents.",
                                            "code": 5
                                        }
                                    ],
                                    "Summary": "Highlights the rapid advancement of AI, its broad recognition, and significant adoption across financial services, indicating a mature state of deployment.",
                                    "code": 2
                                },
                                {
                                    "Subnet": "Key Trends & Focus Areas",
                                    "Categories": [
                                        {
                                            "Category": "Multimodal Adoption",
                                            "Summary": "Significant focus on using multiple AI models to accomplish specific tasks or process flows.",
                                            "code": 7
                                        },
                                        {
                                            "Category": "Context Engineering",
                                            "Summary": "Emphasizes the critical need to build relevant context before engaging with AI models for effective outcomes.",
                                            "code": 8
                                        },
                                        {
                                            "Category": "Enterprise-Specific Models",
                                            "Summary": "Strong demand for secure, accurate, and fine-tuned models tailored for regulated financial industries.",
                                            "code": 9
                                        }
                                    ],
                                    "Summary": "Explores emerging trends like multimodal adoption, the crucial role of context engineering, and the increasing importance of specialized enterprise AI models.",
                                    "code": 6
                                },
                                {
                                    "Subnet": "India-Specific Nuances",
                                    "Categories": [
                                        {
                                            "Category": "Voice-Led Market",
                                            "Summary": "India is identified as a predominantly voice-led country, requiring voice-centric AI solutions.",
                                            "code": 11
                                        },
                                        {
                                            "Category": "Linguistic Variation",
                                            "Summary": "The need to address diverse languages and dialects to ensure broader reach and financial inclusion across India.",
                                            "code": 12
                                        },
                                        {
                                            "Category": "Financial Inclusion",
                                            "Summary": "AI's role in making financial services accessible, understood, and reachable to everyone, everywhere in India.",
                                            "code": 13
                                        }
                                    ],
                                    "Summary": "Addresses the unique opportunities and challenges of AI adoption in India, particularly regarding linguistic diversity and its impact on financial inclusion.",
                                    "code": 10
                                }
                            ],
                            "Summary": "Discusses the current hyper-acceleration phase of AI, key shifts in financial services adoption, and survey findings regarding customer readiness and focus areas globally and in India.",
                            "code": 1
                        },
                        {
                            "Net": "AI Technology & Implementation",
                            "Subnets": [
                                {
                                    "Subnet": "Evolution of AI Models",
                                    "Categories": [
                                        {
                                            "Category": "Rapid Model Releases",
                                            "Summary": "New AI models are being launched frequently, improving quality and driving competition.",
                                            "code": 16
                                        },
                                        {
                                            "Category": "Hybrid AI Solutions",
                                            "Summary": "Strategic use of both large language models (LLMs) for broad tasks and fine-tuned/domain-adapted models for specific needs.",
                                            "code": 17
                                        },
                                        {
                                            "Category": "Model Selection & Context",
                                            "Summary": "Importance of judiciously selecting the appropriate AI model for specific enterprise contexts to maximize effectiveness.",
                                            "code": 18
                                        }
                                    ],
                                    "Summary": "Covers the rapid pace of model development and the strategic imperative of utilizing different model types effectively within an enterprise.",
                                    "code": 15
                                },
                                {
                                    "Subnet": "Fine-tuning & Domain Adaptation",
                                    "Categories": [
                                        {
                                            "Category": "Improved Accuracy & Precision",
                                            "Summary": "Fine-tuning significantly enhances model accuracy and precision for specialized, domain-specific tasks.",
                                            "code": 20
                                        },
                                        {
                                            "Category": "Domain Curation",
                                            "Summary": "Building robust instruction sets for fine-tuning requires deep domain understanding to ensure relevance and performance.",
                                            "code": 21
                                        },
                                        {
                                            "Category": "Hybrid Scenarios for Accuracy",
                                            "Summary": "Combining LLMs with fine-tuned models to achieve high accuracy for critical functions like compliance and advisory.",
                                            "code": 22
                                        }
                                    ],
                                    "Summary": "Emphasizes the critical role of fine-tuning for achieving accuracy and relevance in specific financial domains, driven by domain expertise.",
                                    "code": 19
                                },
                                {
                                    "Subnet": "Vertical Convergence & Infrastructure",
                                    "Categories": [
                                        {
                                            "Category": "Infrastructure & Hardware Focus",
                                            "Summary": "Growing global and domestic investments in AI data centers and underlying hardware infrastructure.",
                                            "code": 24
                                        },
                                        {
                                            "Category": "Full Vertical Stack Integration",
                                            "Summary": "The necessity to view AI implementation as an integrated process across the entire vertical stack, from infrastructure to applications.",
                                            "code": 25
                                        },
                                        {
                                            "Category": "Platform-Based Approach",
                                            "Summary": "The need for an engineered platform thought process to effectively scale AI solutions across the enterprise beyond individual POCs.",
                                            "code": 26
                                        }
                                    ],
                                    "Summary": "Highlights the trend of integrating AI across the entire technology stack, from hardware to applications, for more impactful deployment.",
                                    "code": 23
                                }
                            ],
                            "Summary": "Covers the technical evolution of AI, including model types, the importance of context and fine-tuning, and the vertical integration of AI infrastructure for scaled deployment.",
                            "code": 14
                        },
                        {
                            "Net": "Challenges in AI Deployment",
                            "Subnets": [
                                {
                                    "Subnet": "Business & Technical Hurdles",
                                    "Categories": [
                                        {
                                            "Category": "ROI Challenges",
                                            "Summary": "Persistent difficulties in demonstrating clear return on investment for AI initiatives.",
                                            "code": 29
                                        },
                                        {
                                            "Category": "Accuracy & Compliance Concerns",
                                            "Summary": "Ongoing challenges related to ensuring AI model accuracy and adherence to regulatory compliance standards.",
                                            "code": 30
                                        },
                                        {
                                            "Category": "Integration with Enterprise Systems",
                                            "Summary": "The critical effort required to integrate AI solutions with existing enterprise systems (CRM, core banking, etc.) for real-world functionality.",
                                            "code": 31
                                        }
                                    ],
                                    "Summary": "Addresses the significant business and technical hurdles encountered during the practical deployment and scaling of AI solutions in enterprises.",
                                    "code": 28
                                },
                                {
                                    "Subnet": "Scaling & Operationalization",
                                    "Categories": [
                                        {
                                            "Category": "Difficulty in Scaling POCs",
                                            "Summary": "Challenges in moving AI proof-of-concepts (POCs) and pilots into full-scale production.",
                                            "code": 33
                                        },
                                        {
                                            "Category": "Lack of AI Foundation Capability",
                                            "Summary": "The absence of robust AI foundation capabilities, control planes, and governance mechanisms hindering enterprise-wide scaling.",
                                            "code": 34
                                        },
                                        {
                                            "Category": "Regulatory & Legal Readiness",
                                            "Summary": "AI scaling is dependent on the readiness of legal, compliance, and regulatory bodies, not just model readiness.",
                                            "code": 35
                                        }
                                    ],
                                    "Summary": "Focuses on the complexities of scaling AI solutions from pilot projects to full production, including operational and governance aspects.",
                                    "code": 32
                                },
                                {
                                    "Subnet": "Strategic Application & Outcome Measurement",
                                    "Categories": [
                                        {
                                            "Category": "Identifying Right Opportunities",
                                            "Summary": "Challenge of determining where AI can provide the most transformative value versus applying it everywhere indiscriminately.",
                                            "code": 37
                                        },
                                        {
                                            "Category": "Measuring Business Outcome",
                                            "Summary": "The critical need to measure the tangible business outcomes and metrics, beyond just POCs and pilots.",
                                            "code": 38
                                        },
                                        {
                                            "Category": "Strategic Application of AI",
                                            "Summary": "Emphasizes the importance of knowing where and where not to apply AI strategically based on desired outcomes.",
                                            "code": 39
                                        }
                                    ],
                                    "Summary": "Discusses the challenges related to strategically applying AI, identifying the right use cases, and effectively measuring the business impact and outcomes.",
                                    "code": 36
                                }
                            ],
                            "Summary": "Addresses the multi-faceted challenges faced during the practical implementation and scaling of AI in an enterprise environment, from ROI to regulatory compliance.",
                            "code": 27
                        },
                        {
                            "Net": "AI Application & Use Cases",
                            "Subnets": [
                                {
                                    "Subnet": "Customer Engagement & Service",
                                    "Categories": [
                                        {
                                            "Category": "Multi-modal Voice AI",
                                            "Summary": "Demonstration of a multi-modal voice AI solution supporting customers across their financial journey, e.g., Tata Capital.",
                                            "code": 42
                                        },
                                        {
                                            "Category": "Conversational AI Flows",
                                            "Summary": "AI-led transformation of customer conversations and engagement, enabling empathy and impact at scale.",
                                            "code": 43
                                        },
                                        {
                                            "Category": "Augmented AI for Service Reps",
                                            "Summary": "AI assisting customer service representatives by capturing information and providing insights during conversations, e.g., insurance claims.",
                                            "code": 44
                                        }
                                    ],
                                    "Summary": "Highlights AI's transformative role in enhancing customer engagement and service through advanced conversational and multi-modal solutions.",
                                    "code": 41
                                },
                                {
                                    "Subnet": "Financial Operations & Advisory",
                                    "Categories": [
                                        {
                                            "Category": "Portfolio Summary Generation",
                                            "Summary": "AI assisting portfolio managers in generating quarterly performance summaries and checking stock sentiment from public domains.",
                                            "code": 46
                                        },
                                        {
                                            "Category": "Litigation Prediction",
                                            "Summary": "Using fine-tuned AI models for predicting litigation in claims value chains, enhancing fraud detection and risk assessment.",
                                            "code": 47
                                        },
                                        {
                                            "Category": "Compliance & Advisory Support",
                                            "Summary": "AI's application in compliance, credit research, and investment advisory for improved decision-making and accuracy.",
                                            "code": 48
                                        }
                                    ],
                                    "Summary": "Showcases AI applications in financial operations, including portfolio management, litigation prediction, and compliance advisory roles.",
                                    "code": 45
                                },
                                {
                                    "Subnet": "IT & Business Function Modernization",
                                    "Categories": [
                                        {
                                            "Category": "Coding Assistance",
                                            "Summary": "AI's role in assisting with coding and technology modernization efforts within IT departments.",
                                            "code": 50
                                        },
                                        {
                                            "Category": "Process Flow Acceleration",
                                            "Summary": "Leveraging AI to accelerate various business and corporate functions beyond just customer service.",
                                            "code": 51
                                        },
                                        {
                                            "Category": "Production-Ready AI",
                                            "Summary": "Moving AI from POCs and pilots to real production environments for significant modernization impact.",
                                            "code": 52
                                        }
                                    ],
                                    "Summary": "Illustrates how AI is being applied to modernize both IT operations (coding, technology) and various horizontal corporate functions.",
                                    "code": 49
                                }
                            ],
                            "Summary": "Highlights practical examples and scenarios where AI is being applied across various functions, from customer service to IT modernization and financial operations.",
                            "code": 40
                        },
                        {
                            "Net": "Business-Led AI Development & Prototyping",
                            "Subnets": [
                                {
                                    "Subnet": "Empowering Business Teams",
                                    "Categories": [
                                        {
                                            "Category": "Business-Driven AI Redesign",
                                            "Summary": "Shifting control to business teams to drive the AI solution design process.",
                                            "code": 55
                                        },
                                        {
                                            "Category": "AI as a Thinking Partner",
                                            "Summary": "Enabling business users to leverage AI for ideation, validating ideas, and creating opportunity canvases.",
                                            "code": 56
                                        },
                                        {
                                            "Category": "AI Day for Business Teams",
                                            "Summary": "Conducting immersive workshops to educate business teams on AI tools and their application within organizational context.",
                                            "code": 57
                                        }
                                    ],
                                    "Summary": "Focuses on strategies and methodologies for enabling business teams to actively lead and drive AI initiatives and design processes.",
                                    "code": 54
                                },
                                {
                                    "Subnet": "Rapid Proof-of-Value Process",
                                    "Categories": [
                                        {
                                            "Category": "Quick Prototyping & Vibe Design",
                                            "Summary": "The ability to rapidly build quick prototypes and design solutions to test ideas efficiently.",
                                            "code": 59
                                        },
                                        {
                                            "Category": "Opportunity Canvas Creation",
                                            "Summary": "AI assisting in curating prompts to help business users convert challenges into structured opportunity canvases.",
                                            "code": 60
                                        },
                                        {
                                            "Category": "UX Prototype Generation",
                                            "Summary": "Converting design ideas into visual UX prototypes in minutes using AI tools, accelerating stakeholder validation.",
                                            "code": 61
                                        }
                                    ],
                                    "Summary": "Describes a streamlined process for rapidly developing proof-of-value, from ideation and design to quick prototyping and validation.",
                                    "code": 58
                                },
                                {
                                    "Subnet": "Low-Code Build & Validation",
                                    "Categories": [
                                        {
                                            "Category": "Low-Code Platform Utilization",
                                            "Summary": "Using low-code platforms (e.g., Ne10) to engineer backend capabilities and visualize designs.",
                                            "code": 63
                                        },
                                        {
                                            "Category": "Guardrails & Business Evals",
                                            "Summary": "Verifying guardrails and building business evaluations within the prototyping phase to ensure responsible AI.",
                                            "code": 64
                                        },
                                        {
                                            "Category": "Working End-to-End Solution",
                                            "Summary": "Generating functional code from validated designs to create a working end-to-end solution rapidly.",
                                            "code": 65
                                        }
                                    ],
                                    "Summary": "Explains the process of leveraging low-code platforms for rapid backend engineering, validating designs, and generating working solutions.",
                                    "code": 62
                                }
                            ],
                            "Summary": "Describes the methodology for involving business users in AI ideation, design, and rapid proof-of-value creation, emphasizing business control and quick iteration.",
                            "code": 53
                        },
                        {
                            "Net": "Future Vision & Strategic Approach",
                            "Subnets": [
                                {
                                    "Subnet": "Outcome-Driven Development",
                                    "Categories": [
                                        {
                                            "Category": "Right-to-Left Approach",
                                            "Summary": "Focusing on desired business outcomes first and then working backward to define AI applications.",
                                            "code": 68
                                        },
                                        {
                                            "Category": "Business Assurance",
                                            "Summary": "Prioritizing the business assurance that AI brings, not just technical feasibility.",
                                            "code": 69
                                        },
                                        {
                                            "Category": "Calibrated Scaling",
                                            "Summary": "Gradually and strategically scaling prototypes into full production with careful calibration.",
                                            "code": 70
                                        }
                                    ],
                                    "Summary": "Emphasizes an approach where AI development is guided by clear business outcomes rather than solely focusing on individual tasks or technologies.",
                                    "code": 67
                                },
                                {
                                    "Subnet": "Strategic AI Application",
                                    "Categories": [
                                        {
                                            "Category": "Judicious Use of Models",
                                            "Summary": "Careful consideration of when and where to apply large language models versus domain-adapted models.",
                                            "code": 72
                                        },
                                        {
                                            "Category": "Breaking Down Process Flows",
                                            "Summary": "Deconstructing traditional process flows into tasks to identify optimal points for AI application.",
                                            "code": 73
                                        },
                                        {
                                            "Category": "Avoiding AI Over-Application",
                                            "Summary": "Recognizing scenarios where AI may not be the optimal solution, avoiding indiscriminate application.",
                                            "code": 74
                                        }
                                    ],
                                    "Summary": "Discusses the strategic considerations for applying AI, including judicious model selection and understanding when and where AI delivers the most value.",
                                    "code": 71
                                },
                                {
                                    "Subnet": "Building for Scale",
                                    "Categories": [
                                        {
                                            "Category": "Rapid Build Phase",
                                            "Summary": "A phase dedicated to engineering a full-fledged solution with added functionality before production deployment.",
                                            "code": 76
                                        },
                                        {
                                            "Category": "AI Foundation Capability",
                                            "Summary": "The necessity of building robust AI foundational capabilities, control planes, and governance for enterprise-wide scaling.",
                                            "code": 77
                                        },
                                        {
                                            "Category": "Infrastructure & Platform Focus",
                                            "Summary": "Emphasizing an infrastructure and platform-based thought process for scalable AI deployment, rather than isolated POCs.",
                                            "code": 78
                                        }
                                    ],
                                    "Summary": "Outlines the long-term perspective and architectural considerations necessary to build, sustain, and scale AI solutions effectively across an enterprise.",
                                    "code": 75
                                }
                            ],
                            "Summary": "Encompasses the long-term perspective, platform thinking, outcome-driven development, and balancing different AI models for sustainable and impactful deployment.",
                            "code": 66
                        }
                    ],
                    "Summary": "The provided text discusses the accelerated adoption of AI in financial services, highlighting key trends, challenges, and practical implementation strategies. It covers the shift from chatbots to AI agents, the importance of multimodal adoption, context engineering, and fine-tuning for enterprise-specific needs, particularly in a diverse market like India. The speaker emphasizes the vertical convergence of AI, from infrastructure to applications, and the need for a platform-based approach to scale solutions. A significant portion details a 'vibe session' workshop designed to empower business users to lead AI ideation, design, and rapid prototyping, demonstrating how AI can aid in creating proof-of-value for complex business challenges like portfolio performance reporting and insurance claims processing. Key challenges such as ROI, accuracy, compliance, integration, and scaling are addressed, advocating for an outcome-driven development process where AI is applied strategically."
                }
            },
            "summaryList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 8542,
                        "CandidatesTokenCount": 2470,
                        "TotalTokenCount": 12400,
                        "thoughtsTokenCount": 1388,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Summary": "<h2>Comprehensive Synthesis Report on AI Adoption in Financial Services</h2><h3>Introduction and Global AI Acceleration</h3><p>The financial services sector is currently experiencing a hyper-acceleration phase in Artificial Intelligence (AI) adoption. This surge is characterized by a significant shift from traditional chatbot interfaces to more sophisticated 'agents in action' that are actively being deployed across the industry, with Banking, Financial Services, and Insurance (BFSI) leading this trend. The current landscape is marked by multiple, simultaneous waves of intervention, presenting both substantial challenges and exciting opportunities. A primary focus is on multimodal adoption, where various AI models are judiciously used to accomplish complex tasks and process flows. Critically, 'context engineering' – the process of building relevant context before engaging with an AI model – is paramount for effective deployment given the rapid influx of competing models (e.g., Codex 5.3 and Sonnet 4.6 launched within minutes of each other). (data)</p><h3>Vertical Convergence and Adoption Landscape</h3><p>Accompanying this flurry of activity is a notable vertical convergence within the AI space. This extends beyond just models and applications to include substantial investments in hardware and infrastructure. Our organization, for instance, has announced a one-gigawatt AI data center over the next five years, underscoring the importance of a holistic 'vertical stack' approach—from infrastructure to specialized and foundational models, and finally to applications—to maximize impact. (data)</p><p>A recent survey of 170 global and Indian BFSI customers revealed strong adoption trends: nearly 90% are actively implementing AI in some capacity, with almost 60% in an advanced state of adoption. A key finding, particularly relevant for the regulated financial services industry, is that 50% of customers emphasize the need for enterprise-specific and fine-tuned models to ensure security and accuracy. Furthermore, an overwhelming majority recognized the critical need to measure business outcomes and metrics beyond mere Proof of Concepts (POCs) and pilots. (data)</p><h3>Challenges and Opportunities in India</h3><p>India presents unique opportunities and challenges, primarily due to its voice-led, multi-lingual, and geographically diverse population with varied financial needs. This necessitates AI solutions that prioritize financial inclusion by being present, understood, and reachable to everyone. (data)</p><h4>Illustrative Solution: Tata Capital's Multi-modal Voice AI</h4><p>Tata Capital has pioneered a multi-modal voice AI solution designed to support customers through every key moment of their financial journey, offering true end-to-end engagement. This solution leverages India-based models like Sarvam to address linguistic variations and enable multi-channel support. The demonstration showcased a conversational flow where the AI assistant effectively navigated complex customer queries (e.g., salaried/self-employed status, business type, form filling assistance) and provided accurate loan details, significantly enhancing customer engagement at scale with responsibility, empathy, and real impact. (data)</p><h3>Ground Realities of AI for Business and IT</h3><p>Our experience over the last 9-12 months highlights some grounded realities concerning AI adoption for both business and IT functions:</p><h4>AI for IT:</h4><ul><li>Primarily focused on coding assistance and technology modernization, accelerating the movement of solutions from POCs to production. (data)</li></ul><h4>AI for Business:</h4><ul><li>Broad application across customer servicing, corporate functions, wealth management, and investment advisory. (data)</li></ul><h4>Persistent Challenges:</h4><ul><li>**ROI Challenges:** Demonstrating clear returns on investment remains an issue. (data)</li><li>**Accuracy and Compliance:** Ensuring high accuracy and adherence to regulatory compliance is paramount in financial services. (data)</li><li>**Integration Complexity:** Building AI solutions is one thing; integrating them with existing enterprise systems (CRM, core banking, legacy applications) is a critical and often underestimated challenge. (data)</li><li>**Workflow vs. Use Cases:** Individual use cases rarely yield significant benefits; true value comes from optimizing entire process flows and workflows, often requiring a 'right-to-left' approach focused on desired outcomes. (data)</li><li>**Fine-tuning and Hybrid Models:** There's a strong belief and observed shift towards fine-tuning models for greater precision and accuracy, leading to hybrid scenarios where both fine-tuned and large language models (LLMs) are judiciously used. (data)</li><li>**Agentic Workflows:** A gradual shift towards agentic workflows with controlled autonomy is emerging, moving beyond high-level autonomy. (data)</li><li>**Platform Engineering:** Scalability necessitates a platform-engineered thought process, moving beyond isolated POCs to enterprise-wide deployment. (data)</li></ul><h4>Fine-tuning Rationale:</h4><p>Fine-tuning is crucial for achieving high accuracy and domain adaptation, particularly in compliance, advisory, and credit research within financial services. While LLMs excel in broad enterprise tasks with proper grounding and orchestration, domain-adapted models built through iterative fine-tuning offer superior performance in specialized contexts. Real-life examples, such as litigation prediction in claims value chains and compliance checks, demonstrate significantly higher accuracy with fine-tuned models compared to base models. The key is not just technology but deep domain curation to build effective instruction sets for fine-tuning and selecting the right models (e.g., Llama, Mistral families). (data)</p><h3>Augmented AI for Insurance Claims</h3><p>Another scenario involved augmented AI in an insurance claims process, specifically the First Notice of Loss (FNOL). In this setup, AI operates behind the scenes, listening to conversations and gleaning additional information to augment the customer service representative's insights. A fine-tuned litigation validator model simultaneously assesses the likelihood of fraud or subsequent litigation. This approach ensures crucial information is captured upfront, preventing delays in claims processing and mitigating risks. The focus here is on augmented AI, where the system supports human agents rather than fully automating the conversation, providing critical insights for decision-making. (data)</p><h3>Strategic Approach to AI Implementation and Scaling</h3><h4>Proper Calibration and Outcome-Backward Thinking:</h4><p>It is vital to properly calibrate AI application, understanding not only where to apply AI but also where not to. The philosophy advocated is an outcome-backward, 'right-to-left' approach, focusing on desired business outcomes and then re-engineering processes to achieve them, rather than blindly applying AI to traditional left-to-right process flows. (data)</p><h4>Three-Phase Adoption Model:</h4><ol><li><strong>Proof of Value (PoV):</strong> Rapid prototyping and 'vibe design and build' sessions allow for quick validation of ideas. This phase helps determine if a problem is worth solving for the organization. (data)</li><li><strong>Rapid Build:</strong> Once PoV is established, the focus shifts to engineering a full-fledged solution by adding more functionality to the initial prototype over 5-6 weeks. (data)</li><li><strong>Scale:</strong> This phase involves building a robust AI foundation, including a control plane and addressing legal, compliance, and regulatory aspects. True AI scaling in financial services occurs not when the model is ready, but when the entire organizational ecosystem—legal, compliance, regulatory bodies—is prepared. (data)</li></ol><h3>Enabling Business-Led AI Redesign (Vibe Design & Build Workshop)</h3><p>A significant challenge in AI adoption has been the inability to fully leverage business teams. Past AI implementations often filled gaps where traditional IT struggled (e.g., unstructured data access) but lacked business-driven transformation. To overcome this, the goal is to empower business users to lead the AI redesign journey.</p><h4>The Workbench Approach:</h4><p>The workshop demonstrated a 'workbench'—not a single tool, but an integration of various AI capabilities—to enable business teams. This workbench facilitates:</p><ul><li>**Identifying Business Challenges:** Starting with a specific business challenge (e.g., portfolio managers generating quarterly summaries, requiring sentiment analysis and data gathering). (data)</li><li>**Ideation and Opportunity Canvas:** Guiding business users through curated prompts to load organizational context, conduct deep research, identify pain points, assess business impact, and define solution directions. AI acts as a 'thinking partner' to co-create an 'opportunity canvas'. (data)</li><li>**Brainstorming Design:** AI generates agentic solution blueprints, breaking down opportunities into a series of agents. This shifts power to business teams to define checks, balances, and solution architecture, moving away from an engineering-centric approach. (data)</li><li>**Rapid Prototyping (UX):** Using tools like Lovable and Google AI Studio, users can generate visual UX prototypes from simple English prompts within minutes, drastically reducing the time previously spent (weeks) on design phases. This allows for quick stakeholder validation. (data)</li><li>**Back-end Capability Engineering (Low-Code):** Integrating with low-code platforms (e.g., Ne10) enables visualization of models, guardrails, and business evaluations, and facilitates code generation for a working end-to-end solution. (data)</li></ul><p>The workshop provided a hands-on experience for participants to try this multi-phase process, allowing them to translate a challenge into a working UX prototype and even explore the build phase. (data)</p><h3>Conclusions and Actionable Recommendations</h3><h4>Conclusions:</h4><p>AI has entered a hyper-acceleration phase in financial services, driven by a shift to agentic and multimodal solutions, underpinned by significant vertical convergence in the technology stack. While adoption is high, challenges remain in ROI, accuracy, compliance, and enterprise integration. The true potential of AI is unlocked through outcome-backward, workflow-centric approaches and the strategic use of fine-tuned, domain-adapted models alongside general LLMs. Crucially, scaling AI in regulated industries requires comprehensive organizational readiness, beyond just technological advancement. Empowering business teams with intuitive AI workbenches is key to accelerating transformation and ensuring solutions address real-world problems effectively. (data)</p><h4>Actionable Recommendations:</h4><ol><li>**Adopt a Hybrid AI Strategy:** Judiciously leverage both broad Large Language Models (LLMs) with proper grounding and highly accurate, fine-tuned domain-adapted models for specific financial services use cases (e.g., compliance, advisory, credit research). (data)</li><li>**Prioritize Context Engineering:** Invest in capabilities to build robust context before AI model engagement to enhance accuracy and relevance, especially in complex financial scenarios. (data)</li><li>**Embrace Outcome-Backward Design:** Shift from traditional left-to-right process thinking to a 'right-to-left' approach, starting with desired business outcomes and then designing AI solutions to achieve them. (data)</li><li>**Implement a Three-Phase AI Adoption Model:** Follow a structured approach of Quick Proof of Value (PoV), Rapid Build, and Strategic Scale, integrating feedback loops at each stage to ensure business assurance and alignment. (data)</li><li>**Empower Business Teams with AI Workbenches:** Provide intuitive platforms and frameworks that enable business users to lead AI ideation, prototyping, and design, fostering co-creation with engineering teams to accelerate solution development and validation. (data)</li><li>**Focus on Enterprise Integration:** Recognize that successful AI deployment hinges on seamless integration with existing core enterprise systems (CRM, core banking, legacy infrastructure) and allocate adequate resources for this critical aspect. (data)</li><li>**Address Regulatory and Compliance Readiness:** Integrate legal, compliance, and regulatory considerations from the outset of AI development, recognizing that true scalability in financial services depends on these factors as much as technological readiness. (data)</li><li>**Foster Immersive AI Experiences:** Conduct 'AI Days' or similar initiatives to provide business teams with practical, hands-on experience and training on AI tools and their application within the organizational context, building internal capability and understanding. (data)</li></ol>"
                            }
                        ]
                    }
                }
            ]
        }
    ],
    "headerCols": {
        "col1": "Part Id",
        "col2": "FileName",
        "col3": "Content",
        "col4": "Status"
    },
    "fileName": "",
    "projectType": "VIDEO",
    "sourceLanguage": "English",
    "targetLanguage": "English",
    "__v": 0
},
    {
    "_id": "699435ea25c9f4e0d7e70420",
    "projectId": "152898",
    "projectName": "AI masterclass in Financial Services - Spanish ",
    "projectDescription": "AI masterclass in Financial Services | AI Summit India 2026",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T09:33:27.691Z",
    "projectUpdatedDate": "2026-02-17T09:33:27.691Z",
    "projectStatus": "Active",
    "companyId": "69453bf2e603a0a401b6f35b",
    "projectCreatedBy": "69453bf2e603a0a401b6f35c",
    "projectUpdatedBy": "69453bf2e603a0a401b6f35c",
    "projectLastAccessDate": "2026-02-17T09:33:27.691Z",
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "Content",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "_AI Transforming BFSI_ Practical Strategies _ Safe Scaling _ India AI Impact Summit 2026_(MP3_128K).mp3",
            "col3": " Quick context setting on what we see from our vantage point as well in financial services in terms of the adoption. What's happening on the ground a lot being talked about it, but it's also fact about making it work in an enterprise in the form. So that's on the first part that will do will then move to the next one, which is about you wanted all of you to experience a bit of especially from a wipe session standpoint more in the area of taking an idea through a proof of value.  More of a vibe design and a vibe build session. So you have a lot of laptops out there. There are a series of instructions. It will be both assisted as well as if some of you wanted to do on your own, more than welcome to do it as well. But we'll assist in the process flow. Get a feel for it. How does it work through, especially the design and the initial prototype build part of it. And then of course we'll wrap it up as a final thing. Let me give a little bit of context setting from our vantage point where we see us.  I think it's all well recognized that AI has really got to a hyper acceleration phase. Maybe two or three key call out that I would do out here. I'm sure most of you are familiar with it, but I'll just make a call out. One is the fact that there is a genuine movement from the traditional chat bot kind of an interface to more of agents in action. It's really happening out there in the market, in the industry as well, and BFSI is definitely in leading that as well. There are multiple...  I would call it as S waves that are happening right now. Multiple interventions happening at the same point. So that kind of giving you a bit of challenge as well as an opportunity. Exciting, but at the same time, it's also giving a lot of challenge in terms of the adoption standpoint. We see a significant focus on multimodal adoption. And believe that's the future in which it will shape up. You would judiciously use multiple models to accomplish a particular task or a process flow. And the most important thing is that  End of the day, it's all about context engineering. You go to build the context before you engage with the model. So three callouts, because there's a lot of flurry of activity happening. As you could see, a series of models that are coming into the market, competing ones, quality is improving. For instance, think about these two. The 5.3 Codex and the Sonnet 4.6 were launched in a space of 20 minutes. And you could expect what would happen down the line as well.  So with this flurry of activity, it's important that we keep those three callouts in mind because we're going to judicially use which models would factor for and what context in an enterprise as well. While that happens, this flurry of activity happening, what we believe is that there is also convergence happening in this space. There is a vertical convergence happening. Right from so far, the focus has not been on models and applications. It's all about hardware and infrastructure as well.  could see a lot of investments being called out in this space globally within India and from our organization standpoint, also in the group, we have announced the launch of an AI data center, a one gigawatt data center over the period of next five years as well. But the point is that there is a vertical alignment that's happening right through infrastructure to the models, both specialized and foundation models, and to the applications. It's very important to think so that it has got off to a  inflection point that cannot be seen in isolation. It has to be seen in terms of the entire vertical stack to make it more impactful. We did a survey among customers both globally and in India. It was 170 of our customers we did a survey and see what's adoption and how is it shaping up. It was a very recent survey that we did. The multiple things that came out, one good news is all about that. Close to 90 odd percentage of the customers are in some form or shape. And this cut across banking,  capital markets and insurance customers, as I said, both India and globally, are in some form or shape adopting it with almost 60% fairly advanced state of the adoption standpoint as well. I just want to make one specific call-out here. It'll be interesting to many of you as well. 50% of the customer base feels that a lot of focus needs to be given on enterprise-specific models and some fine-tuned models as well. This is something which is very important given the regulated nature of the financial service industry.  how much of secure and how much of accurate models are also needed from an enterprise standpoint. Then of course, a majority, overwhelming majority also felt there is a good need to measure the outcome. Metrics are needed. It's one to talk about POC and pilots, but another to talk about the business outcome. In India, I think it's also interesting to see India, the opportunity and challenges, the nuances around the language. India, I would call it, is a very voice-led  India is a land of many voices, spoken in different languages,  across diverse geographies and, most importantly, with unique financial needs. At Tata Capital, financial inclusion means more than just access to finance. It means being present, being understood and being reachable to everyone, everywhere. Today, we are proud to showcase a multi-modal voice AI solution supporting our customers across every key moment of their financial journey. Truly,  end to end.  Okay, so first of all, are you salaried or self-employed? I don't understand, madam. Excuse me, I mean, do you work somewhere or do you have a business of your own? I have my own shop. Okay, so you have your own shop. Can you tell me what your business is about? Madam, I have a rental shop. Okay, so you have a grocery shop. So instead of your shop, do you have your own shop or do you have it on rent? My rental shop is on rent. Okay, so your shop is on rent?  I have sent you a personal information form on WhatsApp which requires some basic details. Fill it in. If you have any doubts, you can ask me. Yes, I have got the form. I will fill it in, madam. It won't take much time, right? No, it won't take much time. Fill it in slowly. I am here. If you have any questions, you can ask me. Madam, what do I have to put in the mother's maiden name? Okay, you are asking about the mother's maiden name. This means the name of your mother before marriage.  Do you have a call? If you need anything, let me know. I will help you. Yes, I am filling the form. Madam, I am in the middle. I am filling the form. Okay, no problem. You can fill the form. I am here. Yes, Madam. I have filled the form. Very good. Thank you for filling the form. So, let me tell you again that you will get a loan of 5 lakhs for 3 years at an interest rate of 12.5%. The EMI of which will be 16,727.  What you just saw. Well, that is the future. The future of conversations, of customer engagement and of AI led transformation. This isn't just a concept. It is already powering millions of customer conversations. This is how Tata Capital is reimagining engagement. At scale.  with responsibility, with empathy, and with real impact. Because when trust finally finds a voice, Bharat engages. And when Bharat engages, Bharat progresses.  India-based model based on Sarvam. So I think a lot of linguistic variation that is addressed, and I think this is important because on one side we have the global models available, but it's also important to address the local nuances out there. And the second thing is it's also now allowing from a multi-channel standpoint. So I think the ability to come together more from a customer servicing, I think that has taken on a different league altogether. While that happens, let me call you out some key pointers from.  What we have been experiencing over the last, I would say, nine to 12 months, some very grounded reality has been popped up here. From our vantage, we were looking at the adoption in terms of AI for business and AI for IT. IT is pretty straightforward in terms of, it's mostly around a lot of focus on coding assistance and plus the fact that you do a lot using modernization. Huge focus in many forms in terms of technology modernization, the ability of AI to accelerate it. These are not POCs and pilots really getting into production.  Then the business side of it, if you look at it, it's got a broad stick happening out there. Right from customer servicing to horizontal functions, corporate functions, and also into wealth and investment advisory as well. So there is a significant progress happening in this space, but just also wanted to call out some of the ground reality that are popping up. ROI challenges still persist. There is still challenges in terms of the accuracy, the.  compliance aspect of it, that's still an important thing to be addressed and factored in building an AI solution. The other part is that building an AI solution is one thing. Making it work in an enterprise in terms of integration is so critical. A lot of ROs have been built around building solution, but not necessarily the effort that you need to make it work. Connecting to your enterprise systems, connecting to CRM, your core banking, your learning application, that's a critical aspect of this whole process flow.  I think it's fairly recognized that individual use case are not giving you the benefit, the real benefit come by process flows and by workflows. And that you're going to look at workflows differently, not to the traditional left to right, but more from right to left in terms of an outcome standpoint. We see and we believe fine tuning is something that financial services would pick up in a significant way. If you're already seeing that, that would really do an uptake in the next six to nine months.  Definitely see that picking up in a big way. Why? Because of the need for much more accurate, much more precise models that need to happen. To say that it will be a hybrid scenario, you would have fine-tuned models on one hand, you'll also have large language models, you would judiciously use it. There is definitely a shift towards Hintiq. Having said that, these are not Hintiq as in high level of autonomy, but a slow.  a pedigree of autonomy is being enabled in it, but agentic workflows are happening. The bottom point, I think we refer to that the vertical stack that's evolving, very important to look at an infrastructure in the platform-based thought process. Again, the whole point is that I can do a great POC pilot shiny object, but how do I scale it across and you need a platform engineered thought process to make it happen. So these are some of the experiences that we're going to picked up, thought I can relate that all of you as well. Just wanted to.  give a quick perspective around the fine tuning that we talked about. This is something which piggyback on the fact that it will be in hybrid AI solutions and models that you would end up with. You would have both the scenarios. You will continue to leverage large language model for broad, measurable enterprise task and activities with proper grounding. I think it's very important I just double down on the fact that it has to be engineered.  How well do you orchestrate your workflows? How well do you build your evals around it? It's very critical. But I think this is an equally important. How well do you build your domain-adapted models in compliance area, in advisory, in credit research? I think it's important to see how well they build appropriate and high accuracy. So our vantage point, we believe, it's going to be convergence between the two, not one over the other because each has its own space. It's very important to traditionally use this. I'll just give you a couple of quick.  and adults around it, these are real life examples. Something that we have done for litigation prediction in a claims value chain. If you look at it, what a base model provided versus what a fine tune model provides. And this is not something that happens straight away, you need to iteratively build it, but it's very important to continue, make sure that right from base model, the different techniques of fine tuning to make it more domain adapted in that context. And you could see the high percentage of  accuracy as you build that incrementally. And this is where business gets a lot glued to it saying that, hey, now it's come to a point. It's not just technical. It's also important to make it work. Same thing in compliance. And we have used two different models just to give a variation. There are many models out there. But these are models that you could enable it. And whether it's Lama or Mistral Family, there are many other models. For instance, as I said, for several models, we've been working on it in terms of voice, AI standpoint. But the key message I want to leave out here is that.  fine-tuning or making it proprietary, it's not just technology. I think it's all about building this domain curation. How well do you build the instruction set for a fine-tuning? That need a lot of domain understanding. Once you build that, the initial set, then I think it's a lot about applying the right techniques. It's also equally important to look at the right models. Your slew of models out there,  and there is a method to the madness in terms of which model would fit into what context. So all the three are very important. Building the right instruction sets for fine-tuning. For all of you who I'm sure fine-tuning is something that all of you are aware of it, but essentially what you're trying to do is that making sure a broader model has been fine-tuned towards specific domain and adaptation so that it get much more relevant in that context and accuracy of the decisions it takes us to higher level.  a series of business scenarios that are evolving in this area. We see this in India, we see this globally. So this is something from our vantage, we see a strong uptick that's happening. So just wanted to give a quick flavor of that before we get into the white thought process. I think it's important to look at it from the context engineering, very important. Enterprise adoption is not just about building use cases, but also making it work in the broad end.  it's very critical to make sure that you judicious use of large language models and domain adapted models. With that, let me just show you one specific scenario. Good afternoon. Thank you for calling. Do I have the pleasure of speaking with you? Hi, Catherine. This is Daniel Thompson. I need to report an accident that happened earlier today morning at 4.30 AM.  I am sorry to hear. Could I please have your policy number? Yes, it is ALC 76-12-39-05. Thanks, Daniel. I thought that's your policy. Everything looks active. Could you describe what happened? I was already on the highway. I was driving southbound on the M1 Pacific motorway past the Gosford exit in New South Wales, maintaining my lane and speed when a silver SUV suddenly emerged from the left.  onto the motorway and site wiped my vehicle so the impact was on the left side yes quite severely it hit around the front left door and fender area i had no chance to react the other driver didn't yield or check for traffic understood um with a police call to the scene  Yes, our patrol unit came by. They've noted the details. I have the event number. I also took several photos of the damage. Great. If you can send us the event number, police notes and any photos that will help us proceed. We'll also need the repair estimate or tow docket if you've received one. I will email all the documents shortly. Perfect.  Drive safe and let us know if you need anything else. All right. As you could relate, I think the accent of this is for a UK-based customer. And the context is a little different from the first one that you saw from a voice here, which was a very specific scenario where, in the Tata Capital's story, it was AI at the thick and center of everything in terms of engaging in a customer service. But not actions are not getting taken. It's more of a conversational flow.  This one is scenario in which a multi-model with a fine-tune model being brought into the picture. So if you could see the focus of the customer service rep is more in terms of looking at the information that can be gleaned around it. So this is an augmented AI. AI was listening to the whole conversation, providing from an augmented sample what additional information. Normally, this whole process called FN oil, the first notice of loss from the insurance background would know.  When you meet with an accident, the first notice of loss that you make a call to the back office to inform. A lot of information you gather at that point. It's very important to capture that information. Today, it's a very back and forth happens, and your claims get delayed because of a simple reason, you haven't captured enough information. In this scenario, behind the scenes, so AI was not in the front-end center, behind the scenes capturing the information, and there was a litigation validator, which I was telling the fine-tune model.  looking at it, is there any fraud in this process so that there is a subsequent litigation possible down the line? That information, I don't know how many of you have watched that, that information was gleaned through in between for what I say, this looks like a good call, let's proceed. So I think that the modality is a little different out there, but the whole point is that it's very important to look from both the scenarios. So to put it in a nutshell, I think it's important from an enterprise form, I think proper calibration is very critical.  great technology, great capabilities evolving, but it's also important, how do you enable that in a context? Where do you not apply AI also? I think it's important to also look where you don't apply AI. Because at times, we have the thought process of applying AI all over the place. I think it's very important to break it and then make sure that the ones, and that's what our philosophy is that, don't look at from a traditional process flow from left to right, which you break it down to series of task and say where I can apply it.  Look at from a right to left and see outcome. What is outcome and how do I reapply back? So what are you going to do now for the next 60 minutes or so? What I thought is that, I believe that this need to be broken into three broad areas. You got to have a quick proof of value. You need to build that proof of value because there are certain things you will just move forward then. You say, okay, this is not working. Let me try the next one. You have the ability right now to do a quick prototyping, a quick vibe design and prototyping.  Then you go into a rapid build over a period of time, five to six weeks, and then you really scale it. This is a model that we would advocate as enterprises start picking up more and more. Because, as I said, it's very important to look at outcome backwards. It's very important to look for what is the business assurance that I bring in. So we have constructed the next 60 minutes or so in that format. We have certain challenge cards on your table. You have laptops out there. We're going to do this in two formats. We'll assist you in the process.  in how do I go through the process, and second is also in terms of those who are more wanting to do on their own, they feel free to try it out. Instructions are also given out there. Let me invite Kavita now to run you through the vibe coding design and prototyping session. Thank you. To continue on the topic of scaling, in financial services, AI doesn't scale when the model is ready. It's when legal compliance...  regulatory, all of the body comes together. When they are ready, it's when AI scales. So I think that's one of the challenges that we face when we look at a scaling AI. This is just one of them. And my name is Kavita, and I lead the AI labs for financial services. And I help customers on their journey to adopt AI and scale AI. And one of the other challenges that really come across  when we start discussing about AI use cases is, how do I get the business teams on board? Because if you look at the kind of implementations we have seen so far in AI, you've got knowledge assistance, you've got chatbots, you've got good document AI capability. And if you look at these implementations, they have actually filled the gap where the traditional AI, traditional IT has failed in the sense...  inability to access unstructured data. There was a gap which AI solved. So most of the use cases have been around these lines, and that's primarily because AI has not been, we have not let the business teams actually drive AI. So what does it take to help enable the business teams to understand the power of AI and look at more transformative use cases?  So that's the construct of what we want to do today. We want you to experience that how a business user or a business team can be enabled with AI to accelerate that whole journey of identifying the right opportunity, doing a working prototype, and actually even testing it out and seeing the proof of value. How do we shift left?  in the sense that how do we give business more control and working with the engineering team, build out a quick proof of value that you can validate and then scale to production. Because in one of the earlier conversations I had with one of you here was that scaling is a challenge. But sometimes you don't get to scale because you've not identified the right opportunity. You try to apply AI everywhere. So how do you use AI to help you in that journey?  And what we want to achieve through this is let business lead the whole AI redesign. So how do we do that? And that's something that we will cover today. And what we're going to demonstrate to you is not a tool, but how we can bring all the AI capability together for a business audience.  Yeah, switching on to that. So what we'll see today, disclaimer before I start, this tool is not a single tool. It's a workbench that we have created that brings together different AI capabilities. And then what we do is that we start off with something called a business challenge, because that's the starting point for a business team. Where do I begin my journey? What is the challenge that I need to address? And for the sake of this demonstration, I'll pick up one use case, something that's  A challenge that has come up in one of our customer conversations was that you have a lot of portfolio managers who need to, at the end of the quarter, generate a summary of how their portfolio has performed. And along with collecting a lot of data, they need to actually  check out the sentiment of the stock on the public domain. So there's a lot of data gathering, a lot of analysis that comes in, and as a result, there's a lot of time that goes into that activity, and that delays the overall report, and sometimes this gets delegated to a junior analyst, the quality of the report is affected. So how do I bring all of that in a way that AI can help the portfolio managers solve the problem? So that was the challenge that we picked up, and I'll show you a quick demonstration of  how the challenge works. The challenge was just creating an assistant to help the AI agent, to help the portfolio manager actually look up what is the right, what is the data that's available out there on the public domain, or even look up Yahoo Finance to see how the stock has performed. So all of that was being done by AI in the background. And this is a solution that...  or when we piloted with our customers, they saw a huge advantage, because today, they were actually outsourcing a lot of this activity to a third party organization, and they were able to bring that back, and bring down the overall time to generate the report, and even the quality of the report started coming out much better. I thought I'd give you a view of the final outcome, and take you back to where we started. And that's the platform that we were trying to show you. So switching on.  to that platform. So what we do as a first step is actually pick up the challenge and how do we use AI to ideate and come up with something called the opportunity canvas. So many of our users have, you know, if you see business users are all equipped with co-pilots because it's already available in the enterprise. But what we've observed is that just not giving them AI tools or training them on prompts is really going to help. It's about  helping them to understand that end-to-end process and letting them have complete control. So that's what we've done over here. So we looked at what could be a set of questions that a business user needs to look at when they see a challenge and convert it to an opportunity. So we've curated a set of prompts which can go through the data and come up with what we call the opportunity canvas. So we'll just walk you through those different steps. So I think one thing that we always need to keep in mind is that  When we give AI to the business users, the research is primarily on, you can use a lot of research tools available, but the organization context is more important. How do we know what the user is? What are the different users going through? What are their challenges and pain points? So loading the context is the first step.  And the reason I'm walking you through this is because once I finish the demo, I want you also to try and just see whether as a business leader or a practitioner how the whole experience will be and how can we really build that capability for business teams to use. So we'll go through this flow. So we have built the context. And then after that, you can even go back and do a deep research. Because sometimes your context is not enough. You want to know how your peers are performing.  pull out a deep research from Gemini or ChachiPT. So put that context also in, and then you have an overall knowledge base that your business team can now use to start ideating. So what we have done is created an ideation framework here. Now, as you go through the ideation process, you'll have a lot of questions. Have I captured all the pain points right? Because you want to be sure that you have actually  solving a problem. So there's pain points that you capture. And there are quite a lot of steps. I won't go through all of them. But if you quickly go through, you will have pain points. You'll want to assess what the business impact is. What's the possible solution direction that you want to take? And as we go through these steps, you will see that it's not enough just to analyze the data that's available.  but to actually start seeing AI as a thinking partner. That's what we want the business to do, be able to bounce ideas, validate them, feed the input, and finally, at the end of it, create what we call an opportunity canvas. So a lot of thought has gone in. So while these demonstrations, while he brings up the opportunity canvas, something that we have observed is that just giving a tool may not be the only solution.  What we do in TC, something that's worked for us, is that we conduct an AI day for business teams, where we try to create that immersive experience where they understand what are the right AI tools and also how to use them and how to apply them in the context. So I'll, if you can just go through the, yeah. So there's a lot of questions, so we don't, we, and it will take a minute to go through, but if our demo is ready, you can show that as well.  Yeah, so this is what the final product looked like. So you give a company name, and it went through the organization's research. It also went to the public domain and started picking up all the stock information and brought that up into an overall summary that can be then used by the portfolio manager to make a decision on how his stock has performed, how do I report that?  to his customers? How do I report that into the performance reporting? And if you see, it was instant. And it's gone through, I think, 15 to 20 sites. We've configured all the financial sites to go through and capture the sentiment, capture key financial metrics, all of that instantly on a very high-speed internet. Now, I think we would be done with the Opportunity Canvas. Last one step.  So it's important that we go through these steps. So that's why we have consciously made these steps so that you don't pick an idea which is not the right one. So identifying the opportunity, what is the possible solution direction, what's the impact, and what your roadmap will be. And there you have an opportunity canvas, which the user has co-created along with AI. Now you go on to the next step. And this is important.  All this is fine. The business team could be comfortable in Opportunity Canvas. But can now the business team, with the help of AI, drive the overall design process? So what you'll see is that you can brainstorm with AI and actually give a direction of what that design should be. And this is what AI has generated. It's kind of broken down that whole opportunity into a set of  different agents that can go in. Because as we covered in the previous session, a lot of attention focuses on moving to agentic. So can we look at building what that agentic solution would look like is what you're going to see here. It's not done the coding yet, but what it has done is broken up that entire opportunity into a series of agents which can address this need. And we'll just.  You can try that. You will get a chance to try this, so I won't go into the detail. But here, it's come up with four or five agents. It's organized. And as a business team, they can actually go and say, I want these kind of checks and balances. I want to do it this way. And now you see the difference, right? Earlier, it was all left to the engineering team to actually build the solution. But now, we are passing the power back to the business to drive all of this. Now, going forward,  What is the next important step is to validate what this opportunity looks like from an experience. So you use a simple wipe coding tool. We have here used both Lovable and Google AI Studio. This tool generates the prompt. You copy the prompt. And then you actually can see it through a visual experience.  simple as putting it and if you see the prompt it's all English and you know when you get into your wipe session you'll see that you can actually edit some of it and see what that experience looks like so this will take how much time maybe a minute or so do you have something do you have something that you've already pre-built yeah so this is something that that came out of the tool a very very nice experience which shows what the portfolio summary should look like what are the stocks etc so  kind of understood the opportunity and translated that. And you can play around with it and tweak it to the experience. This step becomes very important because what earlier took two to three weeks of building a UX prototype as the end of a design phase actually got converted to minutes. And this is something you take to these stakeholders, validate it, and you're ready with what is that real challenge you want to address, what's the real opportunity comes out so quickly.  We can't stop with just a UX prototype, because that just gives you an experience. You need to actually engineer some of the back-end capability. And there are low-core platforms that are available. So in today's live session, we will show you that as well. And one of the examples that we have shown here is a solution called Ne10. It's a low-core platform building. It's not that we are saying this is the right tool. As per your enterprise decision, you could have different options.  But this really helps, in our view, to ideate and understand or discuss with your engineering teams how your design should be. You can easily visualize what are the models being used here. Have I thought through all the guardrails? Have I built my business evaluations? All of this can be verified in one shot. And then from this, the code can get generated, and you'll have a working end-to-end solution.  So this is the process we followed for quickly prototyping. And the demo that you saw was an example of something that was built using this approach. So what we'd like to do now is shift back to you and see if you'd like to try this. I mean, some of you could. We have laptops configured with.  with user IDs, we have all the tools available. It's an opportunity for you to try out and we have mentors. We have mentors available for each table. A lot of our teams here to support you and we would really encourage you to try and we can guide you here as you go along. Explore, yeah. Okay, so let me just walk you through what you would do. So all of your laptops will have a handbook already preloaded.  which talks about the use case that you would be picking up. Okay, so, and that is in green, it's tagged as challenge. So that's the use case that you will start with. And you have different phases and you have resources for each of the phase. So feel free to go through this handbook, take it step by step, and we'll help you with it. You are workspace five.  So log into your workspace based on the number that's on your table. So when you start, it will be the discovery and design. Yeah, you can start the solution with the resource present there. Okay, so just a quick time check. We just have five minutes. So if you want some help on any particular section, just raise your hand. We'll come and help. But I'm so happy to see many of you have generated very nice UX prototypes.  And you can also move into that build phase. That's an optional one, but try it out. We'll have mentors to support you.  uh...  So I think we'll just conclude.  So, so far we spoke about, I only covered, I'll hand it over to him, but what you did essentially comes as step one. You actually did, and I'm so happy that many of you got challenge completed, and you saw your code getting converted, and I see a lot of happy faces because it was, I heard words like, amazing, I can do the same thing with my tools, so very happy to hear all that. So what we did,  was the first step, proof of value. We've tried to establish this is something, a problem worth solving for my organization and you may have an answer then, you have an answer at the end of the stage. But the journey to take it to production involves something called rapid building. What you've built is a foundation which you can add on more functionality and then create it into a full-fledged solution. And that's where we call the rapid build.  which is about engineering the full-fledged solution before you go to production. And then finally, the scale. And when you look at scale, and someone asked me that question, you have to build a lot of AI foundation capability, the control plane, a lot of things need to go in. And that's important if you want to bring that scale. I'll just hand it over to Onni to close. I think thank you. Thank you for being a real sport out there. I think all of you joining in, being through the wipe session as well. Great to see all of you continue out there.  Once again, thanks for being participating. I think this is one important thing that we believe AI would make a difference in the ability to quickly translate an idea to prototype. Not to say that that prototype is what you would eventually scale to production, but that gives you the feel of what can be done. And then you naturally scale it into production in a very calibrated way. But again, thanks for your time. And then being part of it, feel free to reach out. Some of these I'm assuming.  would have been captured in your email, et cetera. So it's available for you to do some prompting and wiping back home as well if you're adventurous and wanting to extend the focus. Thank you once again and have a good rest of the day.",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771320765456.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:01",
                    "end": "00:00:29",
                    "text": " Quick context setting on what we see from our vantage point as well in financial services in terms of the adoption. What's happening on the ground a lot being talked about it, but it's also fact about making it work in an enterprise in the form. So that's on the first part that will do will then move to the next one, which is about you wanted all of you to experience a bit of especially from a wipe session standpoint more in the area of taking an idea through a proof of value."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:29",
                    "end": "00:00:59",
                    "text": " More of a vibe design and a vibe build session. So you have a lot of laptops out there. There are a series of instructions. It will be both assisted as well as if some of you wanted to do on your own, more than welcome to do it as well. But we'll assist in the process flow. Get a feel for it. How does it work through, especially the design and the initial prototype build part of it. And then of course we'll wrap it up as a final thing. Let me give a little bit of context setting from our vantage point where we see us."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:59",
                    "end": "00:01:28",
                    "text": " I think it's all well recognized that AI has really got to a hyper acceleration phase. Maybe two or three key call out that I would do out here. I'm sure most of you are familiar with it, but I'll just make a call out. One is the fact that there is a genuine movement from the traditional chat bot kind of an interface to more of agents in action. It's really happening out there in the market, in the industry as well, and BFSI is definitely in leading that as well. There are multiple..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:29",
                    "end": "00:01:59",
                    "text": " I would call it as S waves that are happening right now. Multiple interventions happening at the same point. So that kind of giving you a bit of challenge as well as an opportunity. Exciting, but at the same time, it's also giving a lot of challenge in terms of the adoption standpoint. We see a significant focus on multimodal adoption. And believe that's the future in which it will shape up. You would judiciously use multiple models to accomplish a particular task or a process flow. And the most important thing is that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:59",
                    "end": "00:02:26",
                    "text": " End of the day, it's all about context engineering. You go to build the context before you engage with the model. So three callouts, because there's a lot of flurry of activity happening. As you could see, a series of models that are coming into the market, competing ones, quality is improving. For instance, think about these two. The 5.3 Codex and the Sonnet 4.6 were launched in a space of 20 minutes. And you could expect what would happen down the line as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:26",
                    "end": "00:02:55",
                    "text": " So with this flurry of activity, it's important that we keep those three callouts in mind because we're going to judicially use which models would factor for and what context in an enterprise as well. While that happens, this flurry of activity happening, what we believe is that there is also convergence happening in this space. There is a vertical convergence happening. Right from so far, the focus has not been on models and applications. It's all about hardware and infrastructure as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:55",
                    "end": "00:03:25",
                    "text": " could see a lot of investments being called out in this space globally within India and from our organization standpoint, also in the group, we have announced the launch of an AI data center, a one gigawatt data center over the period of next five years as well. But the point is that there is a vertical alignment that's happening right through infrastructure to the models, both specialized and foundation models, and to the applications. It's very important to think so that it has got off to a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:25",
                    "end": "00:03:57",
                    "text": " inflection point that cannot be seen in isolation. It has to be seen in terms of the entire vertical stack to make it more impactful. We did a survey among customers both globally and in India. It was 170 of our customers we did a survey and see what's adoption and how is it shaping up. It was a very recent survey that we did. The multiple things that came out, one good news is all about that. Close to 90 odd percentage of the customers are in some form or shape. And this cut across banking,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:57",
                    "end": "00:04:27",
                    "text": " capital markets and insurance customers, as I said, both India and globally, are in some form or shape adopting it with almost 60% fairly advanced state of the adoption standpoint as well. I just want to make one specific call-out here. It'll be interesting to many of you as well. 50% of the customer base feels that a lot of focus needs to be given on enterprise-specific models and some fine-tuned models as well. This is something which is very important given the regulated nature of the financial service industry."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:27",
                    "end": "00:04:57",
                    "text": " how much of secure and how much of accurate models are also needed from an enterprise standpoint. Then of course, a majority, overwhelming majority also felt there is a good need to measure the outcome. Metrics are needed. It's one to talk about POC and pilots, but another to talk about the business outcome. In India, I think it's also interesting to see India, the opportunity and challenges, the nuances around the language. India, I would call it, is a very voice-led"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:57",
                    "end": "00:05:26",
                    "text": " India is a land of many voices, spoken in different languages,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:26",
                    "end": "00:05:56",
                    "text": " across diverse geographies and, most importantly, with unique financial needs. At Tata Capital, financial inclusion means more than just access to finance. It means being present, being understood and being reachable to everyone, everywhere. Today, we are proud to showcase a multi-modal voice AI solution supporting our customers across every key moment of their financial journey. Truly,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:56",
                    "end": "00:06:26",
                    "text": " end to end."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:26",
                    "end": "00:06:55",
                    "text": " Okay, so first of all, are you salaried or self-employed? I don't understand, madam. Excuse me, I mean, do you work somewhere or do you have a business of your own? I have my own shop. Okay, so you have your own shop. Can you tell me what your business is about? Madam, I have a rental shop. Okay, so you have a grocery shop. So instead of your shop, do you have your own shop or do you have it on rent? My rental shop is on rent. Okay, so your shop is on rent?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:55",
                    "end": "00:07:26",
                    "text": " I have sent you a personal information form on WhatsApp which requires some basic details. Fill it in. If you have any doubts, you can ask me. Yes, I have got the form. I will fill it in, madam. It won't take much time, right? No, it won't take much time. Fill it in slowly. I am here. If you have any questions, you can ask me. Madam, what do I have to put in the mother's maiden name? Okay, you are asking about the mother's maiden name. This means the name of your mother before marriage."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:32",
                    "end": "00:08:06",
                    "text": " Do you have a call? If you need anything, let me know. I will help you. Yes, I am filling the form. Madam, I am in the middle. I am filling the form. Okay, no problem. You can fill the form. I am here. Yes, Madam. I have filled the form. Very good. Thank you for filling the form. So, let me tell you again that you will get a loan of 5 lakhs for 3 years at an interest rate of 12.5%. The EMI of which will be 16,727."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:06",
                    "end": "00:08:35",
                    "text": " What you just saw. Well, that is the future. The future of conversations, of customer engagement and of AI led transformation. This isn't just a concept. It is already powering millions of customer conversations. This is how Tata Capital is reimagining engagement. At scale."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:35",
                    "end": "00:09:05",
                    "text": " with responsibility, with empathy, and with real impact. Because when trust finally finds a voice, Bharat engages. And when Bharat engages, Bharat progresses."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:05",
                    "end": "00:09:33",
                    "text": " India-based model based on Sarvam. So I think a lot of linguistic variation that is addressed, and I think this is important because on one side we have the global models available, but it's also important to address the local nuances out there. And the second thing is it's also now allowing from a multi-channel standpoint. So I think the ability to come together more from a customer servicing, I think that has taken on a different league altogether. While that happens, let me call you out some key pointers from."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:33",
                    "end": "00:10:03",
                    "text": " What we have been experiencing over the last, I would say, nine to 12 months, some very grounded reality has been popped up here. From our vantage, we were looking at the adoption in terms of AI for business and AI for IT. IT is pretty straightforward in terms of, it's mostly around a lot of focus on coding assistance and plus the fact that you do a lot using modernization. Huge focus in many forms in terms of technology modernization, the ability of AI to accelerate it. These are not POCs and pilots really getting into production."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:03",
                    "end": "00:10:28",
                    "text": " Then the business side of it, if you look at it, it's got a broad stick happening out there. Right from customer servicing to horizontal functions, corporate functions, and also into wealth and investment advisory as well. So there is a significant progress happening in this space, but just also wanted to call out some of the ground reality that are popping up. ROI challenges still persist. There is still challenges in terms of the accuracy, the."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:28",
                    "end": "00:10:56",
                    "text": " compliance aspect of it, that's still an important thing to be addressed and factored in building an AI solution. The other part is that building an AI solution is one thing. Making it work in an enterprise in terms of integration is so critical. A lot of ROs have been built around building solution, but not necessarily the effort that you need to make it work. Connecting to your enterprise systems, connecting to CRM, your core banking, your learning application, that's a critical aspect of this whole process flow."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:56",
                    "end": "00:11:25",
                    "text": " I think it's fairly recognized that individual use case are not giving you the benefit, the real benefit come by process flows and by workflows. And that you're going to look at workflows differently, not to the traditional left to right, but more from right to left in terms of an outcome standpoint. We see and we believe fine tuning is something that financial services would pick up in a significant way. If you're already seeing that, that would really do an uptake in the next six to nine months."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:25",
                    "end": "00:11:51",
                    "text": " Definitely see that picking up in a big way. Why? Because of the need for much more accurate, much more precise models that need to happen. To say that it will be a hybrid scenario, you would have fine-tuned models on one hand, you'll also have large language models, you would judiciously use it. There is definitely a shift towards Hintiq. Having said that, these are not Hintiq as in high level of autonomy, but a slow."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:51",
                    "end": "00:12:21",
                    "text": " a pedigree of autonomy is being enabled in it, but agentic workflows are happening. The bottom point, I think we refer to that the vertical stack that's evolving, very important to look at an infrastructure in the platform-based thought process. Again, the whole point is that I can do a great POC pilot shiny object, but how do I scale it across and you need a platform engineered thought process to make it happen. So these are some of the experiences that we're going to picked up, thought I can relate that all of you as well. Just wanted to."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:21",
                    "end": "00:12:48",
                    "text": " give a quick perspective around the fine tuning that we talked about. This is something which piggyback on the fact that it will be in hybrid AI solutions and models that you would end up with. You would have both the scenarios. You will continue to leverage large language model for broad, measurable enterprise task and activities with proper grounding. I think it's very important I just double down on the fact that it has to be engineered."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:48",
                    "end": "00:13:18",
                    "text": " How well do you orchestrate your workflows? How well do you build your evals around it? It's very critical. But I think this is an equally important. How well do you build your domain-adapted models in compliance area, in advisory, in credit research? I think it's important to see how well they build appropriate and high accuracy. So our vantage point, we believe, it's going to be convergence between the two, not one over the other because each has its own space. It's very important to traditionally use this. I'll just give you a couple of quick."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:18",
                    "end": "00:13:48",
                    "text": " and adults around it, these are real life examples. Something that we have done for litigation prediction in a claims value chain. If you look at it, what a base model provided versus what a fine tune model provides. And this is not something that happens straight away, you need to iteratively build it, but it's very important to continue, make sure that right from base model, the different techniques of fine tuning to make it more domain adapted in that context. And you could see the high percentage of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:48",
                    "end": "00:14:17",
                    "text": " accuracy as you build that incrementally. And this is where business gets a lot glued to it saying that, hey, now it's come to a point. It's not just technical. It's also important to make it work. Same thing in compliance. And we have used two different models just to give a variation. There are many models out there. But these are models that you could enable it. And whether it's Lama or Mistral Family, there are many other models. For instance, as I said, for several models, we've been working on it in terms of voice, AI standpoint. But the key message I want to leave out here is that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:18",
                    "end": "00:14:44",
                    "text": " fine-tuning or making it proprietary, it's not just technology. I think it's all about building this domain curation. How well do you build the instruction set for a fine-tuning? That need a lot of domain understanding. Once you build that, the initial set, then I think it's a lot about applying the right techniques. It's also equally important to look at the right models. Your slew of models out there,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:44",
                    "end": "00:15:14",
                    "text": " and there is a method to the madness in terms of which model would fit into what context. So all the three are very important. Building the right instruction sets for fine-tuning. For all of you who I'm sure fine-tuning is something that all of you are aware of it, but essentially what you're trying to do is that making sure a broader model has been fine-tuned towards specific domain and adaptation so that it get much more relevant in that context and accuracy of the decisions it takes us to higher level."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:14",
                    "end": "00:15:41",
                    "text": " a series of business scenarios that are evolving in this area. We see this in India, we see this globally. So this is something from our vantage, we see a strong uptick that's happening. So just wanted to give a quick flavor of that before we get into the white thought process. I think it's important to look at it from the context engineering, very important. Enterprise adoption is not just about building use cases, but also making it work in the broad end."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:41",
                    "end": "00:16:09",
                    "text": " it's very critical to make sure that you judicious use of large language models and domain adapted models. With that, let me just show you one specific scenario. Good afternoon. Thank you for calling. Do I have the pleasure of speaking with you? Hi, Catherine. This is Daniel Thompson. I need to report an accident that happened earlier today morning at 4.30 AM."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:09",
                    "end": "00:16:38",
                    "text": " I am sorry to hear. Could I please have your policy number? Yes, it is ALC 76-12-39-05. Thanks, Daniel. I thought that's your policy. Everything looks active. Could you describe what happened? I was already on the highway. I was driving southbound on the M1 Pacific motorway past the Gosford exit in New South Wales, maintaining my lane and speed when a silver SUV suddenly emerged from the left."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:38",
                    "end": "00:17:01",
                    "text": " onto the motorway and site wiped my vehicle so the impact was on the left side yes quite severely it hit around the front left door and fender area i had no chance to react the other driver didn't yield or check for traffic understood um with a police call to the scene"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:01",
                    "end": "00:17:29",
                    "text": " Yes, our patrol unit came by. They've noted the details. I have the event number. I also took several photos of the damage. Great. If you can send us the event number, police notes and any photos that will help us proceed. We'll also need the repair estimate or tow docket if you've received one. I will email all the documents shortly. Perfect."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:29",
                    "end": "00:18:07",
                    "text": " Drive safe and let us know if you need anything else. All right. As you could relate, I think the accent of this is for a UK-based customer. And the context is a little different from the first one that you saw from a voice here, which was a very specific scenario where, in the Tata Capital's story, it was AI at the thick and center of everything in terms of engaging in a customer service. But not actions are not getting taken. It's more of a conversational flow."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:07",
                    "end": "00:18:32",
                    "text": " This one is scenario in which a multi-model with a fine-tune model being brought into the picture. So if you could see the focus of the customer service rep is more in terms of looking at the information that can be gleaned around it. So this is an augmented AI. AI was listening to the whole conversation, providing from an augmented sample what additional information. Normally, this whole process called FN oil, the first notice of loss from the insurance background would know."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:32",
                    "end": "00:19:00",
                    "text": " When you meet with an accident, the first notice of loss that you make a call to the back office to inform. A lot of information you gather at that point. It's very important to capture that information. Today, it's a very back and forth happens, and your claims get delayed because of a simple reason, you haven't captured enough information. In this scenario, behind the scenes, so AI was not in the front-end center, behind the scenes capturing the information, and there was a litigation validator, which I was telling the fine-tune model."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:00",
                    "end": "00:19:28",
                    "text": " looking at it, is there any fraud in this process so that there is a subsequent litigation possible down the line? That information, I don't know how many of you have watched that, that information was gleaned through in between for what I say, this looks like a good call, let's proceed. So I think that the modality is a little different out there, but the whole point is that it's very important to look from both the scenarios. So to put it in a nutshell, I think it's important from an enterprise form, I think proper calibration is very critical."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:28",
                    "end": "00:19:56",
                    "text": " great technology, great capabilities evolving, but it's also important, how do you enable that in a context? Where do you not apply AI also? I think it's important to also look where you don't apply AI. Because at times, we have the thought process of applying AI all over the place. I think it's very important to break it and then make sure that the ones, and that's what our philosophy is that, don't look at from a traditional process flow from left to right, which you break it down to series of task and say where I can apply it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:56",
                    "end": "00:20:26",
                    "text": " Look at from a right to left and see outcome. What is outcome and how do I reapply back? So what are you going to do now for the next 60 minutes or so? What I thought is that, I believe that this need to be broken into three broad areas. You got to have a quick proof of value. You need to build that proof of value because there are certain things you will just move forward then. You say, okay, this is not working. Let me try the next one. You have the ability right now to do a quick prototyping, a quick vibe design and prototyping."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:26",
                    "end": "00:20:55",
                    "text": " Then you go into a rapid build over a period of time, five to six weeks, and then you really scale it. This is a model that we would advocate as enterprises start picking up more and more. Because, as I said, it's very important to look at outcome backwards. It's very important to look for what is the business assurance that I bring in. So we have constructed the next 60 minutes or so in that format. We have certain challenge cards on your table. You have laptops out there. We're going to do this in two formats. We'll assist you in the process."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:55",
                    "end": "00:21:24",
                    "text": " in how do I go through the process, and second is also in terms of those who are more wanting to do on their own, they feel free to try it out. Instructions are also given out there. Let me invite Kavita now to run you through the vibe coding design and prototyping session. Thank you. To continue on the topic of scaling, in financial services, AI doesn't scale when the model is ready. It's when legal compliance..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:24",
                    "end": "00:21:53",
                    "text": " regulatory, all of the body comes together. When they are ready, it's when AI scales. So I think that's one of the challenges that we face when we look at a scaling AI. This is just one of them. And my name is Kavita, and I lead the AI labs for financial services. And I help customers on their journey to adopt AI and scale AI. And one of the other challenges that really come across"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:53",
                    "end": "00:22:20",
                    "text": " when we start discussing about AI use cases is, how do I get the business teams on board? Because if you look at the kind of implementations we have seen so far in AI, you've got knowledge assistance, you've got chatbots, you've got good document AI capability. And if you look at these implementations, they have actually filled the gap where the traditional AI, traditional IT has failed in the sense..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:20",
                    "end": "00:22:48",
                    "text": " inability to access unstructured data. There was a gap which AI solved. So most of the use cases have been around these lines, and that's primarily because AI has not been, we have not let the business teams actually drive AI. So what does it take to help enable the business teams to understand the power of AI and look at more transformative use cases?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:48",
                    "end": "00:23:15",
                    "text": " So that's the construct of what we want to do today. We want you to experience that how a business user or a business team can be enabled with AI to accelerate that whole journey of identifying the right opportunity, doing a working prototype, and actually even testing it out and seeing the proof of value. How do we shift left?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:15",
                    "end": "00:23:43",
                    "text": " in the sense that how do we give business more control and working with the engineering team, build out a quick proof of value that you can validate and then scale to production. Because in one of the earlier conversations I had with one of you here was that scaling is a challenge. But sometimes you don't get to scale because you've not identified the right opportunity. You try to apply AI everywhere. So how do you use AI to help you in that journey?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:43",
                    "end": "00:24:09",
                    "text": " And what we want to achieve through this is let business lead the whole AI redesign. So how do we do that? And that's something that we will cover today. And what we're going to demonstrate to you is not a tool, but how we can bring all the AI capability together for a business audience."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:11",
                    "end": "00:24:41",
                    "text": " Yeah, switching on to that. So what we'll see today, disclaimer before I start, this tool is not a single tool. It's a workbench that we have created that brings together different AI capabilities. And then what we do is that we start off with something called a business challenge, because that's the starting point for a business team. Where do I begin my journey? What is the challenge that I need to address? And for the sake of this demonstration, I'll pick up one use case, something that's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:42",
                    "end": "00:25:00",
                    "text": " A challenge that has come up in one of our customer conversations was that you have a lot of portfolio managers who need to, at the end of the quarter, generate a summary of how their portfolio has performed. And along with collecting a lot of data, they need to actually"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:00",
                    "end": "00:25:30",
                    "text": " check out the sentiment of the stock on the public domain. So there's a lot of data gathering, a lot of analysis that comes in, and as a result, there's a lot of time that goes into that activity, and that delays the overall report, and sometimes this gets delegated to a junior analyst, the quality of the report is affected. So how do I bring all of that in a way that AI can help the portfolio managers solve the problem? So that was the challenge that we picked up, and I'll show you a quick demonstration of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:30",
                    "end": "00:25:56",
                    "text": " how the challenge works. The challenge was just creating an assistant to help the AI agent, to help the portfolio manager actually look up what is the right, what is the data that's available out there on the public domain, or even look up Yahoo Finance to see how the stock has performed. So all of that was being done by AI in the background. And this is a solution that..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:56",
                    "end": "00:26:25",
                    "text": " or when we piloted with our customers, they saw a huge advantage, because today, they were actually outsourcing a lot of this activity to a third party organization, and they were able to bring that back, and bring down the overall time to generate the report, and even the quality of the report started coming out much better. I thought I'd give you a view of the final outcome, and take you back to where we started. And that's the platform that we were trying to show you. So switching on."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:25",
                    "end": "00:26:54",
                    "text": " to that platform. So what we do as a first step is actually pick up the challenge and how do we use AI to ideate and come up with something called the opportunity canvas. So many of our users have, you know, if you see business users are all equipped with co-pilots because it's already available in the enterprise. But what we've observed is that just not giving them AI tools or training them on prompts is really going to help. It's about"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:54",
                    "end": "00:27:25",
                    "text": " helping them to understand that end-to-end process and letting them have complete control. So that's what we've done over here. So we looked at what could be a set of questions that a business user needs to look at when they see a challenge and convert it to an opportunity. So we've curated a set of prompts which can go through the data and come up with what we call the opportunity canvas. So we'll just walk you through those different steps. So I think one thing that we always need to keep in mind is that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:25",
                    "end": "00:27:45",
                    "text": " When we give AI to the business users, the research is primarily on, you can use a lot of research tools available, but the organization context is more important. How do we know what the user is? What are the different users going through? What are their challenges and pain points? So loading the context is the first step."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:45",
                    "end": "00:28:15",
                    "text": " And the reason I'm walking you through this is because once I finish the demo, I want you also to try and just see whether as a business leader or a practitioner how the whole experience will be and how can we really build that capability for business teams to use. So we'll go through this flow. So we have built the context. And then after that, you can even go back and do a deep research. Because sometimes your context is not enough. You want to know how your peers are performing."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:15",
                    "end": "00:28:43",
                    "text": " pull out a deep research from Gemini or ChachiPT. So put that context also in, and then you have an overall knowledge base that your business team can now use to start ideating. So what we have done is created an ideation framework here. Now, as you go through the ideation process, you'll have a lot of questions. Have I captured all the pain points right? Because you want to be sure that you have actually"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:43",
                    "end": "00:29:08",
                    "text": " solving a problem. So there's pain points that you capture. And there are quite a lot of steps. I won't go through all of them. But if you quickly go through, you will have pain points. You'll want to assess what the business impact is. What's the possible solution direction that you want to take? And as we go through these steps, you will see that it's not enough just to analyze the data that's available."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:08",
                    "end": "00:29:36",
                    "text": " but to actually start seeing AI as a thinking partner. That's what we want the business to do, be able to bounce ideas, validate them, feed the input, and finally, at the end of it, create what we call an opportunity canvas. So a lot of thought has gone in. So while these demonstrations, while he brings up the opportunity canvas, something that we have observed is that just giving a tool may not be the only solution."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:36",
                    "end": "00:30:03",
                    "text": " What we do in TC, something that's worked for us, is that we conduct an AI day for business teams, where we try to create that immersive experience where they understand what are the right AI tools and also how to use them and how to apply them in the context. So I'll, if you can just go through the, yeah. So there's a lot of questions, so we don't, we, and it will take a minute to go through, but if our demo is ready, you can show that as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:08",
                    "end": "00:30:37",
                    "text": " Yeah, so this is what the final product looked like. So you give a company name, and it went through the organization's research. It also went to the public domain and started picking up all the stock information and brought that up into an overall summary that can be then used by the portfolio manager to make a decision on how his stock has performed, how do I report that?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:37",
                    "end": "00:31:10",
                    "text": " to his customers? How do I report that into the performance reporting? And if you see, it was instant. And it's gone through, I think, 15 to 20 sites. We've configured all the financial sites to go through and capture the sentiment, capture key financial metrics, all of that instantly on a very high-speed internet. Now, I think we would be done with the Opportunity Canvas. Last one step."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:10",
                    "end": "00:31:39",
                    "text": " So it's important that we go through these steps. So that's why we have consciously made these steps so that you don't pick an idea which is not the right one. So identifying the opportunity, what is the possible solution direction, what's the impact, and what your roadmap will be. And there you have an opportunity canvas, which the user has co-created along with AI. Now you go on to the next step. And this is important."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:39",
                    "end": "00:32:08",
                    "text": " All this is fine. The business team could be comfortable in Opportunity Canvas. But can now the business team, with the help of AI, drive the overall design process? So what you'll see is that you can brainstorm with AI and actually give a direction of what that design should be. And this is what AI has generated. It's kind of broken down that whole opportunity into a set of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:08",
                    "end": "00:32:33",
                    "text": " different agents that can go in. Because as we covered in the previous session, a lot of attention focuses on moving to agentic. So can we look at building what that agentic solution would look like is what you're going to see here. It's not done the coding yet, but what it has done is broken up that entire opportunity into a series of agents which can address this need. And we'll just."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:33",
                    "end": "00:33:01",
                    "text": " You can try that. You will get a chance to try this, so I won't go into the detail. But here, it's come up with four or five agents. It's organized. And as a business team, they can actually go and say, I want these kind of checks and balances. I want to do it this way. And now you see the difference, right? Earlier, it was all left to the engineering team to actually build the solution. But now, we are passing the power back to the business to drive all of this. Now, going forward,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:01",
                    "end": "00:33:33",
                    "text": " What is the next important step is to validate what this opportunity looks like from an experience. So you use a simple wipe coding tool. We have here used both Lovable and Google AI Studio. This tool generates the prompt. You copy the prompt. And then you actually can see it through a visual experience."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:33",
                    "end": "00:34:03",
                    "text": " simple as putting it and if you see the prompt it's all English and you know when you get into your wipe session you'll see that you can actually edit some of it and see what that experience looks like so this will take how much time maybe a minute or so do you have something do you have something that you've already pre-built yeah so this is something that that came out of the tool a very very nice experience which shows what the portfolio summary should look like what are the stocks etc so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:03",
                    "end": "00:34:32",
                    "text": " kind of understood the opportunity and translated that. And you can play around with it and tweak it to the experience. This step becomes very important because what earlier took two to three weeks of building a UX prototype as the end of a design phase actually got converted to minutes. And this is something you take to these stakeholders, validate it, and you're ready with what is that real challenge you want to address, what's the real opportunity comes out so quickly."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:32",
                    "end": "00:35:02",
                    "text": " We can't stop with just a UX prototype, because that just gives you an experience. You need to actually engineer some of the back-end capability. And there are low-core platforms that are available. So in today's live session, we will show you that as well. And one of the examples that we have shown here is a solution called Ne10. It's a low-core platform building. It's not that we are saying this is the right tool. As per your enterprise decision, you could have different options."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:02",
                    "end": "00:35:30",
                    "text": " But this really helps, in our view, to ideate and understand or discuss with your engineering teams how your design should be. You can easily visualize what are the models being used here. Have I thought through all the guardrails? Have I built my business evaluations? All of this can be verified in one shot. And then from this, the code can get generated, and you'll have a working end-to-end solution."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:30",
                    "end": "00:35:52",
                    "text": " So this is the process we followed for quickly prototyping. And the demo that you saw was an example of something that was built using this approach. So what we'd like to do now is shift back to you and see if you'd like to try this. I mean, some of you could. We have laptops configured with."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:52",
                    "end": "00:36:46",
                    "text": " with user IDs, we have all the tools available. It's an opportunity for you to try out and we have mentors. We have mentors available for each table. A lot of our teams here to support you and we would really encourage you to try and we can guide you here as you go along. Explore, yeah. Okay, so let me just walk you through what you would do. So all of your laptops will have a handbook already preloaded."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:46",
                    "end": "00:37:15",
                    "text": " which talks about the use case that you would be picking up. Okay, so, and that is in green, it's tagged as challenge. So that's the use case that you will start with. And you have different phases and you have resources for each of the phase. So feel free to go through this handbook, take it step by step, and we'll help you with it. You are workspace five."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:16",
                    "end": "01:07:22",
                    "text": " So log into your workspace based on the number that's on your table. So when you start, it will be the discovery and design. Yeah, you can start the solution with the resource present there. Okay, so just a quick time check. We just have five minutes. So if you want some help on any particular section, just raise your hand. We'll come and help. But I'm so happy to see many of you have generated very nice UX prototypes."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:07:22",
                    "end": "01:09:27",
                    "text": " And you can also move into that build phase. That's an optional one, but try it out. We'll have mentors to support you."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:09:27",
                    "end": "01:10:13",
                    "text": " uh..."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:10:13",
                    "end": "01:11:40",
                    "text": " So I think we'll just conclude."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:11:45",
                    "end": "01:12:14",
                    "text": " So, so far we spoke about, I only covered, I'll hand it over to him, but what you did essentially comes as step one. You actually did, and I'm so happy that many of you got challenge completed, and you saw your code getting converted, and I see a lot of happy faces because it was, I heard words like, amazing, I can do the same thing with my tools, so very happy to hear all that. So what we did,"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:12:14",
                    "end": "01:12:43",
                    "text": " was the first step, proof of value. We've tried to establish this is something, a problem worth solving for my organization and you may have an answer then, you have an answer at the end of the stage. But the journey to take it to production involves something called rapid building. What you've built is a foundation which you can add on more functionality and then create it into a full-fledged solution. And that's where we call the rapid build."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:12:43",
                    "end": "01:13:14",
                    "text": " which is about engineering the full-fledged solution before you go to production. And then finally, the scale. And when you look at scale, and someone asked me that question, you have to build a lot of AI foundation capability, the control plane, a lot of things need to go in. And that's important if you want to bring that scale. I'll just hand it over to Onni to close. I think thank you. Thank you for being a real sport out there. I think all of you joining in, being through the wipe session as well. Great to see all of you continue out there."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:13:14",
                    "end": "01:13:41",
                    "text": " Once again, thanks for being participating. I think this is one important thing that we believe AI would make a difference in the ability to quickly translate an idea to prototype. Not to say that that prototype is what you would eventually scale to production, but that gives you the feel of what can be done. And then you naturally scale it into production in a very calibrated way. But again, thanks for your time. And then being part of it, feel free to reach out. Some of these I'm assuming."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:13:41",
                    "end": "01:13:54",
                    "text": " would have been captured in your email, et cetera. So it's available for you to do some prompting and wiping back home as well if you're adventurous and wanting to extend the focus. Thank you once again and have a good rest of the day."
                }
            ]
        }
    ],
    "projectDeleted": false,
    "codeFrameColumn": [
        {
            "col1": "col1",
            "col2": "col3",
            "researchObjective": "",
            "filterColumn": [],
            "netSetting": true,
            "subNetSetting": true,
            "categorySetting": true,
            "sentimentSetting": true,
            "thinkingBudgetSetting": 2000,
            "questionType": "OPENEND",
            "otherBrand": false,
            "existingCodeFrame": [],
            "isList": false,
            "codeFrameWithDataList": [],
            "codeFrameList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 8881,
                        "CandidatesTokenCount": 2912,
                        "TotalTokenCount": 13688,
                        "thoughtsTokenCount": 1895,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Net": "Adopción y Desafíos de la IA en Servicios Financieros",
                                "Subnets": [
                                    {
                                        "Subnet": "Tendencias de Adopción",
                                        "Categories": [
                                            {
                                                "Category": "Fase de Hiperaceleración de la IA",
                                                "Summary": "La IA ha entrado en una fase de rápida aceleración, impulsando nuevas capacidades y usos."
                                            },
                                            {
                                                "Category": "Agentes en Acción vs. Chatbots Tradicionales",
                                                "Summary": "Cambio de interfaces de chatbot tradicionales a agentes de IA más proactivos y orientados a la acción."
                                            },
                                            {
                                                "Category": "Adopción Multimodal",
                                                "Summary": "Enfoque en el uso de múltiples modelos de IA (texto, voz, imagen) para completar tareas complejas."
                                            }
                                        ],
                                        "Summary": "Explora el estado actual y las direcciones futuras de la implementación de la IA en el sector financiero."
                                    },
                                    {
                                        "Subnet": "Desafíos de la Adopción",
                                        "Categories": [
                                            {
                                                "Category": "ROI Desafiante",
                                                "Summary": "Dificultades para demostrar y medir el retorno de la inversión de las soluciones de IA."
                                            },
                                            {
                                                "Category": "Precisión y Cumplimiento Normativo",
                                                "Summary": "Necesidad de modelos de IA altamente precisos que cumplan con las estrictas regulaciones del sector financiero."
                                            },
                                            {
                                                "Category": "Nuances Lingüísticos Regionales",
                                                "Summary": "Desafíos específicos en mercados como India debido a la diversidad de idiomas y dialectos, requiriendo modelos localizados."
                                            }
                                        ],
                                        "Summary": "Identifica los principales obstáculos que enfrentan las empresas al implementar y escalar la IA."
                                    },
                                    {
                                        "Subnet": "Requerimientos Específicos de la Industria",
                                        "Categories": [
                                            {
                                                "Category": "Modelos Empresariales Específicos",
                                                "Summary": "Demanda de modelos de IA adaptados a las necesidades y datos internos de cada empresa."
                                            },
                                            {
                                                "Category": "Modelos Afinados para Precisión",
                                                "Summary": "Uso de modelos pre-entrenados afinados para tareas y dominios específicos, garantizando mayor exactitud."
                                            },
                                            {
                                                "Category": "Regulaciones y Seguridad",
                                                "Summary": "Importancia de la seguridad y el cumplimiento normativo en el diseño y despliegue de soluciones de IA."
                                            }
                                        ],
                                        "Summary": "Detalla las exigencias particulares que la industria financiera impone a las soluciones de IA."
                                    }
                                ],
                                "Summary": "Abarca la forma en que la IA está siendo adoptada en servicios financieros y los obstáculos inherentes a este proceso."
                            },
                            {
                                "Net": "Desarrollo y Prototipado de Soluciones de IA",
                                "Subnets": [
                                    {
                                        "Subnet": "Metodología de Prototipado",
                                        "Categories": [
                                            {
                                                "Category": "Prueba de Valor Rápida",
                                                "Summary": "Enfoque en la creación rápida de prototipos para validar la viabilidad y el valor de una idea de IA."
                                            },
                                            {
                                                "Category": "Diseño y Prototipado Ágil",
                                                "Summary": "Proceso iterativo que permite a los equipos de negocio participar activamente en el diseño y la creación de prototipos."
                                            },
                                            {
                                                "Category": "Desarrollo Rápido e Iterativo",
                                                "Summary": "Métodos que permiten pasar de una idea a un prototipo funcional en semanas, seguido de escalado."
                                            }
                                        ],
                                        "Summary": "Describe los enfoques y procesos para construir y validar rápidamente soluciones de IA."
                                    },
                                    {
                                        "Subnet": "Empoderamiento del Usuario de Negocio",
                                        "Categories": [
                                            {
                                                "Category": "IA como Socio de Pensamiento",
                                                "Summary": "Posicionar la IA como una herramienta que ayuda a los usuarios de negocio a idear y validar conceptos."
                                            },
                                            {
                                                "Category": "Control del Proceso de Diseño por Negocios",
                                                "Summary": "Permitir que los equipos de negocio lideren y dirijan la creación y el diseño de soluciones de IA."
                                            },
                                            {
                                                "Category": "Validación de la Experiencia del Usuario",
                                                "Summary": "Reducir el tiempo de creación de prototipos de UX para obtener una validación rápida de los stakeholders."
                                            }
                                        ],
                                        "Summary": "Se centra en cómo involucrar y capacitar a los equipos de negocio para impulsar la innovación con IA."
                                    },
                                    {
                                        "Subnet": "Herramientas y Plataformas",
                                        "Categories": [
                                            {
                                                "Category": "Bancos de Trabajo de IA",
                                                "Summary": "Uso de plataformas integradas que combinan diversas capacidades de IA para usuarios de negocio."
                                            },
                                            {
                                                "Category": "Plataformas de Código Bajo/Sin Código",
                                                "Summary": "Herramientas que permiten a los usuarios no técnicos construir y probar soluciones de IA rápidamente."
                                            },
                                            {
                                                "Category": "Integración de Capacidades de IA",
                                                "Summary": "Combinación de diferentes funcionalidades y modelos de IA para crear soluciones completas."
                                            }
                                        ],
                                        "Summary": "Explora las herramientas y tecnologías utilizadas para facilitar el prototipado y desarrollo de IA."
                                    }
                                ],
                                "Summary": "Se enfoca en las estrategias y el rol de los usuarios en la creación de prototipos y soluciones de IA."
                            },
                            {
                                "Net": "Aplicación y Casos de Uso de la IA",
                                "Subnets": [
                                    {
                                        "Subnet": "Casos de Uso en Negocios",
                                        "Categories": [
                                            {
                                                "Category": "Atención al Cliente Conversacional",
                                                "Summary": "Uso de IA para mejorar y automatizar las interacciones de servicio al cliente a través de conversaciones."
                                            },
                                            {
                                                "Category": "Asesoramiento de Inversiones",
                                                "Summary": "Implementación de IA para proporcionar orientación y recomendaciones en el ámbito de las finanzas y las inversiones."
                                            },
                                            {
                                                "Category": "Predicción de Litigios y Detección de Fraude",
                                                "Summary": "Modelos de IA afinados para identificar riesgos legales y patrones de fraude en transacciones y reclamaciones."
                                            }
                                        ],
                                        "Summary": "Describe ejemplos concretos de cómo la IA se aplica para resolver problemas y mejorar procesos en áreas de negocio."
                                    },
                                    {
                                        "Subnet": "Casos de Uso en TI",
                                        "Categories": [
                                            {
                                                "Category": "Asistencia en Codificación",
                                                "Summary": "IA que ayuda a los desarrolladores con la generación, depuración y optimización de código."
                                            },
                                            {
                                                "Category": "Modernización Tecnológica",
                                                "Summary": "Uso de IA para acelerar la actualización y mejora de sistemas y procesos tecnológicos existentes."
                                            },
                                            {
                                                "Category": "Automatización de Operaciones",
                                                "Summary": "Aplicación de IA para optimizar y automatizar tareas y flujos de trabajo en operaciones de TI."
                                            }
                                        ],
                                        "Summary": "Presenta aplicaciones de la IA que benefician directamente a las funciones y operaciones de tecnología de la información."
                                    },
                                    {
                                        "Subnet": "Modelos y Enfoques de IA",
                                        "Categories": [
                                            {
                                                "Category": "Modelos de Lenguaje Grandes (LLMs)",
                                                "Summary": "Utilización de modelos de IA de propósito general para tareas amplias y escalables."
                                            },
                                            {
                                                "Category": "Modelos Afinados y Adaptados al Dominio",
                                                "Summary": "Implementación de modelos especializados para contextos específicos, aumentando la precisión y relevancia."
                                            },
                                            {
                                                "Category": "Modelos Híbridos",
                                                "Summary": "Combinación de LLMs y modelos afinados para aprovechar las fortalezas de ambos en diferentes tareas."
                                            }
                                        ],
                                        "Summary": "Detalla los diferentes tipos de modelos de IA y las estrategias para su aplicación efectiva."
                                    }
                                ],
                                "Summary": "Presenta diversas aplicaciones y escenarios donde la IA puede generar valor, tanto en áreas de negocio como de TI."
                            },
                            {
                                "Net": "Escalabilidad e Integración Empresarial",
                                "Subnets": [
                                    {
                                        "Subnet": "Barreras para la Escalabilidad",
                                        "Categories": [
                                            {
                                                "Category": "Cumplimiento Legal y Regulatorio",
                                                "Summary": "Obstáculos relacionados con las normativas legales y la conformidad de la IA en entornos regulados."
                                            },
                                            {
                                                "Category": "Alineación de Equipos de Negocio y TI",
                                                "Summary": "Desafíos en la colaboración y el entendimiento mutuo entre los equipos de negocio y tecnología para la implementación de IA."
                                            },
                                            {
                                                "Category": "Ausencia de Oportunidades Claras",
                                                "Summary": "Dificultad para identificar los casos de uso de IA más impactantes y adecuados para la escala."
                                            }
                                        ],
                                        "Summary": "Identifica los desafíos clave que impiden la expansión y adopción a gran escala de soluciones de IA."
                                    },
                                    {
                                        "Subnet": "Integración de Sistemas",
                                        "Categories": [
                                            {
                                                "Category": "Conexión a Sistemas Empresariales Heredados",
                                                "Summary": "La dificultad de integrar soluciones de IA con la infraestructura tecnológica preexistente de la empresa."
                                            },
                                            {
                                                "Category": "Orquestación de Flujos de Trabajo",
                                                "Summary": "Diseño y gestión de cómo las soluciones de IA interactúan con los procesos de negocio existentes."
                                            },
                                            {
                                                "Category": "Integración con CRM/Core Banking",
                                                "Summary": "Conectar la IA con sistemas críticos como la gestión de relaciones con clientes y los sistemas bancarios centrales."
                                            }
                                        ],
                                        "Summary": "Describe los retos y soluciones relacionados con la interconexión de la IA con la infraestructura tecnológica de una organización."
                                    },
                                    {
                                        "Subnet": "Infraestructura y Plataforma",
                                        "Categories": [
                                            {
                                                "Category": "Inversiones en Hardware de IA",
                                                "Summary": "La necesidad de invertir en hardware y centros de datos especializados para soportar las cargas de trabajo de IA."
                                            },
                                            {
                                                "Category": "Plataformas para Escalado",
                                                "Summary": "Diseño y uso de plataformas tecnológicas que faciliten el despliegue y la gestión de la IA a escala."
                                            },
                                            {
                                                "Category": "Pila Vertical de IA",
                                                "Summary": "Enfoque integral que considera la IA desde la infraestructura hasta las aplicaciones finales para un mayor impacto."
                                            }
                                        ],
                                        "Summary": "Se enfoca en los requisitos de hardware, software y arquitectura necesarios para soportar y escalar la IA."
                                    }
                                ],
                                "Summary": "Aborda los factores críticos para llevar las soluciones de IA de prototipos a implementaciones productivas y a gran escala."
                            },
                            {
                                "Net": "Implementación Estratégica de la IA",
                                "Subnets": [
                                    {
                                        "Subnet": "Enfoque Centrado en el Resultado",
                                        "Categories": [
                                            {
                                                "Category": "Medición de Resultados de Negocio",
                                                "Summary": "Énfasis en el establecimiento de métricas claras para evaluar el impacto real de la IA en los objetivos empresariales."
                                            },
                                            {
                                                "Category": "Diseño \"de Derecha a Izquierda\"",
                                                "Summary": "Metodología que comienza con el resultado deseado y luego retrocede para definir la implementación de IA."
                                            },
                                            {
                                                "Category": "Calibración de Aplicación de IA",
                                                "Summary": "Determinación de dónde y cómo aplicar la IA de manera más efectiva, y dónde abstenerse."
                                            }
                                        ],
                                        "Summary": "Destaca la importancia de alinear las iniciativas de IA con los objetivos de negocio y medir su impacto."
                                    },
                                    {
                                        "Subnet": "Colaboración entre Equipos",
                                        "Categories": [
                                            {
                                                "Category": "Habilitación de Equipos Interdisciplinarios",
                                                "Summary": "Fomentar la interacción entre expertos en IA, negocios y reguladores para una implementación exitosa."
                                            },
                                            {
                                                "Category": "Entrenamiento y Capacitación en IA",
                                                "Summary": "Programas para educar a los equipos de negocio sobre las capacidades y el potencial de la IA."
                                            },
                                            {
                                                "Category": "Co-creación con IA",
                                                "Summary": "Uso de la IA como un socio colaborativo para idear y desarrollar soluciones conjuntamente con los equipos."
                                            }
                                        ],
                                        "Summary": "Subraya la necesidad de una cooperación fluida entre los diferentes departamentos y roles en la empresa para implementar la IA."
                                    },
                                    {
                                        "Subnet": "Evaluación y Medición",
                                        "Categories": [
                                            {
                                                "Category": "Métricas de Adopción y Rendimiento",
                                                "Summary": "Establecimiento de indicadores clave para monitorear el progreso y la eficacia de la IA."
                                            },
                                            {
                                                "Category": "Validación de Valor Empresarial",
                                                "Summary": "Proceso para asegurar que las soluciones de IA generen un valor tangible y medible para el negocio."
                                            },
                                            {
                                                "Category": "Bucles de Retroalimentación Iterativos",
                                                "Summary": "Sistemas para recopilar información continuamente y mejorar las soluciones de IA a lo largo del tiempo."
                                            }
                                        ],
                                        "Summary": "Se centra en los métodos para evaluar la efectividad y el éxito de las implementaciones de IA."
                                    }
                                ],
                                "Summary": "Describe las directrices y principios para una adopción de IA exitosa y sostenible en la empresa."
                            },
                            {
                                "Net": "996 > No específico",
                                "Subnets": [
                                    {
                                        "Subnet": "Contenido Irrelevante",
                                        "Categories": [
                                            {
                                                "Category": "Respuestas fuera de tema",
                                                "Summary": "Respuestas que no tienen relación con la pregunta o el contexto de la encuesta."
                                            }
                                        ],
                                        "Summary": "Respuestas que son irrelevantes o no encajan fácilmente en las categorías definidas."
                                    }
                                ],
                                "Summary": "Para respuestas irrelevantes o que no pueden ser fácilmente codificadas en las categorías temáticas."
                            },
                            {
                                "Net": "997 > No sé / No puedo decir",
                                "Subnets": [
                                    {
                                        "Subnet": "Respuestas Indecisas",
                                        "Categories": [
                                            {
                                                "Category": "Dato desconocido",
                                                "Summary": "El encuestado indica que no posee la información o el conocimiento necesario."
                                            },
                                            {
                                                "Category": "Opinión vaga",
                                                "Summary": "El encuestado expresa una opinión no comprometida o poco clara."
                                            }
                                        ],
                                        "Summary": "Respuestas no comprometidas, ambiguas o que expresan falta de conocimiento."
                                    }
                                ],
                                "Summary": "Para respuestas que indican incertidumbre, falta de conocimiento o indecisión por parte del encuestado."
                            },
                            {
                                "Net": "998 > No aplica",
                                "Subnets": [
                                    {
                                        "Subnet": "Fuera de Alcance",
                                        "Categories": [
                                            {
                                                "Category": "Condición no aplicable",
                                                "Summary": "La pregunta no es pertinente para la situación o experiencia del encuestado."
                                            },
                                            {
                                                "Category": "Situación inexistente",
                                                "Summary": "El encuestado no ha experimentado lo que se pregunta."
                                            }
                                        ],
                                        "Summary": "Respuestas que son inaplicables a la situación o experiencia del encuestado."
                                    }
                                ],
                                "Summary": "Para respuestas que no son pertinentes o aplicables a la pregunta formulada."
                            },
                            {
                                "Net": "999 > Nada / Sin sentido",
                                "Subnets": [
                                    {
                                        "Subnet": "Datos Vacíos o Incomprensibles",
                                        "Categories": [
                                            {
                                                "Category": "Caracteres aleatorios",
                                                "Summary": "Respuestas que consisten en letras o símbolos sin significado."
                                            },
                                            {
                                                "Category": "Sin respuesta",
                                                "Summary": "El encuestado no proporcionó ninguna información o dejó el campo en blanco."
                                            },
                                            {
                                                "Category": "Texto ilegible",
                                                "Summary": "Contenido que no puede ser entendido o interpretado."
                                            }
                                        ],
                                        "Summary": "Respuestas que son gibberish, caracteres no lingüísticos o carecen de significado alguno."
                                    }
                                ],
                                "Summary": "Para respuestas que son incomprensibles, no tienen sentido o están vacías."
                            }
                        ],
                        "Summary": "Este documento presenta una discusión detallada sobre la adopción de la Inteligencia Artificial (IA) en la industria de servicios financieros, destacando las tendencias actuales, los desafíos y las metodologías para el desarrollo, prototipado y escalado de soluciones de IA. Se enfatiza la necesidad de empoderar a los equipos de negocio en el proceso de diseño de IA, la importancia de los modelos adaptados al dominio y la integración de la IA con los sistemas empresariales existentes. Además, se exploran casos de uso prácticos tanto en operaciones de negocio como en TI, subrayando la importancia de un enfoque estratégico y calibrado para la implementación de la IA."
                    }
                }
            ],
            "onlyCodeFrame": true,
            "errorList": [],
            "codeFrameGenerateType": "CODE_FRAME_ONLY",
            "requestType": "CODEFRAME",
            "requestMode": "DEVELOPMENT",
            "sourceLanguage": "English",
            "targetLanguage": "Spanish",
            "response": {
                "status": "OK",
                "data": {
                    "status": "OK",
                    "PromptTokenCount": 8881,
                    "CandidatesTokenCount": 2912,
                    "TotalTokenCount": 13688,
                    "thoughtsTokenCount": 1895,
                    "finishReason": "STOP",
                    "data": [
                        {
                            "Net": "Adopción y Desafíos de la IA en Servicios Financieros",
                            "Subnets": [
                                {
                                    "Subnet": "Tendencias de Adopción",
                                    "Categories": [
                                        {
                                            "Category": "Fase de Hiperaceleración de la IA",
                                            "Summary": "La IA ha entrado en una fase de rápida aceleración, impulsando nuevas capacidades y usos.",
                                            "code": 3
                                        },
                                        {
                                            "Category": "Agentes en Acción vs. Chatbots Tradicionales",
                                            "Summary": "Cambio de interfaces de chatbot tradicionales a agentes de IA más proactivos y orientados a la acción.",
                                            "code": 4
                                        },
                                        {
                                            "Category": "Adopción Multimodal",
                                            "Summary": "Enfoque en el uso de múltiples modelos de IA (texto, voz, imagen) para completar tareas complejas.",
                                            "code": 5
                                        }
                                    ],
                                    "Summary": "Explora el estado actual y las direcciones futuras de la implementación de la IA en el sector financiero.",
                                    "code": 2
                                },
                                {
                                    "Subnet": "Desafíos de la Adopción",
                                    "Categories": [
                                        {
                                            "Category": "ROI Desafiante",
                                            "Summary": "Dificultades para demostrar y medir el retorno de la inversión de las soluciones de IA.",
                                            "code": 7
                                        },
                                        {
                                            "Category": "Precisión y Cumplimiento Normativo",
                                            "Summary": "Necesidad de modelos de IA altamente precisos que cumplan con las estrictas regulaciones del sector financiero.",
                                            "code": 8
                                        },
                                        {
                                            "Category": "Nuances Lingüísticos Regionales",
                                            "Summary": "Desafíos específicos en mercados como India debido a la diversidad de idiomas y dialectos, requiriendo modelos localizados.",
                                            "code": 9
                                        }
                                    ],
                                    "Summary": "Identifica los principales obstáculos que enfrentan las empresas al implementar y escalar la IA.",
                                    "code": 6
                                },
                                {
                                    "Subnet": "Requerimientos Específicos de la Industria",
                                    "Categories": [
                                        {
                                            "Category": "Modelos Empresariales Específicos",
                                            "Summary": "Demanda de modelos de IA adaptados a las necesidades y datos internos de cada empresa.",
                                            "code": 11
                                        },
                                        {
                                            "Category": "Modelos Afinados para Precisión",
                                            "Summary": "Uso de modelos pre-entrenados afinados para tareas y dominios específicos, garantizando mayor exactitud.",
                                            "code": 12
                                        },
                                        {
                                            "Category": "Regulaciones y Seguridad",
                                            "Summary": "Importancia de la seguridad y el cumplimiento normativo en el diseño y despliegue de soluciones de IA.",
                                            "code": 13
                                        }
                                    ],
                                    "Summary": "Detalla las exigencias particulares que la industria financiera impone a las soluciones de IA.",
                                    "code": 10
                                }
                            ],
                            "Summary": "Abarca la forma en que la IA está siendo adoptada en servicios financieros y los obstáculos inherentes a este proceso.",
                            "code": 1
                        },
                        {
                            "Net": "Desarrollo y Prototipado de Soluciones de IA",
                            "Subnets": [
                                {
                                    "Subnet": "Metodología de Prototipado",
                                    "Categories": [
                                        {
                                            "Category": "Prueba de Valor Rápida",
                                            "Summary": "Enfoque en la creación rápida de prototipos para validar la viabilidad y el valor de una idea de IA.",
                                            "code": 16
                                        },
                                        {
                                            "Category": "Diseño y Prototipado Ágil",
                                            "Summary": "Proceso iterativo que permite a los equipos de negocio participar activamente en el diseño y la creación de prototipos.",
                                            "code": 17
                                        },
                                        {
                                            "Category": "Desarrollo Rápido e Iterativo",
                                            "Summary": "Métodos que permiten pasar de una idea a un prototipo funcional en semanas, seguido de escalado.",
                                            "code": 18
                                        }
                                    ],
                                    "Summary": "Describe los enfoques y procesos para construir y validar rápidamente soluciones de IA.",
                                    "code": 15
                                },
                                {
                                    "Subnet": "Empoderamiento del Usuario de Negocio",
                                    "Categories": [
                                        {
                                            "Category": "IA como Socio de Pensamiento",
                                            "Summary": "Posicionar la IA como una herramienta que ayuda a los usuarios de negocio a idear y validar conceptos.",
                                            "code": 20
                                        },
                                        {
                                            "Category": "Control del Proceso de Diseño por Negocios",
                                            "Summary": "Permitir que los equipos de negocio lideren y dirijan la creación y el diseño de soluciones de IA.",
                                            "code": 21
                                        },
                                        {
                                            "Category": "Validación de la Experiencia del Usuario",
                                            "Summary": "Reducir el tiempo de creación de prototipos de UX para obtener una validación rápida de los stakeholders.",
                                            "code": 22
                                        }
                                    ],
                                    "Summary": "Se centra en cómo involucrar y capacitar a los equipos de negocio para impulsar la innovación con IA.",
                                    "code": 19
                                },
                                {
                                    "Subnet": "Herramientas y Plataformas",
                                    "Categories": [
                                        {
                                            "Category": "Bancos de Trabajo de IA",
                                            "Summary": "Uso de plataformas integradas que combinan diversas capacidades de IA para usuarios de negocio.",
                                            "code": 24
                                        },
                                        {
                                            "Category": "Plataformas de Código Bajo/Sin Código",
                                            "Summary": "Herramientas que permiten a los usuarios no técnicos construir y probar soluciones de IA rápidamente.",
                                            "code": 25
                                        },
                                        {
                                            "Category": "Integración de Capacidades de IA",
                                            "Summary": "Combinación de diferentes funcionalidades y modelos de IA para crear soluciones completas.",
                                            "code": 26
                                        }
                                    ],
                                    "Summary": "Explora las herramientas y tecnologías utilizadas para facilitar el prototipado y desarrollo de IA.",
                                    "code": 23
                                }
                            ],
                            "Summary": "Se enfoca en las estrategias y el rol de los usuarios en la creación de prototipos y soluciones de IA.",
                            "code": 14
                        },
                        {
                            "Net": "Aplicación y Casos de Uso de la IA",
                            "Subnets": [
                                {
                                    "Subnet": "Casos de Uso en Negocios",
                                    "Categories": [
                                        {
                                            "Category": "Atención al Cliente Conversacional",
                                            "Summary": "Uso de IA para mejorar y automatizar las interacciones de servicio al cliente a través de conversaciones.",
                                            "code": 29
                                        },
                                        {
                                            "Category": "Asesoramiento de Inversiones",
                                            "Summary": "Implementación de IA para proporcionar orientación y recomendaciones en el ámbito de las finanzas y las inversiones.",
                                            "code": 30
                                        },
                                        {
                                            "Category": "Predicción de Litigios y Detección de Fraude",
                                            "Summary": "Modelos de IA afinados para identificar riesgos legales y patrones de fraude en transacciones y reclamaciones.",
                                            "code": 31
                                        }
                                    ],
                                    "Summary": "Describe ejemplos concretos de cómo la IA se aplica para resolver problemas y mejorar procesos en áreas de negocio.",
                                    "code": 28
                                },
                                {
                                    "Subnet": "Casos de Uso en TI",
                                    "Categories": [
                                        {
                                            "Category": "Asistencia en Codificación",
                                            "Summary": "IA que ayuda a los desarrolladores con la generación, depuración y optimización de código.",
                                            "code": 33
                                        },
                                        {
                                            "Category": "Modernización Tecnológica",
                                            "Summary": "Uso de IA para acelerar la actualización y mejora de sistemas y procesos tecnológicos existentes.",
                                            "code": 34
                                        },
                                        {
                                            "Category": "Automatización de Operaciones",
                                            "Summary": "Aplicación de IA para optimizar y automatizar tareas y flujos de trabajo en operaciones de TI.",
                                            "code": 35
                                        }
                                    ],
                                    "Summary": "Presenta aplicaciones de la IA que benefician directamente a las funciones y operaciones de tecnología de la información.",
                                    "code": 32
                                },
                                {
                                    "Subnet": "Modelos y Enfoques de IA",
                                    "Categories": [
                                        {
                                            "Category": "Modelos de Lenguaje Grandes (LLMs)",
                                            "Summary": "Utilización de modelos de IA de propósito general para tareas amplias y escalables.",
                                            "code": 37
                                        },
                                        {
                                            "Category": "Modelos Afinados y Adaptados al Dominio",
                                            "Summary": "Implementación de modelos especializados para contextos específicos, aumentando la precisión y relevancia.",
                                            "code": 38
                                        },
                                        {
                                            "Category": "Modelos Híbridos",
                                            "Summary": "Combinación de LLMs y modelos afinados para aprovechar las fortalezas de ambos en diferentes tareas.",
                                            "code": 39
                                        }
                                    ],
                                    "Summary": "Detalla los diferentes tipos de modelos de IA y las estrategias para su aplicación efectiva.",
                                    "code": 36
                                }
                            ],
                            "Summary": "Presenta diversas aplicaciones y escenarios donde la IA puede generar valor, tanto en áreas de negocio como de TI.",
                            "code": 27
                        },
                        {
                            "Net": "Escalabilidad e Integración Empresarial",
                            "Subnets": [
                                {
                                    "Subnet": "Barreras para la Escalabilidad",
                                    "Categories": [
                                        {
                                            "Category": "Cumplimiento Legal y Regulatorio",
                                            "Summary": "Obstáculos relacionados con las normativas legales y la conformidad de la IA en entornos regulados.",
                                            "code": 42
                                        },
                                        {
                                            "Category": "Alineación de Equipos de Negocio y TI",
                                            "Summary": "Desafíos en la colaboración y el entendimiento mutuo entre los equipos de negocio y tecnología para la implementación de IA.",
                                            "code": 43
                                        },
                                        {
                                            "Category": "Ausencia de Oportunidades Claras",
                                            "Summary": "Dificultad para identificar los casos de uso de IA más impactantes y adecuados para la escala.",
                                            "code": 44
                                        }
                                    ],
                                    "Summary": "Identifica los desafíos clave que impiden la expansión y adopción a gran escala de soluciones de IA.",
                                    "code": 41
                                },
                                {
                                    "Subnet": "Integración de Sistemas",
                                    "Categories": [
                                        {
                                            "Category": "Conexión a Sistemas Empresariales Heredados",
                                            "Summary": "La dificultad de integrar soluciones de IA con la infraestructura tecnológica preexistente de la empresa.",
                                            "code": 46
                                        },
                                        {
                                            "Category": "Orquestación de Flujos de Trabajo",
                                            "Summary": "Diseño y gestión de cómo las soluciones de IA interactúan con los procesos de negocio existentes.",
                                            "code": 47
                                        },
                                        {
                                            "Category": "Integración con CRM/Core Banking",
                                            "Summary": "Conectar la IA con sistemas críticos como la gestión de relaciones con clientes y los sistemas bancarios centrales.",
                                            "code": 48
                                        }
                                    ],
                                    "Summary": "Describe los retos y soluciones relacionados con la interconexión de la IA con la infraestructura tecnológica de una organización.",
                                    "code": 45
                                },
                                {
                                    "Subnet": "Infraestructura y Plataforma",
                                    "Categories": [
                                        {
                                            "Category": "Inversiones en Hardware de IA",
                                            "Summary": "La necesidad de invertir en hardware y centros de datos especializados para soportar las cargas de trabajo de IA.",
                                            "code": 50
                                        },
                                        {
                                            "Category": "Plataformas para Escalado",
                                            "Summary": "Diseño y uso de plataformas tecnológicas que faciliten el despliegue y la gestión de la IA a escala.",
                                            "code": 51
                                        },
                                        {
                                            "Category": "Pila Vertical de IA",
                                            "Summary": "Enfoque integral que considera la IA desde la infraestructura hasta las aplicaciones finales para un mayor impacto.",
                                            "code": 52
                                        }
                                    ],
                                    "Summary": "Se enfoca en los requisitos de hardware, software y arquitectura necesarios para soportar y escalar la IA.",
                                    "code": 49
                                }
                            ],
                            "Summary": "Aborda los factores críticos para llevar las soluciones de IA de prototipos a implementaciones productivas y a gran escala.",
                            "code": 40
                        },
                        {
                            "Net": "Implementación Estratégica de la IA",
                            "Subnets": [
                                {
                                    "Subnet": "Enfoque Centrado en el Resultado",
                                    "Categories": [
                                        {
                                            "Category": "Medición de Resultados de Negocio",
                                            "Summary": "Énfasis en el establecimiento de métricas claras para evaluar el impacto real de la IA en los objetivos empresariales.",
                                            "code": 55
                                        },
                                        {
                                            "Category": "Diseño \"de Derecha a Izquierda\"",
                                            "Summary": "Metodología que comienza con el resultado deseado y luego retrocede para definir la implementación de IA.",
                                            "code": 56
                                        },
                                        {
                                            "Category": "Calibración de Aplicación de IA",
                                            "Summary": "Determinación de dónde y cómo aplicar la IA de manera más efectiva, y dónde abstenerse.",
                                            "code": 57
                                        }
                                    ],
                                    "Summary": "Destaca la importancia de alinear las iniciativas de IA con los objetivos de negocio y medir su impacto.",
                                    "code": 54
                                },
                                {
                                    "Subnet": "Colaboración entre Equipos",
                                    "Categories": [
                                        {
                                            "Category": "Habilitación de Equipos Interdisciplinarios",
                                            "Summary": "Fomentar la interacción entre expertos en IA, negocios y reguladores para una implementación exitosa.",
                                            "code": 59
                                        },
                                        {
                                            "Category": "Entrenamiento y Capacitación en IA",
                                            "Summary": "Programas para educar a los equipos de negocio sobre las capacidades y el potencial de la IA.",
                                            "code": 60
                                        },
                                        {
                                            "Category": "Co-creación con IA",
                                            "Summary": "Uso de la IA como un socio colaborativo para idear y desarrollar soluciones conjuntamente con los equipos.",
                                            "code": 61
                                        }
                                    ],
                                    "Summary": "Subraya la necesidad de una cooperación fluida entre los diferentes departamentos y roles en la empresa para implementar la IA.",
                                    "code": 58
                                },
                                {
                                    "Subnet": "Evaluación y Medición",
                                    "Categories": [
                                        {
                                            "Category": "Métricas de Adopción y Rendimiento",
                                            "Summary": "Establecimiento de indicadores clave para monitorear el progreso y la eficacia de la IA.",
                                            "code": 63
                                        },
                                        {
                                            "Category": "Validación de Valor Empresarial",
                                            "Summary": "Proceso para asegurar que las soluciones de IA generen un valor tangible y medible para el negocio.",
                                            "code": 64
                                        },
                                        {
                                            "Category": "Bucles de Retroalimentación Iterativos",
                                            "Summary": "Sistemas para recopilar información continuamente y mejorar las soluciones de IA a lo largo del tiempo.",
                                            "code": 65
                                        }
                                    ],
                                    "Summary": "Se centra en los métodos para evaluar la efectividad y el éxito de las implementaciones de IA.",
                                    "code": 62
                                }
                            ],
                            "Summary": "Describe las directrices y principios para una adopción de IA exitosa y sostenible en la empresa.",
                            "code": 53
                        }
                    ],
                    "Summary": "Este documento presenta una discusión detallada sobre la adopción de la Inteligencia Artificial (IA) en la industria de servicios financieros, destacando las tendencias actuales, los desafíos y las metodologías para el desarrollo, prototipado y escalado de soluciones de IA. Se enfatiza la necesidad de empoderar a los equipos de negocio en el proceso de diseño de IA, la importancia de los modelos adaptados al dominio y la integración de la IA con los sistemas empresariales existentes. Además, se exploran casos de uso prácticos tanto en operaciones de negocio como en TI, subrayando la importancia de un enfoque estratégico y calibrado para la implementación de la IA."
                }
            },
            "summaryList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 8542,
                        "CandidatesTokenCount": 1193,
                        "TotalTokenCount": 11085,
                        "thoughtsTokenCount": 1350,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Summary": "<h2>Síntesis Integral: Aceleración y Escalado de la IA en Servicios Financieros</h2><h3>1. Introducción y Contexto de la Adopción de la IA en Servicios Financieros</h3><p>La inteligencia artificial (IA) ha entrado en una fase de hiperaceleración, especialmente evidente en el sector de Servicios Financieros, Banca y Seguros (BFSI). Se observa una transición significativa desde los <em>chatbots</em> tradicionales hacia <b>agentes de IA en acción</b>, lo que representa un cambio fundamental en cómo las empresas interactúan con los clientes y gestionan los procesos. Este entorno dinámico presenta tanto oportunidades como desafíos para la adopción.</p><h4>1.1. Tendencias Clave en la Adopción de la IA</h4><ul><li><b>Adopción Multimodal</b>: La capacidad de utilizar múltiples modelos de manera juiciosa para lograr tareas específicas o flujos de proceso es vista como el futuro. Esto implica la integración de diversas capacidades de IA (voz, texto, visión) para soluciones más completas.</li><li><b>Ingeniería de Contexto</b>: La preparación y construcción de un contexto rico antes de interactuar con los modelos de IA es crucial. Dada la proliferación de modelos en el mercado (como Codex 5.3 y Sonnet 4.6, lanzados en cuestión de minutos), la selección y el uso contextual de los modelos son primordiales.</li><li><b>Convergencia Vertical</b>: Se está produciendo una convergencia en el espacio de la IA, que abarca desde la infraestructura de hardware (con grandes inversiones como el centro de datos de IA de 1 gigavatio anunciado por la organización) hasta los modelos (especializados y fundamentales) y las aplicaciones. Abordar el “stack” vertical completo es esencial para un impacto significativo.</li></ul><h3>2. Resultados de la Encuesta y Casos de Uso en India</h3><p>Una encuesta reciente a 170 clientes de banca, mercados de capitales y seguros, tanto a nivel global como en India, reveló patrones clave de adopción.</p><h4>2.1. Hallazgos de la Encuesta</h4><ul><li><b>Alta Tasa de Adopción</b>: Aproximadamente el 90% de los clientes están adoptando la IA de alguna forma, con casi el 60% en un estado avanzado de adopción.</li><li><b>Énfasis en Modelos Específicos de Empresa</b>: El 50% de los clientes considera vital centrarse en modelos específicos de la empresa y modelos afinados, dada la naturaleza regulada y sensible a la seguridad de la industria de servicios financieros.</li><li><b>Medición de Resultados</b>: Una abrumadora mayoría enfatiza la necesidad de medir los resultados empresariales, yendo más allá de las Pruebas de Concepto (POC) y los pilotos para demostrar un impacto comercial real.</li></ul><h4>2.2. Oportunidades y Desafíos en India</h4><p>India presenta un contexto único: un mercado dominado por la voz, con diversas lenguas y necesidades financieras. Para Tata Capital, la inclusión financiera implica estar presente, ser comprendido y accesible para todos. Un ejemplo destacado es la solución multimodal de IA de voz, construida sobre un modelo basado en Sarvam, que atiende las variaciones lingüísticas y permite una interacción multicanal para una experiencia de cliente de extremo a extremo.</p><p><b>Ejemplo Ilustrativo: Asistente Financiero de Tata Capital</b></p><p>Se demostró una interacción en la que un cliente, que inicialmente no comprendía el inglés, fue asistido en su idioma local. El asistente de IA guio al cliente a través de un formulario de información personal en WhatsApp, resolviendo dudas sobre campos específicos (como el apellido de soltera de la madre) y finalmente ofreciendo un préstamo con detalles de EMI, todo de manera fluida y empática. Esta solución ya está impulsando millones de conversaciones con clientes, redefiniendo el compromiso a escala con responsabilidad e impacto real.</p><h3>3. Realidades y Desafíos en la Adopción de IA en la Empresa</h3><p>La experiencia de los últimos 9 a 12 meses revela realidades concretas en la implementación de la IA:</p><h4>3.1. Adopción de IA para Negocios y TI</h4><ul><li><b>IA para TI</b>: Principalmente enfocada en asistencia de codificación y modernización tecnológica, con soluciones que ya están en producción.</li><li><b>IA para Negocios</b>: Abarca desde el servicio al cliente hasta funciones horizontales, funciones corporativas y asesoramiento de inversiones/patrimonio.</li></ul><h4>3.2. Desafíos Persistentes</h4><ul><li><b>ROI y Precisión</b>: Persisten los desafíos para demostrar el Retorno de la Inversión (ROI) y garantizar la precisión de las soluciones de IA.</li><li><b>Cumplimiento y Seguridad</b>: La seguridad y el cumplimiento normativo son aspectos críticos que deben abordarse al construir soluciones de IA, especialmente en un sector tan regulado.</li><li><b>Complejidad de la Integración</b>: Construir una solución de IA es una cosa; hacer que funcione en un entorno empresarial a través de la integración con sistemas existentes (CRM, banca central, aplicaciones heredadas) es un aspecto crítico y a menudo subestimado.</li><li><b>Flujos de Proceso vs. Casos de Uso Individuales</b>: Los casos de uso individuales no proporcionan el beneficio real; los beneficios sustanciales provienen de la optimización de flujos de proceso y flujos de trabajo completos, adoptando una perspectiva "
                            }
                        ]
                    }
                }
            ]
        }
    ],
    "headerCols": {
        "col1": "Part Id",
        "col2": "FileName",
        "col3": "Content",
        "col4": "Status"
    },
    "fileName": "",
    "projectType": "VIDEO",
    "sourceLanguage": "English",
    "targetLanguage": "Spanish",
    "__v": 0
},
{
    "_id": "6994371225c9f4e0d7e7047a",
    "projectId": "679765",
    "projectName": "AI masterclass in Finance - Hindi",
    "projectDescription": "AI masterclass in Financial Service | AI impact Summit 2026",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T09:38:23.788Z",
    "projectUpdatedDate": "2026-02-17T09:38:23.788Z",
    "projectStatus": "Active",
    "companyId": "69453bf2e603a0a401b6f35b",
    "projectCreatedBy": "69453bf2e603a0a401b6f35c",
    "projectUpdatedBy": "69453bf2e603a0a401b6f35c",
    "projectLastAccessDate": "2026-02-17T09:38:23.788Z",
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "Content",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "AI Masterclass in finance.docx",
            "col3": "Quick context setting on what we see from our vantage point as well in financial services in terms of the adoption. What's happening on the ground a lot being talked about it, but it's also fact about making it work in an enterprise in the form. So that's on the first part that will do will then move to the next one, which is about you wanted all of you to experience a bit of especially from a wipe session standpoint more in the area of taking an idea through a proof of value.  More of a vibe design and a vibe build session. So you have a lot of laptops out there. There are a series of instructions. It will be both assisted as well as if some of you wanted to do on your own, more than welcome to do it as well. But we'll assist in the process flow. Get a feel for it. How does it work through, especially the design and the initial prototype build part of it. And then of course we'll wrap it up as a final thing. Let me give a little bit of context setting from our vantage point where we see us.  I think it's all well recognized that AI has really got to a hyper acceleration phase. Maybe two or three key call out that I would do out here. I'm sure most of you are familiar with it, but I'll just make a call out. One is the fact that there is a genuine movement from the traditional chat bot kind of an interface to more of agents in action. It's really happening out there in the market, in the industry as well, and BFSI is definitely in leading that as well. There are multiple...  I would call it as S waves that are happening right now. Multiple interventions happening at the same point. So that kind of giving you a bit of challenge as well as an opportunity. Exciting, but at the same time, it's also giving a lot of challenge in terms of the adoption standpoint. We see a significant focus on multimodal adoption. And believe that's the future in which it will shape up. You would judiciously use multiple models to accomplish a particular task or a process flow. And the most important thing is that  End of the day, it's all about context engineering. You go to build the context before you engage with the model. So three callouts, because there's a lot of flurry of activity happening. As you could see, a series of models that are coming into the market, competing ones, quality is improving. For instance, think about these two. The 5.3 Codex and the Sonnet 4.6 were launched in a space of 20 minutes. And you could expect what would happen down the line as well.  So with this flurry of activity, it's important that we keep those three callouts in mind because we're going to judicially use which models would factor for and what context in an enterprise as well. While that happens, this flurry of activity happening, what we believe is that there is also convergence happening in this space. There is a vertical convergence happening. Right from so far, the focus has not been on models and applications. It's all about hardware and infrastructure as well.  could see a lot of investments being called out in this space globally within India and from our organization standpoint, also in the group, we have announced the launch of an AI data center, a one gigawatt data center over the period of next five years as well. But the point is that there is a vertical alignment that's happening right through infrastructure to the models, both specialized and foundation models, and to the applications. It's very important to think so that it has got off to a  inflection point that cannot be seen in isolation. It has to be seen in terms of the entire vertical stack to make it more impactful. We did a survey among customers both globally and in India. It was 170 of our customers we did a survey and see what's adoption and how is it shaping up. It was a very recent survey that we did. The multiple things that came out, one good news is all about that. Close to 90 odd percentage of the customers are in some form or shape. And this cut across banking,  capital markets and insurance customers, as I said, both India and globally, are in some form or shape adopting it with almost 60% fairly advanced state of the adoption standpoint as well. I just want to make one specific call-out here. It'll be interesting to many of you as well. 50% of the customer base feels that a lot of focus needs to be given on enterprise-specific models and some fine-tuned models as well. This is something which is very important given the regulated nature of the financial service industry.  how much of secure and how much of accurate models are also needed from an enterprise standpoint. Then of course, a majority, overwhelming majority also felt there is a good need to measure the outcome. Metrics are needed. It's one to talk about POC and pilots, but another to talk about the  business outcome. In India, I think it's also interesting to see India, the opportunity and challenges, the nuances around the language. India, I would call it, is a very voice-led  India is a land of many voices, spoken in different languages,  across diverse geographies and, most importantly, with unique financial needs. At Tata Capital, financial inclusion means more than just access to finance. It means being present, being understood and being reachable to everyone, everywhere. Today, we are proud to showcase a multi-modal voice AI solution supporting our customers across every key moment of their financial journey. Truly,  end to end.  Okay, so first of all, are you salaried or self-employed? I don't understand, madam. Excuse me, I mean, do you work somewhere or do you have a business of your own? I have my own shop. Okay, so you have your own shop. Can you tell me what your business is about? Madam, I have a rental shop. Okay, so you have a grocery shop. So instead of your shop, do you have your own shop or do you have it on rent? My rental shop is on rent. Okay, so your shop is on rent?  I have sent you a personal information form on WhatsApp which requires some basic details. Fill it in. If you have any doubts, you can ask me. Yes, I have got the form. I will fill it in, madam. It won't take much time, right? No, it won't take much time. Fill it in slowly. I am here. If you have any questions, you can ask me. Madam, what do I have to put in the mother's maiden name? Okay, you are asking about the mother's maiden name. This means the name of your mother before marriage.  Do you have a call? If you need anything, let me know. I will help you. Yes, I am filling the form. Madam, I am in the middle. I am filling the form. Okay, no problem. You can fill the form. I am here. Yes, Madam. I have filled the form. Very good. Thank you for filling the form. So, let me tell you again that you will get a loan of 5 lakhs for 3 years at an interest rate of 12.5%. The EMI of which will be 16,727.  What you just saw. Well, that is the future. The future of conversations, of customer engagement and of AI led transformation. This isn't just a concept. It is already powering millions of customer conversations. This is how Tata Capital is reimagining engagement. At scale.  with responsibility, with empathy, and with real impact. Because when trust finally finds a voice, Bharat engages. And when Bharat engages, Bharat progresses.  India-based model based on Sarvam. So I think a lot of linguistic variation that is addressed, and I think this is important because on one side we have the global models available, but it's also important to address the local nuances out there. And the second thing is it's also now allowing from a multi-channel standpoint. So I think the ability to come together more from a customer servicing, I think that has taken on a different league altogether. While that happens, let me call you out some key pointers from.  What we have been experiencing over the last, I would say, nine to 12 months, some very grounded reality has been popped up here. From our vantage, we were looking at the adoption in terms of AI for business and AI for IT. IT is pretty straightforward in terms of, it's mostly around a lot of focus on coding assistance and plus the fact that you do a lot using modernization. Huge focus in many forms in terms of technology modernization, the ability of AI to accelerate it. These are not POCs and pilots really getting into production.  Then the business side of it, if you look at it, it's got a broad stick happening out there. Right from customer servicing to horizontal functions, corporate functions, and also into wealth and investment advisory as well. So there is a significant progress happening in this space, but just also wanted to call out some of the ground reality that are popping up. ROI challenges still persist. There is still challenges in terms of the accuracy, the.  compliance aspect of it, that's still an important thing to be addressed and factored in building an AI solution. The other part is that building an AI solution is one thing. Making it work in an enterprise in terms of integration is so critical. A lot of ROs have been built around building solution, but not necessarily the effort that you need to make it work. Connecting to your enterprise systems, connecting to CRM, your core banking, your learning application, that's a critical aspect of this whole process flow.  I think it's fairly recognized that individual use case are not giving you the benefit, the real benefit come by process flows and by workflows. And that you're going to look at workflows differently, not to the traditional left to right, but more from right to left in terms of an outcome standpoint. We see and we believe fine tuning is something that financial services would pick up in a significant way. If you're already seeing that, that would really do an uptake in the next six to nine months.  Definitely see that picking up in a big way. Why? Because of the need for much more accurate, much more precise models that need to happen. To say that it will be a hybrid scenario, you would have fine-tuned models on one hand, you'll also have large language models, you would judiciously use it. There is definitely a shift towards Hintiq. Having said that, these are not Hintiq as in high level of autonomy, but a slow.  a pedigree of autonomy is being enabled in it, but agentic workflows are happening. The bottom point, I think we refer to that the vertical stack that's evolving, very important to look at an infrastructure in the platform-based thought process. Again, the whole point is that I can do a great POC pilot shiny object, but how do I scale it across and you need a platform engineered thought process to make it happen. So these are some of the experiences that we're going to picked up, thought I can relate that all of you as well. Just wanted to.  give a quick perspective around the fine tuning that we talked about. This is something which piggyback on the fact that it will be in hybrid AI solutions and models that you would end up with. You would have both the scenarios. You will continue to leverage large language model for broad, measurable enterprise task and activities with proper grounding. I think it's very important I just double down on the fact that it has to be engineered.  How well do you orchestrate your workflows? How well do you build your evals around it? It's very critical. But I think this is an equally important. How well do you build your domain-adapted models in compliance area, in advisory, in credit research? I think it's important to see how well they build appropriate and high accuracy. So our vantage point, we believe, it's going to be convergence between the two, not one over the other because each has its own space. It's very important to traditionally use this. I'll just give you a couple of quick.  and adults around it, these are real life examples. Something that we have done for litigation prediction in a claims value chain. If you look at it, what a base model provided versus what a fine tune model provides. And this is not something that happens straight away, you need to iteratively build it, but it's very important to continue, make sure that right from base model, the different techniques of fine tuning to make it more domain adapted in that context. And you could see the high percentage of  accuracy as you build that incrementally. And this is where business gets a lot glued to it saying that, hey, now it's come to a point. It's not just technical. It's also important to make it work. Same thing in compliance. And we have used two different models just to give a variation. There are many models out there. But these are models that you could enable it. And whether it's Lama or Mistral Family, there are many other models. For instance, as I said, for several models, we've been working on it in terms of voice, AI standpoint. But the key message I want to leave out here is that.  fine-tuning or making it proprietary, it's not just technology. I think it's all about building this domain curation. How well do you build the instruction set for a fine-tuning? That need a lot of domain understanding. Once you build that, the initial set, then I think it's a lot about applying the right techniques. It's also equally important to look at the right models. Your slew of models out there,  and there is a method to the madness in terms of which model would fit into what context. So all the three are very important. Building the right instruction sets for fine-tuning. For all of you who I'm sure fine-tuning is something that all of you are aware of it, but essentially what you're trying to do is that making sure a broader model has been fine-tuned towards specific domain and adaptation so that it get much more relevant in that context and accuracy of the decisions it takes us to higher level.  a series of business scenarios that are evolving in this area. We see this in India, we see this globally. So this is something from our vantage, we see a strong uptick that's happening. So just wanted to give a quick flavor of that before we get into the white thought process. I think it's important to look at it from the context engineering, very important. Enterprise adoption is not just about building use cases, but also making it work in the broad end.  it's very critical to make sure that you judicious use of large language models and domain adapted models. With that, let me just show you one specific scenario. Good afternoon. Thank you for calling. Do I have the pleasure of speaking with you? Hi, Catherine. This is Daniel Thompson. I need to report an accident that happened earlier today morning at 4.30 AM.  I am sorry to hear. Could I please have your policy number? Yes, it is ALC 76-12-39-05. Thanks, Daniel. I thought that's your policy. Everything looks active. Could you describe what happened? I was already on the highway. I was driving southbound on the M1 Pacific motorway past the Gosford exit in New South Wales, maintaining my lane and speed when a silver SUV suddenly emerged from the left.  onto the motorway and site wiped my vehicle so the impact was on the left side yes quite severely it hit around the front left door and fender area i had no chance to react the other driver didn't yield or check for traffic understood um with a police call to the scene  Yes, our patrol unit came by. They've noted the details. I have the event number. I also took several photos of the damage. Great. If you can send us the event number, police notes and any photos that will help us proceed. We'll also need the repair estimate or tow docket if you've received one. I will email all the documents shortly. Perfect.  Drive safe and let us know if you need anything else. All right. As you could relate, I think the accent of this is for a UK-based customer. And the context is a little different from the first one that you saw from a voice here, which was a very specific scenario where, in the Tata Capital's story, it was AI at the thick and center of everything in terms of engaging in a customer service. But not actions are not getting taken. It's more of a conversational flow.  This one is scenario in which a multi-model with a fine-tune model being brought into the picture. So if you could see the focus of the customer service rep is more in terms of looking at the information that can be gleaned around it. So this is an augmented AI. AI was listening to the whole conversation, providing from an augmented sample what additional information. Normally, this whole process called FN oil, the first notice of loss from the insurance background would know.  When you meet with an accident, the first notice of loss that you make a call to the back office to inform. A lot of information you gather at that point. It's very important to capture that information. Today, it's a very back and forth happens, and your claims get delayed because of a simple reason, you haven't captured enough information. In this scenario, behind the scenes, so AI was not in the front-end center, behind the scenes capturing the information, and there was a litigation validator, which I was telling the fine-tune model.  looking at it, is there any fraud in this process so that there is a subsequent litigation possible down the line? That information, I don't know how many of you have watched that, that information was gleaned through in between for what I say, this looks like a good call, let's proceed. So I think that the modality is a little different out there, but the whole point is that it's very important to look from both the scenarios. So to put it in a nutshell, I think it's important from an enterprise form, I think proper calibration is very critical.  great technology, great capabilities evolving, but it's also important, how do you enable that in a context? Where do you not apply AI also? I think it's important to also look where you don't apply AI. Because at times, we have the thought process of applying AI all over the place. I think it's very important to break it and then make sure that the ones, and that's what our philosophy is that, don't look at from a traditional process flow from left to right, which you break it down to series of task and say where I can apply it.  Look at from a right to left and see outcome. What is outcome and how do I reapply back? So what are you going to do now for the next 60 minutes or so? What I thought is that, I believe that this need to be broken into three broad areas. You got to have a quick proof of value. You need to build that proof of value because there are certain things you will just move forward then. You say, okay, this is not working. Let me try the next one. You have the ability right now to do a quick prototyping, a quick vibe design and prototyping.  Then you go into a rapid build over a period of time, five to six weeks, and then you really scale it. This is a model that we would advocate as enterprises start picking up more and more. Because, as I said, it's very important to look at outcome backwards. It's very important to look for what is the business assurance that I bring in. So we have constructed the next 60 minutes or so in that format. We have certain challenge cards on your table. You have laptops out there. We're going to do this in two formats. We'll assist you in the process.  in how do I go through the process, and second is also in terms of those who are more wanting to do on their own, they feel free to try it out. Instructions are also given out there. Let me invite Kavita now to run you through the vibe coding design and prototyping session. Thank you. To continue on the topic of scaling, in financial services, AI doesn't scale when the model is ready. It's when legal compliance...  regulatory, all of the body comes together. When they are ready, it's when AI scales. So I think that's one of the challenges that we face when we look at a scaling AI. This is just one of them. And my name is Kavita, and I lead the AI labs for financial services. And I help customers on their journey to adopt AI and scale AI. And one of the other challenges that really come across  when we start discussing about AI use cases is, how do I get the business teams on board? Because if you look at the kind of implementations we have seen so far in AI, you've got knowledge assistance, you've got chatbots, you've got good document AI capability. And if you look at these implementations, they have actually filled the gap where the traditional AI, traditional IT has failed in the sense...  inability to access unstructured data. There was a gap which AI solved. So most of the use cases have been around these lines, and that's primarily because AI has not been, we have not let the business teams actually drive AI. So what does it take to help enable the business teams to understand the power of AI and look at more transformative use cases?  So that's the construct of what we want to do today. We want you to experience that how a business user or a business team can be enabled with AI to accelerate that whole journey of identifying the right opportunity, doing a working prototype, and actually even testing it out and seeing the proof of value. How do we shift left?  in the sense that how do we give business more control and working with the engineering team, build out a quick proof of value that you can validate and then scale to production. Because in one of the earlier conversations I had with one of you here was that scaling is a challenge. But sometimes you don't get to scale because you've not identified the right opportunity. You try to apply AI everywhere. So how do you use AI to help you in that journey?  And what we want to achieve through this is let business lead the whole AI redesign. So how do we do that? And that's something that we will cover today. And what we're going to demonstrate to you is not a tool, but how we can bring all the AI capability together for a business audience.  Yeah, switching on to that. So what we'll see today, disclaimer before I start, this tool is not a single tool. It's a workbench that we have created that brings together different AI capabilities. And then what we do is that we start off with something called a business challenge, because that's the starting point for a business team. Where do I begin my journey? What is the challenge that I need to address? And for the sake of this demonstration, I'll pick up one use case, something that's  A challenge that has come up in one of our customer conversations was that you have a lot of portfolio managers who need to, at the end of the quarter, generate a summary of how their portfolio has performed. And along with collecting a lot of data, they need to actually  check out the sentiment of the stock on the public domain. So there's a lot of data gathering, a lot of analysis that comes in, and as a result, there's a lot of time that goes into that activity, and that delays the overall report, and sometimes this gets delegated to a junior analyst, the quality of the report is affected. So how do I bring all of that in a way that AI can help the portfolio managers solve the problem? So that was the challenge that we picked up, and I'll show you a quick demonstration of  how the challenge works. The challenge was just creating an assistant to help the AI agent, to help the portfolio manager actually look up what is the right, what is the data that's available out there on the public domain, or even look up Yahoo Finance to see how the stock has performed. So all of that was being done by AI in the background. And this is a solution that...  or when we piloted with our customers, they saw a huge advantage, because today, they were actually outsourcing a lot of this activity to a third party organization, and they were able to bring that back, and bring down the overall time to generate the report, and even the quality of the report started coming out much better. I thought I'd give you a view of the final outcome, and take you back to where we started. And that's the platform that we were trying to show you. So switching on.  to that platform. So what we do as a first step is actually pick up the challenge and how do we use AI to ideate and come up with something called the opportunity canvas. So many of our users have, you know, if you see business users are all equipped with co-pilots because it's already available in the enterprise. But what we've observed is that just not giving them AI tools or training them on prompts is really going to help. It's about  helping them to understand that end-to-end process and letting them have complete control. So that's what we've done over here. So we looked at what could be a set of questions that a business user needs to look at when they see a challenge and convert it to an opportunity. So we've curated a set of prompts which can go through the data and come up with what we call the opportunity canvas. So we'll just walk you through those different steps. So I think one thing that we always need to keep in mind is that  When we give AI to the business users, the research is primarily on, you can use a lot of research tools available, but the organization context is more important. How do we know what the user is? What are the different users going through? What are their challenges and pain points? So loading the context is the first step.  And the reason I'm walking you through this is because once I finish the demo, I want you also to try and just see whether as a business leader or a practitioner how the whole experience will be and how can we really build that capability for business teams to use. So we'll go through this flow. So we have built the context. And then after that, you can even go back and do a deep research. Because sometimes your context is not enough. You want to know how your peers are performing.  pull out a deep research from Gemini or ChachiPT. So put that context also in, and then you have an overall knowledge base that your business team can now use to start ideating. So what we have done is created an ideation framework here. Now, as you go through the ideation process, you'll have a lot of questions. Have I captured all the pain points right? Because you want to be sure that you have actually  solving a problem. So there's pain points that you capture. And there are quite a lot of steps. I won't go through all of them. But if you quickly go through, you will have pain points. You'll want to assess what the business impact is. What's the possible solution direction that you want to take? And as we go through these steps, you will see that it's not enough just to analyze the data that's available.  but to actually start seeing AI as a thinking partner. That's what we want the business to do, be able to bounce ideas, validate them, feed the input, and finally, at the end of it, create what we call an opportunity canvas. So a lot of thought has gone in. So while these demonstrations, while he brings up the opportunity canvas, something that we have observed is that just giving a tool may not be the only solution.  What we do in TC, something that's worked for us, is that we conduct an AI day for business teams, where we try to create that immersive experience where they understand what are the right AI tools and also how to use them and how to apply them in the context. So I'll, if you can just go through the, yeah. So there's a lot of questions, so we don't, we, and it will take a minute to go through, but if our demo is ready, you can show that as well.  Yeah, so this is what the final product looked like. So you give a company name, and it went through the organization's research. It also went to the public domain and started picking up all the stock information and brought that up into an overall summary that can be then used by the portfolio manager to make a decision on how his stock has performed, how do I report that?  to his customers? How do I report that into the performance reporting? And if you see, it was instant. And it's gone through, I think, 15 to 20 sites. We've configured all the financial sites to go through and capture the sentiment, capture key financial metrics, all of that instantly on a very high-speed internet. Now, I think we would be done with the Opportunity Canvas. Last one step.  So it's important that we go through these steps. So that's why we have consciously made these steps so that you don't pick an idea which is not the right one. So identifying the opportunity, what is the possible solution direction, what's the impact, and what your roadmap will be. And there you have an opportunity canvas, which the user has co-created along with AI. Now you go on to the next step. And this is important.  All this is fine. The business team could be comfortable in Opportunity Canvas. But can now the business team, with the help of AI, drive the overall design process? So what you'll see is that you can brainstorm with AI and actually give a direction of what that design should be. And this is what AI has generated. It's kind of broken down that whole opportunity into a set of  different agents that can go in. Because as we covered in the previous session, a lot of attention focuses on moving to agentic. So can we look at building what that agentic solution would look like is what you're going to see here. It's not done the coding yet, but what it has done is broken up that entire opportunity into a series of agents which can address this need. And we'll just.  You can try that. You will get a chance to try this, so I won't go into the detail. But here, it's come up with four or five agents. It's organized. And as a business team, they can actually go and say, I want these kind of checks and balances. I want to do it this way. And now you see the difference, right? Earlier, it was all left to the engineering team to actually build the solution. But now, we are passing the power back to the business to drive all of this. Now, going forward,  What is the next important step is to validate what this opportunity looks like from an experience. So you use a simple wipe coding tool. We have here used both Lovable and Google AI Studio. This tool generates the prompt. You copy the prompt. And then you actually can see it through a visual experience.  simple as putting it and if you see the prompt it's all English and you know when you get into your wipe session you'll see that you can actually edit some of it and see what that experience looks like so this will take how much time maybe a minute or so do you have something do you have something that you've already pre-built yeah so this is something that that came out of the tool a very very nice experience which shows what the portfolio summary should look like what are the stocks etc so  kind of understood the opportunity and translated that. And you can play around with it and tweak it to the experience. This step becomes very important because what earlier took two to three weeks of building a UX prototype as the end of a design phase actually got converted to minutes. And this is something you take to these stakeholders, validate it, and you're ready with what is that real challenge you want to address, what's the real opportunity comes out so quickly.  We can't stop with just a UX prototype, because that just gives you an experience. You need to actually engineer some of the back-end capability. And there are low-core platforms that are available. So in today's live session, we will show you that as well. And one of the examples that we have shown here is a solution called Ne10. It's a low-core platform building. It's not that we are saying this is the right tool. As per your enterprise decision, you could have different options.  But this really helps, in our view, to ideate and understand or discuss with your engineering teams how your design should be. You can easily visualize what are the models being used here. Have I thought through all the guardrails? Have I built my business evaluations? All of this can be verified in one shot. And then from this, the code can get generated, and you'll have a working end-to-end solution.  So this is the process we followed for quickly prototyping. And the demo that you saw was an example of something that was built using this approach. So what we'd like to do now is shift back to you and see if you'd like to try this. I mean, some of you could. We have laptops configured with.  with user IDs, we have all the tools available. It's an opportunity for you to try out and we have mentors. We have mentors available for each table. A lot of our teams here to support you and we would really encourage you to try and we can guide you here as you go along. Explore, yeah. Okay, so let me just walk you through what you would do. So all of your laptops will have a handbook already preloaded.  which talks about the use case that you would be picking up. Okay, so, and that is in green, it's tagged as challenge. So that's the use case that you will start with. And you have different phases and you have resources for each of the phase. So feel free to go through this handbook, take it step by step, and we'll help you with it. You are workspace five.  So log into your workspace based on the number that's on your table. So when you start, it will be the discovery and design. Yeah, you can start the solution with the resource present there. Okay, so just a quick time check. We just have five minutes. So if you want some help on any particular section, just raise your hand. We'll come and help. But I'm so happy to see many of you have generated very nice UX prototypes.  And you can also move into that build phase. That's an optional one, but try it out. We'll have mentors to support you.  uh...  So I think we'll just conclude.  So, so far we spoke about, I only covered, I'll hand it over to him, but what you did essentially comes as step one. You actually did, and I'm so happy that many of you got challenge completed, and you saw your code getting converted, and I see a lot of happy faces because it was, I heard words like, amazing, I can do the same thing with my tools, so very happy to hear all that. So what we did,  was the first step, proof of value. We've tried to establish this is something, a problem worth solving for my organization and you may have an answer then, you have an answer at the end of the stage. But the journey to take it to production involves something called rapid building. What you've built is a foundation which you can add on more functionality and then create it into a full-fledged solution. And that's where we call the rapid build.  which is about engineering the full-fledged solution before you go to production. And then finally, the scale. And when you look at scale, and someone asked me that question, you have to build a lot of AI foundation capability, the control plane, a lot of things need to go in. And that's important if you want to bring that scale. I'll just hand it over to Onni to close. I think thank you. Thank you for being a real sport out there. I think all of you joining in, being through the wipe session as well. Great to see all of you continue out there.  Once again, thanks for being participating. I think this is one important thing that we believe AI would make a difference in the ability to quickly translate an idea to prototype. Not to say that that prototype is what you would eventually scale to production, but that gives you the feel of what can be done. And then you naturally scale it into production in a very calibrated way. But again, thanks for your time. And then being part of it, feel free to reach out. Some of these I'm assuming.  would have been captured in your email, et cetera. So it's available for you to do some prompting and wiping back home as well if you're adventurous and wanting to extend the focus. Thank you once again and have a good rest of the day. ",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "projectDeleted": false,
    "codeFrameColumn": [
        {
            "col1": "col1",
            "col2": "col3",
            "researchObjective": "",
            "filterColumn": [],
            "netSetting": true,
            "subNetSetting": true,
            "categorySetting": true,
            "sentimentSetting": true,
            "thinkingBudgetSetting": 2000,
            "questionType": "OPENEND",
            "otherBrand": false,
            "existingCodeFrame": [],
            "isList": false,
            "codeFrameWithDataList": [],
            "codeFrameList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 8883,
                        "CandidatesTokenCount": 2619,
                        "TotalTokenCount": 12887,
                        "thoughtsTokenCount": 1385,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Net": "AI का वित्तीय सेवाओं में अपनाना और रुझान",
                                "Subnets": [
                                    {
                                        "Subnet": "AI का वर्तमान परिदृश्य",
                                        "Categories": [
                                            {
                                                "Category": "तीव्र विकास और AI का प्रभाव",
                                                "Summary": "AI के तेजी से विकास और इसके उद्योग पर पड़ने वाले बड़े प्रभाव का उल्लेख।"
                                            },
                                            {
                                                "Category": "चैटबॉट से एजेंटिक वर्कफ़्लो तक",
                                                "Summary": "पारंपरिक चैटबॉट्स से अधिक स्वायत्त एजेंट-आधारित AI सिस्टम की ओर बदलाव।"
                                            },
                                            {
                                                "Category": "मल्टीमॉडल अडॉप्शन",
                                                "Summary": "विभिन्न मॉडलों का बुद्धिमानी से उपयोग करके एक कार्य को पूरा करना।"
                                            }
                                        ],
                                        "Summary": "AI का वर्तमान चरण, जिसमें तेजी से विकास, एजेंटिक वर्कफ़्लो और मल्टीमॉडल अडॉप्शन शामिल हैं।"
                                    },
                                    {
                                        "Subnet": "AI के उपयोग के क्षेत्र",
                                        "Categories": [
                                            {
                                                "Category": "BFSI में AI का अग्रणी उपयोग",
                                                "Summary": "बैंकिंग, वित्तीय सेवाओं और बीमा क्षेत्र में AI का अग्रणी उपयोग।"
                                            },
                                            {
                                                "Category": "व्यवसाय और IT के लिए AI",
                                                "Summary": "AI का उपयोग व्यवसाय कार्यों (ग्राहक सेवा, कॉर्पोरेट कार्य) और IT (कोडिंग सहायता, आधुनिकीकरण) दोनों में।"
                                            },
                                            {
                                                "Category": "ऊर्ध्वाधर अभिसरण",
                                                "Summary": "इंफ्रास्ट्रक्चर से मॉडल और एप्लीकेशन तक AI स्टैक का समेकन।"
                                            }
                                        ],
                                        "Summary": "वित्तीय सेवाओं में AI के विभिन्न उपयोग और तकनीकी अभिसरण।"
                                    },
                                    {
                                        "Subnet": "ग्राहक सर्वेक्षण और AI स्वीकृति",
                                        "Categories": [
                                            {
                                                "Category": "AI अपनाने की उच्च दर",
                                                "Summary": "सर्वेक्षण में 90% ग्राहकों द्वारा AI को अपनाना, 60% उन्नत स्थिति में।"
                                            },
                                            {
                                                "Category": "एंटरप्राइज़-विशिष्ट मॉडल की आवश्यकता",
                                                "Summary": "50% ग्राहकों द्वारा एंटरप्राइज़-विशिष्ट और फ़ाइन-ट्यून किए गए मॉडलों पर ध्यान केंद्रित करने की मांग।"
                                            },
                                            {
                                                "Category": "परिणामों को मापने की आवश्यकता",
                                                "Summary": "अधिकांश ग्राहकों द्वारा AI समाधानों के व्यावसायिक परिणामों को मापने के महत्व पर जोर।"
                                            }
                                        ],
                                        "Summary": "ग्राहकों के बीच AI अपनाने की स्थिति और एंटरप्राइज़-विशिष्ट मॉडलों व परिणाम मापन की आवश्यकता।"
                                    }
                                ],
                                "Summary": "वित्तीय सेवाओं में AI को अपनाने की वर्तमान स्थिति, मुख्य रुझान और ग्राहकों के दृष्टिकोण।"
                            },
                            {
                                "Net": "AI कार्यान्वयन में चुनौतियाँ",
                                "Subnets": [
                                    {
                                        "Subnet": "व्यावसायिक और तकनीकी बाधाएँ",
                                        "Categories": [
                                            {
                                                "Category": "ROI (निवेश पर लाभ) की कमी",
                                                "Summary": "AI समाधानों के लिए निवेश पर लाभ प्राप्त करने में चुनौतियाँ।"
                                            },
                                            {
                                                "Category": "सटीकता और अनुपालन के मुद्दे",
                                                "Summary": "AI मॉडल की सटीकता और नियामक अनुपालन सुनिश्चित करने में कठिनाइयाँ।"
                                            },
                                            {
                                                "Category": "एंटरप्राइज़ एकीकरण",
                                                "Summary": "AI समाधानों को मौजूदा एंटरप्राइज़ सिस्टम (CRM, कोर बैंकिंग) के साथ एकीकृत करने में चुनौतियाँ।"
                                            }
                                        ],
                                        "Summary": "AI समाधानों को व्यावसायिक रूप से व्यवहार्य और तकनीकी रूप से एकीकृत करने में आने वाली मुख्य बाधाएँ।"
                                    },
                                    {
                                        "Subnet": "स्केलिंग और कार्यान्वयन चुनौतियाँ",
                                        "Categories": [
                                            {
                                                "Category": "व्यक्तिगत उपयोग के मामलों की सीमाएँ",
                                                "Summary": "व्यक्तिगत AI उपयोग के मामलों के बजाय प्रक्रिया प्रवाह और वर्कफ़्लो से वास्तविक लाभ।"
                                            },
                                            {
                                                "Category": "कानूनी, नियामक और अनुपालन की भूमिका",
                                                "Summary": "AI को बड़े पैमाने पर लागू करने के लिए कानूनी, अनुपालन और नियामक निकायों की तैयारी महत्वपूर्ण है।"
                                            },
                                            {
                                                "Category": "बिजनेस टीमों को शामिल करना",
                                                "Summary": "बिजनेस टीमों को AI अपनाने और परिवर्तनकारी उपयोग के मामलों को चलाने के लिए सशक्त बनाने की चुनौती।"
                                            }
                                        ],
                                        "Summary": "AI समाधानों को बड़े पैमाने पर लागू करने और व्यावसायिक टीमों को प्रक्रिया में शामिल करने से संबंधित चुनौतियाँ।"
                                    }
                                ],
                                "Summary": "AI समाधानों को लागू करने, बड़े पैमाने पर करने और व्यावसायिक परिणामों को प्राप्त करने में आने वाली प्रमुख बाधाएँ।"
                            },
                            {
                                "Net": "AI समाधान विकास और प्रोटोटाइप",
                                "Subnets": [
                                    {
                                        "Subnet": "मॉडल्स का उपयोग",
                                        "Categories": [
                                            {
                                                "Category": "फ़ाइन-ट्यून किए गए बनाम बड़े भाषा मॉडल",
                                                "Summary": "एंटरप्राइज़ कार्यों के लिए बड़े भाषा मॉडल और सटीक, डोमेन-अनुकूलित मॉडलों का विवेकपूर्ण उपयोग।"
                                            },
                                            {
                                                "Category": "हाइब्रिड AI समाधान",
                                                "Summary": "डोमेन-अनुकूलित और LLM का एक साथ उपयोग करने वाले हाइब्रिड AI आर्किटेक्चर।"
                                            },
                                            {
                                                "Category": "डोमेन क्यूरेशन और इंस्ट्रक्शन सेट",
                                                "Summary": "फ़ाइन-ट्यूनिंग के लिए सही डोमेन समझ और इंस्ट्रक्शन सेट बनाने का महत्व।"
                                            }
                                        ],
                                        "Summary": "विभिन्न प्रकार के AI मॉडलों का रणनीतिक उपयोग और फ़ाइन-ट्यूनिंग की भूमिका।"
                                    },
                                    {
                                        "Subnet": "विकास कार्यप्रवाह",
                                        "Categories": [
                                            {
                                                "Category": "विचार से प्रोटोटाइप तक",
                                                "Summary": "एक विचार को जल्दी से प्रोटोटाइप में बदलने की AI की क्षमता।"
                                            },
                                            {
                                                "Category": "प्रक्रिया प्रवाह और वर्कफ़्लो पर ध्यान",
                                                "Summary": "वास्तविक लाभ के लिए व्यक्तिगत उपयोग के मामलों के बजाय संपूर्ण प्रक्रिया प्रवाह पर ध्यान देना।"
                                            },
                                            {
                                                "Category": "प्लेटफ़ॉर्म-आधारित दृष्टिकोण",
                                                "Summary": "AI समाधानों को बड़े पैमाने पर लागू करने के लिए एक प्लेटफ़ॉर्म-इंजीनियर्ड सोच प्रक्रिया की आवश्यकता।"
                                            }
                                        ],
                                        "Summary": "विचार से उत्पादन तक AI समाधान विकसित करने के लिए अनुशंसित कार्यप्रवाह और दृष्टिकोण।"
                                    }
                                ],
                                "Summary": "AI समाधानों को विकसित करने के लिए मॉडल चयन, फ़ाइन-ट्यूनिंग और एक संरचित विकास कार्यप्रवाह का महत्व।"
                            },
                            {
                                "Net": "व्यावसायिक टीमों को AI से सशक्त बनाना",
                                "Subnets": [
                                    {
                                        "Subnet": "\"वाइब\" डिज़ाइन और प्रोटोटाइप",
                                        "Categories": [
                                            {
                                                "Category": "AI-संचालित अवसर कैनवास",
                                                "Summary": "AI का उपयोग करके व्यावसायिक चुनौतियों को अवसरों में बदलने और 'अवसर कैनवास' बनाने के लिए।"
                                            },
                                            {
                                                "Category": "बिजनेस-नेतृत्व वाले AI रीडिज़ाइन",
                                                "Summary": "AI के साथ काम करके व्यावसायिक टीमों को समाधानों के डिज़ाइन प्रक्रिया का नेतृत्व करने में सशक्त बनाना।"
                                            },
                                            {
                                                "Category": "UX प्रोटोटाइप का त्वरित निर्माण",
                                                "Summary": "डिजाइन चरण में UX प्रोटोटाइप के निर्माण में लगने वाले समय को मिनटों तक कम करना।"
                                            }
                                        ],
                                        "Summary": "एक AI वर्कबेंच का उपयोग करके व्यावसायिक उपयोगकर्ताओं को AI समाधानों को डिज़ाइन और प्रोटोटाइप करने में सक्षम बनाना।"
                                    },
                                    {
                                        "Subnet": "व्यावहारिक कार्यशाला अनुभव",
                                        "Categories": [
                                            {
                                                "Category": "हैंड्स-ऑन वाइब सेशन",
                                                "Summary": "AI समाधानों के डिज़ाइन और प्रोटोटाइप के लिए एक व्यावहारिक, निर्देशित सत्र।"
                                            },
                                            {
                                                "Category": "एआई वर्कबेंच का उपयोग",
                                                "Summary": "AI क्षमताओं को एक साथ लाने वाला एक उपकरण जो व्यावसायिक उपयोगकर्ताओं को विचार-मंथन, डिजाइन और मान्य करने में मदद करता है।"
                                            },
                                            {
                                                "Category": "लो-कोड प्लेटफ़ॉर्म एकीकरण",
                                                "Summary": "UX प्रोटोटाइप के बाद बैक-एंड क्षमता के लिए लो-कोड प्लेटफ़ॉर्म का उपयोग।"
                                            }
                                        ],
                                        "Summary": "व्यवसाय उपयोगकर्ताओं को AI की क्षमता का अनुभव कराने के लिए डिज़ाइन किए गए व्यावहारिक सत्र और उपकरणों का प्रदर्शन।"
                                    }
                                ],
                                "Summary": "व्यावसायिक उपयोगकर्ताओं को AI समाधानों को पहचानने, डिज़ाइन करने, प्रोटोटाइप करने और मान्य करने के लिए सशक्त बनाने पर ध्यान केंद्रित करना।"
                            },
                            {
                                "Net": "विशिष्ट AI उपयोग के मामले",
                                "Subnets": [
                                    {
                                        "Subnet": "भारतीय संदर्भ में AI",
                                        "Categories": [
                                            {
                                                "Category": "बहुभाषी वॉइस AI समाधान (टाटा कैपिटल)",
                                                "Summary": "भारत की भाषाई विविधता को संबोधित करने वाला टाटा कैपिटल का मल्टी-मॉडल वॉइस AI समाधान।"
                                            },
                                            {
                                                "Category": "वित्तीय समावेशन",
                                                "Summary": "टाटा कैपिटल का AI का उपयोग करके वित्तीय सेवाओं को सभी तक पहुँचाना।"
                                            }
                                        ],
                                        "Summary": "भारत की अद्वितीय भाषाई और वित्तीय आवश्यकताओं के लिए AI के अनुकूलन का प्रदर्शन।"
                                    },
                                    {
                                        "Subnet": "दावों के प्रसंस्करण में AI (UK ग्राहक)",
                                        "Categories": [
                                            {
                                                "Category": "दावा रिपोर्टिंग और जानकारी कैप्चर",
                                                "Summary": "एक यूके ग्राहक के लिए दुर्घटना दावे की रिपोर्टिंग में AI का उपयोग करके जानकारी को कुशलता से कैप्चर करना।"
                                            },
                                            {
                                                "Category": "संवर्धित AI (Augmented AI)",
                                                "Summary": "AI का उपयोग बातचीत के दौरान सहायक जानकारी प्रदान करने और दावे प्रतिनिधि को सहायता करने के लिए।"
                                            },
                                            {
                                                "Category": "मुकदमेबाजी की भविष्यवाणी और धोखाधड़ी का पता लगाना",
                                                "Summary": "फ़ाइन-ट्यून किए गए मॉडल का उपयोग करके धोखाधड़ी या मुकदमेबाजी की संभावना का पता लगाना।"
                                            }
                                        ],
                                        "Summary": "बीमा दावों के प्रसंस्करण में AI के उपयोग का एक उदाहरण, जिसमें संवर्धित AI और धोखाधड़ी का पता लगाना शामिल है।"
                                    },
                                    {
                                        "Subnet": "पोर्टफोलियो सारांश सहायक",
                                        "Categories": [
                                            {
                                                "Category": "पोर्टफोलियो प्रदर्शन सारांश",
                                                "Summary": "पोर्टफोलियो प्रबंधकों के लिए स्टॉक प्रदर्शन और भावना का त्वरित सारांश उत्पन्न करने के लिए AI का उपयोग।"
                                            },
                                            {
                                                "Category": "डेटा संग्रह और विश्लेषण का स्वचालन",
                                                "Summary": "सार्वजनिक और वित्तीय डोमेन से डेटा एकत्र करने और विश्लेषण करने के कार्य को स्वचालित करना।"
                                            },
                                            {
                                                "Category": "रिपोर्ट की गुणवत्ता और समय में सुधार",
                                                "Summary": "आउटसोर्सिंग को कम करके और रिपोर्ट बनाने के समय और गुणवत्ता में सुधार करके।"
                                            }
                                        ],
                                        "Summary": "पोर्टफोलियो प्रबंधकों के लिए AI-संचालित सहायक का प्रदर्शन, जो स्टॉक प्रदर्शन का त्वरित सारांश प्रदान करता है।"
                                    }
                                ],
                                "Summary": "विभिन्न वास्तविक दुनिया के परिदृश्यों में AI के व्यावहारिक अनुप्रयोगों का प्रदर्शन, जिसमें वित्तीय समावेशन और दावा प्रसंस्करण शामिल है।"
                            },
                            {
                                "Net": "996 गैर-विशिष्ट विषय",
                                "Subnets": [
                                    {
                                        "Subnet": "अन्य/अवर्गीकृत",
                                        "Categories": [
                                            {
                                                "Category": "अवर्गीकृत प्रतिक्रियाएँ",
                                                "Summary": "वे प्रतिक्रियाएँ जो प्रदान किए गए कोडफ़्रेम में आसानी से फिट नहीं होती हैं और अधिक विश्लेषण की आवश्यकता हो सकती है।"
                                            }
                                        ],
                                        "Summary": "वे प्रतिक्रियाएँ जिन्हें वर्तमान कोडफ़्रेम के विषयों में आसानी से वर्गीकृत नहीं किया जा सकता है।"
                                    }
                                ],
                                "Summary": "अवर्गीकृत और अप्रासंगिक प्रतिक्रियाओं के लिए श्रेणी।"
                            },
                            {
                                "Net": "997 पता नहीं/कह नहीं सकता",
                                "Subnets": [
                                    {
                                        "Subnet": "अस्पष्ट या अनिर्णायक",
                                        "Categories": [
                                            {
                                                "Category": "अस्पष्ट/सामान्य प्रतिक्रिया",
                                                "Summary": "वे प्रतिक्रियाएँ जो अनिर्णायक, अस्पष्ट या स्पष्ट जानकारी प्रदान नहीं करती हैं।"
                                            }
                                        ],
                                        "Summary": "उत्तरदाता द्वारा दी गई गैर-प्रतिबद्ध और अस्पष्ट प्रतिक्रियाएँ।"
                                    }
                                ],
                                "Summary": "अनिर्णायक या अस्पष्ट प्रतिक्रियाओं के लिए श्रेणी।"
                            },
                            {
                                "Net": "998 लागू नहीं",
                                "Subnets": [
                                    {
                                        "Subnet": "प्रश्न के लिए अप्रासंगिक",
                                        "Categories": [
                                            {
                                                "Category": "अप्रासंगिक/अनुत्तरदायी",
                                                "Summary": "वे प्रतिक्रियाएँ जो दिए गए प्रश्न या परियोजना से पूरी तरह से अप्रासंगिक हैं।"
                                            }
                                        ],
                                        "Summary": "वे प्रतिक्रियाएँ जो प्रश्न या परियोजना के लिए लागू नहीं हैं।"
                                    }
                                ],
                                "Summary": "उन प्रतिक्रियाओं के लिए श्रेणी जो प्रश्न के लिए प्रासंगिक नहीं हैं।"
                            },
                            {
                                "Net": "999 कुछ नहीं",
                                "Subnets": [
                                    {
                                        "Subnet": "शून्य/अर्थहीन",
                                        "Categories": [
                                            {
                                                "Category": "बेतुकी/अमानक वर्ण",
                                                "Summary": "वे प्रतिक्रियाएँ जिनमें निरर्थक, बेतुके वर्ण या भाषा नहीं है।"
                                            }
                                        ],
                                        "Summary": "वे प्रतिक्रियाएँ जिनमें निरर्थक या अर्थहीन वर्ण शामिल हैं।"
                                    }
                                ],
                                "Summary": "उन प्रतिक्रियाओं के लिए श्रेणी जो निरर्थक या अर्थहीन हैं।"
                            }
                        ],
                        "Summary": "यह प्रस्तुति वित्तीय सेवाओं में आर्टिफिशियल इंटेलिजेंस (AI) को अपनाने, उसके वर्तमान रुझानों, चुनौतियों और अवसरों पर केंद्रित है। इसमें AI की हाइपर-एक्सेलरेशन चरण, एजेंटिक वर्कफ़्लो में बदलाव, मल्टीमॉडल अडॉप्शन और संदर्भ इंजीनियरिंग जैसे प्रमुख पहलुओं पर प्रकाश डाला गया है। वक्ता ने BFSI (बैंकिंग, वित्तीय सेवाएँ और बीमा) क्षेत्र में AI के व्यापक उपयोग का उल्लेख किया और एक सर्वे के परिणामों को साझा किया जिसमें ग्राहकों के बीच AI अपनाने की उच्च दर दिखाई गई। प्रस्तुति में भारत में AI के लिए विशेष भाषा संबंधी चुनौतियों और अवसर पर भी जोर दिया गया, जैसे टाटा कैपिटल का मल्टी-मॉडल वॉइस AI समाधान। मुख्य चुनौतियों में ROI (निवेश पर लाभ) की कमी, सटीकता, नियामक अनुपालन, और एंटरप्राइज़ सिस्टम के साथ एकीकरण शामिल हैं। कार्यशाला का एक महत्वपूर्ण हिस्सा \"AI फॉर बिज़नेस\" पर केंद्रित था, जिसमें व्यावसायिक उपयोगकर्ताओं को AI समाधानों को डिज़ाइन करने, प्रोटोटाइप बनाने और मान्य करने में सशक्त बनाने के लिए एक \"वाइब\" (VIBE) डिज़ाइन और प्रोटोटाइप कार्यप्रवाह का प्रदर्शन किया गया। इसमें एक AI वर्कबेंच का उपयोग करके पोर्टफोलियो प्रबंधकों के लिए स्टॉक प्रदर्शन सारांश बनाने का एक डेमो भी शामिल था, जिससे व्यावसायिक टीमों को AI-संचालित समाधानों को सह-निर्मित करने का अवसर मिला। अंत में, AI समाधानों को जल्दी से विचार से प्रोटोटाइप तक ले जाने और फिर उत्पादन में बड़े पैमाने पर लागू करने के महत्व पर जोर दिया गया, जिसमें कानूनी, अनुपालन और नियामक पहलुओं को भी ध्यान में रखा गया।"
                    }
                }
            ],
            "onlyCodeFrame": true,
            "errorList": [],
            "codeFrameGenerateType": "CODE_FRAME_ONLY",
            "requestType": "CODEFRAME",
            "requestMode": "DEVELOPMENT",
            "sourceLanguage": "English",
            "targetLanguage": "Hindi",
            "response": {
                "status": "OK",
                "data": {
                    "status": "OK",
                    "PromptTokenCount": 8883,
                    "CandidatesTokenCount": 2619,
                    "TotalTokenCount": 12887,
                    "thoughtsTokenCount": 1385,
                    "finishReason": "STOP",
                    "data": [
                        {
                            "Net": "AI का वित्तीय सेवाओं में अपनाना और रुझान",
                            "Subnets": [
                                {
                                    "Subnet": "AI का वर्तमान परिदृश्य",
                                    "Categories": [
                                        {
                                            "Category": "तीव्र विकास और AI का प्रभाव",
                                            "Summary": "AI के तेजी से विकास और इसके उद्योग पर पड़ने वाले बड़े प्रभाव का उल्लेख।",
                                            "code": 3
                                        },
                                        {
                                            "Category": "चैटबॉट से एजेंटिक वर्कफ़्लो तक",
                                            "Summary": "पारंपरिक चैटबॉट्स से अधिक स्वायत्त एजेंट-आधारित AI सिस्टम की ओर बदलाव।",
                                            "code": 4
                                        },
                                        {
                                            "Category": "मल्टीमॉडल अडॉप्शन",
                                            "Summary": "विभिन्न मॉडलों का बुद्धिमानी से उपयोग करके एक कार्य को पूरा करना।",
                                            "code": 5
                                        }
                                    ],
                                    "Summary": "AI का वर्तमान चरण, जिसमें तेजी से विकास, एजेंटिक वर्कफ़्लो और मल्टीमॉडल अडॉप्शन शामिल हैं।",
                                    "code": 2
                                },
                                {
                                    "Subnet": "AI के उपयोग के क्षेत्र",
                                    "Categories": [
                                        {
                                            "Category": "BFSI में AI का अग्रणी उपयोग",
                                            "Summary": "बैंकिंग, वित्तीय सेवाओं और बीमा क्षेत्र में AI का अग्रणी उपयोग।",
                                            "code": 7
                                        },
                                        {
                                            "Category": "व्यवसाय और IT के लिए AI",
                                            "Summary": "AI का उपयोग व्यवसाय कार्यों (ग्राहक सेवा, कॉर्पोरेट कार्य) और IT (कोडिंग सहायता, आधुनिकीकरण) दोनों में।",
                                            "code": 8
                                        },
                                        {
                                            "Category": "ऊर्ध्वाधर अभिसरण",
                                            "Summary": "इंफ्रास्ट्रक्चर से मॉडल और एप्लीकेशन तक AI स्टैक का समेकन।",
                                            "code": 9
                                        }
                                    ],
                                    "Summary": "वित्तीय सेवाओं में AI के विभिन्न उपयोग और तकनीकी अभिसरण।",
                                    "code": 6
                                },
                                {
                                    "Subnet": "ग्राहक सर्वेक्षण और AI स्वीकृति",
                                    "Categories": [
                                        {
                                            "Category": "AI अपनाने की उच्च दर",
                                            "Summary": "सर्वेक्षण में 90% ग्राहकों द्वारा AI को अपनाना, 60% उन्नत स्थिति में।",
                                            "code": 11
                                        },
                                        {
                                            "Category": "एंटरप्राइज़-विशिष्ट मॉडल की आवश्यकता",
                                            "Summary": "50% ग्राहकों द्वारा एंटरप्राइज़-विशिष्ट और फ़ाइन-ट्यून किए गए मॉडलों पर ध्यान केंद्रित करने की मांग।",
                                            "code": 12
                                        },
                                        {
                                            "Category": "परिणामों को मापने की आवश्यकता",
                                            "Summary": "अधिकांश ग्राहकों द्वारा AI समाधानों के व्यावसायिक परिणामों को मापने के महत्व पर जोर।",
                                            "code": 13
                                        }
                                    ],
                                    "Summary": "ग्राहकों के बीच AI अपनाने की स्थिति और एंटरप्राइज़-विशिष्ट मॉडलों व परिणाम मापन की आवश्यकता।",
                                    "code": 10
                                }
                            ],
                            "Summary": "वित्तीय सेवाओं में AI को अपनाने की वर्तमान स्थिति, मुख्य रुझान और ग्राहकों के दृष्टिकोण।",
                            "code": 1
                        },
                        {
                            "Net": "AI कार्यान्वयन में चुनौतियाँ",
                            "Subnets": [
                                {
                                    "Subnet": "व्यावसायिक और तकनीकी बाधाएँ",
                                    "Categories": [
                                        {
                                            "Category": "ROI (निवेश पर लाभ) की कमी",
                                            "Summary": "AI समाधानों के लिए निवेश पर लाभ प्राप्त करने में चुनौतियाँ।",
                                            "code": 16
                                        },
                                        {
                                            "Category": "सटीकता और अनुपालन के मुद्दे",
                                            "Summary": "AI मॉडल की सटीकता और नियामक अनुपालन सुनिश्चित करने में कठिनाइयाँ।",
                                            "code": 17
                                        },
                                        {
                                            "Category": "एंटरप्राइज़ एकीकरण",
                                            "Summary": "AI समाधानों को मौजूदा एंटरप्राइज़ सिस्टम (CRM, कोर बैंकिंग) के साथ एकीकृत करने में चुनौतियाँ।",
                                            "code": 18
                                        }
                                    ],
                                    "Summary": "AI समाधानों को व्यावसायिक रूप से व्यवहार्य और तकनीकी रूप से एकीकृत करने में आने वाली मुख्य बाधाएँ।",
                                    "code": 15
                                },
                                {
                                    "Subnet": "स्केलिंग और कार्यान्वयन चुनौतियाँ",
                                    "Categories": [
                                        {
                                            "Category": "व्यक्तिगत उपयोग के मामलों की सीमाएँ",
                                            "Summary": "व्यक्तिगत AI उपयोग के मामलों के बजाय प्रक्रिया प्रवाह और वर्कफ़्लो से वास्तविक लाभ।",
                                            "code": 20
                                        },
                                        {
                                            "Category": "कानूनी, नियामक और अनुपालन की भूमिका",
                                            "Summary": "AI को बड़े पैमाने पर लागू करने के लिए कानूनी, अनुपालन और नियामक निकायों की तैयारी महत्वपूर्ण है।",
                                            "code": 21
                                        },
                                        {
                                            "Category": "बिजनेस टीमों को शामिल करना",
                                            "Summary": "बिजनेस टीमों को AI अपनाने और परिवर्तनकारी उपयोग के मामलों को चलाने के लिए सशक्त बनाने की चुनौती।",
                                            "code": 22
                                        }
                                    ],
                                    "Summary": "AI समाधानों को बड़े पैमाने पर लागू करने और व्यावसायिक टीमों को प्रक्रिया में शामिल करने से संबंधित चुनौतियाँ।",
                                    "code": 19
                                }
                            ],
                            "Summary": "AI समाधानों को लागू करने, बड़े पैमाने पर करने और व्यावसायिक परिणामों को प्राप्त करने में आने वाली प्रमुख बाधाएँ।",
                            "code": 14
                        },
                        {
                            "Net": "AI समाधान विकास और प्रोटोटाइप",
                            "Subnets": [
                                {
                                    "Subnet": "मॉडल्स का उपयोग",
                                    "Categories": [
                                        {
                                            "Category": "फ़ाइन-ट्यून किए गए बनाम बड़े भाषा मॉडल",
                                            "Summary": "एंटरप्राइज़ कार्यों के लिए बड़े भाषा मॉडल और सटीक, डोमेन-अनुकूलित मॉडलों का विवेकपूर्ण उपयोग।",
                                            "code": 25
                                        },
                                        {
                                            "Category": "हाइब्रिड AI समाधान",
                                            "Summary": "डोमेन-अनुकूलित और LLM का एक साथ उपयोग करने वाले हाइब्रिड AI आर्किटेक्चर।",
                                            "code": 26
                                        },
                                        {
                                            "Category": "डोमेन क्यूरेशन और इंस्ट्रक्शन सेट",
                                            "Summary": "फ़ाइन-ट्यूनिंग के लिए सही डोमेन समझ और इंस्ट्रक्शन सेट बनाने का महत्व।",
                                            "code": 27
                                        }
                                    ],
                                    "Summary": "विभिन्न प्रकार के AI मॉडलों का रणनीतिक उपयोग और फ़ाइन-ट्यूनिंग की भूमिका।",
                                    "code": 24
                                },
                                {
                                    "Subnet": "विकास कार्यप्रवाह",
                                    "Categories": [
                                        {
                                            "Category": "विचार से प्रोटोटाइप तक",
                                            "Summary": "एक विचार को जल्दी से प्रोटोटाइप में बदलने की AI की क्षमता।",
                                            "code": 29
                                        },
                                        {
                                            "Category": "प्रक्रिया प्रवाह और वर्कफ़्लो पर ध्यान",
                                            "Summary": "वास्तविक लाभ के लिए व्यक्तिगत उपयोग के मामलों के बजाय संपूर्ण प्रक्रिया प्रवाह पर ध्यान देना।",
                                            "code": 30
                                        },
                                        {
                                            "Category": "प्लेटफ़ॉर्म-आधारित दृष्टिकोण",
                                            "Summary": "AI समाधानों को बड़े पैमाने पर लागू करने के लिए एक प्लेटफ़ॉर्म-इंजीनियर्ड सोच प्रक्रिया की आवश्यकता।",
                                            "code": 31
                                        }
                                    ],
                                    "Summary": "विचार से उत्पादन तक AI समाधान विकसित करने के लिए अनुशंसित कार्यप्रवाह और दृष्टिकोण।",
                                    "code": 28
                                }
                            ],
                            "Summary": "AI समाधानों को विकसित करने के लिए मॉडल चयन, फ़ाइन-ट्यूनिंग और एक संरचित विकास कार्यप्रवाह का महत्व।",
                            "code": 23
                        },
                        {
                            "Net": "व्यावसायिक टीमों को AI से सशक्त बनाना",
                            "Subnets": [
                                {
                                    "Subnet": "\"वाइब\" डिज़ाइन और प्रोटोटाइप",
                                    "Categories": [
                                        {
                                            "Category": "AI-संचालित अवसर कैनवास",
                                            "Summary": "AI का उपयोग करके व्यावसायिक चुनौतियों को अवसरों में बदलने और 'अवसर कैनवास' बनाने के लिए।",
                                            "code": 34
                                        },
                                        {
                                            "Category": "बिजनेस-नेतृत्व वाले AI रीडिज़ाइन",
                                            "Summary": "AI के साथ काम करके व्यावसायिक टीमों को समाधानों के डिज़ाइन प्रक्रिया का नेतृत्व करने में सशक्त बनाना।",
                                            "code": 35
                                        },
                                        {
                                            "Category": "UX प्रोटोटाइप का त्वरित निर्माण",
                                            "Summary": "डिजाइन चरण में UX प्रोटोटाइप के निर्माण में लगने वाले समय को मिनटों तक कम करना।",
                                            "code": 36
                                        }
                                    ],
                                    "Summary": "एक AI वर्कबेंच का उपयोग करके व्यावसायिक उपयोगकर्ताओं को AI समाधानों को डिज़ाइन और प्रोटोटाइप करने में सक्षम बनाना।",
                                    "code": 33
                                },
                                {
                                    "Subnet": "व्यावहारिक कार्यशाला अनुभव",
                                    "Categories": [
                                        {
                                            "Category": "हैंड्स-ऑन वाइब सेशन",
                                            "Summary": "AI समाधानों के डिज़ाइन और प्रोटोटाइप के लिए एक व्यावहारिक, निर्देशित सत्र।",
                                            "code": 38
                                        },
                                        {
                                            "Category": "एआई वर्कबेंच का उपयोग",
                                            "Summary": "AI क्षमताओं को एक साथ लाने वाला एक उपकरण जो व्यावसायिक उपयोगकर्ताओं को विचार-मंथन, डिजाइन और मान्य करने में मदद करता है।",
                                            "code": 39
                                        },
                                        {
                                            "Category": "लो-कोड प्लेटफ़ॉर्म एकीकरण",
                                            "Summary": "UX प्रोटोटाइप के बाद बैक-एंड क्षमता के लिए लो-कोड प्लेटफ़ॉर्म का उपयोग।",
                                            "code": 40
                                        }
                                    ],
                                    "Summary": "व्यवसाय उपयोगकर्ताओं को AI की क्षमता का अनुभव कराने के लिए डिज़ाइन किए गए व्यावहारिक सत्र और उपकरणों का प्रदर्शन।",
                                    "code": 37
                                }
                            ],
                            "Summary": "व्यावसायिक उपयोगकर्ताओं को AI समाधानों को पहचानने, डिज़ाइन करने, प्रोटोटाइप करने और मान्य करने के लिए सशक्त बनाने पर ध्यान केंद्रित करना।",
                            "code": 32
                        },
                        {
                            "Net": "विशिष्ट AI उपयोग के मामले",
                            "Subnets": [
                                {
                                    "Subnet": "भारतीय संदर्भ में AI",
                                    "Categories": [
                                        {
                                            "Category": "बहुभाषी वॉइस AI समाधान (टाटा कैपिटल)",
                                            "Summary": "भारत की भाषाई विविधता को संबोधित करने वाला टाटा कैपिटल का मल्टी-मॉडल वॉइस AI समाधान।",
                                            "code": 43
                                        },
                                        {
                                            "Category": "वित्तीय समावेशन",
                                            "Summary": "टाटा कैपिटल का AI का उपयोग करके वित्तीय सेवाओं को सभी तक पहुँचाना।",
                                            "code": 44
                                        }
                                    ],
                                    "Summary": "भारत की अद्वितीय भाषाई और वित्तीय आवश्यकताओं के लिए AI के अनुकूलन का प्रदर्शन।",
                                    "code": 42
                                },
                                {
                                    "Subnet": "दावों के प्रसंस्करण में AI (UK ग्राहक)",
                                    "Categories": [
                                        {
                                            "Category": "दावा रिपोर्टिंग और जानकारी कैप्चर",
                                            "Summary": "एक यूके ग्राहक के लिए दुर्घटना दावे की रिपोर्टिंग में AI का उपयोग करके जानकारी को कुशलता से कैप्चर करना।",
                                            "code": 46
                                        },
                                        {
                                            "Category": "संवर्धित AI (Augmented AI)",
                                            "Summary": "AI का उपयोग बातचीत के दौरान सहायक जानकारी प्रदान करने और दावे प्रतिनिधि को सहायता करने के लिए।",
                                            "code": 47
                                        },
                                        {
                                            "Category": "मुकदमेबाजी की भविष्यवाणी और धोखाधड़ी का पता लगाना",
                                            "Summary": "फ़ाइन-ट्यून किए गए मॉडल का उपयोग करके धोखाधड़ी या मुकदमेबाजी की संभावना का पता लगाना।",
                                            "code": 48
                                        }
                                    ],
                                    "Summary": "बीमा दावों के प्रसंस्करण में AI के उपयोग का एक उदाहरण, जिसमें संवर्धित AI और धोखाधड़ी का पता लगाना शामिल है।",
                                    "code": 45
                                },
                                {
                                    "Subnet": "पोर्टफोलियो सारांश सहायक",
                                    "Categories": [
                                        {
                                            "Category": "पोर्टफोलियो प्रदर्शन सारांश",
                                            "Summary": "पोर्टफोलियो प्रबंधकों के लिए स्टॉक प्रदर्शन और भावना का त्वरित सारांश उत्पन्न करने के लिए AI का उपयोग।",
                                            "code": 50
                                        },
                                        {
                                            "Category": "डेटा संग्रह और विश्लेषण का स्वचालन",
                                            "Summary": "सार्वजनिक और वित्तीय डोमेन से डेटा एकत्र करने और विश्लेषण करने के कार्य को स्वचालित करना।",
                                            "code": 51
                                        },
                                        {
                                            "Category": "रिपोर्ट की गुणवत्ता और समय में सुधार",
                                            "Summary": "आउटसोर्सिंग को कम करके और रिपोर्ट बनाने के समय और गुणवत्ता में सुधार करके।",
                                            "code": 52
                                        }
                                    ],
                                    "Summary": "पोर्टफोलियो प्रबंधकों के लिए AI-संचालित सहायक का प्रदर्शन, जो स्टॉक प्रदर्शन का त्वरित सारांश प्रदान करता है।",
                                    "code": 49
                                }
                            ],
                            "Summary": "विभिन्न वास्तविक दुनिया के परिदृश्यों में AI के व्यावहारिक अनुप्रयोगों का प्रदर्शन, जिसमें वित्तीय समावेशन और दावा प्रसंस्करण शामिल है।",
                            "code": 41
                        }
                    ],
                    "Summary": "यह प्रस्तुति वित्तीय सेवाओं में आर्टिफिशियल इंटेलिजेंस (AI) को अपनाने, उसके वर्तमान रुझानों, चुनौतियों और अवसरों पर केंद्रित है। इसमें AI की हाइपर-एक्सेलरेशन चरण, एजेंटिक वर्कफ़्लो में बदलाव, मल्टीमॉडल अडॉप्शन और संदर्भ इंजीनियरिंग जैसे प्रमुख पहलुओं पर प्रकाश डाला गया है। वक्ता ने BFSI (बैंकिंग, वित्तीय सेवाएँ और बीमा) क्षेत्र में AI के व्यापक उपयोग का उल्लेख किया और एक सर्वे के परिणामों को साझा किया जिसमें ग्राहकों के बीच AI अपनाने की उच्च दर दिखाई गई। प्रस्तुति में भारत में AI के लिए विशेष भाषा संबंधी चुनौतियों और अवसर पर भी जोर दिया गया, जैसे टाटा कैपिटल का मल्टी-मॉडल वॉइस AI समाधान। मुख्य चुनौतियों में ROI (निवेश पर लाभ) की कमी, सटीकता, नियामक अनुपालन, और एंटरप्राइज़ सिस्टम के साथ एकीकरण शामिल हैं। कार्यशाला का एक महत्वपूर्ण हिस्सा \"AI फॉर बिज़नेस\" पर केंद्रित था, जिसमें व्यावसायिक उपयोगकर्ताओं को AI समाधानों को डिज़ाइन करने, प्रोटोटाइप बनाने और मान्य करने में सशक्त बनाने के लिए एक \"वाइब\" (VIBE) डिज़ाइन और प्रोटोटाइप कार्यप्रवाह का प्रदर्शन किया गया। इसमें एक AI वर्कबेंच का उपयोग करके पोर्टफोलियो प्रबंधकों के लिए स्टॉक प्रदर्शन सारांश बनाने का एक डेमो भी शामिल था, जिससे व्यावसायिक टीमों को AI-संचालित समाधानों को सह-निर्मित करने का अवसर मिला। अंत में, AI समाधानों को जल्दी से विचार से प्रोटोटाइप तक ले जाने और फिर उत्पादन में बड़े पैमाने पर लागू करने के महत्व पर जोर दिया गया, जिसमें कानूनी, अनुपालन और नियामक पहलुओं को भी ध्यान में रखा गया।"
                }
            },
            "summaryList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 8544,
                        "CandidatesTokenCount": 2424,
                        "TotalTokenCount": 12275,
                        "thoughtsTokenCount": 1307,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Summary": "<html><body><h2>कृत्रिम बुद्धिमत्ता (AI) अपनाने की व्यापक रिपोर्ट और कार्यशाला का सारांश</h2><h3>परिचय और AI की वर्तमान स्थिति</h3><p>वित्तीय सेवा क्षेत्र में कृत्रिम बुद्धिमत्ता (AI) का अपनाना एक तीव्र गति से हो रहा है, जिसे 'हाइपर एक्सेलेरेशन' चरण के रूप में वर्णित किया गया है। वर्तमान में, पारंपरिक चैटबॉट इंटरफ़ेस से हटकर 'एजेंट्स इन एक्शन' (क्रियाशील एजेंट) की ओर एक महत्वपूर्ण बदलाव देखा जा रहा है, जिसमें BFSI (बैंकिंग, वित्तीय सेवाएँ और बीमा) उद्योग अग्रणी है। कई 'एस-वेव' और हस्तक्षेप एक साथ हो रहे हैं, जो AI को अपनाने के लिए चुनौतियाँ और अवसर दोनों प्रदान करते हैं। भविष्य में मल्टीमॉडल AI को अपनाने पर जोर दिया जा रहा है, जहाँ विशिष्ट कार्यों या प्रक्रिया प्रवाह को पूरा करने के लिए कई मॉडलों का न्यायसंगत उपयोग किया जाएगा। इसके मूल में 'कॉन्टेक्स्ट इंजीनियरिंग' है, जिसका अर्थ है कि मॉडल के साथ जुड़ने से पहले उचित संदर्भ का निर्माण करना अत्यंत महत्वपूर्ण है। बाजार में कई प्रतिस्पर्धी मॉडल तेज़ी से आ रहे हैं, जिनकी गुणवत्ता लगातार सुधर रही है, जैसा कि 5.3 कोडेक्स और सोननेट 4.6 के 20 मिनट के भीतर लॉन्च होने से पता चलता है।</p><p>AI पारिस्थितिकी तंत्र में एक 'वर्टिकल कन्वर्जेंस' भी हो रहा है, जिसमें हार्डवेयर और बुनियादी ढाँचे से लेकर मॉडलों (विशिष्ट और फाउंडेशन मॉडल दोनों) और अनुप्रयोगों तक पूरा स्टैक शामिल है। इस क्षेत्र में वैश्विक स्तर पर और भारत में भी भारी निवेश देखा जा रहा है, जिसमें AI डेटा सेंटर (जैसे अगले पाँच वर्षों में 1 गीगावाट डेटा सेंटर) लॉन्च करने की घोषणाएँ शामिल हैं।</p><h3>ग्राहक सर्वेक्षण और AI अपनाने के रुझान</h3><p>एक हालिया सर्वेक्षण, जिसमें भारत और वैश्विक स्तर पर 170 ग्राहक शामिल थे, ने AI अपनाने के संबंध में महत्वपूर्ण अंतर्दृष्टि प्रदान की:</p><ul><li>लगभग 90% ग्राहक किसी न किसी रूप में AI को अपना रहे हैं, जिनमें से लगभग 60% उन्नत स्थिति में हैं। यह बैंकिंग, पूंजी बाजार और बीमा ग्राहकों में देखा गया।</li><li>50% ग्राहक इस बात पर जोर देते हैं कि वित्तीय सेवा उद्योग के विनियमित स्वरूप के कारण एंटरप्राइज़-विशिष्ट और फाइन-ट्यून्ड मॉडलों पर ध्यान केंद्रित करने की आवश्यकता है, जिससे उच्च सुरक्षा और सटीकता सुनिश्चित हो सके।</li><li>अधिकांश ग्राहकों ने व्यावसायिक परिणामों को मापने की आवश्यकता पर बल दिया, यह दर्शाता है कि POCs (प्रूफ ऑफ कॉन्सेप्ट्स) और पायलटों से परे वास्तविक व्यावसायिक प्रभाव की आवश्यकता है।</li></ul><h3>भारत में AI अपनाने की बारीकियां और प्रदर्शन</h3><p>भारत में AI के लिए अद्वितीय अवसर और चुनौतियाँ हैं, खासकर भाषा और आवाज़ के संदर्भ में। भारत एक 'वॉइस-लेड' देश है जहाँ कई भाषाएँ बोली जाती हैं, विविध भौगोलिक क्षेत्रों में फैली हुई हैं, और अद्वितीय वित्तीय आवश्यकताएँ हैं। टाटा कैपिटल ने वित्तीय समावेशन को बढ़ावा देने के लिए एक मल्टीमॉडल वॉयस AI समाधान प्रदर्शित किया।</p><h4>टाटा कैपिटल वॉयस AI समाधान का प्रदर्शन:</h4><p>एक ग्राहक के साथ हिंदी में बातचीत का प्रदर्शन किया गया, जिसमें AI ने ग्राहक के प्रश्नों को समझा, व्यक्तिगत जानकारी फॉर्म भेजने में सहायता की, और ग्राहक द्वारा फॉर्म भरने के बाद ऋण विवरण (जैसे 5 लाख रुपये का ऋण, 3 साल के लिए, 12.5% ब्याज दर, 16,727 रुपये की EMI) प्रदान किया। इस समाधान ने ग्राहक जुड़ाव को 'जिम्मेदारी, सहानुभूति और वास्तविक प्रभाव' के साथ फिर से परिभाषित किया है। यह 'सर्वम' नामक एक भारत-आधारित मॉडल पर आधारित है, जो भाषाई विविधताओं को संबोधित करता है और ग्राहक सेवा के लिए बहु-चैनल क्षमता को बढ़ाता है।</p><h3>AI को अपनाने में जमीनी हकीकत और चुनौतियाँ</h3><p>पिछले 9-12 महीनों के अनुभव से AI को अपनाने में कुछ महत्वपूर्ण जमीनी हकीकत सामने आई हैं:</p><h4>AI फॉर बिजनेस बनाम AI फॉर IT:</h4><ul><li><strong>AI फॉर IT:</strong> मुख्य रूप से कोडिंग सहायता और प्रौद्योगिकी आधुनिकीकरण पर केंद्रित है, जिसमें POCs और पायलटों से उत्पादन में जाने पर जोर दिया जाता है।</li><li><strong>AI फॉर बिजनेस:</strong> ग्राहक सेवा, क्षैतिज कार्य (जैसे कॉर्पोरेट कार्य), और धन व निवेश सलाहकार सेवाओं सहित विभिन्न क्षेत्रों में प्रगति देखी जा रही है।</li></ul><h4>प्रमुख चुनौतियाँ:</h4><ul><li><strong>ROI चुनौतियाँ:</strong> AI समाधानों के लिए निवेश पर रिटर्न को मापना अभी भी एक चुनौती है।</li><li><strong>सटीकता और अनुपालन:</strong> AI समाधानों के निर्माण में सटीकता और अनुपालन पहलू महत्वपूर्ण बने हुए हैं।</li><li><strong>एंटरप्राइज़ इंटीग्रेशन:</strong> AI समाधानों का निर्माण एक बात है, लेकिन उन्हें CRM, कोर बैंकिंग और अन्य एंटरप्राइज़ प्रणालियों के साथ एकीकृत करना और काम करवाना अत्यंत महत्वपूर्ण है।</li><li><strong>प्रक्रिया-प्रवाह और कार्यप्रवाह पर ध्यान दें:</strong> व्यक्तिगत उपयोग के मामलों की बजाय, वास्तविक लाभ प्रक्रिया प्रवाह और कार्यप्रवाहों से आता है। पारंपरिक 'बाएँ से दाएँ' के बजाय 'दाएँ से बाएँ' (परिणाम-केंद्रित) दृष्टिकोण को अपनाना महत्वपूर्ण है।</li></ul><h3>फाइन-ट्यूनिंग और हाइब्रिड AI मॉडल</h3><p>वित्तीय सेवाओं में फाइन-ट्यूनिंग में उल्लेखनीय वृद्धि देखी जाएगी, जो अगले 6-9 महीनों में जोर पकड़ेगी। ऐसा इसलिए है क्योंकि बहुत अधिक सटीक और सटीक मॉडल की आवश्यकता है। एक हाइब्रिड परिदृश्य की उम्मीद है जहाँ फाइन-ट्यून्ड मॉडल और लार्ज लैंग्वेज मॉडल (LLMs) दोनों का न्यायसंगत उपयोग किया जाएगा।</p><h4>मुख्य बिंदु:</h4><ul><li><strong>हाइब्रिड AI समाधान:</strong> LLMs का उपयोग व्यापक, मापने योग्य एंटरप्राइज़ कार्यों के लिए किया जाएगा (उचित ग्राउंडिंग और वर्कफ़्लो ऑर्केस्ट्रेशन के साथ), जबकि डोमेन-अनुकूलित मॉडल का उपयोग अनुपालन, सलाहकार और क्रेडिट अनुसंधान जैसे विशिष्ट क्षेत्रों में किया जाएगा।</li><li><strong>अभिसरण:</strong> LLMs और फाइन-ट्यून्ड मॉडल दोनों का अपना स्थान है, और दोनों के बीच अभिसरण महत्वपूर्ण है।</li><li><strong>वास्तविक जीवन के उदाहरण:</strong></li><ul><li><strong>दावों में मुकदमेबाजी की भविष्यवाणी:</strong> फाइन-ट्यून्ड मॉडल ने बेस मॉडल की तुलना में सटीकता में उल्लेखनीय सुधार दिखाया, जो पुनरावृत्तीय निर्माण के माध्यम से प्राप्त किया गया।</li><li><strong>अनुपालन:</strong> लामा या मिस्ट्राल फैमिली जैसे विभिन्न मॉडलों का उपयोग करके अनुपालन में सुधार।</li></ul><li><strong>फाइन-ट्यूनिंग का सार:</strong> यह केवल प्रौद्योगिकी नहीं है, बल्कि 'डोमेन क्यूरेशन' है। फाइन-ट्यूनिंग के लिए सही निर्देश सेट बनाने में गहरा डोमेन ज्ञान शामिल है। इसके लिए सही तकनीकों और सही मॉडलों का चुनाव भी महत्वपूर्ण है।</li><li><strong>व्यावसायिक परिदृश्य:</strong> फाइन-ट्यूनिंग के माध्यम से व्यवसाय-विशिष्ट AI समाधानों में वैश्विक और भारत दोनों में मजबूत वृद्धि देखी जा रही है।</li></ul><h3>ऑगमेंटेड AI का प्रदर्शन (बीमा दावे)</h3><p>एक अलग प्रदर्शन में, एक UK-आधारित ग्राहक ने दुर्घटना की सूचना दी। यहाँ AI सामने नहीं था, बल्कि पृष्ठभूमि में काम कर रहा था, ग्राहक सेवा प्रतिनिधि को संवर्धित जानकारी प्रदान कर रहा था। यह 'फर्स्ट नोटिस ऑफ लॉस' (FNOL) परिदृश्य था जहाँ AI ने बातचीत को सुनकर महत्वपूर्ण जानकारी एकत्र की।</p><h4>ऑगमेंटेड AI के लाभ:</h4><ul><li>दावों में देरी को रोकने के लिए महत्वपूर्ण जानकारी को कैप्चर करना।</li><li>पृष्ठभूमि में AI द्वारा जानकारी एकत्र की गई, जिसमें एक फाइन-ट्यून्ड 'मुकदमेबाजी सत्यापनकर्ता' मॉडल भी शामिल था जो संभावित धोखाधड़ी का पता लगाता था।</li><li>यह AI मानव एजेंटों को संवर्धित करके निर्णय लेने में सहायता करता है, यह सुनिश्चित करता है कि सही निर्णय लेने के लिए सभी आवश्यक जानकारी उपलब्ध हो।</li></ul><h3>AI अपनाने के लिए उचित कैलिब्रेशन और कार्यप्रणाली</h3><p>एंटरप्राइज़ AI के लिए उचित कैलिब्रेशन महत्वपूर्ण है। यह सिर्फ महान प्रौद्योगिकी और क्षमताओं के बारे में नहीं है, बल्कि यह भी जानना है कि AI को कहाँ लागू नहीं करना है। AI को हर जगह लागू करने के बजाय, इसे 'दाएँ से बाएँ' दृष्टिकोण के साथ परिणाम-आधारित तरीके से लागू करना चाहिए, यह देखने के लिए कि वांछित परिणाम क्या हैं और उन्हें कैसे प्राप्त किया जा सकता है।</p><h3>प्रूफ ऑफ वैल्यू (PoV) कार्यशाला</h3><p>कार्यशाला का उद्देश्य व्यवसायों को AI की शक्ति को समझने और अधिक परिवर्तनकारी उपयोग के मामलों की पहचान करने में सक्षम बनाना था। लक्ष्य एक 'प्रूफ ऑफ वैल्यू' बनाना है जो व्यावसायिक आश्वासन प्रदान करता है।</p><h4>PoV कार्यशाला की संरचना:</h4><ul><li><strong>क्विक प्रूफ ऑफ वैल्यू (PoV):</strong> उपयोगकर्ता त्वरित प्रोटोटाइप, वाइब डिज़ाइन और प्रोटोटाइप बनाने में सक्षम होंगे।</li><li><strong>रैपिड बिल्ड:</strong> 5-6 सप्ताह की अवधि में कार्यक्षमता जोड़कर पूर्ण समाधान विकसित करना।</li><li><strong>स्केलिंग:</strong> उत्पादन के लिए समाधान को कैलिब्रेटेड तरीके से स्केल करना, जिसमें AI फाउंडेशन क्षमता और नियंत्रण विमान का निर्माण शामिल है।</li></ul><h4>कार्यशाला गतिविधि:</h4><p>भाग लेने वालों को लैपटॉप और 'चैलेंज कार्ड' प्रदान किए गए। उन्होंने एक पोर्टफोलियो मैनेजर के लिए एक सहायक बनाने का उपयोग का मामला चुना, जिसे तिमाही रिपोर्ट के लिए स्टॉक के प्रदर्शन और भावना का सारांश उत्पन्न करना था। AI का उपयोग चुनौती से एक अवसर कैनवास (opportunity canvas) बनाने के लिए किया गया था, जिसमें दर्द बिंदुओं का विश्लेषण करना, व्यावसायिक प्रभाव का आकलन करना और समाधान दिशाओं की पहचान करना शामिल था।</p><h4>कार्यशाला के परिणाम:</h4><ul><li><strong>व्यवसाय-नेतृत्व वाला डिज़ाइन:</strong> व्यवसाय उपयोगकर्ताओं को AI के साथ डिज़ाइन प्रक्रिया को चलाने के लिए सशक्त बनाया गया, एजेंटिक समाधानों को तोड़कर।</li><li><strong>तेज़ UX प्रोटोटाइपिंग:</strong> Lovable और Google AI Studio जैसे लो-कोड टूल का उपयोग करके, UX प्रोटोटाइप मिनटों में बनाए गए, जिससे हितधारक सत्यापन और त्वरित पुनरावृति सक्षम हुई।</li><li><strong>बैक-एंड इंजीनियरिंग:</strong> Ne10 जैसे लो-कोड प्लेटफॉर्म का उपयोग करके बैक-एंड क्षमताओं को इंजीनियर किया गया, जिससे मॉडल उपयोग, सुरक्षा उपायों और व्यावसायिक मूल्यांकन का विज़ुअलाइज़ेशन और सत्यापन संभव हुआ।</li></ul><p>कई प्रतिभागियों ने अपने स्वयं के UX प्रोटोटाइप सफलतापूर्वक उत्पन्न किए, यह दर्शाता है कि यह प्रक्रिया व्यापार टीमों को AI क्षमताओं को समझने और उपयोग करने में कैसे मदद कर सकती है।</p><h3>निष्कर्ष</h3><p>AI विचार को प्रोटोटाइप में तेजी से बदलने की क्षमता रखता है, हालांकि वह प्रोटोटाइप सीधे उत्पादन में नहीं जा सकता है, यह क्या किया जा सकता है इसका अनुभव प्रदान करता है। उत्पादन में AI को स्केल करने के लिए एक कैलिब्रेटेड दृष्टिकोण की आवश्यकता होती है, जिसमें कानूनी, अनुपालन और नियामक आवश्यकताओं को पूरा करना और व्यापार टीमों को शामिल करना शामिल है। सफल AI अपनाने के लिए एक मंच-आधारित विचार प्रक्रिया और एंड-टू-एंड व्यावसायिक परिणामों पर ध्यान केंद्रित करना आवश्यक है।</p></body></html>"
                            }
                        ]
                    }
                }
            ]
        }
    ],
    "headerCols": {
        "col1": "Part Id",
        "col2": "FileName",
        "col3": "Content"
    },
    "fileName": "",
    "projectType": "DOCUMENT",
    "sourceLanguage": "English",
    "targetLanguage": "Hindi",
    "__v": 0
},
    {
    "_id": "69941c4525c9f4e0d7e70127",
    "projectId": "874585",
    "projectName": "AI in Health Saving Lives at Scale - English",
    "projectDescription": "AI in Health Saving Lives at Scale - English",
        "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T07:44:09.926Z",
    "projectUpdatedDate": "2026-02-17T07:44:09.926Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T08:08:22.457Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "AI in Health Saving Lives at Scale.mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771314204468.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:59",
                    "end": "00:02:42",
                    "text": " Excuse me, we request you to please exit quickly so that we can start our next session. Welcome back to J-PAL's AI for Social Good Impact That Works seminar. Our next session is AI in Health, Saving Lives at Scale. Please join me in welcoming Ziad Obermeyer. Ziad is an associate professor of health policy and management at the UC Berkeley School of Public Health, where he works at the intersection of machine learning, medicine, and health policy."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:42",
                    "end": "00:03:09",
                    "text": " He helped set up the computational precision health program and his research focuses on how machine learning can help clinicians make better decisions. A fun fact about Ziad is that he was named one of the 100 most influential people in AI by 10 magazine. Ziad, over to you. Thank you so much. So I thought I would tell you a story about when I first came to my job at Berkeley. So this was in 2018."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:09",
                    "end": "00:03:39",
                    "text": " I was a junior, untenured faculty member, and I just signed up to give the first seminar of the year in the economics department. Now, the economics department at Berkeley is friendlier than the economics department at other institutions affiliated with J-PAL, but they are still economists and still very hardcore, so I was nervous. And about three or four days before the seminar, I started feeling a little unwell."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:39",
                    "end": "00:04:07",
                    "text": " My stomach was hurting. I just didn't feel right. And I made the mistake of mentioning this to my wife, who said something like, oh, is your tummy hurting? Are you a little nervous about your... Anyway, she was not very sympathetic. So I prepared for my talk. I prepared for my first class that I was teaching. And I gave the seminar. And during the seminar, instead of feeling just globally not well,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:07",
                    "end": "00:04:36",
                    "text": " I started feeling unwell in a particular place in my stomach, which is the bottom right side of my stomach. Some of you might know what I have already, but I want to first make an observation that this made me realize, which is that when people make really important health decisions, like do I come into the hospital, do I ask a doctor about this, people are making those decisions that are very, very important with very little data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:36",
                    "end": "00:05:03",
                    "text": " So at the end of the seminar, I actually drove myself to the emergency room near Berkeley and I got a CT scan. And this is my CT scan. And some of you might be able to see what I have or guess what I have, but I had acute appendicitis and it had perforated and I needed to have my appendix removed. I think what makes this story surprising and a little bit embarrassing for me"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:03",
                    "end": "00:05:26",
                    "text": " is that I trained as an emergency room doctor. So there's a few things that you're not supposed to miss as an emergency room doctor, and this is one of them. And yet, I missed it in myself for four or five days. So I think what this highlighted to me is how critically important it is."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:26",
                    "end": "00:05:48",
                    "text": " to be able to have access to this kind of data. The kind of data that normally live only inside of the health system and that you can only have access to if you're lucky enough to be close to an emergency room with a CT scanner to have access both financial and linguistic and socioeconomic to these kinds of technologies. Because without data like this,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:48",
                    "end": "00:06:17",
                    "text": " There's no diagnosis of critical illnesses. And without diagnoses, there's no treatments. So this data is really the key to a lot of things that happen in the health system. And yet, in order to get access to this, you have to be part of the health system. You have to have access to begin with. So there's a circularity to this problem. And it locks out many, many people, of course, not just in the US, but even more around the world."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:17",
                    "end": "00:06:34",
                    "text": " This kind of underdiagnosis is, of course, not just limited to acute surgical things like appendicitis, but to many, many different conditions. And a particularly salient example to many people happens even in high income countries, which is heart attack. So there are these studies that follow."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:34",
                    "end": "00:06:53",
                    "text": " big cohorts of people across time in the US, and every so often they do an MRI of the heart to detect scarring in the wall of the heart that indicates that this person had a heart attack. And in these studies, when they look at all of the people who have scar in their heart consistent with a heart attack,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:53",
                    "end": "00:07:18",
                    "text": " Most studies find that the majority of those heart attacks are silent. They are not known to the patient and not known to the doctor and not diagnosed, either because the symptoms were subtle or because the person never sought care to get those symptoms diagnosed. So as you can imagine, this is a huge problem even in high income countries. It's an even bigger problem elsewhere in the world. And so there's data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:18",
                    "end": "00:07:33",
                    "text": " from South India, from Tamil Nadu, that asks people about things like high blood pressure and diabetes and asks them, do you have these things? And then does the objective measure of those things. And even for these basic risk factors for heart attacks, we..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:33",
                    "end": "00:08:02",
                    "text": " We can't do cardiac MRIs in a lot of places, but even for these risk factors, there's dramatic underdiagnosis and under-awareness of the things that cause heart attack. So this is a problem everywhere, and it's a problem that really urgently needs solutions. So I thought I could tell you, I'm going to tell you about half of the solution, and instead of just telling you, I'm going to show you, and you'll see more of this from my colleague Shredda in a few moments."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:02",
                    "end": "00:08:18",
                    "text": " This device that I'm holding in my hand is a mobile handheld electrocardiogram. So the two silver sides are electrodes. You put your fingers on it and it records the electrical field generated by your heart as it beats."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:18",
                    "end": "00:08:43",
                    "text": " And I think these technologies are really miraculous because they cost so little. That device costs $60. In the US, you can order it on Amazon and get it delivered in the same day. It connects to your phone or tablet via Bluetooth with no wires. And it requires absolutely no training to do an electrocardiogram. So this is really an incredible technology that's become commoditized over the past five to 10 years."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:43",
                    "end": "00:09:06",
                    "text": " But the reason that these kinds of devices have not had the impact that any of us would have hoped is because there's a key constraint, which is that what do you do once you generate this waveform? Well, you have to send it to a doctor for interpretation. This is too complex for patients, and even for some doctors like myself have trouble interpreting electrocardiogram waveforms."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:06",
                    "end": "00:09:35",
                    "text": " And so it's great that you can generate this data wherever you are in the world. But if you still need to send it to a doctor to get interpreted and to translate those data into decision making, you haven't solved the key bottleneck, which is access to the health care system. So even though these devices are wonderful and miraculous, they're only part of the solution to democratizing access to diagnosis around the world. So this is an AI summit. Could AI help?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:35",
                    "end": "00:10:01",
                    "text": " Yes. But it's important to remember that there's nothing magical about artificial intelligence. And in fact, artificial intelligence is just data in a very literal sense. So in an ideal world, what you'd have as far as a data set to train your AI algorithm is a data set that links data from these cheap devices to some ground truth outcome about heart attack, appendicitis, et cetera, et cetera."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:01",
                    "end": "00:10:25",
                    "text": " Unfortunately, those data on the ground truth are quite hard to find even in high-income countries and certainly in low- and middle-income countries. And so what we often do when we're training AI tools is we make a key and very pernicious substitution, which is we say, okay, well, we can't get the ground truth linked to these electrocardiogram waveforms."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:25",
                    "end": "00:10:52",
                    "text": " So instead, we're going to ask doctors to look at the waveforms and tell us what the doctor thinks about that waveform. So we substitute human judgment and human interpretation for some sense of the ground truth of, does this person really have a heart attack or not? Let's just ask the human. And the problem with humans, many of my good friends are humans, so this is nothing about humans in general, but humans have biases and humans make mistakes. And human knowledge..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:52",
                    "end": "00:11:22",
                    "text": " about the human body and many other things is incomplete. And so when we automate human judgment, we also run the risk of automating all of the problems with human judgment along with it. And I think that's what makes it so critically important to train artificial intelligence not on doctors, but on patients, on what happens to patients and their outcomes. So let me tell you about some work that I am doing in collaboration with J-PAL South Asia in a few"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:22",
                    "end": "00:11:37",
                    "text": " villages in Tamil Nadu. So this is a line of people queuing up for a district health center. And what we did over the summer of 2023 is set up health camps, a long-standing tradition where we just..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:37",
                    "end": "00:12:05",
                    "text": " had a facility outside of this district health center where, thanks to the tireless work of our colleagues at J-PAL South Asia, we rented out a bunch of very fancy equipment that you can normally only get inside of hospitals. So for example, we rented out a cardiac ultrasound machine and a technologist to acquire the images of cardiac ultrasounds. We hired cardiologists to interpret those images and look."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:05",
                    "end": "00:12:33",
                    "text": " among other things, for signs that this person had had a heart attack, signs of scarring in their heart. And at the same time, we did data collection from these cheap $50 devices. So here you see two patients that were participants in this health camp. This young lady is having an electrocardiogram done and sent to the tablet in front of her. And the gentleman over here is getting a cardiac ultrasound, which serves as our way of understanding."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:33",
                    "end": "00:12:53",
                    "text": " if this person has had a heart attack by looking at wall motion abnormalities in the walls of his heart. So every patient got both the fancy expensive tests from the hospital and the cheap low cost tests. And of course, if we found any abnormalities, we communicated those back to the patient and their doctors as part of this study."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:54",
                    "end": "00:13:18",
                    "text": " So so the eventual goal of this is to create a way to screen people for a number of serious health conditions including prior potentially silent heart attacks in the comfort and safety of their own home either by doing the electrocardiogram themselves or with the assistance of a community health worker and Just to give you a sense of how well even this early version of the algorithm is working"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:18",
                    "end": "00:13:40",
                    "text": " when we flag the highest risk, two, three, four, five percent of the population, just general population of people visiting an ambulatory health center, 10% have actually had signs of a prior heart attack, and that's compared to about 0.2% in the general population of people that we were looking at. So we're finding very high-risk people, and..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:40",
                    "end": "00:13:54",
                    "text": " people who, even though the algorithm is saying that they're high risk and they do indeed have silent heart attack on their cardiac ultrasound, they lack the traditional risk factors like blood pressure and cholesterol and diabetes."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:54",
                    "end": "00:14:22",
                    "text": " that are traditional because they were defined in Western populations. And as we know, South Asian populations have different risk profiles and different risk factors that this electrocardiographic approach is able to find despite the fact that they lack those traditional risk factors. So I think this highlights the fact that these algorithms can be trained and tuned to populations all over the world that might be different from the populations in which most of our medical knowledge is developed."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:22",
                    "end": "00:14:51",
                    "text": " When we put this through the cost-effectiveness of screening, in other words, sending high-risk people for confirmatory tests at the health center and treating them with the medications that you need to be on for people who have had prior heart attacks, this comes out to be cost-effective, even using Indian guidelines. So it's around $2,000 per disability-adjusted life year, even for this very early version of the algorithm, which is continuously improving as we collect more data. This is very exciting."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:51",
                    "end": "00:15:06",
                    "text": " And the next step is actually, this wouldn't be J-PAL unless the next step were a randomized evaluation. And so we're doing a randomized evaluation where we're gonna start screening people either with traditional risk factors like blood pressure, et cetera."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:06",
                    "end": "00:15:28",
                    "text": " or this AI enhanced electrocardiogram screening approach and referring the high risk people for confirmatory testing in both groups and seeing which approach finds more people that we want to find, people with prior heart attacks. So one of the exciting things about doing this project is that unlike a lot of the research that I've done in the past,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:28",
                    "end": "00:15:58",
                    "text": " This is a real thing. It's a device. And it's already interacting with real people. And so to give you a sense of that, if you don't mind, I'm going to introduce Ms. Shreda Rajesh, who is an instrumental part of our team with J-PAL South Asia. And we're going to violate every principle of giving a talk by doing a live demo and hoping that it works. So if you'll bear with us for a second, we're going to get this set up. And then you will see my own electrocardiogram being put through."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:58",
                    "end": "00:16:29",
                    "text": " our web application and screening me in real time for my own risk of prior heart attack. Luckily, since I'm a physician, nobody needs to take responsibility for the outcome of the test besides myself. All right. So we're getting it hooked up. And I'm hoping it works because we tried it before the talk. But of course, live demos have a way of frustrating your expectations."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:40",
                    "end": "00:17:53",
                    "text": " We're gonna switch to the other podium. One thing I'll just mention as we're getting set up is that this approach is not limited to mobile electrocardiograms."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:53",
                    "end": "00:18:21",
                    "text": " those of you who have Apple watches, of course, know that there's increasing amounts of mobile technology built into your watch that can be fed in, but these electrocardiograms are not the only low-cost data collection devices that we have available to us. So in the course of this study in Tamil Nadu, we also collected a number of other low-cost devices. So those of you who sadly had to live through COVID here, no doubt had"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:21",
                    "end": "00:18:40",
                    "text": " mobile portable pulse oximeters. Those pulse oximeters, in addition to generating oxygen percentage numbers, generate a waveform that actually represents the blood flow through your finger that the pulse oximeter is measuring to get that fraction of oxygen."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:40",
                    "end": "00:19:07",
                    "text": " But that waveform has an enormous amount of information about how your heart is working, how your capillaries are working, how your blood vessels are allowing the flow of blood through all of these channels. So that also can plug into algorithms. The gentleman in the front row is wearing an aura ring or a related ring that can also measure this kind of thing. Continuous temperature monitoring is another feature of these rings. There are smartphone-based attachments."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:07",
                    "end": "00:19:10",
                    "text": " that can take a picture of your retina."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:10",
                    "end": "00:19:39",
                    "text": " A fun fact about the retina, you got a lot of fun facts about the speakers from Audrey. A fun fact about the retina is that it is part of your central nervous system. It's the only exposed part of your central nervous system. And with a $200 smartphone attachment, we can actually get a view into the central nervous system that can be useful for diagnosing stroke, intracranial hemorrhage from trauma, retinal disease, other kinds of diseases of the eye. Okay."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:39",
                    "end": "00:20:10",
                    "text": " Thank you so much for getting this working, Shyada. Would you mind telling us a little bit about what we're going to do? Of course. Thanks so much, Ziad. Sorry, this took a bit of time. Good morning, everyone. We'd now like to demonstrate what this looks like in practice. So all a participant has to do is basically place their fingers on this very small device that's connected to my phone, as you see. And that's it. No wires, no hospitals, and just 30 seconds. Ziad, I hope you don't mind."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:25",
                    "end": "00:20:46",
                    "text": " Sounds good. So what this device does is it records a single-led ECG, which is the electrical signal of your heart. And once we capture that, I enter the participant's ID onto our app that we developed in-house. And behind the scenes, our AI model, it analyzes this waveform, the pattern of the heart, almost instantly."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:46",
                    "end": "00:21:40",
                    "text": " which it has learned to do from the thousands of real life cases that we've collected from the field. And within seconds, you get a risk score that we'll show you just in a bit. So what this risk score tells us is based on, just again, 30 seconds with this single LED ECG alone, how likely it is that this participant has had a silent heart attack in the past. The stress of the live demo is elevating my... I hope it hasn't provoked a heart attack. That would be unfortunate."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:41",
                    "end": "00:22:41",
                    "text": " Are there any other doctors in the audience? All right. I think we're ready to go. So what you're seeing now is the ECG was automatically uploaded into a web application and generating a low risk score. Thank goodness for that. Thank you so much. And thank you so much to our tech team here. Live demos are an unpleasant ask for the tech team. And I really appreciate your help. Thank you. OK."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:41",
                    "end": "00:23:08",
                    "text": " So as I mentioned, this is not just about ECGs. You could imagine a setup where a very small amount of hardware totaling under 20,000 US dollars comprised of an ECG, a retinal photo, a mobile chest x-ray unit, a handheld ultrasound device can actually be put together in a district health center or even in a little kiosk in a shopping center somewhere else."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:08",
                    "end": "00:23:30",
                    "text": " and could generate an enormous amount of raw device data that can feed into a number of diagnoses that are of critical importance to lots of causes of burden of disease in low and middle income countries, but in high income countries as well. So I think the scope and the scale of these kinds of diagnostic devices and approaches is really enormous."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:30",
                    "end": "00:23:60",
                    "text": " And I'll just mention one very personal application of this for me to give you a sense of what this can do in life. So my wife and I have twin daughters who are almost two years old. And the way that we found out that we were having twin daughters is because I, as part of my practice in emergency medicine, I own a handheld ultrasound device. So this costs $2,000. And it can generate pictures."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:60",
                    "end": "00:24:24",
                    "text": " like this of our two future daughters. So we knew that we were having twins well before most people know because we are lucky enough to have access to this technology. And I think democratizing and broadening access to both the hardware and the software components is really incredibly important as we seek to diagnose illnesses across the world. The last thing I'll say is that I think one of the really exciting parts of this work"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:24",
                    "end": "00:24:42",
                    "text": " is that it represents a fundamentally new way that we can develop AI algorithms in a sense that many people I think in well-resourced settings find counterintuitive. Because this is an innovation that starts here in India. Why does it start here?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:42",
                    "end": "00:25:04",
                    "text": " because there's very high volume of data at very low cost, but also because there's an enormous amount of human capital here. Engineering and AI talent, people who know how to build hardware and train AI products, there's the ability to move quickly, and at the same time, and especially thanks to the team at J-PAL, there's scope for very rigorous testing."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:04",
                    "end": "00:25:25",
                    "text": " This kind of innovation can scale seamlessly to the rest of the world. One interesting fact about our algorithm is that we pre-trained our AI model on electrocardiogram data from Sweden and the US. And that actually helped a lot for improving the model performance in India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:25",
                    "end": "00:25:47",
                    "text": " And that interoperability, if you will, of the ECG waveform is a really exciting part of building algorithms that run on this kind of physiological data. Because as we've tested algorithms that are trained in one place, for example, in Sweden, they generalize incredibly well to the US, to India, to Taiwan, to wherever we've tested them. Because..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:47",
                    "end": "00:25:57",
                    "text": " we do electrocardiograms the same way in all of those places, and because the heart is fundamentally wired in the same physiological way in all of those places."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:57",
                    "end": "00:26:23",
                    "text": " And so those algorithms can generalize and they can solve barriers to access that are widespread not just here in India but all over the world where people struggle to get access to the formal health system because of distance, because of poverty, because of discrimination, because of cultural or linguistic barriers. So I'm very excited about that and I would urge anyone who's working on this kind of work here with data is to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:23",
                    "end": "00:26:59",
                    "text": " protect and steward the secret sauce that you have here in India. Because it's really such a valuable resource, but not just to India, but to the rest of the world, to be able to develop AI products in a way that is safe, ethical, but also rigorous and fast. Thank you very much. Thank you so much, Siyad, for that insightful presentation. And may your risk of heart attack remain low for many years. Please rejoin the stage for a group photo with the rest of the session speakers."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:59",
                    "end": "00:27:27",
                    "text": " It's my pleasure to invite to the stage the panelists for the AI and Health Saving Lives at Scale session to join Ziad on stage. The panel will be moderated by Amy Barnes. Amy, please join us on stage. Amy is a policy manager at J-PAL Global, where she leads J-PAL's health sector. With all the promising applications of AI and health, Amy has increasingly been pushing to increase the available evidence on the real-world effects of AI solutions on both providers and their patients."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:27",
                    "end": "00:27:56",
                    "text": " A fun fact is that Amy started her time at J-PAL focused on policies to strengthen governance and public service delivery and to reduce crime and conflict. Now please join me in welcoming our panelists, starting with Shahed Alam. Shahed is the co-founder and co-CEO of Nura Health. Since its founding in 2014, Nura Health has been transferring health skills to family caregivers and patients. So far, Nura has trained over 43 million caregivers and patients across India, Bangladesh,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:56",
                    "end": "00:28:27",
                    "text": " Indonesia, and Nepal. And now Shahed is using AI to take Noor's mission to the next level. A fun fact about Shahed is that he earned his MD from Stanford. He's an associate faculty member at Ariadne Labs at Harvard Chan School of Public Health. And he's also been a TED speaker. Also on stage, Zamir Bray. Zamir is the deputy director of technology diffusion at the Gates Foundation. In this capacity, Zamir leads the foundation's strategic coordination on the equitable and safe use of AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:27",
                    "end": "00:28:55",
                    "text": " including work like the internal AI task force, AI related grant challenges, and external AI advisory efforts. A fun fact about Zamir is that he trained as a medical doctor, and he also earned an MBA. And on top of that, he has a PhD in health systems and innovation, and a master of law from the University of Cape Town. And he serves on the board of Groot Sure Hospital in Cape Town. Finally, Rob Sherman is the vice president of policy and deputy chief privacy officer at MEDA."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:55",
                    "end": "00:29:28",
                    "text": " where he works to set the governance and safeguards for how AI-powered products are designed and deployed responsibly. A fun fact about Rob is that before joining Meta, he was a lawyer advising major tech and digital tech companies on privacy and data security, and he was recognized by Chambers USA as one of the nation's leading media regulatory lawyers. We have very great people on stage with us today. Amy, over to you. Thank you so much, Audrey. The audience can send in questions by scanning this mentee code."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:28",
                    "end": "00:29:53",
                    "text": " And thank you, Amy. Awesome. First, just so excited to be here with all of you today. Hear your thoughts around the rapid evolution of AI in health systems, how it can be integrated, your thoughts on evaluation, of course. Audience members, please do submit your questions. But I'll just take the moderator privilege and start diving in with one. I want to start locally with Shahed. I know we're so far, but."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:53",
                    "end": "00:30:22",
                    "text": " Going to start with you and your organization, Nora Health, which does a lot of work, including in India, to train family members to advocate more for their loved ones in hospitals and to provide better care when they're leaving the hospitals to return home together. So Shahed, Nora is already working across multiple countries and already recognized for its impact. At this stage in Nora Health's growth, what problem are you trying to solve and what's the role of AI in your solution moving forward?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:22",
                    "end": "00:30:52",
                    "text": " Thanks so much for having us and also for that question. At Norah Health, again, we support the family caregiver of the patient with the knowledge, skills, and confidence they need to care for their loved ones. We work with governments and health systems in order to incorporate these programs that really support caregivers. The anchor of our work really starts in healthcare facilities when families are anxiously waiting by their loved one"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:52",
                    "end": "00:31:17",
                    "text": " And we provide them a skill-based training by a trusted health worker in person to engage them in their loved one's care. Now, once they go home, families may, they'll face some sort of a challenge. They may need a reminder or a nudge to remember what to do in order to care for their loved ones. And so that's where..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:17",
                    "end": "00:31:36",
                    "text": " For us, technology and our mobile messaging service steps in. So after an in-person training to continue to receive nudges and reminders for families once they go home. From the very beginning, as we've developed this service, we've not developed this as a chat bot because we wanted to always keep..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:36",
                    "end": "00:32:05",
                    "text": " that human connection between patients and families at home and a health worker at the back end as a critical and core part of what we do. But as we've scaled it, we have seen that that whole process can become very constrained. We went from response times that were pretty quick to response times that needed to take more than 10 hours. So that's where AI really stepped in for us."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:05",
                    "end": "00:32:28",
                    "text": " to be able to address this challenge of keeping high quality support to families once they're at home, but also ensuring that response times don't balloon to something like 10 hours. We had families who loved the service. They would reach out to us and they would ask our nurses to name their newborn, but they were getting increasingly frustrated with how long it took."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:28",
                    "end": "00:32:58",
                    "text": " So we sat down with our health workers, and that's really where it begins, is to define what are the challenges that people are facing, and for us those are health workers, and started to look at a lot of manual tasks that they were doing, looking up things in an FAQ database, annotating things manually, and then with them, alongside us, we started to co-develop an AI co-pilot that supports health workers in answering questions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:58",
                    "end": "00:33:18",
                    "text": " providing them context on the patient and family that they're supporting. And importantly, also tracking operational and other data so that we understand the quality of the responses in the system overall. And with that in place, we've been able to reduce those 10 hours down to 30 minutes."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:18",
                    "end": "00:33:49",
                    "text": " We've improved retention, so families, more than 30% of families keep coming back to us for more support. And importantly, the health workers are happier. They're getting to focus on what motivates them, which is to provide care versus to fill out a bunch of forms on an Excel sheet. And to follow up on that, I've heard rumors at least that you're planning a randomized evaluation later this year on this. Could you say more about why you're exploring an RCT now?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:49",
                    "end": "00:34:17",
                    "text": " given that you already have some evidence of Nora's impact. You are correct. For us, impact and evaluations is not a one-time thing. As we evolve, as the world changes, we need to continually update the evidence that we have and the knowledge that we have. The RCT is really critical for us because while we've done evaluations on the model, some of the product metrics, how it's affecting users,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:17",
                    "end": "00:34:31",
                    "text": " Ultimately, what we hold ourselves accountable to is having an impact on health outcomes, things like newborn morbidity and mortality. And the RCT is a way for us to actually, in a more definitive way, answer that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:31",
                    "end": "00:35:00",
                    "text": " In a simple way, it is to understand the impact on this incredibly important outcome, which is newborn mortality, and to understand also incrementally what does the AI-supported tool provide versus our standard program, which is the in-person initiative. And hopefully with that data, ultimately what we want to do is to have family caregiver support be the standard of care across health systems globally."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:00",
                    "end": "00:35:29",
                    "text": " RCTs and this evidence that we're generating will allow us to do is to provide information on what works, what doesn't, to governments so that they can use their scarce resources in supporting caregivers most effectively. Thank you so much. Zamir, I want to bring you in next and hear from you from the perspective of a funder who's thinking about technology diffusion a lot in health systems around the world. Because AI tools are evolving so rapidly, how should funders and implementers like Nora Health"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:29",
                    "end": "00:35:58",
                    "text": " design evaluations that can keep pace with that reality, while still producing evidence that's credible enough to drive adoption and policy. Thanks, Amy. I must just depart for a second, and I wonder if Ziad is still around, because when Ziad made his admission about being a clinician that missed the diagnosis, I realized that's why I was invited to this panel. I too missed the diagnosis, and I was just remembering that..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:58",
                    "end": "00:36:24",
                    "text": " It was actually my first trip to India 15 years ago when I came here, had a great time and went back home and had a classic swinging fever for six days and completely missed that it was typhoid fever and too ended up in the surgery room with a cholecystitis."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:24",
                    "end": "00:36:53",
                    "text": " To Ziad, this is a bit of a clinician anonymous admission session, so thanks for creating a safe space. But Amy, to get to your question and also just to pick up on where Shayid left off, I think the role of evaluations for decision makers for governments where resources are scarce is absolutely critical. And I can already tell from..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:53",
                    "end": "00:37:22",
                    "text": " the proceedings this week, that this will be one of the key messages we take back, which is we need to double down on evaluation, particularly giving decision-makers the right kind of evidence at the right time. I'm a big fan of RCTs. They have fundamentally changed the global health landscape. But there are two kind of issues we need to grapple with. The one is..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:22",
                    "end": "00:37:52",
                    "text": " the models change very fast. And I see some colleagues smiling with me because this is a very real issue. You know, I was doing a little bit of homework on, like, how long does this take? It takes us six months to design the RCT. Let's do the RCT for maybe 24 months and then take 18 months to publish the data. That means that if we all left here today and we started designing, Shahid..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:52",
                    "end": "00:38:20",
                    "text": " maybe 2029, we have some good results that you can put out there, right? I'm not saying that that's what we should do. The other reality is that policymakers, like in India, like in many countries we work in as the Gates Foundation, are really hoping that AI brings new promise, brings new tools to frontline workers that it's never been able to do before. And so the policy windows"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:20",
                    "end": "00:38:46",
                    "text": " the policy cycles don't track to publication cycles. Policy decisions are often made because of resource constraints, or there's an outbreak, or there's some political economy issue that drives the decision. And so being receptive to that means that we're going to have to challenge ourselves on conventional models of doing..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:46",
                    "end": "00:39:16",
                    "text": " RCTs or long trial designs. That shouldn't mean that we compromise the evidence generation. So just to be clear, we have to have good evidence. But in the early phases, we can start thinking about, does this give us a safety signal? Can we do something in short step, in three to six months that says, this looks like this could be safe, could be interesting?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:16",
                    "end": "00:39:46",
                    "text": " and very much to what the previous speaker said. Maybe those are short, inexpensive, quick experiments that we can run. Then we can move into kind of pragmatic trials, adaptive trial designs that don't take us two or three years, but give us the kind of evidence around adoption. And this is really key. We've seen lots of models perform really well in labs and test stops."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:46",
                    "end": "00:40:17",
                    "text": " they beat every single benchmark. And there's no doubt that most models today are better than clinicians. Those results don't hold up when you hit the field. Why? There's some fundamental issues. You know, one trial we saw was that the clinicians actually didn't trust the AI. So adoptions were stuck at about four, five, six percent until"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:17",
                    "end": "00:40:47",
                    "text": " someone actually sat with the clinicians, went through their own clinical cases and their mistakes, and showed them that somebody in the room, there were 100 clinicians in a trial in Kenya, and one of them was making the mistakes. And those mistakes were significant and catastrophic, that they then sat around the table and said, hang on, there's something we need to embrace about the technology. And then we saw adoption surge to upwards of 60%."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:47",
                    "end": "00:41:16",
                    "text": " still not at 100%. So I think that how we think about evidence is what's the right methodology for the right time. And again, I want to just applaud the Indian government and the India AI mission for proactively recognizing that evaluation evidence will need to be a key part of the AI strategy for the country, setting up these centers of excellence on health."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:16",
                    "end": "00:41:44",
                    "text": " education, agriculture, et cetera. And then bringing really strong partners, like J-PAL, who has the experience, but also locally, like ICMR and many other folks who can kind of put our heads together, and I'm sure folks in the room, about what are the best ways to evaluate AI in a period where if we were stuck on one method that took us that long, we'd likely..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:44",
                    "end": "00:42:03",
                    "text": " work ourselves into irrelevance and increase the inequality of how AI plays out. Thank you so much for sharing that perspective, Zameer. Rob, I want to bring you in next from a global governance vantage point as we're thinking about what's needed to really drive adoption and policy and impact."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:03",
                    "end": "00:42:32",
                    "text": " What are the biggest privacy and governance mistakes that you see when health AI projects are scaling? And how can programs avoid them without stalling innovation? So thanks for the question. I mean, I think it's really an exciting opportunity and an exciting moment where we are sitting here, where the technology that is being built is capable of solving lots of different problems. We've heard about a bunch of them from this panel, from the speakers already."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:32",
                    "end": "00:43:01",
                    "text": " from preventing heart attacks, to diagnosing anemia, to giving personalized health guidance to people who otherwise wouldn't be served by the traditional healthcare system. So these are really exciting opportunities. Before I answer the question about privacy and security, I wanna say, I was just sitting here listening to this conversation and thinking back to the past couple of global AI summits that we've had. Four years ago, the first one was in the UK."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:01",
                    "end": "00:43:24",
                    "text": " And I think I wouldn't have been able to say the same thing four years ago, that we build general purpose technology, like the kind of technology that my company builds, and then people use it for all of these compelling use cases that actually enable health delivery to scale. And to enable the kind of health delivery that would previously have been limited to."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:24",
                    "end": "00:43:34",
                    "text": " more wealthy nations to more limited populations is something that everyone can have. And I actually think that's a really exciting opportunity right now. So."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:34",
                    "end": "00:44:03",
                    "text": " What are the big mistakes? The first one is just treating governance as a gate rather than a guide. I think very often you get into this mode where people say, hey, we're really worried about X issue. We're worried about privacy. I spent my career thinking about privacy. I worry about privacy a lot too. But I think it's important to realize that the fact that we are worried about privacy doesn't mean that we shouldn't be able to use data in novel ways in order to..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:03",
                    "end": "00:44:32",
                    "text": " deliver really positive health outcomes. We actually did a study with Stanford University on something called deliberative democracy, basically looking at broadly representative groups of people in the US and in India and asking them for their attitudes about frontier AI. And one of the things that we found is here in India, 90 plus percent of people said that they want AI to be used for cutting edge cures and that they're willing to contribute their data in privacy preserving ways to have that happen."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:32",
                    "end": "00:44:40",
                    "text": " As a starting point, it's important to know that if you're not..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:40",
                    "end": "00:45:08",
                    "text": " If you're stopping the development or stopping deployment because you're worried about those issues, rather than trying to build in a way that is privacy safe and that is secure, you're making a mistake. The other big thing, which I think we also heard, and I think you talked about earlier, is the idea of human in a loop. The traditional understanding is always, humans are the gold standard. We should make sure that we have humans in the loop and making decisions at all times. That has two consequences."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:08",
                    "end": "00:45:13",
                    "text": " You talked about mistakes, and you talked about people not getting access to care because they're bottlenecks."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:13",
                    "end": "00:45:39",
                    "text": " And so I've come to think about this problem, especially as we're getting to a point where the AI is potentially more capable than humans, faster than humans, and more reliable than humans. Thinking about humans as the architect, we should be driving the way that the technology should be used and thinking about AI as a tool that can help us scale our impact so that we don't run into this case where people who could be helped by the technology aren't receiving it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:40",
                    "end": "00:46:11",
                    "text": " And Shahed or Zamir, do you have any reactions to Rob's comments here? Rob, you remind me of a real encounter I had speaking to a skeptic, head of neurosurgery at a very big hospital. I won't mention the name. He looked at me and he said, this AI thing is a complete waste of time. And it's not going to change anything in my practice. And it was in South Africa."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:12",
                    "end": "00:46:41",
                    "text": " And he said, the only reason I'll give you five minutes is because you are my student at one point in time. And put me on the spot and said, there's the CT scan, now show me. And I was like, I didn't come and do a demo, I didn't have anything special in my pocket, I didn't have a trained model, you know, and I took out the general purpose LLM, I won't say the name, took an image, it's multimodal, you know which one, maybe."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:43",
                    "end": "00:47:18",
                    "text": " And I turned the phone around, and he looked me dead in the eye and he said, hang on, actually, this is crazy, because he then told me the clinical case. This was a case, a very real case, of a 12 year old child who was admitted to a district hospital, who the clinician on duty was late in the morning, I don't know, looked at the child and said,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:18",
                    "end": "00:47:50",
                    "text": " Child's got raised in the cranial pressure of unknown cause and for whatever reason decided to do a lumbar puncture Clinicians in the audience are gasping You know what happened next? Unfortunately the child cone by the morning the child did not survive and When he looked at the LLM result it said to him Very clearly. It was the first time I saw bold and capital letters do not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:50",
                    "end": "00:48:22",
                    "text": " lumbar puncture, whatever you do. That clinician turned around and said to me, this is something I've never seen before. And just last week, he emailed me to say he submitted his first publication to the New England Journal of Medicine on the power of AI for African communities. And that's where I think we're gonna get the change, is when clinicians start to see the evidence and trust that this will make a change for their patients."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:22",
                    "end": "00:48:44",
                    "text": " in a kind of a real life situation. Anything that you would briefly add in 30 seconds there? Before we get to one last question, I just want to say that this discussion has been really exciting for me personally, because while I can't say too much yet, J-PAL is going to very soon be..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:44",
                    "end": "00:49:09",
                    "text": " really massively expanding our work through a new initiative to support research on AI and health, specifically in low and middle income countries. And that's going to be formally announced on the 20th at the summit. So stay tuned for additional details there. But I just want to flag that with that new initiative in mind, we would be really eager to reconnect with the researchers and implementers and policymakers in the room here today."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:09",
                    "end": "00:49:36",
                    "text": " after the announcement on the 20th to speak more about how we can build the evidence base on health and AI together and make those new evaluations possible. So with that in mind, as J-PAL is looking to support even more research on AI and how to integrate promising tools into the health system, I want to hear from each of you one more time. So in one to two minutes each, based on your experience,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:36",
                    "end": "00:50:03",
                    "text": " What differentiates AI applications that succeed from those that fail? That could be data availability, regulation, evaluation, other factors. And what advice would you give to implementers who want to build AI into programs iteratively, while also staying grounded and learning what works? So I'll start with Rob first maybe, and then Shahed and Zamir, feel free to bounce off of that. Sure."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:03",
                    "end": "00:50:29",
                    "text": " So I think that, you know, I talked earlier about the downside risk of not deploying. And I think that, you know, by thinking about things in binary terms, you often run into this case where you leave opportunity on the table. And I think the projects that are successful are ones that have a dual focus. They are thinking about the way to, the upside. Who are they helping?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:29",
                    "end": "00:50:58",
                    "text": " What is the specific mechanism by which they're going to deliver that? They're thinking about governance up front. They're making sure that they're identifying the downside risks and they're building them in. So as an example, when we do our data modeling, we look at anonymization technology. So we take data and we build the data sets in ways that protect the anonymity of the people that are in it. So we don't have problems later on. And then thinking about access to data, this is something that came up earlier too."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:58",
                    "end": "00:51:20",
                    "text": " making sure that the data that's being used is broadly representative of the people that we're trying to serve. I think if you build a project and you build a model based on data from people in the US, you're going to end up with something that serves people in the US and doesn't serve people in the rest of the world. And so I think thinking about both a very clear targeted upside."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:20",
                    "end": "00:51:48",
                    "text": " very clear mechanism of action, very clear governance for how to build it responsibly, and broadly representative data so that the technology and the models that you're building can serve as many people as possible. I think those are the key attributes. Shahed, maybe over to you. Sure, yeah, and thank you. Building on that, maybe would share three things. For implementers who are thinking about incorporating AI,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:48",
                    "end": "00:52:18",
                    "text": " First would be starting with people, not the technology. For us that meant health workers understanding their needs, their challenges, their motivations as professionals deeply, and then seeing how this tool can unlock potential. Again, for us that was about unlocking moments that they provide care. Second, it was really important for us to get very precise with our definitions. What do we even mean by good care?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:18",
                    "end": "00:52:43",
                    "text": " what's in the bounds of the service, what's out of the bounds of the service. Both of those things helped essentially provide a north star for things like when we developed out the knowledge base, built in guardrails for our prompting. And again, do that in a way where we iteratively come back to it. And finally, especially because we had humans in the loop,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:43",
                    "end": "00:53:10",
                    "text": " putting that product or that tool out there in a small geography for us to learn and understand and then iterate on. We learned so much more when we actually started to communicate with families and see what were the challenges that our health workers were facing when we put it out there. So again, of course, do that in a safe way. For us, it was keeping people in the loop, health workers specifically."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:10",
                    "end": "00:53:41",
                    "text": " and monitoring all the conversations, but what we learned was tremendous in those initial days. Thanks, Amy. I think a couple of things. I think we're going to hear this week lots about models and the cool tricks they're going to do and so on. I think it's a 10% problem. I think models, the accuracy will be 10% of what we need."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:41",
                    "end": "00:54:11",
                    "text": " I think there's a 20% on how does this actually integrate, like what sort of infrastructure do you need? And then I think 70% of this is gonna come down to change management and how we engage people and take people along on the journey. And I think if we neglect that, I think this is gonna be an uphill battle for some time. I also think that as we think about evidence and adoption, there's a very real cost issue."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:11",
                    "end": "00:54:38",
                    "text": " One of the studies we did in partnership with Bath over here showed that the clinician assist consult actually only costs four US cents and the accuracy was up to 97%. That may be expensive when you expand it over thousands of users, but actually the cost of compute will continue to come down. It's 240 times less than what it was two and a half years ago."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:38",
                    "end": "00:55:00",
                    "text": " So you can easily start to see that this becomes scalable and sustainable. I think that one of the biggest game changers, if I was forced to bet, I don't bet, but if I was forced to bet on what I think could be the biggest game changer for development on AI, I think it's personalized health coaches."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:00",
                    "end": "00:55:25",
                    "text": " in the pocket of every single individual that can speak in their language, that has context and memory over time, that eventually doesn't only serve health needs, but is your coach for education, is your coach for financial, but just a coach that you can trust with decision-making over time."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:25",
                    "end": "00:55:58",
                    "text": " Thank you so much, Samir. That's a really exciting vision to lay out, and I think a good way to close. It was so great from hearing from each of you. I think this was just a great way to highlight the need for more evidence, the excitement around AI and health, and just really appreciate you taking the time to be here, and all of you for attending. Thank you. Thank you so much to all of our AI and health speakers. It's really just amazing to hear their perspectives from every type of organization from all around the world. I hope you're all learning a lot."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:58",
                    "end": "00:56:07",
                    "text": " We'll take a short break and we'll rejoin at 11.40. Please try to exit from the right side of the stage. Thank you very much."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6994425e25c9f4e0d7e706d4",
    "projectId": "446259",
    "projectName": "From Insights to Action for Resilient_ High-Performance Data Centres - French",
    "projectDescription": "From Insights to Action for Resilient_ High-Performance Data Centres - French",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T10:26:42.563Z",
    "projectUpdatedDate": "2026-02-17T10:26:42.563Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T10:26:42.563Z",
    "projectDeleted": false,
    "projectDataLength": 3,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "From Insights to Action for Resilient_ High-Performance Data Centres.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "From Insights to Action for Resilient_ High-Performance Data Centres.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "French"
},
{
    "_id": "6994408825c9f4e0d7e70670",
    "projectId": "212694",
    "projectName": "From Insights to Action for Resilient_ High-Performance Data Centres - Hindi",
    "projectDescription": "From Insights to Action for Resilient_ High-Performance Data Centres - Hindi",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T10:18:53.404Z",
    "projectUpdatedDate": "2026-02-17T10:18:53.404Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T10:18:53.404Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "From Insights to Action for Resilient_ High-Performance Data Centres.mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771323517625.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:30",
                    "text": " What we all know is people think, oh, AI doesn't run on algorithms alone, right? It runs on electricity. And the objective of this session is simple but urgent. As AI scales globally, data centers are becoming foundations for economic growth and national resilience. And how AI infrastructure, grid integration, policy coordination can work together to build resilient economies. So today is about leadership."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:30",
                    "end": "00:00:51",
                    "text": " technical leadership, system leadership, and policy leadership. So let's explore how we can move forward. So to kick things off, I would like to invite Dan Oates on the stage, who is the Acting Deputy Director of the US Department of State Bureau of Cyberspace and Digital Policy, Office of Emerging and Critical Technology Policy."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:54",
                    "end": "00:01:15",
                    "text": " Good morning, everyone. And I want to thank the National Lab of the Rockies for organizing this important discussion. As stated, my name's Dan Oates. I'm the Deputy Director in the State Department's Bureau of Cyberspace and Digital Policy. Our team really advances an open, secure, interoperable tech ecosystem."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:15",
                    "end": "00:01:41",
                    "text": " to support economic growth, protect national security, and reflect democratic values. And today I'm going to help set some of the broader context for what are US AI priorities before we dive into the infrastructure questions that will drive this panel. So as the global leader in AI development and industry, the US is committed to sharing its technologies to enable countries around the world to realize the economic and societal benefits of AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:41",
                    "end": "00:02:03",
                    "text": " Last July President Trump released America's AI action plan that reflects a fundamental truth AI is a foundational technology that will define economic growth national security and global competitiveness for generations of in the future The action plan is built on three pillars first strengthening American AI innovation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:03",
                    "end": "00:02:30",
                    "text": " We're promoting US AI leadership to ensure Americans benefit from AI's transformative potential while avoiding overly precautionary regulation. AI systems must be built from the ground up with freedom of expression in mind. Second, we're building American AI infrastructure. President Trump signed an executive order to accelerate federal permitting of data center infrastructure. Recognizing the AI leadership requires the physical foundation to support it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:30",
                    "end": "00:02:54",
                    "text": " This is directly relevant to today's discussion around energy systems and data centers. Third, leading in international AI diplomacy and security. We launched the American AI exports program to promote full stack American technology packages. And we want to make it easy for friends and partners around the world to adopt US AI technologies that have been proven through performance, trust, and security."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:54",
                    "end": "00:03:24",
                    "text": " So AI infrastructure isn't just about compute. It's also about the entire ecosystem that makes AI possible, including secure, reliable energy systems. So as AI scales globally, data centers are becoming critical infrastructure and catalysts for energy system modernization. The AI race will determine not just our technological leadership, but also the balance of power and wealth of nations. It will determine whether democratic values are open markets."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:24",
                    "end": "00:03:40",
                    "text": " It will determine whether democratic values in open markets or authoritarian models guide AI's development worldwide. And the US wants to partner with countries around the world to build long-term capabilities and become AI creators, not just consumers, contributing to a vibrant global ecosystem."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:40",
                    "end": "00:04:19",
                    "text": " I think the technical challenges that you'll discuss today on this panel on high-efficiency data centers, intelligent power systems, energy optimization, they're not just engineering problems. They're really strategic imperatives that shape which countries can fully participate in the AI revolution. So I look forward to the panel, and I'll turn it over to the moderator. Thank you. Good morning. My name is Jacqueline Cochran, and I am Associate Lab Director of the National Lab of the Rockies, formerly known as NREL."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:19",
                    "end": "00:04:60",
                    "text": " And I'm really excited to bring up my panelists. Could you all please come up? AI infrastructure is scaling at an unprecedented pace, not only in my country, here, around the world. The question is no longer whether data centers will strain power systems, but whether leadership, technology, and policy can ensure that they strengthen power systems and economic resilience."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:60",
                    "end": "00:05:31",
                    "text": " So before we dive into questions, let me quickly introduce our esteemed panelists for today's sessions. We will start with my old friend, Mr. Abhishek Renton. He is from his CEO of BSES, Rajdhani Power, one of Delhi's primary electricity distribution companies, which brings deep experience in utility operations and system reforms. Next to him, another old friend, Mr. Samir Saxena."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:31",
                    "end": "00:05:56",
                    "text": " chairman and managing director of Grid India, responsible for national-level transmission planning and real-time system operations, where large-scale, high-density demands such as AI data centers meet grid stability and coordination. And then from the technology side, we have Somer Foukan, director and head of customer and partner engineering at Intel."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:56",
                    "end": "00:06:26",
                    "text": " working at the forefront of compute architecture, performance per watt innovation, and hardware software integration. And then representing the hyper scale perspective is Kartik Krishna, principal for energy projects at Amazon Web Services, where he focuses on integrating large scale AI data centers with local energy systems and long-term infrastructure planning. And finally, we have Mr. Deepesh Nanda."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:26",
                    "end": "00:06:54",
                    "text": " CEO of TCS's AI data center business, leading the development of next generation AI native hyperscale ready data center platforms, bringing a unique blend of power sector leadership and digital execution. Together, this group represents the full AI infrastructure ecosystem from grid planning and semiconductor design to hyperscale deployment and execution on the ground."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:54",
                    "end": "00:07:24",
                    "text": " All right. Let's begin looking at how AI data centers are already reshaping power systems today and how this growth is being experienced on the ground. So I think I'll sit. OK. Back. OK. We're going to start with you, Mr. Ranjan. VRPL manages one of India's most densely loaded urban distribution networks."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:24",
                    "end": "00:07:54",
                    "text": " How are AI data centers changing how you approach load forecasting, capital investment, and reliability planning at the city level? Thank you, Jacqueline. And thank you to the National Lab of Rockies for having me here, organizing this wonderful panel on a beautiful morning. See, as far as hyperscalers data centers are concerned, as of now, we do not see anything in Delhi."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:54",
                    "end": "00:08:24",
                    "text": " In Delhi NCR, we do. For example, in Noida, there are certain hyperscalers, which is maybe hardly 20, 25 kilometers away from South Delhi or East Delhi. They're coming up. Now, if we do not plan them according to resource planning in the sense like grid, where it is to be sited, it is going to be a strain on the grid. Because the kind of gigawatt requirements from each of the hyperscalers is very, very huge. One gigawatt."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:24",
                    "end": "00:08:55",
                    "text": " One hyperscaler, something like that. So planning, when we do forecasting of the demand and the planning as such, it's not only from the content perspective. It's not about power as such. Power can be because India's national grid is there, which is a one asynchronous grid. As you have already done the study, I think that was in 2017, that resource planning study that was done by then USAID. It's more about the network."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:55",
                    "end": "00:09:25",
                    "text": " the carriage part of it. Suppose you have in the heart of the demand center at 11 KV or 23 KV. In fact, 1 gigawatt will not be at 11 KV 33 KV. This will be at I mean 220 KV or something in that or 66 KV 220 KV and above. Therefore a it is not going to impact the medium and low voltage grid the Giga scale hyperscalers, but then the way in India how we organized that is central level transmission."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:25",
                    "end": "00:09:56",
                    "text": " operators like CTU grid India or maybe state level STUs in India at that level the planning needs to be there where how and where the grid substitution 220 by 66 or 220 by 33 has to be created substitution were dedicated to the same apart from this since the base load is very very you know stable in a hyperscaler the initial sources"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:56",
                    "end": "00:10:24",
                    "text": " So would have to be also long duration stable generation sources in us. You're going for nuclear India. We are just starting maybe it will take another 10 years 15 years to reach there. So for the time being there has to be a mix of approach. Are you going to dependent on grid completely which may not be possible? So apart from great you should have a local say long duration, maybe 6 hour 8 hour or 9 hour 10 hour battery storage."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:24",
                    "end": "00:10:53",
                    "text": " Some technology co-located with that center. Some captive generation as well, maybe renewable generation with solar and wind alike. Nuclear comes in later on. So in a nutshell, the planning has to be done through meticulous modeling exercise. It's not like you simply apply for connectivity, and it's not like a small connectivity. It's a big one. One gigawatt is a very big one."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:53",
                    "end": "00:11:21",
                    "text": " And it has to be comprehensive. It can't be piecemeal transactional. So just like the CEA creates a national plan for transmission for normal loads. I believe there would be a national plan for sighting of possible sites where the data centers can open and that has to be pre-decided. Otherwise, it may lead to chaos and unnecessary adding of wires here and there suboptimal optimization and therefore."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:21",
                    "end": "00:11:46",
                    "text": " the burden of tariff on the end consumers. Thank you so much. All right, next we'll turn to Mr. Saxena. From a national system operator perspective, how does the rise of AI-driven data centers change transmission planning and real-time grid balancing? And what new risks and opportunities does this create at the national level? Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:46",
                    "end": "00:12:15",
                    "text": " First of all, I would like to thank National Laboratory of the Rockies for inviting me and having me here today. Of course, this is a paradigm shift that we are going to see over the next few years. In fact, these are data centers we consider as very dense and intense loads that are coming, unlike bulk loads that we used to have at the distribution level. Now these would be at the transmission level because sub-transmission they will not be able to..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:15",
                    "end": "00:12:41",
                    "text": " cater this kind of requirement. In fact, the scale at which these are growing presently by 2030, the size of data center penetration would probably be something equivalent to that of one of the state demands that we are seeing, maybe 8 to 10 gigawatts. So even that size is going to be very significant. Now, this load manifests on the grid. It is variable. It is spiky."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:41",
                    "end": "00:13:10",
                    "text": " It has sharp ramps, and it is so much variable. And in fact, there are silent exists also from the system. Anything happens on the grid sides, the data centers prefer to quietly isolate themselves. They are inverter-based. So a big chunk of load like 1 to 1.5 gigawatt or maybe a couple of gigawatts simply walking out of the system quietly creates a disturbance kind of situation for the grid."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:10",
                    "end": "00:13:40",
                    "text": " And we need to plan for it. We need to be able to handle this in a more planned manner rather than having it in a random fashion. So from our operator's perspective, the planning for the infrastructure, firstly, it has to be very reliable. Robust transmission system has to be there. Along with the connectivity, like I think Abhishek has already mentioned, reactive power compliance, fault right through, all those kind of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:40",
                    "end": "00:14:06",
                    "text": " things must be put in place. That is one side. The other side is the resource adequacy side. Such large chunks of demand, they must be made obligated to meet the resource adequacy requirements not only in terms of the primary energy requirements but also for the reserves and balancing requirements. So here comes scheduling and forecasting and scheduling. I understand that the AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:06",
                    "end": "00:14:35",
                    "text": " especially during the prompt phase that we understand that it is highly unpredictable. So forecasting and scheduling would be a challenge for these entities. And here again, I believe that this is something that needs to be handled at the demand side itself. Because everything cannot be passed on to the grid. The grid cannot absorb all these risks. The other side I think is another important aspect is that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:35",
                    "end": "00:15:04",
                    "text": " Many jurisdictions across the world require these data centers to come mandatorily with some sort of generation to be able to handle these variations and ensure adequacy part of it. So I think in India also, our systems and the codes and standards need to evolve further to be able to cater to this kind of new requirement before this large chunk comes into the system. And of course, energy storage."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:04",
                    "end": "00:15:24",
                    "text": " And all need to be integrated with that. So I would stop at that. Thank you. I'm just going to add that our laboratory partnered with Grid India 10 years ago. And it's amazing how many comments now today are similar to back then, right? It's like, how do you plan for all these changes, for something large?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:24",
                    "end": "00:15:52",
                    "text": " that's variable, that's coming out of the grid, that has impacts to the grid. And it's a lot of the same tools today that we were using then, like what regulatory changes, what new balancing structures. But one thing that gets me really excited is that India has had so much change in the last 10 years. And so it's very exciting to think of how easily you're going to manage all these changes going forward."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:52",
                    "end": "00:16:22",
                    "text": " OK, so now Mr. Bhukan, as AI workloads explode, how is Intel improving performance per watt at both the chip and system level? And why does that matter for data centers operating at scale? So I think in the whole scheme of things, the silicon providers are the problem child. They are the ones that are creating the whole power issue, right? So I think from an Intel perspective, we are trying to address it at three levels. First, of course, is at the silicon level itself."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:22",
                    "end": "00:16:51",
                    "text": " So if you look at the 18A process, if you're reading about it, we have introduced three technologies. First one is called ribbon FET. So what's happening is that it allows the stacking of the gates closer to one another. So it essentially allows to switch transistors at lower voltages. That area uses power. That's kind of one. Second, we're also introducing something called power via. So typically in silicon design, for those who are interested, the complexity of power and data lines is very, very kind of interdependent."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:51",
                    "end": "00:17:21",
                    "text": " So with PowerVR, we are able to get the power in the back plane rather than the front plane. It is increasing the efficiency of the silicon by 15%. So that's kind of number two. The third piece that we talk about is something on packaging. We have something called forwarders that's reasonably popular right now across hyperscalers as well, if you see. What it does is it allows the stacking of memory and silicon. And if you really look at power generation in the silicon level, the IO between the memory and silicon is the biggest thing that comes in."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:21",
                    "end": "00:17:49",
                    "text": " So to be able to put it together, it reduces the IO distance, hence reduces the power. So these three things at a cynical level on 18A is something that we are introducing in the platform. Now, coming at a system level, we have always been a big proponent of heterogeneous AI. Every workload does not require the guzzlers of GPUs and accelerators we talk about. So Intel has something which we call as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:49",
                    "end": "00:18:12",
                    "text": " Appropriate silicon for the workload, including custom silicon. So we work with platform providers and hyperscalers to say, based on your workload, what silicon do you need? Including, for example, a lot of workloads can run on your PC client nowadays. If you're able to distribute that workload across the kind of silicon you want to use, that's the second thing on power you really want to do."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:12",
                    "end": "00:18:40",
                    "text": " The third thing, of course, is the workload itself. And AI is a problematic workload. But if you really look at this workload, what you really need is observability. So I'll give an example. We ran a workload end-to-end across CPUs and GPUs, as an example. The end-to-end latency when you run on a mix of CPUs and GPUs is about 15% or 15 seconds between"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:40",
                    "end": "00:19:09",
                    "text": " end-to-end, start from where you have the CPU and where you have the end output. But if you take a single task, which is very comparable to run it on a CPU versus GPU mix, heterogeneous or just a GPU mix. But if you want to just compare a single component saying, hey, what is my training workload on this? Obviously, the heterogeneity doesn't work. So the way we are looking at it is these three things. A, transistor level, make sure it counts. B, workshop at the system level, you're talking about."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:09",
                    "end": "00:19:32",
                    "text": " the designs or custom silicon to be able to meet the workloads and power requirements that you have. And third, work with the customers and partners so they understand this heterogeneity and be able to dissipate the power so that people treat us less as a problem child or more as a problem solver. Thank you. OK, Mr. Krishnan, as AWS scales AI data centers globally,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:32",
                    "end": "00:20:03",
                    "text": " How do you decide where and how to deploy AI data centers so that it aligns with local energy systems, regulatory environments, and long-term economic priorities? What has worked and what hasn't worked? First of all, thanks to NLR for this opportunity. Good discussion to have, right? And at the right opportunity. AI is just scaling in India. It's not scaled yet. A lot of these problems are in the future. So we are not facing it today. We might face it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:03",
                    "end": "00:20:32",
                    "text": " 18 months, 24 months, 36 months from now. And we still have an opportunity to correct it or to address it. Coming to what we as hyperscalers look for, three, four things. One is speed and certainty of power. So we have our load projections. And let's say we need power in 18 months. So we need certainty that the power can be delivered in the 18 months."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:32",
                    "end": "00:21:00",
                    "text": " and then some visibility on the ultimate capacity. So let's say the data center is 150 or 200 megawatt capacity. We obviously are not going to consume that power on day one. We will ramp to meet that demand, but we should have visibility of, hey, is there a network augmentation that needs to be done, whether that will get done on time. So that's one. Second is visibility to 100% renewable energy. So we want to run our data center sustainably."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:00",
                    "end": "00:21:29",
                    "text": " So visibility to 100% renewable is extremely important. Third that we look at is regulatory coherence. Like if a country or a state has a data center policy, there are electricity regulations, then there are open access policy, then there are certain permits and transmission planning. We see whether they are coherent. It should not happen that one is saying something else as compared to the other."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:29",
                    "end": "00:21:59",
                    "text": " The last but not the least is a certainty of power prices over a longer period. So we are, I mean, our decisions are based on a 15, 20 year projections. And we need reasonable certainty of power prices. This is not to say that the power prices have to remain flat for 15, 20 years, but we see that it has to be reasonably predictable. So that's about, that's four or five things that we will typically look at when we are setting up an AI dataset. Right, thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:59",
                    "end": "00:22:22",
                    "text": " Okay, finally, Mr. Nanda, at TCS, with platforms like Hypervault, you operate where AI strategy meets real world execution. From a systems integration perspective, where do you see the biggest gap between AI data center ambition and day-to-day operations? And how are integrated platforms helping to close that gap at scale?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:22",
                    "end": "00:22:52",
                    "text": " Thank you very much. Very happy to be here and with this eminent panel. I think a lot has been said here in terms of how infrastructure planning is a key element of getting into the AI data center part of the business. India, as was mentioned, is very early in its AI data center journey. We have 1.6 gigawatts of data centers for sure, but those are mostly to serve the needs of edge and cloud storage."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:52",
                    "end": "00:23:17",
                    "text": " Now, as you move towards AI, the big difference, as all of us know, is the dense requirement for power. So you're looking at anywhere between 10 to 12 X additional power that's needed. India, I think, is very well positioned at this point in time as compared to the rest of the world. So if I were to pick India versus the United States of America, our ability in the next 24 to 36 months to add capacity"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:17",
                    "end": "00:23:47",
                    "text": " in parallel with the AI deployment that will happen in India, more for inference load will probably be faster than what the US can do at the same time because the US is just maxed out right now in terms of power. Power both in terms of generation capacity as well as in terms of transmission and distribution capacity. So that's a big advantage for India. And that comes from the fact that the policies that the Indian government has embarked upon in the last 10-15 years, especially on renewable energy, is now coming to fore."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:47",
                    "end": "00:24:15",
                    "text": " So, we have the ability to produce renewable energy and then supply through open access to the zones which are data center popular. Now, most of the hyperscalers like, for example, Amazon would work around a concept called zones. Once they have established a position in a particular zone, they tend to expand in that same zone. And hence, policies that are written towards making sure that those zones are adequately planned well in advance."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:15",
                    "end": "00:24:39",
                    "text": " for the physical infrastructure. Because remember, the AI models are coming at a pace which is much faster than physical infrastructure in terms of setting up a power plant transmission lines can catch up. That gap is anywhere between 30 to 36 months, which is the issue right now in the US. In India, we have a head start because we have additional power available at this point in time."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:39",
                    "end": "00:25:08",
                    "text": " We should do well in terms of making sure that those data center zones, whether it's Navi Mumbai, for example, or whether it's Hyderabad, for example, in the next few quarters, there should be robust planning in place to make sure that those zones are adequately catered to. And then the concept of coming up with newer zones in the country. It'll be a factor, not just of power infrastructure, but also how can you network these sites."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:08",
                    "end": "00:25:31",
                    "text": " because ultimately all the data that gets stored in these sites will have to go to an end consumer outside of the site and hence that network connectivity is important. The other element in terms of energy systems that these large data centers will need is the element of cooling and hence to make sure that on day one the planning process is very well integrated."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:31",
                    "end": "00:25:57",
                    "text": " Also to ensure that the direct to chip cooling that will be required is given the quantum of heat that will get generated I think that's an element that we are working on very very closely. So what I expect is that in the next Few years India will probably add in the range of two to three gigawatt. We have adequate renewable energy capacity even our pumped hydro projects probably will come in line to coincide with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:57",
                    "end": "00:26:18",
                    "text": " the build out of the data centers and hence our ability to offer round the clock green electricity at a rate that's affordable for the term of the contract will be important. Now what the hyperscalers need to do is that the term of the contract that the hyperscalers need to contract for utilities in India has to be long term."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:18",
                    "end": "00:26:47",
                    "text": " Currently that's in the range of say seven years, but that's not adequate. You can't recover your capital cost within that period. Then there's an optionality of extension. So again from a policy point of view and hyperscalers get together, I think they need to underwrite utility capacity on power, water and networking for that tenure. Only then you will see that these business models will become more and more predictable. There will be a decent return for the investor and hence that's how the sector will grow."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:47",
                    "end": "00:27:17",
                    "text": " The gap between ambition that you mentioned and what's happening on the ground right now, I think it's all about timing and how efficiently we get our planning process in place and how intelligently we design the systems. Points were made on this panel on say for example, battery energy storage system. It's a good concept, but then the area required for a best system as compared to the density of power needed for a backup system in that area"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:17",
                    "end": "00:27:45",
                    "text": " are two different things. It's like 2x more when you use batteries. In the world of data centers, what I've quickly understood is that it's about high quality utilities. It is lesser about cost and it is about high quality utilities. Lastly, I would say is that the element of safety, for example, making sure that there are codes and standards that speak very well to all aspects of safety."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:45",
                    "end": "00:28:14",
                    "text": " not just of the data security, but the safety of constructions. These are gonna be massive sites. So if you look at a typical data center rack, maybe that of say Intel or any of the players, it's a massive piece of equipment. Each weighs around 5,000 pounds. If you look at the architecture on power that needs to support that, there is a UPS that backs it up. At the back of the UPS, there's a large diesel engine that backs it up."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:14",
                    "end": "00:28:43",
                    "text": " At the back of that, you have two grid independent lines on an N plus one configuration. So you have so much of electrical gear that goes in. And then it was mentioned earlier on this panel that the load curve is not consistent. You have spikes, you have harmonics. To cure that, you have additional electrical equipment that goes in. And hence, the aspect of safety when we write the policy or when we get together to think about what else is important."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:43",
                    "end": "00:29:12",
                    "text": " for this sector to be sustainable in the long term. I think it's all elements that go around safety. Then I would say it's about quality. Third is execution. Now, the expectation, there is this massive race that's going on, right? Everybody wants to overtake the other person. You have new models that come up, you know, every few weeks, right? Quote unquote, they drop a model. Now, if your rack density"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:12",
                    "end": "00:29:42",
                    "text": " as it's defined, if it changes say from 90 kilowatt to 160 kilowatt, physical infrastructure is not yet designed to adapt to that change. So there are big questions around obsolescence risk for physical hard infrastructure that will be put up and hence that's another important debate that forums like these need to discuss. Once we have safety in place, once we have gone into details of quality, once we have known about at what pace we need to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:42",
                    "end": "00:30:09",
                    "text": " execute and yet not make these large assets obsolete, then is the element of cost. Again, on that parameter, I think India is very well placed in terms of cost, because the cost per megawatt of build out in India currently is highly competitive. And the fact that this summit is happening in India, I'd just like to also make that case that this is the market for us to be in. Great. OK."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:11",
                    "end": "00:30:35",
                    "text": " So we have heard across the suite of panelists about some many of the changes that are happening that But now what what else needs to change? So I'm going to talk to each panelist about what they would like to see changed in terms of coordination Planning policy or something else that can help them keep pace with growth Okay, so we're going to go back in the same order"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:35",
                    "end": "00:30:59",
                    "text": " All right, Mr. Vandan, from an urban distribution utility perspective, what needs to change for distribution infrastructure to keep pace with rapidly growing AI data center demand? Thanks, Jacqueline. And Deepesh mentioned some of the good points about design of hyperscalers and then how integrated it would be to the network and also the business models. Now, in a distribution perspective,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:60",
                    "end": "00:31:28",
                    "text": " The distribution resource planning becomes very very important the resource adequacy planning was introduced in India couple of years ago. Maybe two three years ago, which was mainly about the supply side, but from the network side integrated resource planning and resource planning distribution resource planning is very very important that needs to change. So that is the amendment that needs to be done in the resource adequacy regulations the grid code where the transmission"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:28",
                    "end": "00:31:59",
                    "text": " plan should not be assumed as a copper plate. So that will give you idea about where to cite flexible resources. Like in case of hyperscalers, it would be more of chemistry change. What is that? So vanadium redox in the recently concluded India energy week there were vanadium redox 20 feet containers on display, which can be, you know, stacked up as a Lego Lego brick just like a Lego brick very compact unlike lithium."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:59",
                    "end": "00:32:24",
                    "text": " lithium ion chemistry. So and which are long duration storage. So the chemistry is completely different. The long duration may be 20 hours, 24 hours, more than that, the kind of base generation. So the planning needs to be changed. The grid code starts with the grid code. So just like we have connectivity regulations or GNA regulations, I think this should be classified as a mass load as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:24",
                    "end": "00:32:50",
                    "text": " My colleague was talking about Mr. Saxena. And in the connectivity regulations, they should be given a special category. And then, therefore, the planning would start from there. Otherwise, this may lead to chaos. And ultimately, the businesses might not be profitable. Or if it is loaded on the end customers, that might not be liked by the regulator as well. So instead of doing all those problems,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:50",
                    "end": "00:33:19",
                    "text": " A, the grid code, B, the regulations need to be changed, and C, the special status to these kind of load hyperscalers in the GNA regulations to be there. And then the systems are already there, and so it will be taken care of in the planning level. Technologies are there which can jump off, especially co-located generation and as balancing reserves, as Mr. Saxena was saying. And this could be co-optimized. Great. Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:19",
                    "end": "00:33:44",
                    "text": " Mr. Siksen, what about you? What changes would be really helpful for your coordination and planning? Yeah. So from an operator's perspective, even during the planning, we start right from the planning phase itself. The information flow, as well as very close coordination from the people who are trying to put in the data centers, right from the demand, the plan, exactly what they are trying to do."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:44",
                    "end": "00:34:12",
                    "text": " and what time frame and, of course, with the planning agencies as well as the transmission utilities and the system operators, it is extremely close coordination that is required. And the more important part of the planning exercise is that the models for studies, system studies, dynamic studies, all those must be very, very clearly mandated and they must be provided by the data centers so that they are able to do that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:12",
                    "end": "00:34:41",
                    "text": " And in fact, empowering the data centers also to be on the electrical side also to develop expertise on the electrical side. We saw this in the RE integration phase. In the RE integration part, we saw that the developers were simply kind of trying to do plug and play. They were not bothered about the technical requirements of the grid. They were not equipped also to be able to carry out the studies, to do fault analysis, and all those kind of things and take corrective action. They were largely dependent on the OEMs."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:41",
                    "end": "00:34:57",
                    "text": " So we would like to prefer that this kind of gap is not there with such large loads. It's not affordable, in fact. So that is one from the planning side. On the operation side, very close coordination with the load dispatch center for forecasting, scheduling."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:57",
                    "end": "00:35:20",
                    "text": " Balancing, providing reserves is also very important. Going forward, the data center should be able to provide load shifting, peak shaving, maybe reactive power support also as and when required. Ancillary services, demand reduction, that can also, they can participate. So all these services would become important because these are large chunks. And they are easily doable."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:20",
                    "end": "00:35:49",
                    "text": " You can directly have a signal sent out to them. So all those things are becoming very, very important. So this kind of close coordination must be developed. And they must come in as responsible citizens towards the grid as well. Great. Thank you. Great. OK. Mr. Phukan, what design or system choices across the stack from silicon to full systems most effectively translates performance per watt gains into real efficiency at data center scale?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:49",
                    "end": "00:36:17",
                    "text": " So I think it's like in any hyper-growth market, people often overlook the need for cost and sustainability versus design. Like I think my esteemed friends represent and say, now people are looking for quality versus cost. That's never true. But to be able to be sustainable, you really need to look at a design that's sustainable over time irrespective of the cost curve. A few things need to happen. I think if you look globally,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:17",
                    "end": "00:36:43",
                    "text": " and you or, you know, Mr. Saxena and et cetera would know more. There are, you know, data centers out there with a PUE of 1.06, 1.1, how have you. But if you look at the India AI mission that is set up, although the goal was set there, we were not able to achieve that to be able to do, and we kind of land up at 1.3, 1.4. So step one, I think data center providers need to relook at data center design."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:43",
                    "end": "00:37:09",
                    "text": " Okay, to be able to make sure, can we meet the PUE from a sustainability perspective? That's one, right? What does it require? It requires, as I said, the heterogeneity of platforms that can give you the power requirement that you're talking about. You really need to look at computation and other platforms across to be able to do that. Second thing, the IO is important, and I think Mr. Nanda touched upon it. I think photonics is the next thing."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:09",
                    "end": "00:37:39",
                    "text": " You should be able to use newer technology like optics to be able to reduce the power and increase the transmission, what have you, to be able to meet the cost. Last but not the least, I think as a country or as a set of people who are looking at sustainability in the long run, the future is always open, right? Supporting for open systems, open ecosystems, for organizations to work together to meet to one common infrastructure."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:39",
                    "end": "00:38:07",
                    "text": " is what has been sustainable in the technology industry. And AI is no different. So if I were to ask the conversation for scale that you're talking about, I think step one, I think the government needs to be firm about setting PUE standards so that it reduces Mr. Saxena's work in terms of making sure it's predictable. B, you really need to look at newer technology like optics to be able to meet the conversation of what kind of transmission power you're looking at, or transmission distance you're looking at. And third,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:07",
                    "end": "00:38:23",
                    "text": " I think there needs to be, and that India has done well. We haven't been very committed to open source, what have you, right? We really need to be very focused on open so that we know that it's sustainable in the long term versus getting locked into one architecture for another. Those are treating that, I would request it out before this to happen in the country."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:23",
                    "end": "00:38:52",
                    "text": " Great, thank you. OK, Mr. Krishnan, from AWS's experience scaling AI data centers worldwide, where do coordination challenges with utilities or regulators most often slow deployment, and what has proven most effective in overcoming those frictions? So I was talking to Saxena Ji backstage, and I think the fundamental difference between traditional"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:52",
                    "end": "00:39:19",
                    "text": " cloud services versus AI. I think there's a lot of confusion. There is no clarity. And I think we as hyperscalers should also do a better job of explaining that. Like today, like Saxenaji rightly said that AI data centers, the load behavior is slightly different as compared to traditional DCs. And it needs to be dealt differently. That's a fact. So that one thing I think it's a challenge in the sense that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:19",
                    "end": "00:39:47",
                    "text": " Both of us have to come together. We have to explain that there is this difference and how it needs to be treated separately and what our expectations are. Our expectations also are slightly different for an AI data center as compared to, let us say, a traditional DC. So that's one. Second is just the planning cycle. As a DC, our planning horizon is 15 to 20 years. The transmission planning cycle at the state level at least is five years. Regulatory certainty is for two years."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:47",
                    "end": "00:40:17",
                    "text": " The planning cycles are different, right? So at somewhere, the planning cycles has to merge. We have to be more long-term in terms of planning. Third is more policy. India has really good policies, to be honest, I think. Deepeshji also touched that. Abundance of renewable, and there are multiple models of securing renewable power. There is green tariffs. There is virtual power purchase agreements. We can buy recs from the market. There is physical PPA."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:17",
                    "end": "00:40:47",
                    "text": " So let's talk about physical PPAs. State-wise, it differs. In some states, there is, let us say, monthly banking. In some states, there is time-of-use banking. In some states, the banking is reduced only to the solar generation. So somewhere, this policy coherence across state should be there to help us to take better decisions. And last but not the least is there are policies in place in terms of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:47",
                    "end": "00:41:15",
                    "text": " incentives for data centers or certain waivers. But on ground, the implementation of these policy is a challenge. That's what we have seen that there could be an electricity duty waiver, or there could be an incentive on energy rebate. But on ground, actually getting that rebate, there are administrative challenges. There is a process in place, but it takes about 18 months or two years to actually get those incentives."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:15",
                    "end": "00:41:45",
                    "text": " So yeah, I think high level these would be the challenges that we see. Thank you. OK, Mr. Nanda, as AI data centers move toward higher power densities and liquid cooling, where do traditional data center models break down? And what needs to be rethought from the ground up? I think between the conventional data centers and the AI data centers, the design philosophy is going to be very different. We're going to get together on day one between the semiconductor"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:45",
                    "end": "00:42:15",
                    "text": " manufacturer between the hyperscaler and the data center designer. That probably did not happen in the erstwhile data centers. It was more looked like a building where you just go and put racks. These are going to be complex energy systems. So that's one big change that will happen. And that will happen just given the kind of power requirement. But I'll also add to the comments that we made earlier in terms of what else can India do. So India is at an advantageous position, as I mentioned earlier."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:15",
                    "end": "00:42:42",
                    "text": " Also given the fact that we have a single national grid, not that in many countries, which also means that there is a different power prize in different parts of those countries, which gives us an advantage. The Indian states currently are competing for investment on data centers. And that's super attractive. That is not the case internationally. And I hope it remains in India until we become"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:42",
                    "end": "00:43:09",
                    "text": " a force to reckon with. So on your question of the coordination on policy, I think if all of us speak in unison with the states who are currently vying for investment and make sure that those investments are translated into infrastructure projects in quick time, I think we would have served ourselves very very well. Third, just learning from the issues that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:09",
                    "end": "00:43:36",
                    "text": " data center build out in the United States of America are facing. And I quote that example because that's the country where maximum build out is happening currently. I think India has a huge manufacturing footprint right now. And a lot of that is underutilized, especially on the power side. So if you say, for example, turbines, if you take, you know, whether it's wind turbines or gas turbines or steam turbines and so on. In the U.S.,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:36",
                    "end": "00:44:05",
                    "text": " That's maxed out and there is no availability till 2030, 2031. Companies like GE or Siemens and so on are sold out completely. We have that capacity, same with electrical gear. So I think this is the moment for us and has been said by my fellow panelists, for us to quantify by state, by location, what is the demand and then kickstart the manufacturing process."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:05",
                    "end": "00:44:32",
                    "text": " so that we have enough gear in place as this exponential growth takes off. Finally, also sources of power. So we've talked about renewables. I'm also a firm believer that we will also need back of the meter power in the sense that these campuses will have to be self-sustained at some point in time. And to that extent, India should rethink the approach towards gas-based power generation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:32",
                    "end": "00:45:01",
                    "text": " We as a country, we have around 45 million tons of import capacity. And that's ideally located across the periphery of the country. We have a very good pipeline network. And hence, again, as hyperscalers look at the zones where they want to build out large data center campuses, it'll all go extremely well for India to kickstart the gas-based power for these plants. Because when it comes to affordability, I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:01",
                    "end": "00:45:25",
                    "text": " the data center operators and the hyperscalers, that may not be the issue as compared to conventional power. Because it was mentioned earlier, I think it was Abhishek who mentioned that ultimately my user should, my consumer, which is the consumer that they serve currently as a distributor, should not get impacted on the cost of electricity. That's another policy dimension that we need to work upon."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:26",
                    "end": "00:45:56",
                    "text": " We can have a whole other session on rate design, but for now let's just we'll have a lightning round and so for each panelist Tell me the theme of this is collaboration So of this panel so what? To address any of your biggest like disconnects today between where you are where you're going what others are doing What would one what is one collaboration that could close the gap fastest?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:56",
                    "end": "00:46:31",
                    "text": " And I will take any order. Lightning Ram, one. Adoption of heterogeneity or silicon diversity. I think that would be one key one. Right. Provision of reliability services by the data centers would definitely improve the system operation side. This is the design of the hyperscalers, the technological operation, which is flexible, grid integrable, just like you have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:31",
                    "end": "00:47:04",
                    "text": " Equalizers which can flex so that kind of technology I'd say hyperscale is taking a bigger bet on India Which I think they already are so I Think the good and the bad thing are the same AI data centers are the power density is very high like in the same footprint you're talking about Gigawatt capacity, I think the next five years will talk about gigawatt capacity why I'm saying it's good and bad it's bad because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:04",
                    "end": "00:47:29",
                    "text": " you would want the data centers to be spread out, right? So that the strain on the network is not that much, but that's not happening. Why I'm saying it is good is you know the pockets that you need to focus on. You know, like Deepesh rightly mentioned, we spoke about Navi Mumbai, or he spoke about Hyderabad. We exactly know the pockets where the data centers are coming up. And we should give special attention to those pockets when it comes to network augmentation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:29",
                    "end": "00:47:59",
                    "text": " much ahead in time, right? Network augmentation or like Saxenaji rightly said that there could be a problem of fault currents being injected or harmonics or voltage flicker. So the good thing is that you can arrest it at that point. Like if you put in proper checks and balances, these things can be arrested at the local grid itself and it will not pass on to the grid. So planning, I think we have to get better at planning, more localized planning. I don't see problem on the generation side."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:59",
                    "end": "00:48:34",
                    "text": " upstream transmission network is also robust. But when it comes to localized network, where in the data center clusters are forming, that is where I think we should collectively focus and plan better. Wonderful. And then on our last few minutes, do any of you have a question for one of the others on the panel? I'd like to take this opportunity to ask you on battery energy storage and the role that it will play."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:34",
                    "end": "00:48:55",
                    "text": " to stabilize grid power for large data center campuses. Any thoughts on that because that's a topic that, you know, we've been looking at. And I know that India has embarked on a journey of buying massive capacity of battery energy storage. How do you see that playing out?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:55",
                    "end": "00:49:10",
                    "text": " is definitely going to be playing a very, very important role, especially in the context of data centers. In fact, that is where you can actually absorb the ramps and maybe provide backup, provide reliability service, provide voltage support."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:10",
                    "end": "00:49:39",
                    "text": " All those kinds of reliability services can also come out from the battery energy storage systems. And in fact, it could offer some sort of shifting of demand as well. So all this, I think, should definitely come through in the form of local support to the data centers. So, Deepesh, just to add, this is another commercial opportunity for you, business opportunity. So you run data centers, right? So that is your main business. But then when you..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:39",
                    "end": "00:50:06",
                    "text": " add these resources, unblocking lithium chemistry. You can talk about any other chemistry which I attribute to maybe the vanadium redox, which is a flow chemistry. There could be many other sodium ion batteries, long duration ones, because this is a very stable base kind of load, data centers are. So if you have this, and then the regulations which are in place, ancillary services regulations, maybe amended, AS market base are getting introduced."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:07",
                    "end": "00:50:36",
                    "text": " Ancillary service on market is already there. And SRAS, that is Secondary Reserve Ancillary Service and Primary, they are going to be market based in short term of future, short term future. So that could be a way where you are making your lives easier. Also, there is a money spinning opportunity where in your valuation sheet, one of the line items would be how I'm flexing my data centers without impacting the reliable operation and making money out of it. But next one thing Jacqueline, I wanted to add here."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:37",
                    "end": "00:51:07",
                    "text": " As we are becoming more and more data-driven data dependent data is just like becoming our oxygen air. So the data centers definitely are going to be very very crucial and critical infrastructure as like in India or anywhere else. We are classifying the power infrastructure as very critical and they are very soft targets. Now the islanding schemes of power in in India, for example for cities like Mumbai Delhi, Hyderabad, Bengaluru."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:07",
                    "end": "00:51:36",
                    "text": " Where these data centers are likely to come up? they need to be brushed and Dusted now because if the islanding schemes are not properly operating In Mumbai there was a case I think in October of 2020 or 2021 where there was a failure because of some transmission glitch something happened and there was cascade failure the Hydro power in the Western guards were not able to sync up or you know supply power to Mumbai"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:36",
                    "end": "00:52:03",
                    "text": " So those things will be catastrophic. Why? The data centers, if they fail, then your critical life support, your police enforcement, maybe airports, hospitals, they're all going to die down. So therefore, the islanding schemes have to be dusted. And there, we need more technological inputs. It's no longer a power play, because it's going to impact your lives. Data centers are now heart of lives, actually."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:04",
                    "end": "00:52:39",
                    "text": " Well, thank you. This has been a fascinating discussion. And it is clear that all these choices that we're making today are going to be shaping our economy for decades to come. And I want to give a round of applause to our panel. OK. Others, do you want to come up? I'm going to invite now my colleague Adarsh Nagarajan to give some closing remarks. And then stay for the next panel. We're going to be going more into data centers."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:42",
                    "end": "00:53:12",
                    "text": " Thank you. Please, please stay back and ask your question, sir. This is a very important topic, but little later, once we end the session. Kuch nahi hoga? Is it? We'll figure it out, sir. Oh my god. OK, we'll figure it out, sir. Thank you. Thanks for your voice. So this whole journey of AI as a data center and the load growth is so rightly mentioned by Jacqueline."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:12",
                    "end": "00:53:41",
                    "text": " What India went through 10, 15 years ago, huge, huge change, huge change in demand. Here in this case, we are adding huge demand of gigawatt scale, adding a new state. Even a variability of five to 10% is hundreds of megawatts. It's huge. And as this committee so rightly pointed out, this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:41",
                    "end": "00:54:10",
                    "text": " will be successful only when there is a knowledge exchange and a handshake across all the stakeholders present here and the categories of stakeholders, the chip manufacturers, as model developers, the solutions, the DISCOMs, and the national grid operators, all put together when there is a harmony of knowledge. That's when it becomes highly successful, organized. And I believe the leaders here have already communicated that. And with that, I want to..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:10",
                    "end": "00:54:42",
                    "text": " take the next few minutes to provide the memento to all the panelists. Let's, let me invite Jacqueline, please, give the memento to Abhishek Ranjan, CEO of BRPL. The memento that we'll give. Oh, we have. Thank you. Thank you, sir. Let's have Mr. Sameer Sathinath, sir, please. You can come stand here. Thank you, sir."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:42",
                    "end": "00:55:13",
                    "text": " You can come in the front, a little better picture for everyone. Thank you. Let's have Mr. Krishnan. Sorry. Phukan, sorry for that. Mr. Krishnan, please. Thank you. And Mr. Nanda. Thanks a lot. Thanks a lot. Please stay back. We have an equally interesting yet more technical session following."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:13",
                    "end": "00:55:21",
                    "text": " Right after this same place will be replaced with more eminent guests as well. So please stay back and listen to that as well. Thank you."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Hindi"
},
{
    "_id": "6994377d25c9f4e0d7e704a6",
    "projectId": "611035",
    "projectName": "From Insights to Action for Resilient_ High-Performance Data Centres - English",
    "projectDescription": "From Insights to Action for Resilient_ High-Performance Data Centres - English",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T09:40:17.575Z",
    "projectUpdatedDate": "2026-02-17T09:40:17.575Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T09:40:17.575Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "From Insights to Action for Resilient_ High-Performance Data Centres.mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771321178411.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:30",
                    "text": " What we all know is people think, oh, AI doesn't run on algorithms alone, right? It runs on electricity. And the objective of this session is simple but urgent. As AI scales globally, data centers are becoming foundations for economic growth and national resilience. And how AI infrastructure, grid integration, policy coordination can work together to build resilient economies. So today is about leadership."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:30",
                    "end": "00:00:51",
                    "text": " technical leadership, system leadership, and policy leadership. So let's explore how we can move forward. So to kick things off, I would like to invite Dan Oates on the stage, who is the Acting Deputy Director of the US Department of State Bureau of Cyberspace and Digital Policy, Office of Emerging and Critical Technology Policy."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:54",
                    "end": "00:01:15",
                    "text": " Good morning, everyone. And I want to thank the National Lab of the Rockies for organizing this important discussion. As stated, my name's Dan Oates. I'm the Deputy Director in the State Department's Bureau of Cyberspace and Digital Policy. Our team really advances an open, secure, interoperable tech ecosystem."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:15",
                    "end": "00:01:41",
                    "text": " to support economic growth, protect national security, and reflect democratic values. And today I'm going to help set some of the broader context for what are US AI priorities before we dive into the infrastructure questions that will drive this panel. So as the global leader in AI development and industry, the US is committed to sharing its technologies to enable countries around the world to realize the economic and societal benefits of AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:41",
                    "end": "00:02:03",
                    "text": " Last July President Trump released America's AI action plan that reflects a fundamental truth AI is a foundational technology that will define economic growth national security and global competitiveness for generations of in the future The action plan is built on three pillars first strengthening American AI innovation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:03",
                    "end": "00:02:30",
                    "text": " We're promoting US AI leadership to ensure Americans benefit from AI's transformative potential while avoiding overly precautionary regulation. AI systems must be built from the ground up with freedom of expression in mind. Second, we're building American AI infrastructure. President Trump signed an executive order to accelerate federal permitting of data center infrastructure. Recognizing the AI leadership requires the physical foundation to support it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:30",
                    "end": "00:02:54",
                    "text": " This is directly relevant to today's discussion around energy systems and data centers. Third, leading in international AI diplomacy and security. We launched the American AI exports program to promote full stack American technology packages. And we want to make it easy for friends and partners around the world to adopt US AI technologies that have been proven through performance, trust, and security."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:54",
                    "end": "00:03:24",
                    "text": " So AI infrastructure isn't just about compute. It's also about the entire ecosystem that makes AI possible, including secure, reliable energy systems. So as AI scales globally, data centers are becoming critical infrastructure and catalysts for energy system modernization. The AI race will determine not just our technological leadership, but also the balance of power and wealth of nations. It will determine whether democratic values are open markets."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:24",
                    "end": "00:03:40",
                    "text": " It will determine whether democratic values in open markets or authoritarian models guide AI's development worldwide. And the US wants to partner with countries around the world to build long-term capabilities and become AI creators, not just consumers, contributing to a vibrant global ecosystem."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:40",
                    "end": "00:04:19",
                    "text": " I think the technical challenges that you'll discuss today on this panel on high-efficiency data centers, intelligent power systems, energy optimization, they're not just engineering problems. They're really strategic imperatives that shape which countries can fully participate in the AI revolution. So I look forward to the panel, and I'll turn it over to the moderator. Thank you. Good morning. My name is Jacqueline Cochran, and I am Associate Lab Director of the National Lab of the Rockies, formerly known as NREL."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:19",
                    "end": "00:04:60",
                    "text": " And I'm really excited to bring up my panelists. Could you all please come up? AI infrastructure is scaling at an unprecedented pace, not only in my country, here, around the world. The question is no longer whether data centers will strain power systems, but whether leadership, technology, and policy can ensure that they strengthen power systems and economic resilience."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:60",
                    "end": "00:05:31",
                    "text": " So before we dive into questions, let me quickly introduce our esteemed panelists for today's sessions. We will start with my old friend, Mr. Abhishek Renton. He is from his CEO of BSES, Rajdhani Power, one of Delhi's primary electricity distribution companies, which brings deep experience in utility operations and system reforms. Next to him, another old friend, Mr. Samir Saxena."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:31",
                    "end": "00:05:56",
                    "text": " chairman and managing director of Grid India, responsible for national-level transmission planning and real-time system operations, where large-scale, high-density demands such as AI data centers meet grid stability and coordination. And then from the technology side, we have Somer Foukan, director and head of customer and partner engineering at Intel."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:56",
                    "end": "00:06:26",
                    "text": " working at the forefront of compute architecture, performance per watt innovation, and hardware software integration. And then representing the hyper scale perspective is Kartik Krishna, principal for energy projects at Amazon Web Services, where he focuses on integrating large scale AI data centers with local energy systems and long-term infrastructure planning. And finally, we have Mr. Deepesh Nanda."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:26",
                    "end": "00:06:54",
                    "text": " CEO of TCS's AI data center business, leading the development of next generation AI native hyperscale ready data center platforms, bringing a unique blend of power sector leadership and digital execution. Together, this group represents the full AI infrastructure ecosystem from grid planning and semiconductor design to hyperscale deployment and execution on the ground."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:54",
                    "end": "00:07:24",
                    "text": " All right. Let's begin looking at how AI data centers are already reshaping power systems today and how this growth is being experienced on the ground. So I think I'll sit. OK. Back. OK. We're going to start with you, Mr. Ranjan. VRPL manages one of India's most densely loaded urban distribution networks."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:24",
                    "end": "00:07:54",
                    "text": " How are AI data centers changing how you approach load forecasting, capital investment, and reliability planning at the city level? Thank you, Jacqueline. And thank you to the National Lab of Rockies for having me here, organizing this wonderful panel on a beautiful morning. See, as far as hyperscalers data centers are concerned, as of now, we do not see anything in Delhi."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:54",
                    "end": "00:08:24",
                    "text": " In Delhi NCR, we do. For example, in Noida, there are certain hyperscalers, which is maybe hardly 20, 25 kilometers away from South Delhi or East Delhi. They're coming up. Now, if we do not plan them according to resource planning in the sense like grid, where it is to be sited, it is going to be a strain on the grid. Because the kind of gigawatt requirements from each of the hyperscalers is very, very huge. One gigawatt."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:24",
                    "end": "00:08:55",
                    "text": " One hyperscaler, something like that. So planning, when we do forecasting of the demand and the planning as such, it's not only from the content perspective. It's not about power as such. Power can be because India's national grid is there, which is a one asynchronous grid. As you have already done the study, I think that was in 2017, that resource planning study that was done by then USAID. It's more about the network."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:55",
                    "end": "00:09:25",
                    "text": " the carriage part of it. Suppose you have in the heart of the demand center at 11 KV or 23 KV. In fact, 1 gigawatt will not be at 11 KV 33 KV. This will be at I mean 220 KV or something in that or 66 KV 220 KV and above. Therefore a it is not going to impact the medium and low voltage grid the Giga scale hyperscalers, but then the way in India how we organized that is central level transmission."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:25",
                    "end": "00:09:56",
                    "text": " operators like CTU grid India or maybe state level STUs in India at that level the planning needs to be there where how and where the grid substitution 220 by 66 or 220 by 33 has to be created substitution were dedicated to the same apart from this since the base load is very very you know stable in a hyperscaler the initial sources"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:56",
                    "end": "00:10:24",
                    "text": " So would have to be also long duration stable generation sources in us. You're going for nuclear India. We are just starting maybe it will take another 10 years 15 years to reach there. So for the time being there has to be a mix of approach. Are you going to dependent on grid completely which may not be possible? So apart from great you should have a local say long duration, maybe 6 hour 8 hour or 9 hour 10 hour battery storage."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:24",
                    "end": "00:10:53",
                    "text": " Some technology co-located with that center. Some captive generation as well, maybe renewable generation with solar and wind alike. Nuclear comes in later on. So in a nutshell, the planning has to be done through meticulous modeling exercise. It's not like you simply apply for connectivity, and it's not like a small connectivity. It's a big one. One gigawatt is a very big one."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:53",
                    "end": "00:11:21",
                    "text": " And it has to be comprehensive. It can't be piecemeal transactional. So just like the CEA creates a national plan for transmission for normal loads. I believe there would be a national plan for sighting of possible sites where the data centers can open and that has to be pre-decided. Otherwise, it may lead to chaos and unnecessary adding of wires here and there suboptimal optimization and therefore."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:21",
                    "end": "00:11:46",
                    "text": " the burden of tariff on the end consumers. Thank you so much. All right, next we'll turn to Mr. Saxena. From a national system operator perspective, how does the rise of AI-driven data centers change transmission planning and real-time grid balancing? And what new risks and opportunities does this create at the national level? Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:46",
                    "end": "00:12:15",
                    "text": " First of all, I would like to thank National Laboratory of the Rockies for inviting me and having me here today. Of course, this is a paradigm shift that we are going to see over the next few years. In fact, these are data centers we consider as very dense and intense loads that are coming, unlike bulk loads that we used to have at the distribution level. Now these would be at the transmission level because sub-transmission they will not be able to..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:15",
                    "end": "00:12:41",
                    "text": " cater this kind of requirement. In fact, the scale at which these are growing presently by 2030, the size of data center penetration would probably be something equivalent to that of one of the state demands that we are seeing, maybe 8 to 10 gigawatts. So even that size is going to be very significant. Now, this load manifests on the grid. It is variable. It is spiky."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:41",
                    "end": "00:13:10",
                    "text": " It has sharp ramps, and it is so much variable. And in fact, there are silent exists also from the system. Anything happens on the grid sides, the data centers prefer to quietly isolate themselves. They are inverter-based. So a big chunk of load like 1 to 1.5 gigawatt or maybe a couple of gigawatts simply walking out of the system quietly creates a disturbance kind of situation for the grid."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:10",
                    "end": "00:13:40",
                    "text": " And we need to plan for it. We need to be able to handle this in a more planned manner rather than having it in a random fashion. So from our operator's perspective, the planning for the infrastructure, firstly, it has to be very reliable. Robust transmission system has to be there. Along with the connectivity, like I think Abhishek has already mentioned, reactive power compliance, fault right through, all those kind of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:40",
                    "end": "00:14:06",
                    "text": " things must be put in place. That is one side. The other side is the resource adequacy side. Such large chunks of demand, they must be made obligated to meet the resource adequacy requirements not only in terms of the primary energy requirements but also for the reserves and balancing requirements. So here comes scheduling and forecasting and scheduling. I understand that the AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:06",
                    "end": "00:14:35",
                    "text": " especially during the prompt phase that we understand that it is highly unpredictable. So forecasting and scheduling would be a challenge for these entities. And here again, I believe that this is something that needs to be handled at the demand side itself. Because everything cannot be passed on to the grid. The grid cannot absorb all these risks. The other side I think is another important aspect is that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:35",
                    "end": "00:15:04",
                    "text": " Many jurisdictions across the world require these data centers to come mandatorily with some sort of generation to be able to handle these variations and ensure adequacy part of it. So I think in India also, our systems and the codes and standards need to evolve further to be able to cater to this kind of new requirement before this large chunk comes into the system. And of course, energy storage."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:04",
                    "end": "00:15:24",
                    "text": " And all need to be integrated with that. So I would stop at that. Thank you. I'm just going to add that our laboratory partnered with Grid India 10 years ago. And it's amazing how many comments now today are similar to back then, right? It's like, how do you plan for all these changes, for something large?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:24",
                    "end": "00:15:52",
                    "text": " that's variable, that's coming out of the grid, that has impacts to the grid. And it's a lot of the same tools today that we were using then, like what regulatory changes, what new balancing structures. But one thing that gets me really excited is that India has had so much change in the last 10 years. And so it's very exciting to think of how easily you're going to manage all these changes going forward."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:52",
                    "end": "00:16:22",
                    "text": " OK, so now Mr. Bhukan, as AI workloads explode, how is Intel improving performance per watt at both the chip and system level? And why does that matter for data centers operating at scale? So I think in the whole scheme of things, the silicon providers are the problem child. They are the ones that are creating the whole power issue, right? So I think from an Intel perspective, we are trying to address it at three levels. First, of course, is at the silicon level itself."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:22",
                    "end": "00:16:51",
                    "text": " So if you look at the 18A process, if you're reading about it, we have introduced three technologies. First one is called ribbon FET. So what's happening is that it allows the stacking of the gates closer to one another. So it essentially allows to switch transistors at lower voltages. That area uses power. That's kind of one. Second, we're also introducing something called power via. So typically in silicon design, for those who are interested, the complexity of power and data lines is very, very kind of interdependent."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:51",
                    "end": "00:17:21",
                    "text": " So with PowerVR, we are able to get the power in the back plane rather than the front plane. It is increasing the efficiency of the silicon by 15%. So that's kind of number two. The third piece that we talk about is something on packaging. We have something called forwarders that's reasonably popular right now across hyperscalers as well, if you see. What it does is it allows the stacking of memory and silicon. And if you really look at power generation in the silicon level, the IO between the memory and silicon is the biggest thing that comes in."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:21",
                    "end": "00:17:49",
                    "text": " So to be able to put it together, it reduces the IO distance, hence reduces the power. So these three things at a cynical level on 18A is something that we are introducing in the platform. Now, coming at a system level, we have always been a big proponent of heterogeneous AI. Every workload does not require the guzzlers of GPUs and accelerators we talk about. So Intel has something which we call as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:49",
                    "end": "00:18:12",
                    "text": " Appropriate silicon for the workload, including custom silicon. So we work with platform providers and hyperscalers to say, based on your workload, what silicon do you need? Including, for example, a lot of workloads can run on your PC client nowadays. If you're able to distribute that workload across the kind of silicon you want to use, that's the second thing on power you really want to do."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:12",
                    "end": "00:18:40",
                    "text": " The third thing, of course, is the workload itself. And AI is a problematic workload. But if you really look at this workload, what you really need is observability. So I'll give an example. We ran a workload end-to-end across CPUs and GPUs, as an example. The end-to-end latency when you run on a mix of CPUs and GPUs is about 15% or 15 seconds between"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:40",
                    "end": "00:19:09",
                    "text": " end-to-end, start from where you have the CPU and where you have the end output. But if you take a single task, which is very comparable to run it on a CPU versus GPU mix, heterogeneous or just a GPU mix. But if you want to just compare a single component saying, hey, what is my training workload on this? Obviously, the heterogeneity doesn't work. So the way we are looking at it is these three things. A, transistor level, make sure it counts. B, workshop at the system level, you're talking about."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:09",
                    "end": "00:19:32",
                    "text": " the designs or custom silicon to be able to meet the workloads and power requirements that you have. And third, work with the customers and partners so they understand this heterogeneity and be able to dissipate the power so that people treat us less as a problem child or more as a problem solver. Thank you. OK, Mr. Krishnan, as AWS scales AI data centers globally,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:32",
                    "end": "00:20:03",
                    "text": " How do you decide where and how to deploy AI data centers so that it aligns with local energy systems, regulatory environments, and long-term economic priorities? What has worked and what hasn't worked? First of all, thanks to NLR for this opportunity. Good discussion to have, right? And at the right opportunity. AI is just scaling in India. It's not scaled yet. A lot of these problems are in the future. So we are not facing it today. We might face it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:03",
                    "end": "00:20:32",
                    "text": " 18 months, 24 months, 36 months from now. And we still have an opportunity to correct it or to address it. Coming to what we as hyperscalers look for, three, four things. One is speed and certainty of power. So we have our load projections. And let's say we need power in 18 months. So we need certainty that the power can be delivered in the 18 months."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:32",
                    "end": "00:21:00",
                    "text": " and then some visibility on the ultimate capacity. So let's say the data center is 150 or 200 megawatt capacity. We obviously are not going to consume that power on day one. We will ramp to meet that demand, but we should have visibility of, hey, is there a network augmentation that needs to be done, whether that will get done on time. So that's one. Second is visibility to 100% renewable energy. So we want to run our data center sustainably."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:00",
                    "end": "00:21:29",
                    "text": " So visibility to 100% renewable is extremely important. Third that we look at is regulatory coherence. Like if a country or a state has a data center policy, there are electricity regulations, then there are open access policy, then there are certain permits and transmission planning. We see whether they are coherent. It should not happen that one is saying something else as compared to the other."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:29",
                    "end": "00:21:59",
                    "text": " The last but not the least is a certainty of power prices over a longer period. So we are, I mean, our decisions are based on a 15, 20 year projections. And we need reasonable certainty of power prices. This is not to say that the power prices have to remain flat for 15, 20 years, but we see that it has to be reasonably predictable. So that's about, that's four or five things that we will typically look at when we are setting up an AI dataset. Right, thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:59",
                    "end": "00:22:22",
                    "text": " Okay, finally, Mr. Nanda, at TCS, with platforms like Hypervault, you operate where AI strategy meets real world execution. From a systems integration perspective, where do you see the biggest gap between AI data center ambition and day-to-day operations? And how are integrated platforms helping to close that gap at scale?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:22",
                    "end": "00:22:52",
                    "text": " Thank you very much. Very happy to be here and with this eminent panel. I think a lot has been said here in terms of how infrastructure planning is a key element of getting into the AI data center part of the business. India, as was mentioned, is very early in its AI data center journey. We have 1.6 gigawatts of data centers for sure, but those are mostly to serve the needs of edge and cloud storage."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:52",
                    "end": "00:23:17",
                    "text": " Now, as you move towards AI, the big difference, as all of us know, is the dense requirement for power. So you're looking at anywhere between 10 to 12 X additional power that's needed. India, I think, is very well positioned at this point in time as compared to the rest of the world. So if I were to pick India versus the United States of America, our ability in the next 24 to 36 months to add capacity"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:17",
                    "end": "00:23:47",
                    "text": " in parallel with the AI deployment that will happen in India, more for inference load will probably be faster than what the US can do at the same time because the US is just maxed out right now in terms of power. Power both in terms of generation capacity as well as in terms of transmission and distribution capacity. So that's a big advantage for India. And that comes from the fact that the policies that the Indian government has embarked upon in the last 10-15 years, especially on renewable energy, is now coming to fore."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:47",
                    "end": "00:24:15",
                    "text": " So, we have the ability to produce renewable energy and then supply through open access to the zones which are data center popular. Now, most of the hyperscalers like, for example, Amazon would work around a concept called zones. Once they have established a position in a particular zone, they tend to expand in that same zone. And hence, policies that are written towards making sure that those zones are adequately planned well in advance."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:15",
                    "end": "00:24:39",
                    "text": " for the physical infrastructure. Because remember, the AI models are coming at a pace which is much faster than physical infrastructure in terms of setting up a power plant transmission lines can catch up. That gap is anywhere between 30 to 36 months, which is the issue right now in the US. In India, we have a head start because we have additional power available at this point in time."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:39",
                    "end": "00:25:08",
                    "text": " We should do well in terms of making sure that those data center zones, whether it's Navi Mumbai, for example, or whether it's Hyderabad, for example, in the next few quarters, there should be robust planning in place to make sure that those zones are adequately catered to. And then the concept of coming up with newer zones in the country. It'll be a factor, not just of power infrastructure, but also how can you network these sites."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:08",
                    "end": "00:25:31",
                    "text": " because ultimately all the data that gets stored in these sites will have to go to an end consumer outside of the site and hence that network connectivity is important. The other element in terms of energy systems that these large data centers will need is the element of cooling and hence to make sure that on day one the planning process is very well integrated."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:31",
                    "end": "00:25:57",
                    "text": " Also to ensure that the direct to chip cooling that will be required is given the quantum of heat that will get generated I think that's an element that we are working on very very closely. So what I expect is that in the next Few years India will probably add in the range of two to three gigawatt. We have adequate renewable energy capacity even our pumped hydro projects probably will come in line to coincide with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:57",
                    "end": "00:26:18",
                    "text": " the build out of the data centers and hence our ability to offer round the clock green electricity at a rate that's affordable for the term of the contract will be important. Now what the hyperscalers need to do is that the term of the contract that the hyperscalers need to contract for utilities in India has to be long term."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:18",
                    "end": "00:26:47",
                    "text": " Currently that's in the range of say seven years, but that's not adequate. You can't recover your capital cost within that period. Then there's an optionality of extension. So again from a policy point of view and hyperscalers get together, I think they need to underwrite utility capacity on power, water and networking for that tenure. Only then you will see that these business models will become more and more predictable. There will be a decent return for the investor and hence that's how the sector will grow."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:47",
                    "end": "00:27:17",
                    "text": " The gap between ambition that you mentioned and what's happening on the ground right now, I think it's all about timing and how efficiently we get our planning process in place and how intelligently we design the systems. Points were made on this panel on say for example, battery energy storage system. It's a good concept, but then the area required for a best system as compared to the density of power needed for a backup system in that area"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:17",
                    "end": "00:27:45",
                    "text": " are two different things. It's like 2x more when you use batteries. In the world of data centers, what I've quickly understood is that it's about high quality utilities. It is lesser about cost and it is about high quality utilities. Lastly, I would say is that the element of safety, for example, making sure that there are codes and standards that speak very well to all aspects of safety."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:45",
                    "end": "00:28:14",
                    "text": " not just of the data security, but the safety of constructions. These are gonna be massive sites. So if you look at a typical data center rack, maybe that of say Intel or any of the players, it's a massive piece of equipment. Each weighs around 5,000 pounds. If you look at the architecture on power that needs to support that, there is a UPS that backs it up. At the back of the UPS, there's a large diesel engine that backs it up."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:14",
                    "end": "00:28:43",
                    "text": " At the back of that, you have two grid independent lines on an N plus one configuration. So you have so much of electrical gear that goes in. And then it was mentioned earlier on this panel that the load curve is not consistent. You have spikes, you have harmonics. To cure that, you have additional electrical equipment that goes in. And hence, the aspect of safety when we write the policy or when we get together to think about what else is important."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:43",
                    "end": "00:29:12",
                    "text": " for this sector to be sustainable in the long term. I think it's all elements that go around safety. Then I would say it's about quality. Third is execution. Now, the expectation, there is this massive race that's going on, right? Everybody wants to overtake the other person. You have new models that come up, you know, every few weeks, right? Quote unquote, they drop a model. Now, if your rack density"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:12",
                    "end": "00:29:42",
                    "text": " as it's defined, if it changes say from 90 kilowatt to 160 kilowatt, physical infrastructure is not yet designed to adapt to that change. So there are big questions around obsolescence risk for physical hard infrastructure that will be put up and hence that's another important debate that forums like these need to discuss. Once we have safety in place, once we have gone into details of quality, once we have known about at what pace we need to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:42",
                    "end": "00:30:09",
                    "text": " execute and yet not make these large assets obsolete, then is the element of cost. Again, on that parameter, I think India is very well placed in terms of cost, because the cost per megawatt of build out in India currently is highly competitive. And the fact that this summit is happening in India, I'd just like to also make that case that this is the market for us to be in. Great. OK."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:11",
                    "end": "00:30:35",
                    "text": " So we have heard across the suite of panelists about some many of the changes that are happening that But now what what else needs to change? So I'm going to talk to each panelist about what they would like to see changed in terms of coordination Planning policy or something else that can help them keep pace with growth Okay, so we're going to go back in the same order"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:35",
                    "end": "00:30:59",
                    "text": " All right, Mr. Vandan, from an urban distribution utility perspective, what needs to change for distribution infrastructure to keep pace with rapidly growing AI data center demand? Thanks, Jacqueline. And Deepesh mentioned some of the good points about design of hyperscalers and then how integrated it would be to the network and also the business models. Now, in a distribution perspective,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:60",
                    "end": "00:31:28",
                    "text": " The distribution resource planning becomes very very important the resource adequacy planning was introduced in India couple of years ago. Maybe two three years ago, which was mainly about the supply side, but from the network side integrated resource planning and resource planning distribution resource planning is very very important that needs to change. So that is the amendment that needs to be done in the resource adequacy regulations the grid code where the transmission"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:28",
                    "end": "00:31:59",
                    "text": " plan should not be assumed as a copper plate. So that will give you idea about where to cite flexible resources. Like in case of hyperscalers, it would be more of chemistry change. What is that? So vanadium redox in the recently concluded India energy week there were vanadium redox 20 feet containers on display, which can be, you know, stacked up as a Lego Lego brick just like a Lego brick very compact unlike lithium."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:59",
                    "end": "00:32:24",
                    "text": " lithium ion chemistry. So and which are long duration storage. So the chemistry is completely different. The long duration may be 20 hours, 24 hours, more than that, the kind of base generation. So the planning needs to be changed. The grid code starts with the grid code. So just like we have connectivity regulations or GNA regulations, I think this should be classified as a mass load as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:24",
                    "end": "00:32:50",
                    "text": " My colleague was talking about Mr. Saxena. And in the connectivity regulations, they should be given a special category. And then, therefore, the planning would start from there. Otherwise, this may lead to chaos. And ultimately, the businesses might not be profitable. Or if it is loaded on the end customers, that might not be liked by the regulator as well. So instead of doing all those problems,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:50",
                    "end": "00:33:19",
                    "text": " A, the grid code, B, the regulations need to be changed, and C, the special status to these kind of load hyperscalers in the GNA regulations to be there. And then the systems are already there, and so it will be taken care of in the planning level. Technologies are there which can jump off, especially co-located generation and as balancing reserves, as Mr. Saxena was saying. And this could be co-optimized. Great. Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:19",
                    "end": "00:33:44",
                    "text": " Mr. Siksen, what about you? What changes would be really helpful for your coordination and planning? Yeah. So from an operator's perspective, even during the planning, we start right from the planning phase itself. The information flow, as well as very close coordination from the people who are trying to put in the data centers, right from the demand, the plan, exactly what they are trying to do."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:44",
                    "end": "00:34:12",
                    "text": " and what time frame and, of course, with the planning agencies as well as the transmission utilities and the system operators, it is extremely close coordination that is required. And the more important part of the planning exercise is that the models for studies, system studies, dynamic studies, all those must be very, very clearly mandated and they must be provided by the data centers so that they are able to do that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:12",
                    "end": "00:34:41",
                    "text": " And in fact, empowering the data centers also to be on the electrical side also to develop expertise on the electrical side. We saw this in the RE integration phase. In the RE integration part, we saw that the developers were simply kind of trying to do plug and play. They were not bothered about the technical requirements of the grid. They were not equipped also to be able to carry out the studies, to do fault analysis, and all those kind of things and take corrective action. They were largely dependent on the OEMs."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:41",
                    "end": "00:34:57",
                    "text": " So we would like to prefer that this kind of gap is not there with such large loads. It's not affordable, in fact. So that is one from the planning side. On the operation side, very close coordination with the load dispatch center for forecasting, scheduling."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:57",
                    "end": "00:35:20",
                    "text": " Balancing, providing reserves is also very important. Going forward, the data center should be able to provide load shifting, peak shaving, maybe reactive power support also as and when required. Ancillary services, demand reduction, that can also, they can participate. So all these services would become important because these are large chunks. And they are easily doable."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:20",
                    "end": "00:35:49",
                    "text": " You can directly have a signal sent out to them. So all those things are becoming very, very important. So this kind of close coordination must be developed. And they must come in as responsible citizens towards the grid as well. Great. Thank you. Great. OK. Mr. Phukan, what design or system choices across the stack from silicon to full systems most effectively translates performance per watt gains into real efficiency at data center scale?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:49",
                    "end": "00:36:17",
                    "text": " So I think it's like in any hyper-growth market, people often overlook the need for cost and sustainability versus design. Like I think my esteemed friends represent and say, now people are looking for quality versus cost. That's never true. But to be able to be sustainable, you really need to look at a design that's sustainable over time irrespective of the cost curve. A few things need to happen. I think if you look globally,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:17",
                    "end": "00:36:43",
                    "text": " and you or, you know, Mr. Saxena and et cetera would know more. There are, you know, data centers out there with a PUE of 1.06, 1.1, how have you. But if you look at the India AI mission that is set up, although the goal was set there, we were not able to achieve that to be able to do, and we kind of land up at 1.3, 1.4. So step one, I think data center providers need to relook at data center design."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:43",
                    "end": "00:37:09",
                    "text": " Okay, to be able to make sure, can we meet the PUE from a sustainability perspective? That's one, right? What does it require? It requires, as I said, the heterogeneity of platforms that can give you the power requirement that you're talking about. You really need to look at computation and other platforms across to be able to do that. Second thing, the IO is important, and I think Mr. Nanda touched upon it. I think photonics is the next thing."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:09",
                    "end": "00:37:39",
                    "text": " You should be able to use newer technology like optics to be able to reduce the power and increase the transmission, what have you, to be able to meet the cost. Last but not the least, I think as a country or as a set of people who are looking at sustainability in the long run, the future is always open, right? Supporting for open systems, open ecosystems, for organizations to work together to meet to one common infrastructure."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:39",
                    "end": "00:38:07",
                    "text": " is what has been sustainable in the technology industry. And AI is no different. So if I were to ask the conversation for scale that you're talking about, I think step one, I think the government needs to be firm about setting PUE standards so that it reduces Mr. Saxena's work in terms of making sure it's predictable. B, you really need to look at newer technology like optics to be able to meet the conversation of what kind of transmission power you're looking at, or transmission distance you're looking at. And third,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:07",
                    "end": "00:38:23",
                    "text": " I think there needs to be, and that India has done well. We haven't been very committed to open source, what have you, right? We really need to be very focused on open so that we know that it's sustainable in the long term versus getting locked into one architecture for another. Those are treating that, I would request it out before this to happen in the country."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:23",
                    "end": "00:38:52",
                    "text": " Great, thank you. OK, Mr. Krishnan, from AWS's experience scaling AI data centers worldwide, where do coordination challenges with utilities or regulators most often slow deployment, and what has proven most effective in overcoming those frictions? So I was talking to Saxena Ji backstage, and I think the fundamental difference between traditional"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:52",
                    "end": "00:39:19",
                    "text": " cloud services versus AI. I think there's a lot of confusion. There is no clarity. And I think we as hyperscalers should also do a better job of explaining that. Like today, like Saxenaji rightly said that AI data centers, the load behavior is slightly different as compared to traditional DCs. And it needs to be dealt differently. That's a fact. So that one thing I think it's a challenge in the sense that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:19",
                    "end": "00:39:47",
                    "text": " Both of us have to come together. We have to explain that there is this difference and how it needs to be treated separately and what our expectations are. Our expectations also are slightly different for an AI data center as compared to, let us say, a traditional DC. So that's one. Second is just the planning cycle. As a DC, our planning horizon is 15 to 20 years. The transmission planning cycle at the state level at least is five years. Regulatory certainty is for two years."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:47",
                    "end": "00:40:17",
                    "text": " The planning cycles are different, right? So at somewhere, the planning cycles has to merge. We have to be more long-term in terms of planning. Third is more policy. India has really good policies, to be honest, I think. Deepeshji also touched that. Abundance of renewable, and there are multiple models of securing renewable power. There is green tariffs. There is virtual power purchase agreements. We can buy recs from the market. There is physical PPA."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:17",
                    "end": "00:40:47",
                    "text": " So let's talk about physical PPAs. State-wise, it differs. In some states, there is, let us say, monthly banking. In some states, there is time-of-use banking. In some states, the banking is reduced only to the solar generation. So somewhere, this policy coherence across state should be there to help us to take better decisions. And last but not the least is there are policies in place in terms of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:47",
                    "end": "00:41:15",
                    "text": " incentives for data centers or certain waivers. But on ground, the implementation of these policy is a challenge. That's what we have seen that there could be an electricity duty waiver, or there could be an incentive on energy rebate. But on ground, actually getting that rebate, there are administrative challenges. There is a process in place, but it takes about 18 months or two years to actually get those incentives."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:15",
                    "end": "00:41:45",
                    "text": " So yeah, I think high level these would be the challenges that we see. Thank you. OK, Mr. Nanda, as AI data centers move toward higher power densities and liquid cooling, where do traditional data center models break down? And what needs to be rethought from the ground up? I think between the conventional data centers and the AI data centers, the design philosophy is going to be very different. We're going to get together on day one between the semiconductor"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:45",
                    "end": "00:42:15",
                    "text": " manufacturer between the hyperscaler and the data center designer. That probably did not happen in the erstwhile data centers. It was more looked like a building where you just go and put racks. These are going to be complex energy systems. So that's one big change that will happen. And that will happen just given the kind of power requirement. But I'll also add to the comments that we made earlier in terms of what else can India do. So India is at an advantageous position, as I mentioned earlier."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:15",
                    "end": "00:42:42",
                    "text": " Also given the fact that we have a single national grid, not that in many countries, which also means that there is a different power prize in different parts of those countries, which gives us an advantage. The Indian states currently are competing for investment on data centers. And that's super attractive. That is not the case internationally. And I hope it remains in India until we become"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:42",
                    "end": "00:43:09",
                    "text": " a force to reckon with. So on your question of the coordination on policy, I think if all of us speak in unison with the states who are currently vying for investment and make sure that those investments are translated into infrastructure projects in quick time, I think we would have served ourselves very very well. Third, just learning from the issues that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:09",
                    "end": "00:43:36",
                    "text": " data center build out in the United States of America are facing. And I quote that example because that's the country where maximum build out is happening currently. I think India has a huge manufacturing footprint right now. And a lot of that is underutilized, especially on the power side. So if you say, for example, turbines, if you take, you know, whether it's wind turbines or gas turbines or steam turbines and so on. In the U.S.,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:36",
                    "end": "00:44:05",
                    "text": " That's maxed out and there is no availability till 2030, 2031. Companies like GE or Siemens and so on are sold out completely. We have that capacity, same with electrical gear. So I think this is the moment for us and has been said by my fellow panelists, for us to quantify by state, by location, what is the demand and then kickstart the manufacturing process."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:05",
                    "end": "00:44:32",
                    "text": " so that we have enough gear in place as this exponential growth takes off. Finally, also sources of power. So we've talked about renewables. I'm also a firm believer that we will also need back of the meter power in the sense that these campuses will have to be self-sustained at some point in time. And to that extent, India should rethink the approach towards gas-based power generation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:32",
                    "end": "00:45:01",
                    "text": " We as a country, we have around 45 million tons of import capacity. And that's ideally located across the periphery of the country. We have a very good pipeline network. And hence, again, as hyperscalers look at the zones where they want to build out large data center campuses, it'll all go extremely well for India to kickstart the gas-based power for these plants. Because when it comes to affordability, I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:01",
                    "end": "00:45:25",
                    "text": " the data center operators and the hyperscalers, that may not be the issue as compared to conventional power. Because it was mentioned earlier, I think it was Abhishek who mentioned that ultimately my user should, my consumer, which is the consumer that they serve currently as a distributor, should not get impacted on the cost of electricity. That's another policy dimension that we need to work upon."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:26",
                    "end": "00:45:56",
                    "text": " We can have a whole other session on rate design, but for now let's just we'll have a lightning round and so for each panelist Tell me the theme of this is collaboration So of this panel so what? To address any of your biggest like disconnects today between where you are where you're going what others are doing What would one what is one collaboration that could close the gap fastest?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:56",
                    "end": "00:46:31",
                    "text": " And I will take any order. Lightning Ram, one. Adoption of heterogeneity or silicon diversity. I think that would be one key one. Right. Provision of reliability services by the data centers would definitely improve the system operation side. This is the design of the hyperscalers, the technological operation, which is flexible, grid integrable, just like you have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:31",
                    "end": "00:47:04",
                    "text": " Equalizers which can flex so that kind of technology I'd say hyperscale is taking a bigger bet on India Which I think they already are so I Think the good and the bad thing are the same AI data centers are the power density is very high like in the same footprint you're talking about Gigawatt capacity, I think the next five years will talk about gigawatt capacity why I'm saying it's good and bad it's bad because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:04",
                    "end": "00:47:29",
                    "text": " you would want the data centers to be spread out, right? So that the strain on the network is not that much, but that's not happening. Why I'm saying it is good is you know the pockets that you need to focus on. You know, like Deepesh rightly mentioned, we spoke about Navi Mumbai, or he spoke about Hyderabad. We exactly know the pockets where the data centers are coming up. And we should give special attention to those pockets when it comes to network augmentation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:29",
                    "end": "00:47:59",
                    "text": " much ahead in time, right? Network augmentation or like Saxenaji rightly said that there could be a problem of fault currents being injected or harmonics or voltage flicker. So the good thing is that you can arrest it at that point. Like if you put in proper checks and balances, these things can be arrested at the local grid itself and it will not pass on to the grid. So planning, I think we have to get better at planning, more localized planning. I don't see problem on the generation side."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:59",
                    "end": "00:48:34",
                    "text": " upstream transmission network is also robust. But when it comes to localized network, where in the data center clusters are forming, that is where I think we should collectively focus and plan better. Wonderful. And then on our last few minutes, do any of you have a question for one of the others on the panel? I'd like to take this opportunity to ask you on battery energy storage and the role that it will play."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:34",
                    "end": "00:48:55",
                    "text": " to stabilize grid power for large data center campuses. Any thoughts on that because that's a topic that, you know, we've been looking at. And I know that India has embarked on a journey of buying massive capacity of battery energy storage. How do you see that playing out?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:55",
                    "end": "00:49:10",
                    "text": " is definitely going to be playing a very, very important role, especially in the context of data centers. In fact, that is where you can actually absorb the ramps and maybe provide backup, provide reliability service, provide voltage support."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:10",
                    "end": "00:49:39",
                    "text": " All those kinds of reliability services can also come out from the battery energy storage systems. And in fact, it could offer some sort of shifting of demand as well. So all this, I think, should definitely come through in the form of local support to the data centers. So, Deepesh, just to add, this is another commercial opportunity for you, business opportunity. So you run data centers, right? So that is your main business. But then when you..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:39",
                    "end": "00:50:06",
                    "text": " add these resources, unblocking lithium chemistry. You can talk about any other chemistry which I attribute to maybe the vanadium redox, which is a flow chemistry. There could be many other sodium ion batteries, long duration ones, because this is a very stable base kind of load, data centers are. So if you have this, and then the regulations which are in place, ancillary services regulations, maybe amended, AS market base are getting introduced."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:07",
                    "end": "00:50:36",
                    "text": " Ancillary service on market is already there. And SRAS, that is Secondary Reserve Ancillary Service and Primary, they are going to be market based in short term of future, short term future. So that could be a way where you are making your lives easier. Also, there is a money spinning opportunity where in your valuation sheet, one of the line items would be how I'm flexing my data centers without impacting the reliable operation and making money out of it. But next one thing Jacqueline, I wanted to add here."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:37",
                    "end": "00:51:07",
                    "text": " As we are becoming more and more data-driven data dependent data is just like becoming our oxygen air. So the data centers definitely are going to be very very crucial and critical infrastructure as like in India or anywhere else. We are classifying the power infrastructure as very critical and they are very soft targets. Now the islanding schemes of power in in India, for example for cities like Mumbai Delhi, Hyderabad, Bengaluru."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:07",
                    "end": "00:51:36",
                    "text": " Where these data centers are likely to come up? they need to be brushed and Dusted now because if the islanding schemes are not properly operating In Mumbai there was a case I think in October of 2020 or 2021 where there was a failure because of some transmission glitch something happened and there was cascade failure the Hydro power in the Western guards were not able to sync up or you know supply power to Mumbai"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:36",
                    "end": "00:52:03",
                    "text": " So those things will be catastrophic. Why? The data centers, if they fail, then your critical life support, your police enforcement, maybe airports, hospitals, they're all going to die down. So therefore, the islanding schemes have to be dusted. And there, we need more technological inputs. It's no longer a power play, because it's going to impact your lives. Data centers are now heart of lives, actually."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:04",
                    "end": "00:52:39",
                    "text": " Well, thank you. This has been a fascinating discussion. And it is clear that all these choices that we're making today are going to be shaping our economy for decades to come. And I want to give a round of applause to our panel. OK. Others, do you want to come up? I'm going to invite now my colleague Adarsh Nagarajan to give some closing remarks. And then stay for the next panel. We're going to be going more into data centers."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:42",
                    "end": "00:53:12",
                    "text": " Thank you. Please, please stay back and ask your question, sir. This is a very important topic, but little later, once we end the session. Kuch nahi hoga? Is it? We'll figure it out, sir. Oh my god. OK, we'll figure it out, sir. Thank you. Thanks for your voice. So this whole journey of AI as a data center and the load growth is so rightly mentioned by Jacqueline."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:12",
                    "end": "00:53:41",
                    "text": " What India went through 10, 15 years ago, huge, huge change, huge change in demand. Here in this case, we are adding huge demand of gigawatt scale, adding a new state. Even a variability of five to 10% is hundreds of megawatts. It's huge. And as this committee so rightly pointed out, this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:41",
                    "end": "00:54:10",
                    "text": " will be successful only when there is a knowledge exchange and a handshake across all the stakeholders present here and the categories of stakeholders, the chip manufacturers, as model developers, the solutions, the DISCOMs, and the national grid operators, all put together when there is a harmony of knowledge. That's when it becomes highly successful, organized. And I believe the leaders here have already communicated that. And with that, I want to..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:10",
                    "end": "00:54:42",
                    "text": " take the next few minutes to provide the memento to all the panelists. Let's, let me invite Jacqueline, please, give the memento to Abhishek Ranjan, CEO of BRPL. The memento that we'll give. Oh, we have. Thank you. Thank you, sir. Let's have Mr. Sameer Sathinath, sir, please. You can come stand here. Thank you, sir."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:42",
                    "end": "00:55:13",
                    "text": " You can come in the front, a little better picture for everyone. Thank you. Let's have Mr. Krishnan. Sorry. Phukan, sorry for that. Mr. Krishnan, please. Thank you. And Mr. Nanda. Thanks a lot. Thanks a lot. Please stay back. We have an equally interesting yet more technical session following."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:13",
                    "end": "00:55:21",
                    "text": " Right after this same place will be replaced with more eminent guests as well. So please stay back and listen to that as well. Thank you."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "69942a5a25c9f4e0d7e70271",
    "projectId": "762001",
    "projectName": "AI in Health Saving Lives at Scale - Spanish",
    "projectDescription": "AI in Health Saving Lives at Scale - Spanish",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T08:44:14.744Z",
    "projectUpdatedDate": "2026-02-17T08:44:14.744Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T08:44:14.744Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "AI in Health Saving Lives at Scale.mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771317822958.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:59",
                    "end": "00:02:42",
                    "text": " Excuse me, we request you to please exit quickly so that we can start our next session. Welcome back to J-PAL's AI for Social Good Impact That Works seminar. Our next session is AI in Health, Saving Lives at Scale. Please join me in welcoming Ziad Obermeyer. Ziad is an associate professor of health policy and management at the UC Berkeley School of Public Health, where he works at the intersection of machine learning, medicine, and health policy."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:42",
                    "end": "00:03:09",
                    "text": " He helped set up the computational precision health program and his research focuses on how machine learning can help clinicians make better decisions. A fun fact about Ziad is that he was named one of the 100 most influential people in AI by 10 magazine. Ziad, over to you. Thank you so much. So I thought I would tell you a story about when I first came to my job at Berkeley. So this was in 2018."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:09",
                    "end": "00:03:39",
                    "text": " I was a junior, untenured faculty member, and I just signed up to give the first seminar of the year in the economics department. Now, the economics department at Berkeley is friendlier than the economics department at other institutions affiliated with J-PAL, but they are still economists and still very hardcore, so I was nervous. And about three or four days before the seminar, I started feeling a little unwell."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:39",
                    "end": "00:04:07",
                    "text": " My stomach was hurting. I just didn't feel right. And I made the mistake of mentioning this to my wife, who said something like, oh, is your tummy hurting? Are you a little nervous about your... Anyway, she was not very sympathetic. So I prepared for my talk. I prepared for my first class that I was teaching. And I gave the seminar. And during the seminar, instead of feeling just globally not well,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:07",
                    "end": "00:04:36",
                    "text": " I started feeling unwell in a particular place in my stomach, which is the bottom right side of my stomach. Some of you might know what I have already, but I want to first make an observation that this made me realize, which is that when people make really important health decisions, like do I come into the hospital, do I ask a doctor about this, people are making those decisions that are very, very important with very little data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:36",
                    "end": "00:05:03",
                    "text": " So at the end of the seminar, I actually drove myself to the emergency room near Berkeley and I got a CT scan. And this is my CT scan. And some of you might be able to see what I have or guess what I have, but I had acute appendicitis and it had perforated and I needed to have my appendix removed. I think what makes this story surprising and a little bit embarrassing for me"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:03",
                    "end": "00:05:26",
                    "text": " is that I trained as an emergency room doctor. So there's a few things that you're not supposed to miss as an emergency room doctor, and this is one of them. And yet, I missed it in myself for four or five days. So I think what this highlighted to me is how critically important it is."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:26",
                    "end": "00:05:48",
                    "text": " to be able to have access to this kind of data. The kind of data that normally live only inside of the health system and that you can only have access to if you're lucky enough to be close to an emergency room with a CT scanner to have access both financial and linguistic and socioeconomic to these kinds of technologies. Because without data like this,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:48",
                    "end": "00:06:17",
                    "text": " There's no diagnosis of critical illnesses. And without diagnoses, there's no treatments. So this data is really the key to a lot of things that happen in the health system. And yet, in order to get access to this, you have to be part of the health system. You have to have access to begin with. So there's a circularity to this problem. And it locks out many, many people, of course, not just in the US, but even more around the world."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:17",
                    "end": "00:06:34",
                    "text": " This kind of underdiagnosis is, of course, not just limited to acute surgical things like appendicitis, but to many, many different conditions. And a particularly salient example to many people happens even in high income countries, which is heart attack. So there are these studies that follow."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:34",
                    "end": "00:06:53",
                    "text": " big cohorts of people across time in the US, and every so often they do an MRI of the heart to detect scarring in the wall of the heart that indicates that this person had a heart attack. And in these studies, when they look at all of the people who have scar in their heart consistent with a heart attack,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:53",
                    "end": "00:07:18",
                    "text": " Most studies find that the majority of those heart attacks are silent. They are not known to the patient and not known to the doctor and not diagnosed, either because the symptoms were subtle or because the person never sought care to get those symptoms diagnosed. So as you can imagine, this is a huge problem even in high income countries. It's an even bigger problem elsewhere in the world. And so there's data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:18",
                    "end": "00:07:33",
                    "text": " from South India, from Tamil Nadu, that asks people about things like high blood pressure and diabetes and asks them, do you have these things? And then does the objective measure of those things. And even for these basic risk factors for heart attacks, we..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:33",
                    "end": "00:08:02",
                    "text": " We can't do cardiac MRIs in a lot of places, but even for these risk factors, there's dramatic underdiagnosis and under-awareness of the things that cause heart attack. So this is a problem everywhere, and it's a problem that really urgently needs solutions. So I thought I could tell you, I'm going to tell you about half of the solution, and instead of just telling you, I'm going to show you, and you'll see more of this from my colleague Shredda in a few moments."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:02",
                    "end": "00:08:18",
                    "text": " This device that I'm holding in my hand is a mobile handheld electrocardiogram. So the two silver sides are electrodes. You put your fingers on it and it records the electrical field generated by your heart as it beats."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:18",
                    "end": "00:08:43",
                    "text": " And I think these technologies are really miraculous because they cost so little. That device costs $60. In the US, you can order it on Amazon and get it delivered in the same day. It connects to your phone or tablet via Bluetooth with no wires. And it requires absolutely no training to do an electrocardiogram. So this is really an incredible technology that's become commoditized over the past five to 10 years."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:43",
                    "end": "00:09:06",
                    "text": " But the reason that these kinds of devices have not had the impact that any of us would have hoped is because there's a key constraint, which is that what do you do once you generate this waveform? Well, you have to send it to a doctor for interpretation. This is too complex for patients, and even for some doctors like myself have trouble interpreting electrocardiogram waveforms."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:06",
                    "end": "00:09:35",
                    "text": " And so it's great that you can generate this data wherever you are in the world. But if you still need to send it to a doctor to get interpreted and to translate those data into decision making, you haven't solved the key bottleneck, which is access to the health care system. So even though these devices are wonderful and miraculous, they're only part of the solution to democratizing access to diagnosis around the world. So this is an AI summit. Could AI help?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:35",
                    "end": "00:10:01",
                    "text": " Yes. But it's important to remember that there's nothing magical about artificial intelligence. And in fact, artificial intelligence is just data in a very literal sense. So in an ideal world, what you'd have as far as a data set to train your AI algorithm is a data set that links data from these cheap devices to some ground truth outcome about heart attack, appendicitis, et cetera, et cetera."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:01",
                    "end": "00:10:25",
                    "text": " Unfortunately, those data on the ground truth are quite hard to find even in high-income countries and certainly in low- and middle-income countries. And so what we often do when we're training AI tools is we make a key and very pernicious substitution, which is we say, okay, well, we can't get the ground truth linked to these electrocardiogram waveforms."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:25",
                    "end": "00:10:52",
                    "text": " So instead, we're going to ask doctors to look at the waveforms and tell us what the doctor thinks about that waveform. So we substitute human judgment and human interpretation for some sense of the ground truth of, does this person really have a heart attack or not? Let's just ask the human. And the problem with humans, many of my good friends are humans, so this is nothing about humans in general, but humans have biases and humans make mistakes. And human knowledge..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:52",
                    "end": "00:11:22",
                    "text": " about the human body and many other things is incomplete. And so when we automate human judgment, we also run the risk of automating all of the problems with human judgment along with it. And I think that's what makes it so critically important to train artificial intelligence not on doctors, but on patients, on what happens to patients and their outcomes. So let me tell you about some work that I am doing in collaboration with J-PAL South Asia in a few"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:22",
                    "end": "00:11:37",
                    "text": " villages in Tamil Nadu. So this is a line of people queuing up for a district health center. And what we did over the summer of 2023 is set up health camps, a long-standing tradition where we just..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:37",
                    "end": "00:12:05",
                    "text": " had a facility outside of this district health center where, thanks to the tireless work of our colleagues at J-PAL South Asia, we rented out a bunch of very fancy equipment that you can normally only get inside of hospitals. So for example, we rented out a cardiac ultrasound machine and a technologist to acquire the images of cardiac ultrasounds. We hired cardiologists to interpret those images and look."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:05",
                    "end": "00:12:33",
                    "text": " among other things, for signs that this person had had a heart attack, signs of scarring in their heart. And at the same time, we did data collection from these cheap $50 devices. So here you see two patients that were participants in this health camp. This young lady is having an electrocardiogram done and sent to the tablet in front of her. And the gentleman over here is getting a cardiac ultrasound, which serves as our way of understanding."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:33",
                    "end": "00:12:53",
                    "text": " if this person has had a heart attack by looking at wall motion abnormalities in the walls of his heart. So every patient got both the fancy expensive tests from the hospital and the cheap low cost tests. And of course, if we found any abnormalities, we communicated those back to the patient and their doctors as part of this study."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:54",
                    "end": "00:13:18",
                    "text": " So so the eventual goal of this is to create a way to screen people for a number of serious health conditions including prior potentially silent heart attacks in the comfort and safety of their own home either by doing the electrocardiogram themselves or with the assistance of a community health worker and Just to give you a sense of how well even this early version of the algorithm is working"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:18",
                    "end": "00:13:40",
                    "text": " when we flag the highest risk, two, three, four, five percent of the population, just general population of people visiting an ambulatory health center, 10% have actually had signs of a prior heart attack, and that's compared to about 0.2% in the general population of people that we were looking at. So we're finding very high-risk people, and..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:40",
                    "end": "00:13:54",
                    "text": " people who, even though the algorithm is saying that they're high risk and they do indeed have silent heart attack on their cardiac ultrasound, they lack the traditional risk factors like blood pressure and cholesterol and diabetes."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:54",
                    "end": "00:14:22",
                    "text": " that are traditional because they were defined in Western populations. And as we know, South Asian populations have different risk profiles and different risk factors that this electrocardiographic approach is able to find despite the fact that they lack those traditional risk factors. So I think this highlights the fact that these algorithms can be trained and tuned to populations all over the world that might be different from the populations in which most of our medical knowledge is developed."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:22",
                    "end": "00:14:51",
                    "text": " When we put this through the cost-effectiveness of screening, in other words, sending high-risk people for confirmatory tests at the health center and treating them with the medications that you need to be on for people who have had prior heart attacks, this comes out to be cost-effective, even using Indian guidelines. So it's around $2,000 per disability-adjusted life year, even for this very early version of the algorithm, which is continuously improving as we collect more data. This is very exciting."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:51",
                    "end": "00:15:06",
                    "text": " And the next step is actually, this wouldn't be J-PAL unless the next step were a randomized evaluation. And so we're doing a randomized evaluation where we're gonna start screening people either with traditional risk factors like blood pressure, et cetera."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:06",
                    "end": "00:15:28",
                    "text": " or this AI enhanced electrocardiogram screening approach and referring the high risk people for confirmatory testing in both groups and seeing which approach finds more people that we want to find, people with prior heart attacks. So one of the exciting things about doing this project is that unlike a lot of the research that I've done in the past,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:28",
                    "end": "00:15:58",
                    "text": " This is a real thing. It's a device. And it's already interacting with real people. And so to give you a sense of that, if you don't mind, I'm going to introduce Ms. Shreda Rajesh, who is an instrumental part of our team with J-PAL South Asia. And we're going to violate every principle of giving a talk by doing a live demo and hoping that it works. So if you'll bear with us for a second, we're going to get this set up. And then you will see my own electrocardiogram being put through."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:58",
                    "end": "00:16:29",
                    "text": " our web application and screening me in real time for my own risk of prior heart attack. Luckily, since I'm a physician, nobody needs to take responsibility for the outcome of the test besides myself. All right. So we're getting it hooked up. And I'm hoping it works because we tried it before the talk. But of course, live demos have a way of frustrating your expectations."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:40",
                    "end": "00:17:53",
                    "text": " We're gonna switch to the other podium. One thing I'll just mention as we're getting set up is that this approach is not limited to mobile electrocardiograms."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:53",
                    "end": "00:18:21",
                    "text": " those of you who have Apple watches, of course, know that there's increasing amounts of mobile technology built into your watch that can be fed in, but these electrocardiograms are not the only low-cost data collection devices that we have available to us. So in the course of this study in Tamil Nadu, we also collected a number of other low-cost devices. So those of you who sadly had to live through COVID here, no doubt had"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:21",
                    "end": "00:18:40",
                    "text": " mobile portable pulse oximeters. Those pulse oximeters, in addition to generating oxygen percentage numbers, generate a waveform that actually represents the blood flow through your finger that the pulse oximeter is measuring to get that fraction of oxygen."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:40",
                    "end": "00:19:07",
                    "text": " But that waveform has an enormous amount of information about how your heart is working, how your capillaries are working, how your blood vessels are allowing the flow of blood through all of these channels. So that also can plug into algorithms. The gentleman in the front row is wearing an aura ring or a related ring that can also measure this kind of thing. Continuous temperature monitoring is another feature of these rings. There are smartphone-based attachments."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:07",
                    "end": "00:19:10",
                    "text": " that can take a picture of your retina."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:10",
                    "end": "00:19:39",
                    "text": " A fun fact about the retina, you got a lot of fun facts about the speakers from Audrey. A fun fact about the retina is that it is part of your central nervous system. It's the only exposed part of your central nervous system. And with a $200 smartphone attachment, we can actually get a view into the central nervous system that can be useful for diagnosing stroke, intracranial hemorrhage from trauma, retinal disease, other kinds of diseases of the eye. Okay."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:39",
                    "end": "00:20:10",
                    "text": " Thank you so much for getting this working, Shyada. Would you mind telling us a little bit about what we're going to do? Of course. Thanks so much, Ziad. Sorry, this took a bit of time. Good morning, everyone. We'd now like to demonstrate what this looks like in practice. So all a participant has to do is basically place their fingers on this very small device that's connected to my phone, as you see. And that's it. No wires, no hospitals, and just 30 seconds. Ziad, I hope you don't mind."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:25",
                    "end": "00:20:46",
                    "text": " Sounds good. So what this device does is it records a single-led ECG, which is the electrical signal of your heart. And once we capture that, I enter the participant's ID onto our app that we developed in-house. And behind the scenes, our AI model, it analyzes this waveform, the pattern of the heart, almost instantly."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:46",
                    "end": "00:21:40",
                    "text": " which it has learned to do from the thousands of real life cases that we've collected from the field. And within seconds, you get a risk score that we'll show you just in a bit. So what this risk score tells us is based on, just again, 30 seconds with this single LED ECG alone, how likely it is that this participant has had a silent heart attack in the past. The stress of the live demo is elevating my... I hope it hasn't provoked a heart attack. That would be unfortunate."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:41",
                    "end": "00:22:41",
                    "text": " Are there any other doctors in the audience? All right. I think we're ready to go. So what you're seeing now is the ECG was automatically uploaded into a web application and generating a low risk score. Thank goodness for that. Thank you so much. And thank you so much to our tech team here. Live demos are an unpleasant ask for the tech team. And I really appreciate your help. Thank you. OK."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:41",
                    "end": "00:23:08",
                    "text": " So as I mentioned, this is not just about ECGs. You could imagine a setup where a very small amount of hardware totaling under 20,000 US dollars comprised of an ECG, a retinal photo, a mobile chest x-ray unit, a handheld ultrasound device can actually be put together in a district health center or even in a little kiosk in a shopping center somewhere else."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:08",
                    "end": "00:23:30",
                    "text": " and could generate an enormous amount of raw device data that can feed into a number of diagnoses that are of critical importance to lots of causes of burden of disease in low and middle income countries, but in high income countries as well. So I think the scope and the scale of these kinds of diagnostic devices and approaches is really enormous."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:30",
                    "end": "00:23:60",
                    "text": " And I'll just mention one very personal application of this for me to give you a sense of what this can do in life. So my wife and I have twin daughters who are almost two years old. And the way that we found out that we were having twin daughters is because I, as part of my practice in emergency medicine, I own a handheld ultrasound device. So this costs $2,000. And it can generate pictures."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:60",
                    "end": "00:24:24",
                    "text": " like this of our two future daughters. So we knew that we were having twins well before most people know because we are lucky enough to have access to this technology. And I think democratizing and broadening access to both the hardware and the software components is really incredibly important as we seek to diagnose illnesses across the world. The last thing I'll say is that I think one of the really exciting parts of this work"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:24",
                    "end": "00:24:42",
                    "text": " is that it represents a fundamentally new way that we can develop AI algorithms in a sense that many people I think in well-resourced settings find counterintuitive. Because this is an innovation that starts here in India. Why does it start here?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:42",
                    "end": "00:25:04",
                    "text": " because there's very high volume of data at very low cost, but also because there's an enormous amount of human capital here. Engineering and AI talent, people who know how to build hardware and train AI products, there's the ability to move quickly, and at the same time, and especially thanks to the team at J-PAL, there's scope for very rigorous testing."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:04",
                    "end": "00:25:25",
                    "text": " This kind of innovation can scale seamlessly to the rest of the world. One interesting fact about our algorithm is that we pre-trained our AI model on electrocardiogram data from Sweden and the US. And that actually helped a lot for improving the model performance in India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:25",
                    "end": "00:25:47",
                    "text": " And that interoperability, if you will, of the ECG waveform is a really exciting part of building algorithms that run on this kind of physiological data. Because as we've tested algorithms that are trained in one place, for example, in Sweden, they generalize incredibly well to the US, to India, to Taiwan, to wherever we've tested them. Because..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:47",
                    "end": "00:25:57",
                    "text": " we do electrocardiograms the same way in all of those places, and because the heart is fundamentally wired in the same physiological way in all of those places."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:57",
                    "end": "00:26:23",
                    "text": " And so those algorithms can generalize and they can solve barriers to access that are widespread not just here in India but all over the world where people struggle to get access to the formal health system because of distance, because of poverty, because of discrimination, because of cultural or linguistic barriers. So I'm very excited about that and I would urge anyone who's working on this kind of work here with data is to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:23",
                    "end": "00:26:59",
                    "text": " protect and steward the secret sauce that you have here in India. Because it's really such a valuable resource, but not just to India, but to the rest of the world, to be able to develop AI products in a way that is safe, ethical, but also rigorous and fast. Thank you very much. Thank you so much, Siyad, for that insightful presentation. And may your risk of heart attack remain low for many years. Please rejoin the stage for a group photo with the rest of the session speakers."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:59",
                    "end": "00:27:27",
                    "text": " It's my pleasure to invite to the stage the panelists for the AI and Health Saving Lives at Scale session to join Ziad on stage. The panel will be moderated by Amy Barnes. Amy, please join us on stage. Amy is a policy manager at J-PAL Global, where she leads J-PAL's health sector. With all the promising applications of AI and health, Amy has increasingly been pushing to increase the available evidence on the real-world effects of AI solutions on both providers and their patients."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:27",
                    "end": "00:27:56",
                    "text": " A fun fact is that Amy started her time at J-PAL focused on policies to strengthen governance and public service delivery and to reduce crime and conflict. Now please join me in welcoming our panelists, starting with Shahed Alam. Shahed is the co-founder and co-CEO of Nura Health. Since its founding in 2014, Nura Health has been transferring health skills to family caregivers and patients. So far, Nura has trained over 43 million caregivers and patients across India, Bangladesh,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:56",
                    "end": "00:28:27",
                    "text": " Indonesia, and Nepal. And now Shahed is using AI to take Noor's mission to the next level. A fun fact about Shahed is that he earned his MD from Stanford. He's an associate faculty member at Ariadne Labs at Harvard Chan School of Public Health. And he's also been a TED speaker. Also on stage, Zamir Bray. Zamir is the deputy director of technology diffusion at the Gates Foundation. In this capacity, Zamir leads the foundation's strategic coordination on the equitable and safe use of AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:27",
                    "end": "00:28:55",
                    "text": " including work like the internal AI task force, AI related grant challenges, and external AI advisory efforts. A fun fact about Zamir is that he trained as a medical doctor, and he also earned an MBA. And on top of that, he has a PhD in health systems and innovation, and a master of law from the University of Cape Town. And he serves on the board of Groot Sure Hospital in Cape Town. Finally, Rob Sherman is the vice president of policy and deputy chief privacy officer at MEDA."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:55",
                    "end": "00:29:28",
                    "text": " where he works to set the governance and safeguards for how AI-powered products are designed and deployed responsibly. A fun fact about Rob is that before joining Meta, he was a lawyer advising major tech and digital tech companies on privacy and data security, and he was recognized by Chambers USA as one of the nation's leading media regulatory lawyers. We have very great people on stage with us today. Amy, over to you. Thank you so much, Audrey. The audience can send in questions by scanning this mentee code."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:28",
                    "end": "00:29:53",
                    "text": " And thank you, Amy. Awesome. First, just so excited to be here with all of you today. Hear your thoughts around the rapid evolution of AI in health systems, how it can be integrated, your thoughts on evaluation, of course. Audience members, please do submit your questions. But I'll just take the moderator privilege and start diving in with one. I want to start locally with Shahed. I know we're so far, but."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:53",
                    "end": "00:30:22",
                    "text": " Going to start with you and your organization, Nora Health, which does a lot of work, including in India, to train family members to advocate more for their loved ones in hospitals and to provide better care when they're leaving the hospitals to return home together. So Shahed, Nora is already working across multiple countries and already recognized for its impact. At this stage in Nora Health's growth, what problem are you trying to solve and what's the role of AI in your solution moving forward?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:22",
                    "end": "00:30:52",
                    "text": " Thanks so much for having us and also for that question. At Norah Health, again, we support the family caregiver of the patient with the knowledge, skills, and confidence they need to care for their loved ones. We work with governments and health systems in order to incorporate these programs that really support caregivers. The anchor of our work really starts in healthcare facilities when families are anxiously waiting by their loved one"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:52",
                    "end": "00:31:17",
                    "text": " And we provide them a skill-based training by a trusted health worker in person to engage them in their loved one's care. Now, once they go home, families may, they'll face some sort of a challenge. They may need a reminder or a nudge to remember what to do in order to care for their loved ones. And so that's where..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:17",
                    "end": "00:31:36",
                    "text": " For us, technology and our mobile messaging service steps in. So after an in-person training to continue to receive nudges and reminders for families once they go home. From the very beginning, as we've developed this service, we've not developed this as a chat bot because we wanted to always keep..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:36",
                    "end": "00:32:05",
                    "text": " that human connection between patients and families at home and a health worker at the back end as a critical and core part of what we do. But as we've scaled it, we have seen that that whole process can become very constrained. We went from response times that were pretty quick to response times that needed to take more than 10 hours. So that's where AI really stepped in for us."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:05",
                    "end": "00:32:28",
                    "text": " to be able to address this challenge of keeping high quality support to families once they're at home, but also ensuring that response times don't balloon to something like 10 hours. We had families who loved the service. They would reach out to us and they would ask our nurses to name their newborn, but they were getting increasingly frustrated with how long it took."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:28",
                    "end": "00:32:58",
                    "text": " So we sat down with our health workers, and that's really where it begins, is to define what are the challenges that people are facing, and for us those are health workers, and started to look at a lot of manual tasks that they were doing, looking up things in an FAQ database, annotating things manually, and then with them, alongside us, we started to co-develop an AI co-pilot that supports health workers in answering questions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:58",
                    "end": "00:33:18",
                    "text": " providing them context on the patient and family that they're supporting. And importantly, also tracking operational and other data so that we understand the quality of the responses in the system overall. And with that in place, we've been able to reduce those 10 hours down to 30 minutes."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:18",
                    "end": "00:33:49",
                    "text": " We've improved retention, so families, more than 30% of families keep coming back to us for more support. And importantly, the health workers are happier. They're getting to focus on what motivates them, which is to provide care versus to fill out a bunch of forms on an Excel sheet. And to follow up on that, I've heard rumors at least that you're planning a randomized evaluation later this year on this. Could you say more about why you're exploring an RCT now?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:49",
                    "end": "00:34:17",
                    "text": " given that you already have some evidence of Nora's impact. You are correct. For us, impact and evaluations is not a one-time thing. As we evolve, as the world changes, we need to continually update the evidence that we have and the knowledge that we have. The RCT is really critical for us because while we've done evaluations on the model, some of the product metrics, how it's affecting users,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:17",
                    "end": "00:34:31",
                    "text": " Ultimately, what we hold ourselves accountable to is having an impact on health outcomes, things like newborn morbidity and mortality. And the RCT is a way for us to actually, in a more definitive way, answer that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:31",
                    "end": "00:35:00",
                    "text": " In a simple way, it is to understand the impact on this incredibly important outcome, which is newborn mortality, and to understand also incrementally what does the AI-supported tool provide versus our standard program, which is the in-person initiative. And hopefully with that data, ultimately what we want to do is to have family caregiver support be the standard of care across health systems globally."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:00",
                    "end": "00:35:29",
                    "text": " RCTs and this evidence that we're generating will allow us to do is to provide information on what works, what doesn't, to governments so that they can use their scarce resources in supporting caregivers most effectively. Thank you so much. Zamir, I want to bring you in next and hear from you from the perspective of a funder who's thinking about technology diffusion a lot in health systems around the world. Because AI tools are evolving so rapidly, how should funders and implementers like Nora Health"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:29",
                    "end": "00:35:58",
                    "text": " design evaluations that can keep pace with that reality, while still producing evidence that's credible enough to drive adoption and policy. Thanks, Amy. I must just depart for a second, and I wonder if Ziad is still around, because when Ziad made his admission about being a clinician that missed the diagnosis, I realized that's why I was invited to this panel. I too missed the diagnosis, and I was just remembering that..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:58",
                    "end": "00:36:24",
                    "text": " It was actually my first trip to India 15 years ago when I came here, had a great time and went back home and had a classic swinging fever for six days and completely missed that it was typhoid fever and too ended up in the surgery room with a cholecystitis."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:24",
                    "end": "00:36:53",
                    "text": " To Ziad, this is a bit of a clinician anonymous admission session, so thanks for creating a safe space. But Amy, to get to your question and also just to pick up on where Shayid left off, I think the role of evaluations for decision makers for governments where resources are scarce is absolutely critical. And I can already tell from..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:53",
                    "end": "00:37:22",
                    "text": " the proceedings this week, that this will be one of the key messages we take back, which is we need to double down on evaluation, particularly giving decision-makers the right kind of evidence at the right time. I'm a big fan of RCTs. They have fundamentally changed the global health landscape. But there are two kind of issues we need to grapple with. The one is..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:22",
                    "end": "00:37:52",
                    "text": " the models change very fast. And I see some colleagues smiling with me because this is a very real issue. You know, I was doing a little bit of homework on, like, how long does this take? It takes us six months to design the RCT. Let's do the RCT for maybe 24 months and then take 18 months to publish the data. That means that if we all left here today and we started designing, Shahid..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:52",
                    "end": "00:38:20",
                    "text": " maybe 2029, we have some good results that you can put out there, right? I'm not saying that that's what we should do. The other reality is that policymakers, like in India, like in many countries we work in as the Gates Foundation, are really hoping that AI brings new promise, brings new tools to frontline workers that it's never been able to do before. And so the policy windows"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:20",
                    "end": "00:38:46",
                    "text": " the policy cycles don't track to publication cycles. Policy decisions are often made because of resource constraints, or there's an outbreak, or there's some political economy issue that drives the decision. And so being receptive to that means that we're going to have to challenge ourselves on conventional models of doing..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:46",
                    "end": "00:39:16",
                    "text": " RCTs or long trial designs. That shouldn't mean that we compromise the evidence generation. So just to be clear, we have to have good evidence. But in the early phases, we can start thinking about, does this give us a safety signal? Can we do something in short step, in three to six months that says, this looks like this could be safe, could be interesting?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:16",
                    "end": "00:39:46",
                    "text": " and very much to what the previous speaker said. Maybe those are short, inexpensive, quick experiments that we can run. Then we can move into kind of pragmatic trials, adaptive trial designs that don't take us two or three years, but give us the kind of evidence around adoption. And this is really key. We've seen lots of models perform really well in labs and test stops."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:46",
                    "end": "00:40:17",
                    "text": " they beat every single benchmark. And there's no doubt that most models today are better than clinicians. Those results don't hold up when you hit the field. Why? There's some fundamental issues. You know, one trial we saw was that the clinicians actually didn't trust the AI. So adoptions were stuck at about four, five, six percent until"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:17",
                    "end": "00:40:47",
                    "text": " someone actually sat with the clinicians, went through their own clinical cases and their mistakes, and showed them that somebody in the room, there were 100 clinicians in a trial in Kenya, and one of them was making the mistakes. And those mistakes were significant and catastrophic, that they then sat around the table and said, hang on, there's something we need to embrace about the technology. And then we saw adoption surge to upwards of 60%."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:47",
                    "end": "00:41:16",
                    "text": " still not at 100%. So I think that how we think about evidence is what's the right methodology for the right time. And again, I want to just applaud the Indian government and the India AI mission for proactively recognizing that evaluation evidence will need to be a key part of the AI strategy for the country, setting up these centers of excellence on health."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:16",
                    "end": "00:41:44",
                    "text": " education, agriculture, et cetera. And then bringing really strong partners, like J-PAL, who has the experience, but also locally, like ICMR and many other folks who can kind of put our heads together, and I'm sure folks in the room, about what are the best ways to evaluate AI in a period where if we were stuck on one method that took us that long, we'd likely..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:44",
                    "end": "00:42:03",
                    "text": " work ourselves into irrelevance and increase the inequality of how AI plays out. Thank you so much for sharing that perspective, Zameer. Rob, I want to bring you in next from a global governance vantage point as we're thinking about what's needed to really drive adoption and policy and impact."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:03",
                    "end": "00:42:32",
                    "text": " What are the biggest privacy and governance mistakes that you see when health AI projects are scaling? And how can programs avoid them without stalling innovation? So thanks for the question. I mean, I think it's really an exciting opportunity and an exciting moment where we are sitting here, where the technology that is being built is capable of solving lots of different problems. We've heard about a bunch of them from this panel, from the speakers already."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:32",
                    "end": "00:43:01",
                    "text": " from preventing heart attacks, to diagnosing anemia, to giving personalized health guidance to people who otherwise wouldn't be served by the traditional healthcare system. So these are really exciting opportunities. Before I answer the question about privacy and security, I wanna say, I was just sitting here listening to this conversation and thinking back to the past couple of global AI summits that we've had. Four years ago, the first one was in the UK."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:01",
                    "end": "00:43:24",
                    "text": " And I think I wouldn't have been able to say the same thing four years ago, that we build general purpose technology, like the kind of technology that my company builds, and then people use it for all of these compelling use cases that actually enable health delivery to scale. And to enable the kind of health delivery that would previously have been limited to."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:24",
                    "end": "00:43:34",
                    "text": " more wealthy nations to more limited populations is something that everyone can have. And I actually think that's a really exciting opportunity right now. So."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:34",
                    "end": "00:44:03",
                    "text": " What are the big mistakes? The first one is just treating governance as a gate rather than a guide. I think very often you get into this mode where people say, hey, we're really worried about X issue. We're worried about privacy. I spent my career thinking about privacy. I worry about privacy a lot too. But I think it's important to realize that the fact that we are worried about privacy doesn't mean that we shouldn't be able to use data in novel ways in order to..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:03",
                    "end": "00:44:32",
                    "text": " deliver really positive health outcomes. We actually did a study with Stanford University on something called deliberative democracy, basically looking at broadly representative groups of people in the US and in India and asking them for their attitudes about frontier AI. And one of the things that we found is here in India, 90 plus percent of people said that they want AI to be used for cutting edge cures and that they're willing to contribute their data in privacy preserving ways to have that happen."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:32",
                    "end": "00:44:40",
                    "text": " As a starting point, it's important to know that if you're not..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:40",
                    "end": "00:45:08",
                    "text": " If you're stopping the development or stopping deployment because you're worried about those issues, rather than trying to build in a way that is privacy safe and that is secure, you're making a mistake. The other big thing, which I think we also heard, and I think you talked about earlier, is the idea of human in a loop. The traditional understanding is always, humans are the gold standard. We should make sure that we have humans in the loop and making decisions at all times. That has two consequences."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:08",
                    "end": "00:45:13",
                    "text": " You talked about mistakes, and you talked about people not getting access to care because they're bottlenecks."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:13",
                    "end": "00:45:39",
                    "text": " And so I've come to think about this problem, especially as we're getting to a point where the AI is potentially more capable than humans, faster than humans, and more reliable than humans. Thinking about humans as the architect, we should be driving the way that the technology should be used and thinking about AI as a tool that can help us scale our impact so that we don't run into this case where people who could be helped by the technology aren't receiving it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:40",
                    "end": "00:46:11",
                    "text": " And Shahed or Zamir, do you have any reactions to Rob's comments here? Rob, you remind me of a real encounter I had speaking to a skeptic, head of neurosurgery at a very big hospital. I won't mention the name. He looked at me and he said, this AI thing is a complete waste of time. And it's not going to change anything in my practice. And it was in South Africa."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:12",
                    "end": "00:46:41",
                    "text": " And he said, the only reason I'll give you five minutes is because you are my student at one point in time. And put me on the spot and said, there's the CT scan, now show me. And I was like, I didn't come and do a demo, I didn't have anything special in my pocket, I didn't have a trained model, you know, and I took out the general purpose LLM, I won't say the name, took an image, it's multimodal, you know which one, maybe."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:43",
                    "end": "00:47:18",
                    "text": " And I turned the phone around, and he looked me dead in the eye and he said, hang on, actually, this is crazy, because he then told me the clinical case. This was a case, a very real case, of a 12 year old child who was admitted to a district hospital, who the clinician on duty was late in the morning, I don't know, looked at the child and said,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:18",
                    "end": "00:47:50",
                    "text": " Child's got raised in the cranial pressure of unknown cause and for whatever reason decided to do a lumbar puncture Clinicians in the audience are gasping You know what happened next? Unfortunately the child cone by the morning the child did not survive and When he looked at the LLM result it said to him Very clearly. It was the first time I saw bold and capital letters do not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:50",
                    "end": "00:48:22",
                    "text": " lumbar puncture, whatever you do. That clinician turned around and said to me, this is something I've never seen before. And just last week, he emailed me to say he submitted his first publication to the New England Journal of Medicine on the power of AI for African communities. And that's where I think we're gonna get the change, is when clinicians start to see the evidence and trust that this will make a change for their patients."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:22",
                    "end": "00:48:44",
                    "text": " in a kind of a real life situation. Anything that you would briefly add in 30 seconds there? Before we get to one last question, I just want to say that this discussion has been really exciting for me personally, because while I can't say too much yet, J-PAL is going to very soon be..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:44",
                    "end": "00:49:09",
                    "text": " really massively expanding our work through a new initiative to support research on AI and health, specifically in low and middle income countries. And that's going to be formally announced on the 20th at the summit. So stay tuned for additional details there. But I just want to flag that with that new initiative in mind, we would be really eager to reconnect with the researchers and implementers and policymakers in the room here today."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:09",
                    "end": "00:49:36",
                    "text": " after the announcement on the 20th to speak more about how we can build the evidence base on health and AI together and make those new evaluations possible. So with that in mind, as J-PAL is looking to support even more research on AI and how to integrate promising tools into the health system, I want to hear from each of you one more time. So in one to two minutes each, based on your experience,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:36",
                    "end": "00:50:03",
                    "text": " What differentiates AI applications that succeed from those that fail? That could be data availability, regulation, evaluation, other factors. And what advice would you give to implementers who want to build AI into programs iteratively, while also staying grounded and learning what works? So I'll start with Rob first maybe, and then Shahed and Zamir, feel free to bounce off of that. Sure."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:03",
                    "end": "00:50:29",
                    "text": " So I think that, you know, I talked earlier about the downside risk of not deploying. And I think that, you know, by thinking about things in binary terms, you often run into this case where you leave opportunity on the table. And I think the projects that are successful are ones that have a dual focus. They are thinking about the way to, the upside. Who are they helping?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:29",
                    "end": "00:50:58",
                    "text": " What is the specific mechanism by which they're going to deliver that? They're thinking about governance up front. They're making sure that they're identifying the downside risks and they're building them in. So as an example, when we do our data modeling, we look at anonymization technology. So we take data and we build the data sets in ways that protect the anonymity of the people that are in it. So we don't have problems later on. And then thinking about access to data, this is something that came up earlier too."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:58",
                    "end": "00:51:20",
                    "text": " making sure that the data that's being used is broadly representative of the people that we're trying to serve. I think if you build a project and you build a model based on data from people in the US, you're going to end up with something that serves people in the US and doesn't serve people in the rest of the world. And so I think thinking about both a very clear targeted upside."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:20",
                    "end": "00:51:48",
                    "text": " very clear mechanism of action, very clear governance for how to build it responsibly, and broadly representative data so that the technology and the models that you're building can serve as many people as possible. I think those are the key attributes. Shahed, maybe over to you. Sure, yeah, and thank you. Building on that, maybe would share three things. For implementers who are thinking about incorporating AI,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:48",
                    "end": "00:52:18",
                    "text": " First would be starting with people, not the technology. For us that meant health workers understanding their needs, their challenges, their motivations as professionals deeply, and then seeing how this tool can unlock potential. Again, for us that was about unlocking moments that they provide care. Second, it was really important for us to get very precise with our definitions. What do we even mean by good care?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:18",
                    "end": "00:52:43",
                    "text": " what's in the bounds of the service, what's out of the bounds of the service. Both of those things helped essentially provide a north star for things like when we developed out the knowledge base, built in guardrails for our prompting. And again, do that in a way where we iteratively come back to it. And finally, especially because we had humans in the loop,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:43",
                    "end": "00:53:10",
                    "text": " putting that product or that tool out there in a small geography for us to learn and understand and then iterate on. We learned so much more when we actually started to communicate with families and see what were the challenges that our health workers were facing when we put it out there. So again, of course, do that in a safe way. For us, it was keeping people in the loop, health workers specifically."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:10",
                    "end": "00:53:41",
                    "text": " and monitoring all the conversations, but what we learned was tremendous in those initial days. Thanks, Amy. I think a couple of things. I think we're going to hear this week lots about models and the cool tricks they're going to do and so on. I think it's a 10% problem. I think models, the accuracy will be 10% of what we need."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:41",
                    "end": "00:54:11",
                    "text": " I think there's a 20% on how does this actually integrate, like what sort of infrastructure do you need? And then I think 70% of this is gonna come down to change management and how we engage people and take people along on the journey. And I think if we neglect that, I think this is gonna be an uphill battle for some time. I also think that as we think about evidence and adoption, there's a very real cost issue."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:11",
                    "end": "00:54:38",
                    "text": " One of the studies we did in partnership with Bath over here showed that the clinician assist consult actually only costs four US cents and the accuracy was up to 97%. That may be expensive when you expand it over thousands of users, but actually the cost of compute will continue to come down. It's 240 times less than what it was two and a half years ago."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:38",
                    "end": "00:55:00",
                    "text": " So you can easily start to see that this becomes scalable and sustainable. I think that one of the biggest game changers, if I was forced to bet, I don't bet, but if I was forced to bet on what I think could be the biggest game changer for development on AI, I think it's personalized health coaches."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:00",
                    "end": "00:55:25",
                    "text": " in the pocket of every single individual that can speak in their language, that has context and memory over time, that eventually doesn't only serve health needs, but is your coach for education, is your coach for financial, but just a coach that you can trust with decision-making over time."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:25",
                    "end": "00:55:58",
                    "text": " Thank you so much, Samir. That's a really exciting vision to lay out, and I think a good way to close. It was so great from hearing from each of you. I think this was just a great way to highlight the need for more evidence, the excitement around AI and health, and just really appreciate you taking the time to be here, and all of you for attending. Thank you. Thank you so much to all of our AI and health speakers. It's really just amazing to hear their perspectives from every type of organization from all around the world. I hope you're all learning a lot."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:58",
                    "end": "00:56:07",
                    "text": " We'll take a short break and we'll rejoin at 11.40. Please try to exit from the right side of the stage. Thank you very much."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Spanish"
},
{
    "_id": "69941d9025c9f4e0d7e70168",
    "projectId": "678630",
    "projectName": "AI in Health Saving Lives at Scale - Hindi",
    "projectDescription": "AI in Health Saving Lives at Scale - Hindi",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T07:49:41.460Z",
    "projectUpdatedDate": "2026-02-17T07:49:41.460Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T08:12:36.941Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "AI in Health Saving Lives at Scale.mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771314561043.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:59",
                    "end": "00:02:42",
                    "text": " Excuse me, we request you to please exit quickly so that we can start our next session. Welcome back to J-PAL's AI for Social Good Impact That Works seminar. Our next session is AI in Health, Saving Lives at Scale. Please join me in welcoming Ziad Obermeyer. Ziad is an associate professor of health policy and management at the UC Berkeley School of Public Health, where he works at the intersection of machine learning, medicine, and health policy."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:42",
                    "end": "00:03:09",
                    "text": " He helped set up the computational precision health program and his research focuses on how machine learning can help clinicians make better decisions. A fun fact about Ziad is that he was named one of the 100 most influential people in AI by 10 magazine. Ziad, over to you. Thank you so much. So I thought I would tell you a story about when I first came to my job at Berkeley. So this was in 2018."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:09",
                    "end": "00:03:39",
                    "text": " I was a junior, untenured faculty member, and I just signed up to give the first seminar of the year in the economics department. Now, the economics department at Berkeley is friendlier than the economics department at other institutions affiliated with J-PAL, but they are still economists and still very hardcore, so I was nervous. And about three or four days before the seminar, I started feeling a little unwell."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:39",
                    "end": "00:04:07",
                    "text": " My stomach was hurting. I just didn't feel right. And I made the mistake of mentioning this to my wife, who said something like, oh, is your tummy hurting? Are you a little nervous about your... Anyway, she was not very sympathetic. So I prepared for my talk. I prepared for my first class that I was teaching. And I gave the seminar. And during the seminar, instead of feeling just globally not well,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:07",
                    "end": "00:04:36",
                    "text": " I started feeling unwell in a particular place in my stomach, which is the bottom right side of my stomach. Some of you might know what I have already, but I want to first make an observation that this made me realize, which is that when people make really important health decisions, like do I come into the hospital, do I ask a doctor about this, people are making those decisions that are very, very important with very little data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:36",
                    "end": "00:05:03",
                    "text": " So at the end of the seminar, I actually drove myself to the emergency room near Berkeley and I got a CT scan. And this is my CT scan. And some of you might be able to see what I have or guess what I have, but I had acute appendicitis and it had perforated and I needed to have my appendix removed. I think what makes this story surprising and a little bit embarrassing for me"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:03",
                    "end": "00:05:26",
                    "text": " is that I trained as an emergency room doctor. So there's a few things that you're not supposed to miss as an emergency room doctor, and this is one of them. And yet, I missed it in myself for four or five days. So I think what this highlighted to me is how critically important it is."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:26",
                    "end": "00:05:48",
                    "text": " to be able to have access to this kind of data. The kind of data that normally live only inside of the health system and that you can only have access to if you're lucky enough to be close to an emergency room with a CT scanner to have access both financial and linguistic and socioeconomic to these kinds of technologies. Because without data like this,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:48",
                    "end": "00:06:17",
                    "text": " There's no diagnosis of critical illnesses. And without diagnoses, there's no treatments. So this data is really the key to a lot of things that happen in the health system. And yet, in order to get access to this, you have to be part of the health system. You have to have access to begin with. So there's a circularity to this problem. And it locks out many, many people, of course, not just in the US, but even more around the world."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:17",
                    "end": "00:06:34",
                    "text": " This kind of underdiagnosis is, of course, not just limited to acute surgical things like appendicitis, but to many, many different conditions. And a particularly salient example to many people happens even in high income countries, which is heart attack. So there are these studies that follow."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:34",
                    "end": "00:06:53",
                    "text": " big cohorts of people across time in the US, and every so often they do an MRI of the heart to detect scarring in the wall of the heart that indicates that this person had a heart attack. And in these studies, when they look at all of the people who have scar in their heart consistent with a heart attack,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:53",
                    "end": "00:07:18",
                    "text": " Most studies find that the majority of those heart attacks are silent. They are not known to the patient and not known to the doctor and not diagnosed, either because the symptoms were subtle or because the person never sought care to get those symptoms diagnosed. So as you can imagine, this is a huge problem even in high income countries. It's an even bigger problem elsewhere in the world. And so there's data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:18",
                    "end": "00:07:33",
                    "text": " from South India, from Tamil Nadu, that asks people about things like high blood pressure and diabetes and asks them, do you have these things? And then does the objective measure of those things. And even for these basic risk factors for heart attacks, we..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:33",
                    "end": "00:08:02",
                    "text": " We can't do cardiac MRIs in a lot of places, but even for these risk factors, there's dramatic underdiagnosis and under-awareness of the things that cause heart attack. So this is a problem everywhere, and it's a problem that really urgently needs solutions. So I thought I could tell you, I'm going to tell you about half of the solution, and instead of just telling you, I'm going to show you, and you'll see more of this from my colleague Shredda in a few moments."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:02",
                    "end": "00:08:18",
                    "text": " This device that I'm holding in my hand is a mobile handheld electrocardiogram. So the two silver sides are electrodes. You put your fingers on it and it records the electrical field generated by your heart as it beats."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:18",
                    "end": "00:08:43",
                    "text": " And I think these technologies are really miraculous because they cost so little. That device costs $60. In the US, you can order it on Amazon and get it delivered in the same day. It connects to your phone or tablet via Bluetooth with no wires. And it requires absolutely no training to do an electrocardiogram. So this is really an incredible technology that's become commoditized over the past five to 10 years."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:43",
                    "end": "00:09:06",
                    "text": " But the reason that these kinds of devices have not had the impact that any of us would have hoped is because there's a key constraint, which is that what do you do once you generate this waveform? Well, you have to send it to a doctor for interpretation. This is too complex for patients, and even for some doctors like myself have trouble interpreting electrocardiogram waveforms."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:06",
                    "end": "00:09:35",
                    "text": " And so it's great that you can generate this data wherever you are in the world. But if you still need to send it to a doctor to get interpreted and to translate those data into decision making, you haven't solved the key bottleneck, which is access to the health care system. So even though these devices are wonderful and miraculous, they're only part of the solution to democratizing access to diagnosis around the world. So this is an AI summit. Could AI help?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:35",
                    "end": "00:10:01",
                    "text": " Yes. But it's important to remember that there's nothing magical about artificial intelligence. And in fact, artificial intelligence is just data in a very literal sense. So in an ideal world, what you'd have as far as a data set to train your AI algorithm is a data set that links data from these cheap devices to some ground truth outcome about heart attack, appendicitis, et cetera, et cetera."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:01",
                    "end": "00:10:25",
                    "text": " Unfortunately, those data on the ground truth are quite hard to find even in high-income countries and certainly in low- and middle-income countries. And so what we often do when we're training AI tools is we make a key and very pernicious substitution, which is we say, okay, well, we can't get the ground truth linked to these electrocardiogram waveforms."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:25",
                    "end": "00:10:52",
                    "text": " So instead, we're going to ask doctors to look at the waveforms and tell us what the doctor thinks about that waveform. So we substitute human judgment and human interpretation for some sense of the ground truth of, does this person really have a heart attack or not? Let's just ask the human. And the problem with humans, many of my good friends are humans, so this is nothing about humans in general, but humans have biases and humans make mistakes. And human knowledge..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:52",
                    "end": "00:11:22",
                    "text": " about the human body and many other things is incomplete. And so when we automate human judgment, we also run the risk of automating all of the problems with human judgment along with it. And I think that's what makes it so critically important to train artificial intelligence not on doctors, but on patients, on what happens to patients and their outcomes. So let me tell you about some work that I am doing in collaboration with J-PAL South Asia in a few"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:22",
                    "end": "00:11:37",
                    "text": " villages in Tamil Nadu. So this is a line of people queuing up for a district health center. And what we did over the summer of 2023 is set up health camps, a long-standing tradition where we just..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:37",
                    "end": "00:12:05",
                    "text": " had a facility outside of this district health center where, thanks to the tireless work of our colleagues at J-PAL South Asia, we rented out a bunch of very fancy equipment that you can normally only get inside of hospitals. So for example, we rented out a cardiac ultrasound machine and a technologist to acquire the images of cardiac ultrasounds. We hired cardiologists to interpret those images and look."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:05",
                    "end": "00:12:33",
                    "text": " among other things, for signs that this person had had a heart attack, signs of scarring in their heart. And at the same time, we did data collection from these cheap $50 devices. So here you see two patients that were participants in this health camp. This young lady is having an electrocardiogram done and sent to the tablet in front of her. And the gentleman over here is getting a cardiac ultrasound, which serves as our way of understanding."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:33",
                    "end": "00:12:53",
                    "text": " if this person has had a heart attack by looking at wall motion abnormalities in the walls of his heart. So every patient got both the fancy expensive tests from the hospital and the cheap low cost tests. And of course, if we found any abnormalities, we communicated those back to the patient and their doctors as part of this study."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:54",
                    "end": "00:13:18",
                    "text": " So so the eventual goal of this is to create a way to screen people for a number of serious health conditions including prior potentially silent heart attacks in the comfort and safety of their own home either by doing the electrocardiogram themselves or with the assistance of a community health worker and Just to give you a sense of how well even this early version of the algorithm is working"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:18",
                    "end": "00:13:40",
                    "text": " when we flag the highest risk, two, three, four, five percent of the population, just general population of people visiting an ambulatory health center, 10% have actually had signs of a prior heart attack, and that's compared to about 0.2% in the general population of people that we were looking at. So we're finding very high-risk people, and..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:40",
                    "end": "00:13:54",
                    "text": " people who, even though the algorithm is saying that they're high risk and they do indeed have silent heart attack on their cardiac ultrasound, they lack the traditional risk factors like blood pressure and cholesterol and diabetes."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:54",
                    "end": "00:14:22",
                    "text": " that are traditional because they were defined in Western populations. And as we know, South Asian populations have different risk profiles and different risk factors that this electrocardiographic approach is able to find despite the fact that they lack those traditional risk factors. So I think this highlights the fact that these algorithms can be trained and tuned to populations all over the world that might be different from the populations in which most of our medical knowledge is developed."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:22",
                    "end": "00:14:51",
                    "text": " When we put this through the cost-effectiveness of screening, in other words, sending high-risk people for confirmatory tests at the health center and treating them with the medications that you need to be on for people who have had prior heart attacks, this comes out to be cost-effective, even using Indian guidelines. So it's around $2,000 per disability-adjusted life year, even for this very early version of the algorithm, which is continuously improving as we collect more data. This is very exciting."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:51",
                    "end": "00:15:06",
                    "text": " And the next step is actually, this wouldn't be J-PAL unless the next step were a randomized evaluation. And so we're doing a randomized evaluation where we're gonna start screening people either with traditional risk factors like blood pressure, et cetera."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:06",
                    "end": "00:15:28",
                    "text": " or this AI enhanced electrocardiogram screening approach and referring the high risk people for confirmatory testing in both groups and seeing which approach finds more people that we want to find, people with prior heart attacks. So one of the exciting things about doing this project is that unlike a lot of the research that I've done in the past,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:28",
                    "end": "00:15:58",
                    "text": " This is a real thing. It's a device. And it's already interacting with real people. And so to give you a sense of that, if you don't mind, I'm going to introduce Ms. Shreda Rajesh, who is an instrumental part of our team with J-PAL South Asia. And we're going to violate every principle of giving a talk by doing a live demo and hoping that it works. So if you'll bear with us for a second, we're going to get this set up. And then you will see my own electrocardiogram being put through."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:58",
                    "end": "00:16:29",
                    "text": " our web application and screening me in real time for my own risk of prior heart attack. Luckily, since I'm a physician, nobody needs to take responsibility for the outcome of the test besides myself. All right. So we're getting it hooked up. And I'm hoping it works because we tried it before the talk. But of course, live demos have a way of frustrating your expectations."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:40",
                    "end": "00:17:53",
                    "text": " We're gonna switch to the other podium. One thing I'll just mention as we're getting set up is that this approach is not limited to mobile electrocardiograms."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:53",
                    "end": "00:18:21",
                    "text": " those of you who have Apple watches, of course, know that there's increasing amounts of mobile technology built into your watch that can be fed in, but these electrocardiograms are not the only low-cost data collection devices that we have available to us. So in the course of this study in Tamil Nadu, we also collected a number of other low-cost devices. So those of you who sadly had to live through COVID here, no doubt had"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:21",
                    "end": "00:18:40",
                    "text": " mobile portable pulse oximeters. Those pulse oximeters, in addition to generating oxygen percentage numbers, generate a waveform that actually represents the blood flow through your finger that the pulse oximeter is measuring to get that fraction of oxygen."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:40",
                    "end": "00:19:07",
                    "text": " But that waveform has an enormous amount of information about how your heart is working, how your capillaries are working, how your blood vessels are allowing the flow of blood through all of these channels. So that also can plug into algorithms. The gentleman in the front row is wearing an aura ring or a related ring that can also measure this kind of thing. Continuous temperature monitoring is another feature of these rings. There are smartphone-based attachments."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:07",
                    "end": "00:19:10",
                    "text": " that can take a picture of your retina."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:10",
                    "end": "00:19:39",
                    "text": " A fun fact about the retina, you got a lot of fun facts about the speakers from Audrey. A fun fact about the retina is that it is part of your central nervous system. It's the only exposed part of your central nervous system. And with a $200 smartphone attachment, we can actually get a view into the central nervous system that can be useful for diagnosing stroke, intracranial hemorrhage from trauma, retinal disease, other kinds of diseases of the eye. Okay."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:39",
                    "end": "00:20:10",
                    "text": " Thank you so much for getting this working, Shyada. Would you mind telling us a little bit about what we're going to do? Of course. Thanks so much, Ziad. Sorry, this took a bit of time. Good morning, everyone. We'd now like to demonstrate what this looks like in practice. So all a participant has to do is basically place their fingers on this very small device that's connected to my phone, as you see. And that's it. No wires, no hospitals, and just 30 seconds. Ziad, I hope you don't mind."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:25",
                    "end": "00:20:46",
                    "text": " Sounds good. So what this device does is it records a single-led ECG, which is the electrical signal of your heart. And once we capture that, I enter the participant's ID onto our app that we developed in-house. And behind the scenes, our AI model, it analyzes this waveform, the pattern of the heart, almost instantly."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:46",
                    "end": "00:21:40",
                    "text": " which it has learned to do from the thousands of real life cases that we've collected from the field. And within seconds, you get a risk score that we'll show you just in a bit. So what this risk score tells us is based on, just again, 30 seconds with this single LED ECG alone, how likely it is that this participant has had a silent heart attack in the past. The stress of the live demo is elevating my... I hope it hasn't provoked a heart attack. That would be unfortunate."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:41",
                    "end": "00:22:41",
                    "text": " Are there any other doctors in the audience? All right. I think we're ready to go. So what you're seeing now is the ECG was automatically uploaded into a web application and generating a low risk score. Thank goodness for that. Thank you so much. And thank you so much to our tech team here. Live demos are an unpleasant ask for the tech team. And I really appreciate your help. Thank you. OK."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:41",
                    "end": "00:23:08",
                    "text": " So as I mentioned, this is not just about ECGs. You could imagine a setup where a very small amount of hardware totaling under 20,000 US dollars comprised of an ECG, a retinal photo, a mobile chest x-ray unit, a handheld ultrasound device can actually be put together in a district health center or even in a little kiosk in a shopping center somewhere else."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:08",
                    "end": "00:23:30",
                    "text": " and could generate an enormous amount of raw device data that can feed into a number of diagnoses that are of critical importance to lots of causes of burden of disease in low and middle income countries, but in high income countries as well. So I think the scope and the scale of these kinds of diagnostic devices and approaches is really enormous."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:30",
                    "end": "00:23:60",
                    "text": " And I'll just mention one very personal application of this for me to give you a sense of what this can do in life. So my wife and I have twin daughters who are almost two years old. And the way that we found out that we were having twin daughters is because I, as part of my practice in emergency medicine, I own a handheld ultrasound device. So this costs $2,000. And it can generate pictures."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:60",
                    "end": "00:24:24",
                    "text": " like this of our two future daughters. So we knew that we were having twins well before most people know because we are lucky enough to have access to this technology. And I think democratizing and broadening access to both the hardware and the software components is really incredibly important as we seek to diagnose illnesses across the world. The last thing I'll say is that I think one of the really exciting parts of this work"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:24",
                    "end": "00:24:42",
                    "text": " is that it represents a fundamentally new way that we can develop AI algorithms in a sense that many people I think in well-resourced settings find counterintuitive. Because this is an innovation that starts here in India. Why does it start here?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:42",
                    "end": "00:25:04",
                    "text": " because there's very high volume of data at very low cost, but also because there's an enormous amount of human capital here. Engineering and AI talent, people who know how to build hardware and train AI products, there's the ability to move quickly, and at the same time, and especially thanks to the team at J-PAL, there's scope for very rigorous testing."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:04",
                    "end": "00:25:25",
                    "text": " This kind of innovation can scale seamlessly to the rest of the world. One interesting fact about our algorithm is that we pre-trained our AI model on electrocardiogram data from Sweden and the US. And that actually helped a lot for improving the model performance in India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:25",
                    "end": "00:25:47",
                    "text": " And that interoperability, if you will, of the ECG waveform is a really exciting part of building algorithms that run on this kind of physiological data. Because as we've tested algorithms that are trained in one place, for example, in Sweden, they generalize incredibly well to the US, to India, to Taiwan, to wherever we've tested them. Because..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:47",
                    "end": "00:25:57",
                    "text": " we do electrocardiograms the same way in all of those places, and because the heart is fundamentally wired in the same physiological way in all of those places."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:57",
                    "end": "00:26:23",
                    "text": " And so those algorithms can generalize and they can solve barriers to access that are widespread not just here in India but all over the world where people struggle to get access to the formal health system because of distance, because of poverty, because of discrimination, because of cultural or linguistic barriers. So I'm very excited about that and I would urge anyone who's working on this kind of work here with data is to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:23",
                    "end": "00:26:59",
                    "text": " protect and steward the secret sauce that you have here in India. Because it's really such a valuable resource, but not just to India, but to the rest of the world, to be able to develop AI products in a way that is safe, ethical, but also rigorous and fast. Thank you very much. Thank you so much, Siyad, for that insightful presentation. And may your risk of heart attack remain low for many years. Please rejoin the stage for a group photo with the rest of the session speakers."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:59",
                    "end": "00:27:27",
                    "text": " It's my pleasure to invite to the stage the panelists for the AI and Health Saving Lives at Scale session to join Ziad on stage. The panel will be moderated by Amy Barnes. Amy, please join us on stage. Amy is a policy manager at J-PAL Global, where she leads J-PAL's health sector. With all the promising applications of AI and health, Amy has increasingly been pushing to increase the available evidence on the real-world effects of AI solutions on both providers and their patients."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:27",
                    "end": "00:27:56",
                    "text": " A fun fact is that Amy started her time at J-PAL focused on policies to strengthen governance and public service delivery and to reduce crime and conflict. Now please join me in welcoming our panelists, starting with Shahed Alam. Shahed is the co-founder and co-CEO of Nura Health. Since its founding in 2014, Nura Health has been transferring health skills to family caregivers and patients. So far, Nura has trained over 43 million caregivers and patients across India, Bangladesh,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:56",
                    "end": "00:28:27",
                    "text": " Indonesia, and Nepal. And now Shahed is using AI to take Noor's mission to the next level. A fun fact about Shahed is that he earned his MD from Stanford. He's an associate faculty member at Ariadne Labs at Harvard Chan School of Public Health. And he's also been a TED speaker. Also on stage, Zamir Bray. Zamir is the deputy director of technology diffusion at the Gates Foundation. In this capacity, Zamir leads the foundation's strategic coordination on the equitable and safe use of AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:27",
                    "end": "00:28:55",
                    "text": " including work like the internal AI task force, AI related grant challenges, and external AI advisory efforts. A fun fact about Zamir is that he trained as a medical doctor, and he also earned an MBA. And on top of that, he has a PhD in health systems and innovation, and a master of law from the University of Cape Town. And he serves on the board of Groot Sure Hospital in Cape Town. Finally, Rob Sherman is the vice president of policy and deputy chief privacy officer at MEDA."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:55",
                    "end": "00:29:28",
                    "text": " where he works to set the governance and safeguards for how AI-powered products are designed and deployed responsibly. A fun fact about Rob is that before joining Meta, he was a lawyer advising major tech and digital tech companies on privacy and data security, and he was recognized by Chambers USA as one of the nation's leading media regulatory lawyers. We have very great people on stage with us today. Amy, over to you. Thank you so much, Audrey. The audience can send in questions by scanning this mentee code."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:28",
                    "end": "00:29:53",
                    "text": " And thank you, Amy. Awesome. First, just so excited to be here with all of you today. Hear your thoughts around the rapid evolution of AI in health systems, how it can be integrated, your thoughts on evaluation, of course. Audience members, please do submit your questions. But I'll just take the moderator privilege and start diving in with one. I want to start locally with Shahed. I know we're so far, but."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:53",
                    "end": "00:30:22",
                    "text": " Going to start with you and your organization, Nora Health, which does a lot of work, including in India, to train family members to advocate more for their loved ones in hospitals and to provide better care when they're leaving the hospitals to return home together. So Shahed, Nora is already working across multiple countries and already recognized for its impact. At this stage in Nora Health's growth, what problem are you trying to solve and what's the role of AI in your solution moving forward?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:22",
                    "end": "00:30:52",
                    "text": " Thanks so much for having us and also for that question. At Norah Health, again, we support the family caregiver of the patient with the knowledge, skills, and confidence they need to care for their loved ones. We work with governments and health systems in order to incorporate these programs that really support caregivers. The anchor of our work really starts in healthcare facilities when families are anxiously waiting by their loved one"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:52",
                    "end": "00:31:17",
                    "text": " And we provide them a skill-based training by a trusted health worker in person to engage them in their loved one's care. Now, once they go home, families may, they'll face some sort of a challenge. They may need a reminder or a nudge to remember what to do in order to care for their loved ones. And so that's where..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:17",
                    "end": "00:31:36",
                    "text": " For us, technology and our mobile messaging service steps in. So after an in-person training to continue to receive nudges and reminders for families once they go home. From the very beginning, as we've developed this service, we've not developed this as a chat bot because we wanted to always keep..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:36",
                    "end": "00:32:05",
                    "text": " that human connection between patients and families at home and a health worker at the back end as a critical and core part of what we do. But as we've scaled it, we have seen that that whole process can become very constrained. We went from response times that were pretty quick to response times that needed to take more than 10 hours. So that's where AI really stepped in for us."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:05",
                    "end": "00:32:28",
                    "text": " to be able to address this challenge of keeping high quality support to families once they're at home, but also ensuring that response times don't balloon to something like 10 hours. We had families who loved the service. They would reach out to us and they would ask our nurses to name their newborn, but they were getting increasingly frustrated with how long it took."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:28",
                    "end": "00:32:58",
                    "text": " So we sat down with our health workers, and that's really where it begins, is to define what are the challenges that people are facing, and for us those are health workers, and started to look at a lot of manual tasks that they were doing, looking up things in an FAQ database, annotating things manually, and then with them, alongside us, we started to co-develop an AI co-pilot that supports health workers in answering questions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:58",
                    "end": "00:33:18",
                    "text": " providing them context on the patient and family that they're supporting. And importantly, also tracking operational and other data so that we understand the quality of the responses in the system overall. And with that in place, we've been able to reduce those 10 hours down to 30 minutes."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:18",
                    "end": "00:33:49",
                    "text": " We've improved retention, so families, more than 30% of families keep coming back to us for more support. And importantly, the health workers are happier. They're getting to focus on what motivates them, which is to provide care versus to fill out a bunch of forms on an Excel sheet. And to follow up on that, I've heard rumors at least that you're planning a randomized evaluation later this year on this. Could you say more about why you're exploring an RCT now?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:49",
                    "end": "00:34:17",
                    "text": " given that you already have some evidence of Nora's impact. You are correct. For us, impact and evaluations is not a one-time thing. As we evolve, as the world changes, we need to continually update the evidence that we have and the knowledge that we have. The RCT is really critical for us because while we've done evaluations on the model, some of the product metrics, how it's affecting users,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:17",
                    "end": "00:34:31",
                    "text": " Ultimately, what we hold ourselves accountable to is having an impact on health outcomes, things like newborn morbidity and mortality. And the RCT is a way for us to actually, in a more definitive way, answer that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:31",
                    "end": "00:35:00",
                    "text": " In a simple way, it is to understand the impact on this incredibly important outcome, which is newborn mortality, and to understand also incrementally what does the AI-supported tool provide versus our standard program, which is the in-person initiative. And hopefully with that data, ultimately what we want to do is to have family caregiver support be the standard of care across health systems globally."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:00",
                    "end": "00:35:29",
                    "text": " RCTs and this evidence that we're generating will allow us to do is to provide information on what works, what doesn't, to governments so that they can use their scarce resources in supporting caregivers most effectively. Thank you so much. Zamir, I want to bring you in next and hear from you from the perspective of a funder who's thinking about technology diffusion a lot in health systems around the world. Because AI tools are evolving so rapidly, how should funders and implementers like Nora Health"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:29",
                    "end": "00:35:58",
                    "text": " design evaluations that can keep pace with that reality, while still producing evidence that's credible enough to drive adoption and policy. Thanks, Amy. I must just depart for a second, and I wonder if Ziad is still around, because when Ziad made his admission about being a clinician that missed the diagnosis, I realized that's why I was invited to this panel. I too missed the diagnosis, and I was just remembering that..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:58",
                    "end": "00:36:24",
                    "text": " It was actually my first trip to India 15 years ago when I came here, had a great time and went back home and had a classic swinging fever for six days and completely missed that it was typhoid fever and too ended up in the surgery room with a cholecystitis."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:24",
                    "end": "00:36:53",
                    "text": " To Ziad, this is a bit of a clinician anonymous admission session, so thanks for creating a safe space. But Amy, to get to your question and also just to pick up on where Shayid left off, I think the role of evaluations for decision makers for governments where resources are scarce is absolutely critical. And I can already tell from..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:53",
                    "end": "00:37:22",
                    "text": " the proceedings this week, that this will be one of the key messages we take back, which is we need to double down on evaluation, particularly giving decision-makers the right kind of evidence at the right time. I'm a big fan of RCTs. They have fundamentally changed the global health landscape. But there are two kind of issues we need to grapple with. The one is..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:22",
                    "end": "00:37:52",
                    "text": " the models change very fast. And I see some colleagues smiling with me because this is a very real issue. You know, I was doing a little bit of homework on, like, how long does this take? It takes us six months to design the RCT. Let's do the RCT for maybe 24 months and then take 18 months to publish the data. That means that if we all left here today and we started designing, Shahid..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:52",
                    "end": "00:38:20",
                    "text": " maybe 2029, we have some good results that you can put out there, right? I'm not saying that that's what we should do. The other reality is that policymakers, like in India, like in many countries we work in as the Gates Foundation, are really hoping that AI brings new promise, brings new tools to frontline workers that it's never been able to do before. And so the policy windows"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:20",
                    "end": "00:38:46",
                    "text": " the policy cycles don't track to publication cycles. Policy decisions are often made because of resource constraints, or there's an outbreak, or there's some political economy issue that drives the decision. And so being receptive to that means that we're going to have to challenge ourselves on conventional models of doing..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:46",
                    "end": "00:39:16",
                    "text": " RCTs or long trial designs. That shouldn't mean that we compromise the evidence generation. So just to be clear, we have to have good evidence. But in the early phases, we can start thinking about, does this give us a safety signal? Can we do something in short step, in three to six months that says, this looks like this could be safe, could be interesting?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:16",
                    "end": "00:39:46",
                    "text": " and very much to what the previous speaker said. Maybe those are short, inexpensive, quick experiments that we can run. Then we can move into kind of pragmatic trials, adaptive trial designs that don't take us two or three years, but give us the kind of evidence around adoption. And this is really key. We've seen lots of models perform really well in labs and test stops."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:46",
                    "end": "00:40:17",
                    "text": " they beat every single benchmark. And there's no doubt that most models today are better than clinicians. Those results don't hold up when you hit the field. Why? There's some fundamental issues. You know, one trial we saw was that the clinicians actually didn't trust the AI. So adoptions were stuck at about four, five, six percent until"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:17",
                    "end": "00:40:47",
                    "text": " someone actually sat with the clinicians, went through their own clinical cases and their mistakes, and showed them that somebody in the room, there were 100 clinicians in a trial in Kenya, and one of them was making the mistakes. And those mistakes were significant and catastrophic, that they then sat around the table and said, hang on, there's something we need to embrace about the technology. And then we saw adoption surge to upwards of 60%."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:47",
                    "end": "00:41:16",
                    "text": " still not at 100%. So I think that how we think about evidence is what's the right methodology for the right time. And again, I want to just applaud the Indian government and the India AI mission for proactively recognizing that evaluation evidence will need to be a key part of the AI strategy for the country, setting up these centers of excellence on health."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:16",
                    "end": "00:41:44",
                    "text": " education, agriculture, et cetera. And then bringing really strong partners, like J-PAL, who has the experience, but also locally, like ICMR and many other folks who can kind of put our heads together, and I'm sure folks in the room, about what are the best ways to evaluate AI in a period where if we were stuck on one method that took us that long, we'd likely..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:44",
                    "end": "00:42:03",
                    "text": " work ourselves into irrelevance and increase the inequality of how AI plays out. Thank you so much for sharing that perspective, Zameer. Rob, I want to bring you in next from a global governance vantage point as we're thinking about what's needed to really drive adoption and policy and impact."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:03",
                    "end": "00:42:32",
                    "text": " What are the biggest privacy and governance mistakes that you see when health AI projects are scaling? And how can programs avoid them without stalling innovation? So thanks for the question. I mean, I think it's really an exciting opportunity and an exciting moment where we are sitting here, where the technology that is being built is capable of solving lots of different problems. We've heard about a bunch of them from this panel, from the speakers already."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:32",
                    "end": "00:43:01",
                    "text": " from preventing heart attacks, to diagnosing anemia, to giving personalized health guidance to people who otherwise wouldn't be served by the traditional healthcare system. So these are really exciting opportunities. Before I answer the question about privacy and security, I wanna say, I was just sitting here listening to this conversation and thinking back to the past couple of global AI summits that we've had. Four years ago, the first one was in the UK."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:01",
                    "end": "00:43:24",
                    "text": " And I think I wouldn't have been able to say the same thing four years ago, that we build general purpose technology, like the kind of technology that my company builds, and then people use it for all of these compelling use cases that actually enable health delivery to scale. And to enable the kind of health delivery that would previously have been limited to."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:24",
                    "end": "00:43:34",
                    "text": " more wealthy nations to more limited populations is something that everyone can have. And I actually think that's a really exciting opportunity right now. So."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:34",
                    "end": "00:44:03",
                    "text": " What are the big mistakes? The first one is just treating governance as a gate rather than a guide. I think very often you get into this mode where people say, hey, we're really worried about X issue. We're worried about privacy. I spent my career thinking about privacy. I worry about privacy a lot too. But I think it's important to realize that the fact that we are worried about privacy doesn't mean that we shouldn't be able to use data in novel ways in order to..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:03",
                    "end": "00:44:32",
                    "text": " deliver really positive health outcomes. We actually did a study with Stanford University on something called deliberative democracy, basically looking at broadly representative groups of people in the US and in India and asking them for their attitudes about frontier AI. And one of the things that we found is here in India, 90 plus percent of people said that they want AI to be used for cutting edge cures and that they're willing to contribute their data in privacy preserving ways to have that happen."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:32",
                    "end": "00:44:40",
                    "text": " As a starting point, it's important to know that if you're not..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:40",
                    "end": "00:45:08",
                    "text": " If you're stopping the development or stopping deployment because you're worried about those issues, rather than trying to build in a way that is privacy safe and that is secure, you're making a mistake. The other big thing, which I think we also heard, and I think you talked about earlier, is the idea of human in a loop. The traditional understanding is always, humans are the gold standard. We should make sure that we have humans in the loop and making decisions at all times. That has two consequences."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:08",
                    "end": "00:45:13",
                    "text": " You talked about mistakes, and you talked about people not getting access to care because they're bottlenecks."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:13",
                    "end": "00:45:39",
                    "text": " And so I've come to think about this problem, especially as we're getting to a point where the AI is potentially more capable than humans, faster than humans, and more reliable than humans. Thinking about humans as the architect, we should be driving the way that the technology should be used and thinking about AI as a tool that can help us scale our impact so that we don't run into this case where people who could be helped by the technology aren't receiving it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:40",
                    "end": "00:46:11",
                    "text": " And Shahed or Zamir, do you have any reactions to Rob's comments here? Rob, you remind me of a real encounter I had speaking to a skeptic, head of neurosurgery at a very big hospital. I won't mention the name. He looked at me and he said, this AI thing is a complete waste of time. And it's not going to change anything in my practice. And it was in South Africa."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:12",
                    "end": "00:46:41",
                    "text": " And he said, the only reason I'll give you five minutes is because you are my student at one point in time. And put me on the spot and said, there's the CT scan, now show me. And I was like, I didn't come and do a demo, I didn't have anything special in my pocket, I didn't have a trained model, you know, and I took out the general purpose LLM, I won't say the name, took an image, it's multimodal, you know which one, maybe."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:43",
                    "end": "00:47:18",
                    "text": " And I turned the phone around, and he looked me dead in the eye and he said, hang on, actually, this is crazy, because he then told me the clinical case. This was a case, a very real case, of a 12 year old child who was admitted to a district hospital, who the clinician on duty was late in the morning, I don't know, looked at the child and said,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:18",
                    "end": "00:47:50",
                    "text": " Child's got raised in the cranial pressure of unknown cause and for whatever reason decided to do a lumbar puncture Clinicians in the audience are gasping You know what happened next? Unfortunately the child cone by the morning the child did not survive and When he looked at the LLM result it said to him Very clearly. It was the first time I saw bold and capital letters do not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:50",
                    "end": "00:48:22",
                    "text": " lumbar puncture, whatever you do. That clinician turned around and said to me, this is something I've never seen before. And just last week, he emailed me to say he submitted his first publication to the New England Journal of Medicine on the power of AI for African communities. And that's where I think we're gonna get the change, is when clinicians start to see the evidence and trust that this will make a change for their patients."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:22",
                    "end": "00:48:44",
                    "text": " in a kind of a real life situation. Anything that you would briefly add in 30 seconds there? Before we get to one last question, I just want to say that this discussion has been really exciting for me personally, because while I can't say too much yet, J-PAL is going to very soon be..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:44",
                    "end": "00:49:09",
                    "text": " really massively expanding our work through a new initiative to support research on AI and health, specifically in low and middle income countries. And that's going to be formally announced on the 20th at the summit. So stay tuned for additional details there. But I just want to flag that with that new initiative in mind, we would be really eager to reconnect with the researchers and implementers and policymakers in the room here today."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:09",
                    "end": "00:49:36",
                    "text": " after the announcement on the 20th to speak more about how we can build the evidence base on health and AI together and make those new evaluations possible. So with that in mind, as J-PAL is looking to support even more research on AI and how to integrate promising tools into the health system, I want to hear from each of you one more time. So in one to two minutes each, based on your experience,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:36",
                    "end": "00:50:03",
                    "text": " What differentiates AI applications that succeed from those that fail? That could be data availability, regulation, evaluation, other factors. And what advice would you give to implementers who want to build AI into programs iteratively, while also staying grounded and learning what works? So I'll start with Rob first maybe, and then Shahed and Zamir, feel free to bounce off of that. Sure."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:03",
                    "end": "00:50:29",
                    "text": " So I think that, you know, I talked earlier about the downside risk of not deploying. And I think that, you know, by thinking about things in binary terms, you often run into this case where you leave opportunity on the table. And I think the projects that are successful are ones that have a dual focus. They are thinking about the way to, the upside. Who are they helping?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:29",
                    "end": "00:50:58",
                    "text": " What is the specific mechanism by which they're going to deliver that? They're thinking about governance up front. They're making sure that they're identifying the downside risks and they're building them in. So as an example, when we do our data modeling, we look at anonymization technology. So we take data and we build the data sets in ways that protect the anonymity of the people that are in it. So we don't have problems later on. And then thinking about access to data, this is something that came up earlier too."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:58",
                    "end": "00:51:20",
                    "text": " making sure that the data that's being used is broadly representative of the people that we're trying to serve. I think if you build a project and you build a model based on data from people in the US, you're going to end up with something that serves people in the US and doesn't serve people in the rest of the world. And so I think thinking about both a very clear targeted upside."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:20",
                    "end": "00:51:48",
                    "text": " very clear mechanism of action, very clear governance for how to build it responsibly, and broadly representative data so that the technology and the models that you're building can serve as many people as possible. I think those are the key attributes. Shahed, maybe over to you. Sure, yeah, and thank you. Building on that, maybe would share three things. For implementers who are thinking about incorporating AI,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:48",
                    "end": "00:52:18",
                    "text": " First would be starting with people, not the technology. For us that meant health workers understanding their needs, their challenges, their motivations as professionals deeply, and then seeing how this tool can unlock potential. Again, for us that was about unlocking moments that they provide care. Second, it was really important for us to get very precise with our definitions. What do we even mean by good care?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:18",
                    "end": "00:52:43",
                    "text": " what's in the bounds of the service, what's out of the bounds of the service. Both of those things helped essentially provide a north star for things like when we developed out the knowledge base, built in guardrails for our prompting. And again, do that in a way where we iteratively come back to it. And finally, especially because we had humans in the loop,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:43",
                    "end": "00:53:10",
                    "text": " putting that product or that tool out there in a small geography for us to learn and understand and then iterate on. We learned so much more when we actually started to communicate with families and see what were the challenges that our health workers were facing when we put it out there. So again, of course, do that in a safe way. For us, it was keeping people in the loop, health workers specifically."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:10",
                    "end": "00:53:41",
                    "text": " and monitoring all the conversations, but what we learned was tremendous in those initial days. Thanks, Amy. I think a couple of things. I think we're going to hear this week lots about models and the cool tricks they're going to do and so on. I think it's a 10% problem. I think models, the accuracy will be 10% of what we need."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:41",
                    "end": "00:54:11",
                    "text": " I think there's a 20% on how does this actually integrate, like what sort of infrastructure do you need? And then I think 70% of this is gonna come down to change management and how we engage people and take people along on the journey. And I think if we neglect that, I think this is gonna be an uphill battle for some time. I also think that as we think about evidence and adoption, there's a very real cost issue."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:11",
                    "end": "00:54:38",
                    "text": " One of the studies we did in partnership with Bath over here showed that the clinician assist consult actually only costs four US cents and the accuracy was up to 97%. That may be expensive when you expand it over thousands of users, but actually the cost of compute will continue to come down. It's 240 times less than what it was two and a half years ago."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:38",
                    "end": "00:55:00",
                    "text": " So you can easily start to see that this becomes scalable and sustainable. I think that one of the biggest game changers, if I was forced to bet, I don't bet, but if I was forced to bet on what I think could be the biggest game changer for development on AI, I think it's personalized health coaches."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:00",
                    "end": "00:55:25",
                    "text": " in the pocket of every single individual that can speak in their language, that has context and memory over time, that eventually doesn't only serve health needs, but is your coach for education, is your coach for financial, but just a coach that you can trust with decision-making over time."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:25",
                    "end": "00:55:58",
                    "text": " Thank you so much, Samir. That's a really exciting vision to lay out, and I think a good way to close. It was so great from hearing from each of you. I think this was just a great way to highlight the need for more evidence, the excitement around AI and health, and just really appreciate you taking the time to be here, and all of you for attending. Thank you. Thank you so much to all of our AI and health speakers. It's really just amazing to hear their perspectives from every type of organization from all around the world. I hope you're all learning a lot."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:58",
                    "end": "00:56:07",
                    "text": " We'll take a short break and we'll rejoin at 11.40. Please try to exit from the right side of the stage. Thank you very much."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Hindi"
},

{
    "_id": "6994116a25c9f4e0d7e70011",
    "projectId": "457027",
    "projectName": "AI for the Last Mile Human-Centred Design for Bharat - Hindi",
    "projectDescription": "AI for the Last Mile Human-Centred Design for Bharat - Hindi",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T06:57:50.692Z",
    "projectUpdatedDate": "2026-02-17T06:57:50.692Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T06:57:50.692Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "AI for the Last Mile_ Human-Centred Design for Bharat.mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771311402803.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:30",
                    "text": " electronics and communications at the government of Telangana. At Telangana, she did something very interesting. She took the data that was in the government archives and made it available for innovators who actually work with that to build new solutions. So welcome. Next to her is Prashan Prakash. He's a founding partner of Excel, and he's also a board member of ACT Foundation. ACT is the one that's a knowledge partner for this."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:30",
                    "end": "00:00:58",
                    "text": " Prashanth has been involved from the very start. He's a pioneering venture capitalist who has helped shape the country's startup ecosystem. He was awarded the Pradham Sri in 2025 for his contribution to trade, industry and entrepreneurship. Welcome Prashanth. On my left, I have James Walsh. He's a senior behavioral scientist with the Agency Fund. He advances social innovation through strategic planning, cross-sector partnerships and resource mobilization. He focuses on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:58",
                    "end": "00:01:28",
                    "text": " building scalable solutions to complex global challenges, helping communities achieve sustainable lasting impact. He's also been involved with some of the work that we have done with one of our portfolio companies at ACT in the area of education. Thanks, James. And then we have the people who are actually doing some serious work on the ground. We have Ramakant. Ramakant is a co-founder of VISA. VISA is one of the leading healthcare bots for mental health."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:28",
                    "end": "00:01:55",
                    "text": " It's working across 100 countries with a billion conversations that are happening across multiple languages. A quick background, VISA used to be in English only. And during COVID, we worked with them to actually bring it to Hindi. And in that process, they made it multilingual. So that has really scaled the business. And last, I have Utkarsh Saxena."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:55",
                    "end": "00:02:27",
                    "text": " Utkarsh is the co-founder and CEO of Adalat AI. As the name suggests, you know what this is about. They work across India's justice and policy ecosystem, including at the Supreme Court of India, on court reform, marriage equality litigation, and reducing judicial backlogs. He holds an LLM from Harvard Law School and a master's from the Harvard Kennedy School. Welcome. So on that note, maybe I'll just give a little bit of a sort of set the stage for what this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:27",
                    "end": "00:02:57",
                    "text": " discussion is about and then we'll do a short video after that about one of our portfolio companies. So the question that is in front of us is as AI adoption accelerates globally, a critical question that we in India have to sort of think through is can AI help narrow our social and economic divides or will it deepen them? The reason why we're having this discussion is ACT as a philanthropic organization has been working very closely with various entrepreneurs"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:57",
                    "end": "00:03:27",
                    "text": " in this area of bringing technology-led innovation to either public education, public health, or the environment. All three areas are big problems. And to do it at scale, we believe that technology, and now AI in particular, has a big role to play. But what we also believe that building for Bharat is not necessarily just copying what is happening in Silicon Valley. And we have to think about it differently. So today's conversation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:27",
                    "end": "00:03:56",
                    "text": " is not what AI can do in theory, but what it takes to deliver real outcomes, not just innovation, but at the last mile in clinics, classrooms, courts, and communities. We'll focus on three things. What breaks when AI meets real users and real systems? You all had lunch outside. Did you see any AI working there? Right? That's the challenge. Second, how does trust, safety, and explainability"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:56",
                    "end": "00:04:27",
                    "text": " get designed from day one, and with what end user in mind and front and center? And lastly, what needs to change across government, capital, and builders for AI to scale responsibly? So those are the three questions that we'll deal with. I'll start with the first question, right? So, ma'am, I'll start with you. You know, you have been very involved as a... Sorry, I've been told, yes, the video. Can somebody play the video?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:27",
                    "end": "00:04:54",
                    "text": " Sorry."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:54",
                    "end": "00:05:38",
                    "text": " A lot of kids have learned, ma'am. My daughter has learned a lot. I think my knowledge has also increased. And the knowledge of the kids is also increasing. In the old days, they used to get a lot of help. In the old days, they used to learn a lot. And they didn't learn. So now the activity is starting through mobile apps. So the kids also get to learn a lot through this. So these are..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:38",
                    "end": "00:06:10",
                    "text": " Rocket learning and the team there has actually scaled very well. James has been very involved in that journey. And James, I'll come to you after I ask this question. So ma'am, from a government perspective, what is the role that you have seen AI play and where have you seen things work and not work? Hello? Yeah, yeah. So post-learn session is a bit difficult, I think. Right? Yeah."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:10",
                    "end": "00:06:34",
                    "text": " Coming to your question, Sandeep, like, currently I'm working, we have implemented a number of, I would say, I've qualified as pilots, you know, AI pilots in different sectors, ranging from healthcare to forestry to education to, you know, mental wellness tool. So while, you know, some of the pilots have been successful, some we couldn't scale it up."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:34",
                    "end": "00:06:59",
                    "text": " Honestly, many of them we couldn't scale it up. That's a different issue. But the underlying core problem that we faced across all these solutions was the availability of quality data. That was a single biggest roadblock for us, in fact, to either initiate the initially even to start up the projects and then going forward even to scale up."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:59",
                    "end": "00:07:27",
                    "text": " This we started way back in around 2021, building these AI solutions. At that point of time, now everybody's talking about AI, but it was too early for the government to even think about AI. So even the startup that worked, now it's a very mature startup in India, in the agriculture sector. Actually we worked with them. Even they were not sure what kind of data they need. So that was, they used to ask us data. We used to go to different departments, get it, give it to them. Then they used to process it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:27",
                    "end": "00:07:56",
                    "text": " But even getting the raw data used to take three to four months time for us. Then the startup used to process it. Another 30% of the time used to go. Finally, they used to come back and say, no, no, this is not the data set we needed. So that really took us six months to actually get the right quality data. So that was the biggest lesson for us. And based on those learnings, in Telangana, we built the agriculture data exchange platform initially. That made all the data related to agriculture."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:56",
                    "end": "00:08:12",
                    "text": " This we did two years back. It's now running up and running and startups and others are using data. Based on the success of that, we actually extended that from agriculture to other sectors as well. Just this July, I think, just before I left Telangana, they've launched the state as long as the..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:12",
                    "end": "00:08:37",
                    "text": " Telangana data exchange platform where data related to all the sectors is now made available. I heard the recent number is about 1100 data sets are made available for, I mean, related to different sectors. So that's from the learning, you know, that's what we built. Yeah. Well, that's great. If there are any founders, entrepreneurs that are sitting here, you know that Telangana is one place to go to go build your AI business."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:37",
                    "end": "00:09:06",
                    "text": " You know James, going to the video that we just saw with Rocket Learning and you know you've been a big supporter of the work that they're doing. What have you seen work on the ground, your thesis, your theory of change and supporting them? What has worked and what do you see going forward? What would work better? Thank you. So we have been working with Rocket Learning for about three years now or maybe four."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:06",
                    "end": "00:09:37",
                    "text": " Sorry, can everyone hear me? Yes, thanks. And so I think there's a couple of factors that have made rocket learning so successful. They now are working with hundreds of thousands of Angamwadi workers around India. They're touching with millions of parents. And so there's several million kids that are ultimately being impacted by the work that rocket do. I think one of the things that rocket have done that it's been so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:37",
                    "end": "00:09:39",
                    "text": " effective is that from the beginning"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:39",
                    "end": "00:10:08",
                    "text": " they took a product first approach. And so they have a suite of engineers that are really kind of driving forward the capabilities of what they do. If you're not familiar with rocket learning, they have a WhatsApp product where they communicate lesson plans to body workers and they provide advice to parents. And the fact that they have such a product led approach, which is quite unusual, actually, in the nonprofit sector. And what that has meant has been that they've been able to expand, solve problems, figure out and track users."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:08",
                    "end": "00:10:28",
                    "text": " engagement and really get, like, develop a solution that works for the folks that they're trying to support. But I think there's another thing that makes Rocket particularly unusual, which is that very early on, they also decided to embed quite rigorous research into the work that they've done."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:28",
                    "end": "00:10:57",
                    "text": " within a year or two, they began a randomized control trial with some of the top economists in the world and to test whether or not they were having the impact that they were hoping to on child development. And they found a bit midway through the course, they were having some impact, but it wasn't the impact that they were hoping for. And actually what they did in the response to that was that they started to make adaptations, they started to A-B test different variants of their solution and ultimately improve the product based on that feedback that they weren't achieving what they wanted to do. So I think research"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:57",
                    "end": "00:11:04",
                    "text": " is the kind of second really exciting thing that they've been able to embed. The fact that they're both product and research makes them"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:04",
                    "end": "00:11:33",
                    "text": " like relatively, kind of in a very small number of nonprofits advancing at the frontier. And I think moving forward, I think there are two real things that could make them even more impactful. The first, of course, is AI and thinking about how can you generate bespoke solutions that support that kind of very individual needs that an Angamwadi worker may need at that given point in time, be able to have a conversation with them, be able to help them, support them with their confidence or engage in the community in particular ways. These are often things that are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:33",
                    "end": "00:12:00",
                    "text": " quite difficult for anybody workers, and they often need a lot of support in customized ways. And I think the second thing that would really excite me, at least from Rocket's perspective, is this is truly one of the best nonprofits in the world. And I think they have been an amazing addition to the ecosystem in India. What's really exciting is could"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:00",
                    "end": "00:12:27",
                    "text": " large countries around the world benefit from these Indian innovations. So could you imagine these kind of supports in Brazil or in Nigeria and other large countries? I think that's the next frontier for some of these amazing innovations that have been developed in India and other countries to start bringing that skill set and that capability that is kind of indigenous to the Indian nonprofit tech sector to the world at large."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:28",
                    "end": "00:12:56",
                    "text": " You know, it's interesting that you say it's indigenous to the Indian nonprofit sector. I think Rocket is a very unique company in the sense that it really behaves like a company. I think Prashan, you know, you've been involved with ACT from day one. You've seen multiple opportunities to work with different founders, right? And obviously with Rocket as well. As you know, we are hearing James speak, it sounds like it's a VC company, right? Rocket is no different than any VC company."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:56",
                    "end": "00:13:28",
                    "text": " Would love to hear what you believe, you know what from everything that you've seen what has worked and where things haven't worked and You know how again as a VC as a capital provider across various areas you I know you're also big in climate and sustainability What are your thinking in AI AI's role what will work and what will not work? So said I think The pilot to scale Has been a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:28",
                    "end": "00:13:55",
                    "text": " has been a challenge if like in the rocket case, it has worked because they've day one thought about it like a product, like you were saying. So it has been engineered like a product and it has been built for scale. Now with AI, this starts getting slightly more, even more complex because AI systems, even at the pilot stage can be slightly compute intensive and the..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:55",
                    "end": "00:14:23",
                    "text": " Capital allocation becomes a problem. So it's a very early days in AI only products. What we are seeing is that unless it is something that is very monetizable, even for Bharat, even for Bharat, unless it's like test prep, very monetizable, different price point, even at a very different price point, you'll be able to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:23",
                    "end": "00:14:46",
                    "text": " kind of quickly do a pilot and see product market fit. But if it's not in those areas, then capital allocation becomes a problem. So that's one challenge that we see. The second challenge is going from pilot to scale. I think there has to be a lot more re-engineering for even more efficiency."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:46",
                    "end": "00:15:09",
                    "text": " with AI compute again. So whether it's bringing in edge compute related stuff, SLMs and your own proprietary LLM technologies. So I think it's not very straightforward from..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:09",
                    "end": "00:15:39",
                    "text": " I mean, if you built technologies in the previous era, I think server costs won't disproportionately increase when you went from pilot to scale. So that time we are starting to see is a problem. The third problem is, I think, very much what I was talking about, which is data. And the data environments in our country are extremely archaic and broken. So when you look at something like health care,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:39",
                    "end": "00:16:08",
                    "text": " So even if you do a pilot in a particular environment, how do you then actually integrate it into a clinical system and make it work across states, across boundaries becomes, you can make it work in Telangana at a pilot level, but then how do you take it to another state? So that's the third problem. And fourth, I think just the engineering that is required to make it easy to evaluate."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:08",
                    "end": "00:16:37",
                    "text": " monitor continuously. So it's not like your old products that you've built, right? So you've built the product and then you could have a next version three months from now. But AI products need continuous monitoring and evaluation. And I think that's one way you can differentiate a good AI team from a not so great AI team in terms of what kind of eval and monitoring mechanisms have they built from day one. I think Prashant, you pick up a very important point that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:37",
                    "end": "00:16:57",
                    "text": " you have to retrain in some ways the donors, because the donors need to understand how capital will get deployed. Yes. Because if you're saying that there's a much higher compute requirement upfront, and donors need to pay for that, and donors don't understand that at this point. Yes. Because as you scale, the cost will only go up."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:57",
                    "end": "00:17:26",
                    "text": " But I think that's where the ingenuity and the ability to build frugally as AI evolves for these impact sectors. I think teams have to, you need to be able to judge the capability of the teams to be able to do that because capital is not going to be available just like you have in these other sectors. You know, I think from Ramakant, your standpoint with VISA, the VISA was a for-profit from day one."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:26",
                    "end": "00:17:53",
                    "text": " but you were solving a problem that was a huge impact, right? How did you think about, and you were AI first, you know, before AI became big, right, in terms of how you built the bot, how did you think about this, you know, AI, for-profit, scaling, who would pay for it, how, you know, how did that all thinking evolve? So I think the answer, Sandeep, obviously is different from India versus probably outside India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:53",
                    "end": "00:18:17",
                    "text": " And we started life as wise as a product launched in 2017 on direct to consumer, almost as an experiment, personal experience with mental health. We said you need to have a mental health intervention which doesn't need you to step out of your home and go to a therapist. Can there be a chatbot? This was five years before chat GPT. So we were literally whistling in the wind."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:17",
                    "end": "00:18:44",
                    "text": " And there was nothing there. And most people used to look at us and saying, when we used to say, oh, we are building a chat bot for mental health. And they used to say, what? And now it's not what, it's how. So the question has changed because now it's become much more accepted and it's almost become, I think, the zeitgeist. I think monetization models have also evolved from 2017 to 2026 now. So nine years initially we tried monetizing in direct to consumer."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:44",
                    "end": "00:19:13",
                    "text": " COVID happened in 2019. Suddenly, I think the world woke up to mental health as an issue. And I remember the email we got from a clinician in the NHS, the National Health Service. That was 2019 saying, who are you guys? What have you invented? Have you any idea what you have done? Because wait lists for therapists in the NHS in the UK are about as much as 12 months long. So you have to literally wait for a year before you can see a therapist. So what do you do in the meantime? So she tried it, tried this with her patients."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:13",
                    "end": "00:19:40",
                    "text": " And she said, this is the only thing which is ticking on their phones and nothing else is. So you have done something special. Have you thought about clinical safety? That was the first time in 2019 we said, oh, we need to think about clinical safety when you're thinking mental health. And now, of course, that's the only thing we probably think about. So but at the same time, the ability of I think advanced public health systems to absorb AI to implement it and also to pay for it has evolved."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:40",
                    "end": "00:20:00",
                    "text": " I think the story became very different when we moved to India. So because the topic of this panel as well as AI for Bharat, so what broke or what changed? So we started life as a chatbot for mental health and then we came to Bharat and one chatbot didn't work, mental health didn't work."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:00",
                    "end": "00:20:24",
                    "text": " So you're like, what do you do now? Basically we wanted to implement this as a mental health intervention for young people saying that it's a very vulnerable time in their lives. There's huge amount of neuroplasticity. The intervention has a generational impact because it impacts families for many, many, especially if you're thinking about young girls. So which is the target group, which we wanted to go after."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:24",
                    "end": "00:20:37",
                    "text": " But then we also realized that you also have to go beyond the English language speaking, you know, all the people in this room. And if you wanted to go to say a municipal school in Raigad in Maharashtra, we are working with the Women and Child Development Ministry in Maharashtra right now."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:37",
                    "end": "00:21:05",
                    "text": " So we realized that a chatbot format doesn't work because one, the girl doesn't have access to a phone. She probably borrows her phone from her mother twice, maybe two hours a day. It's also very closely monitored. And the consumption pattern is very different. The user behavior is very different and nobody's willing to pay for it. So you cannot build a business model around here. So then obviously lots of bloody noses later. We, I think a chance to upon her, we created a product with said that this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:05",
                    "end": "00:21:25",
                    "text": " should not be a digital intervention. It has to be a physical intervention, not digital, which was combination of a physical plus digital saying that the actual intervention, which the child is receiving is a workbook, which looks like a normal scrapbook or a journal, which the say, imagine a 13 year old girl."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:25",
                    "end": "00:21:46",
                    "text": " which is getting access to a colorful book like a scrapbook given to her by her teacher, which has QR codes embedded inside it, which take the child, even if she's on a borrowed phone, they take her to a very private anonymous chatbot interaction on the web. So you're still delivering digital, but you're delivering it in this blended format."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:46",
                    "end": "00:22:15",
                    "text": " And the second thing is nobody's going to pay for a mental health intervention. So we said this actually has to become social emotional learning. So this is a skill building intervention and there are budgets for books. So schools pay for books. They don't pay for digital apps. So that's the direction we're going in saying that our intervention for mental health for India, for Bharat is actually education. So ma'am, I'll come to you before I'll go to Utkash later, because I think what Ramakant is outlining here."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:15",
                    "end": "00:22:44",
                    "text": " in terms of how AI will come to India, you mentioned this a little bit, but the government will have a big role to play because to the point that Prashanth made that the cost of rolling out AI interventions will be expensive and the government is already doing some work in this. But where do you see, like for example, if Ramakant were to come with Vaisa to Telangana or to come to Niti Aayog, what would your reaction be? How would you see that being taken by the government in India?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:44",
                    "end": "00:23:13",
                    "text": " I can actually share one example. We worked with one of the mental wellness, forgot the name actually in Telangana. So and government of India is also doing that currently subsidizing the compute. That's one way of, you know, helping the startups where the cost of compute is born by the government because at the end of the day, as someone said, you know, we don't look as a government, we don't look at profits, but we look at impact or, you know, solving a certain problem. So that's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:13",
                    "end": "00:23:42",
                    "text": " the main thing that the government looks at. So, the computer is one thing. And one of the biggest challenge start-ups face, you know, is access to, if you're, it's an agri-tech solution, access to farmers. Because at the ground level, even for the mental health solutions that we implemented for substance abuse, you know, I mean, in Telangana, many of the states, this is one of the biggest problem with children are actually getting addicted to these drugs. So, we actually implemented an AI solution."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:42",
                    "end": "00:24:10",
                    "text": " where which is administered for children below 18 years. So one of the biggest problem the startup faced was, well, the government also wanted to implement that. How do we reach out to the students who are actually out there who really need the most? That's where the entire government machinery can help the startup. It could be you may call it as a first customer or whatever it is, but that scale and size the government can really handle."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:10",
                    "end": "00:24:39",
                    "text": " Number one, infrastructure. Number two, you may call it customer base or, you know. Second, third, you know, at the ground level, there are certain nuances. You may have the best product in the world. We just rolled out. We actually had this experience in the agritech sector, you know. It was a marketing, you know, access to market solution, like it's a very good product. We actually rolled it out to the, in one of the districts in Telangana. So what happened was nobody was using it except very few farmers."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:39",
                    "end": "00:25:07",
                    "text": " we didn't understand why they you know they're getting access to market very good information that they can actually get price higher than what they're paying so then we did a you know the ground study and all our team went there and said finally what they found out is the farmers they rely on local suppliers for available to credit you know getting credit and all so they really I mean they don't see actually going beyond what is available because of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:07",
                    "end": "00:25:36",
                    "text": " those already relationships that they've built. So, these kind of nuances, which the startups may not know, you know, that's another thing that the government can help them with. Great. Thank you. I just want to add here, when I spoke about data before, we are working on a very interesting project in Odisha. In fact, I was talking to you, right? So, it's in the health sector where we're trying to actually understand the entire life cycle of data in the health sector because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:36",
                    "end": "00:26:03",
                    "text": " for any government, we don't have the right data for decision making right from the ASHA worker to even the chief secretary level. So there, the government actually wanted us to identify the prioritized list of challenges that they're facing in Odisha and come out with suitable solutions in health care that help ASHA workers to deliver better services. Just wanted to mention this. Great."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:03",
                    "end": "00:26:33",
                    "text": " Again, taking off on that point about data, Utkash, I think I would love if you could talk a little bit about what Adalat does, because you are fundamentally based on data, right? And you have taken on the responsibility of digitizing this and bringing it into AI. So we will talk a little bit about that. And what have been the challenges as you've gone down the journey? Yeah. So Adalat AI is a legal tech nonprofit that builds tech solutions for district courts in India to help them process their cases faster so that we can sunset that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:33",
                    "end": "00:26:57",
                    "text": " dialogue of Tariq per Tariq, which is so iconic of our courts. And the idea is to identify the manual and clerical processes that become pain points in the plumbing of a court. So we are the proud plumbers of the courts that are trying to unclog the pipes of all these processes and paperwork. I think in terms of the thinking and what worked and what started, it's very sector-specific."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:57",
                    "end": "00:27:25",
                    "text": " I'm quite envious of education and health care because there are all these incredible nonprofits that have done a lot of incredible work that's paved the way for impact. For us, we don't know other nonprofits that work directly with courts. That's a big blind spot. And so we had some responsibility to educate the sector on the role of civil society and doing this kind of work. So our ethos was identify painkillers and not multivitamins. A lot of solutions are multivitamins, the green vegetables that would be lovely to have."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:25",
                    "end": "00:27:53",
                    "text": " But I think we thought that let's identify the high fever points in the courts and go with a painkiller to develop that credibility that hey, civil society can actually help you. Give me an example of that. Exactly. So our first solution was stenography, which confused a lot of people. They were like, you're going to solve delays and backlogs by expediting stenography in courtrooms. It's actually a big pain point because in a cross examination in a courtroom, if I ask you a question and you answered it and Prashant is the judge,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:53",
                    "end": "00:28:15",
                    "text": " There are no stenographers in most of India's courts, so Prashanth will write the question and answer. Then Judge Prashanth will get transferred, and Judge Jamie will come in. And Judge Jamie will be like, I can't read Prashanth's handwriting. So judges have WhatsApp groups where they send each other images of depositions saying, hey, what have you written? And Judge Prashanth will say, it's been two years. I can't read my own handwriting."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:15",
                    "end": "00:28:39",
                    "text": " That's the kind of challenges the courts work with. Some states, which were innovative, came forward with a solution called the Fair Copy Department. And as the word suggests, there's a department which will take Judge Prashan's depositions every day and fair them out in writing by hand. And so we asked them, saying, why are you writing this? You can type it out. And they're like, we don't have Tamil keyboards. We don't have Hindi keyboards."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:39",
                    "end": "00:28:54",
                    "text": " And so we were like, this is clearly a high fever point. Judges are developing spondylitis. It's physical manual work. They will say, we're not judges. We are scribes in the system. And so we built a speech-to-text legal transcription tool powered by AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:54",
                    "end": "00:29:13",
                    "text": " Scaling was not by push, which is what often happens in the social impact space. It was pulling us into the system. We were getting calls from states every day saying, hey, we need this painkiller as well. And we were like, wait, we have to build capacity, get more compute, build more language models. This will take time. And I think once we established that successfully,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:13",
                    "end": "00:29:34",
                    "text": " the states then started coming to us for the multivitamins. There was a relationship of trust saying, hey, you actually helped me address a fever in the system. Can you now help me build paperless filings? Can you now help me build chatbots? Because we've proven ourselves with the painkiller. So I think that was our approach. And I think it's proven. It's worked well to some extent."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:34",
                    "end": "00:30:02",
                    "text": " In terms of challenges, as you said, we work in an extremely confidential setting. This is a courtroom. This is a judge dictating a judgment before the world finds out. Stock markets can move on that kind of information. And so security is the highest priority. And we have a three-pronged approach to security. One is we don't use any third-party APIs. So no OpenAI, Google, nothing. These are Adalat AI's models. We are an AI native company, which means a lot of compute."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:02",
                    "end": "00:30:30",
                    "text": " for both experimentation, for the fine tuning of the models, but also for the inferencing. The unit economics changes when the API call is not going to Google, but is coming to Adalat AI. And that's obviously a big challenge. The second is we have to think about encryption systems. This is extremely sensitive databases. And so we've kind of built a blockchain kind of decentralized encryption system where each judge gets an encryption key. And to decrypt the database, you need every judge's localized local machine encryption key."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:30",
                    "end": "00:30:54",
                    "text": " And so security is a big challenge. And the third is just the peculiar nature of the domain and the language we use. As per the doctrine of res ipsa loquitar, per order 39, rule 1 and 2, whatever that means, all conversational speech to text collapses in a courtroom because the language is so pedantic, so technical, that we are kind of."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:54",
                    "end": "00:31:19",
                    "text": " building our own speech to text models for our own languages. Law is its own language. One is Tamil, and then one is legal Tamil. And so I think all of these complicate the work that we do, but I think this is unique for sovereign AI. And so we are really looking forward to more support on compute, number one, that if someone asked me what's the one thing that you would like more from the system, it's compute. And I would say"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:19",
                    "end": "00:31:48",
                    "text": " To the C, I would add coordination in the compute. You know, I was just standing outside and I was talking to someone from an IIT, I won't name which one, and they were like, we're building this amazing chat bot that does this thing to demystify laws. And I was like, I know five other IITs and three other triple IITs that are doing the exact same thing. Why aren't the organizations talking to each other if compute is so precious? I'm not sure that there's as much thinking going into like using it in the most optimized and efficient way. So yeah, I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:48",
                    "end": "00:32:18",
                    "text": " We've now kind of got a foot in the door with the painkiller. The system is now opening the door for us to build more end-to-end systems, but for that to happen at scale, compute is pre-acquisite. Prashant, I'm going to come back to you and to you, Jimmy, as donors, funders in this. You've heard both, I think all three in some ways talk about compute being a big sort of bottleneck for impact, at least impact. You know, you're getting VCs to put money into"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:18",
                    "end": "00:32:51",
                    "text": " other companies but on the impact side how do you see your organizations you know act or in your case agency fund actually playing a role in in driving making that change you know either of you can go. So I think like Utkush was saying I think this requires a very sovereign approach right so it's not even when you look at how to optimize some of this compute so there's a company called E6 data for example right so you're aware of that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:51",
                    "end": "00:33:20",
                    "text": " Now they are doing a lot of this optimization for the larger companies, the Flipkarts, the Swiggies. I don't know what it would take to get some of that into a government stack, sovereign stack, which would be able to provide the same compute at a very differential cost. So it's not just compute. Can the government AI stack look at compute at a very differential, optimize it to a very differential cost? That's one."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:20",
                    "end": "00:33:50",
                    "text": " And I think over time, I would say, I think the solution is edge compute. I don't think there's, I mean, if this has to drop down to, you know, one 10th, kind of a price point. So, and we just launched something in Karnataka, the first AI kind of risk computer, right? Yeah, so that this is an AI edge computer where it runs on Chrome, the models run on Chrome."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:50",
                    "end": "00:34:10",
                    "text": " 15,000 rupees. So it's going to be first deployed in schools for the teachers to personalize curriculums for the classroom they're teaching. So you can't give one of these for every student, but can you give it to every teacher?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:10",
                    "end": "00:34:28",
                    "text": " so that the lessons are personalized at Microsoft. You can run like models on that. Yeah, models run on that. I mean, it's totally off the grid. You can run the model. So I'm seeing that come down to 5,000 rupees to 6,000 rupees, such edge AI compute."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:28",
                    "end": "00:34:58",
                    "text": " computers, right? And then I can see in a courtroom, it's going to be very, very practical to roll it out across the country. So I think there'll be innovation on the central AI stack, the API-driven stuff that's going to be, of course, made more optimal. But there's more category-specific compute solutions on the edge that will also hopefully play a role."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:59",
                    "end": "00:35:29",
                    "text": " Thanks, so that very much resonates. I think there's a need for technological innovation to come. In terms of how we, the agency funds, approach this, I don't think we're radically different to other funders in the philanthropic space. We see our role very early stage, and so where we pay attention is trying to help early stage non-profits."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:29",
                    "end": "00:35:55",
                    "text": " Identify a model a product that works and that might mean model evaluation to get the actual AI to be behaving as it should be, but all the way up to kind of getting user engagement effectively. This is something that is well understood in the private sector, but not always that kind of muscle doesn't exist in the nonprofit sector in the same way. And then ultimately helping nonprofits run randomized trials, impact evaluations to show impact."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:55",
                    "end": "00:36:15",
                    "text": " Because I think ultimately there we want to be able to work with nonprofits to help them demonstrate to the ultimate kind of payer, which is going to be government, that this is a cost effective, high impact solution. And I think that's that kind of is the second part of where we see the solution."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:15",
                    "end": "00:36:43",
                    "text": " From because we're relatively small funder to these big questions of large-scale fund a kind of large-scale costs Ultimately, like I would love to see a world where the user and the payer are much closer I don't think we want a world where like the Gates Foundation or the agency fund or whoever else are paying for indefinitely for things not only because of the kind of the sustainability question, but also because exactly to with crushes point about"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:43",
                    "end": "00:36:45",
                    "text": " You know pain points"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:45",
                    "end": "00:37:14",
                    "text": " You know, if the funder is the pair, the pain points that are going to be often solved are the funders pain points. But really, what you want is the nonprofit or the organization, the product to be solving the pain points off the government who are essentially acting on behalf of citizens. And so getting that much more close to citizens, I think, is ultimately better as well. So those are the two things. One is just helping nonprofits build products that work through evaluations, whether that's technical AI model evaluation or it's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:14",
                    "end": "00:37:38",
                    "text": " standard kind of evaluate, impact evaluation. And then the second is really helping find models where the user is the payer. And I loved hearing about the two stories that you guys had. I think that's a good either of you take on what he just said. I'd like to probably add a different dimension. One is definitely because we are talking about AI and the cost of compute."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:38",
                    "end": "00:37:54",
                    "text": " But I think the cost of implementation or the cost of actually rolling this out in real life is not just compute or my sense is compute is probably a very small fraction of the cost you would need to incur to actually make this rollout at scale in government."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:54",
                    "end": "00:38:22",
                    "text": " or in the public sphere in real life. And I'm talking about the transformation, I'm talking about the fact that you need to train people, you need to make sure that they're comfortable with whatever you're implementing. That's a non-trivial cost, not just in terms of energy and attention, but also in terms of money. And I had two or three different, just in terms of our work on the ground, both with in Maharashtra and now we are about to do something very significant in Meghalaya as well. And we're talking to the government of Punjab and we've just..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:22",
                    "end": "00:38:51",
                    "text": " We just won a $7 million five-year grant with the Wellcome Trust as a clinical trial to actually implement this in UP. So there are so multiple places where we are doing this. And I think we are very quickly realizing that AI is not just tech. It is tech plus people. And in the mental health space or in the emotional resilience and skill building space, you do need to, for example, gain the young girl's trust."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:51",
                    "end": "00:39:21",
                    "text": " And how do you do that? You need to think laterally or you need to think very non tech wise saying, go with the teacher and don't go out and say, hey, here is an app or here is a digital solution which you need. You need to get the parent on your side. So how do you do that? So you need to first get the government to say, look, we are backing this. And then we are now sending letters to each of the parents of all the children in school saying this is actually supported by the government of Maharashtra so that they allow their children to actually use the service."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:21",
                    "end": "00:39:40",
                    "text": " So these are non-trivial problems and if you don't think about them then the tech will just be a piece of tech lying on the shelf. That's a very good point that the systemic change actually ends up in many cases stopping things and that probably is true for you more than anybody else right? Oh yeah absolutely I think any impact has behavioral change behind it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:40",
                    "end": "00:40:09",
                    "text": " and convincing judges who are pretty set in their ways, who aren't usually told how to do their work to now be told how to change their workflows is very radical. And so it's a twin engine mission. I completely agree. Part of it is tech, and that's only a part of it. The other big part of it is programming to actually train and workshop with the audience and the population that needs to use the technology because otherwise the product is not going to see that kind of uptake. So we actually work and workshop with the judges."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:09",
                    "end": "00:40:33",
                    "text": " But then with the stenographers, the court clerk, the court record keeper, the whole team around the judge that operates in a courtroom so that the technology is internalized in their workflow and is actually implemented in court. And that requires sometimes bottom-up approaches where you can kind of go and work with courts and do all of it. But honestly, in our setting, because it's the law,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:33",
                    "end": "00:41:01",
                    "text": " top-down approaches also work very effectively. So we've seen very clear impact of things like mandates. Kerala became the first state in the world a few months ago to mandate us as an AI solution in every courtroom to transcribe court proceedings. And while we had seen very healthy uptake till then, the level up, I mean, after that has been absolutely transformational. Does government pay for that? Partly. Not entirely, but partly."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:01",
                    "end": "00:41:12",
                    "text": " Now it's become a precedent for other states. But I was going to come to the second point, which is that the end goal for us and many other organizations is government adoption. And today they don't have the compute to run these systems."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:12",
                    "end": "00:41:32",
                    "text": " and not just compute they don't have internal tech teams to actually productionize models and scale them and so we are filling that gap now but the end goal is at some point to migrate these solutions to state systems just like governments have very routine IT teams will they have routine AI teams in five years ten years that actually run these systems and so how do you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:32",
                    "end": "00:41:59",
                    "text": " build systems in a way from the outset that prepare you for that migration and future, I think keeping that direction in mind completely changes how you design and implement solutions and it changes the engineering. On that point, I think that's a very interesting point he brings up that the government IT teams effectively have to become government AI teams going forward, right? Are you, Niti Ayog, how are you driving this?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:59",
                    "end": "00:42:26",
                    "text": " I mean, basically there is a shift in the mindset of the CDAC or any of these organizations that have been building your issues and other applications in the past. Now I have to start thinking AI first. How are you seeing or how are you going to see something like Niti Aayog is at the center of this, right? And working with all the states, how do you see that driving that change from an IT first mindset to an AI first mindset?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:26",
                    "end": "00:42:55",
                    "text": " In fact, I mean, I work with Frontier Tech Hub in India. So, which actually looks at different technologies, how it's going to impact, you know, India for the next 10 years and we come out with the roadmap. So, one of the roadmap that we have recently launched a couple of months back, two months back is AI for jobs, you know, wherein we have identified actually different profiles of individuals, you know, what are the roles that are going to change, different roles that are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:55",
                    "end": "00:43:25",
                    "text": " new rules that are going to be created and what are the jobs that are going to be lost and we actually came out with certain recommendations to the government as to how the you know government should be prepared for it. I mean since it was part of the budget I'm not sure if I can speak it now speak now about it we actually recommended setting up a talent AI talent machine you know across the governments so but that recommendation they actually it came out in a different way which government is setting up a standing AI committee."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:25",
                    "end": "00:43:49",
                    "text": " they are going to and there's so many fragmented initiatives across the country you know in AI scaling like so how do you actually bring all those into one umbrella and create that AI talent pool for the government as well as for the private sector is what you were thinking it's that's a very good report that has been launched I think I think we need a AI just for coordinating government that I think is a starting Bharat we can start with that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:49",
                    "end": "00:44:07",
                    "text": " You know, what I'll do, thank you very much. I can keep asking questions, but I'm sure people over here have questions as well of the panel. We'll give some mics if anybody has any questions. Please, just if you have a specific question to a panel member, please ask them specifically."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:07",
                    "end": "00:44:30",
                    "text": " Yes. So when we are talking about education and a person who is working in education, I want you to understand from law to health care to the education system that we are talking about, any individual that is going to be like taking a job in future, how skills and capacities are being changed from 21st century skills to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:30",
                    "end": "00:44:53",
                    "text": " AI skills, what will be replaced by the skills that are required to for a person to let's say work with AI or what what are all sort of changes in curriculum are required per se to know where the students are going to head in future. So that is one question that I wanted to ask. Yeah."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:53",
                    "end": "00:45:23",
                    "text": " That's exactly what I said before, you know, we came out with a roadmap, which actually identifies all that, you know, and what level we need to actually introduce courses, you know, so that students right from certain level understand AI first, as you rightly said, you know, so, and government of India and the minister of education, they also actually set up NSDs national skill development, you know, centers where they're actually offering these courses. That's one aspect of second MSM is."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:23",
                    "end": "00:45:50",
                    "text": " That's one of the biggest sector in India, you know, they contribute almost more than 50% of the GDP, but they don't actually use technology and forget about AI. So that is one sector where the government is also focusing on to build skills so that their productivity increases. So one study that was done earlier long back, you know, said that why is that the MSMEs remain without, you know, improving their income and returns on."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:50",
                    "end": "00:46:18",
                    "text": " One of the biggest, you know, roadblock for them is they just don't use technology. Forget about AI. Even basic IT tools also they don't use. So, you need to start from there, you know, build the technology and then come to AI. So, but again, MSMEs are into different levels of maturity. Certain MSMEs, they do have technology. Then how do you introduce AI to them? That's a planned approach, strategic approach, you know, you need to actually adopt. So, right from schools to colleges to undergrads to..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:18",
                    "end": "00:46:48",
                    "text": " to fellowships there's a whole life cycle you know level of educational programs we are actually recommending. Hi I wanted to ask about the like how do you think the the the police or like the people who are protecting us should be trained in order to like find the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:48",
                    "end": "00:47:20",
                    "text": " upcoming uh state of the art uh like uh cyber hacks that are happening plus like how do you the my uh second query will be it should be in those regional languages and like how should like uh not just the training but uh uh the uh inference from from the infinite sides of things how should that be done yeah"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:20",
                    "end": "00:47:49",
                    "text": " No, I think, you know, this is something that recognizes is a, is a need. This issue of, you know, obviously there has been cyber hacks. And particularly those that have been targeted towards financial crimes and so on. And today I would say it's probably the number 1 of the 1st, top most things that the government is looking at. From a protection perspective, you're seeing the RBI is involved in this, the whole ministry is involved in this."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:49",
                    "end": "00:48:17",
                    "text": " The various banks, there's a lot more. It starts with awareness. Number one is, you know, in cyber, the likely victim is the one that has to be made aware that these things can happen. And there is a huge budget that is now has been created around awareness in particular. But beyond that, you know, 25,000 or something is given as insight, you know."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:17",
                    "end": "00:48:44",
                    "text": " I mean, people who lose money, they're going to government is going to pay something like that. So there is understanding of that. Now, the problem that you are highlighting is a global problem. It is not just an India-specific problem. And the solutions also have to be done at that level of scale. So I don't think this is something that is just a Bharat problem. The advantage that, and obviously you are aware of what Sarvam and others are doing,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:44",
                    "end": "00:49:14",
                    "text": " But the level of voice capabilities that are getting built in AI and solutions in India is actually a very big positive for us. Vaisa explained that. But one of the things that we are seeing is voice is going to become the primary means of communicating with technology in India. We're already seeing a lot of that happen. Adalat is a great example of all transcription. Everything is happening now. Voice makes technology accessible."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:14",
                    "end": "00:49:42",
                    "text": " And these problems require that accessibility, right? Somebody should feel comfortable being able to communicate and say, this is what has happened to me. So I think- And just to add, most of these police departments, they have cybersecurity cells within their organizations. And they're well-equipped, you know, both technically. They have the kind of expertise also built within the organizations that's being done."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:42",
                    "end": "00:50:00",
                    "text": " Hi, sir. My question is to Mr. Utkarsh. Sir, I was reading a newspaper report recently in which there have been two startups which have come up in which they have built an AI model which studies the material which is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:00",
                    "end": "00:50:25",
                    "text": " published by the companies which are listed on the stock market. So they do all those stuff to analyze. So they say there are 3000 to 5000 paper report that a single company submits to the stock exchange by which people can actually analyze and make the decision to kind of forecast what and when they do need to put in money when they do need to take it back. So I was"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:25",
                    "end": "00:50:44",
                    "text": " uh seeing a startup pitch also on a similar note in which these AI generated models are actually transcripting it making it short and making it worthwhile for the people to actually analyze and then make a decision but still people find it very what's the question"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:44",
                    "end": "00:51:10",
                    "text": " Yeah, but still people think that when they transcript all of this and make it in a smaller decision-making forum, do we still think that AI is equipped enough alone technologically to make it as a decision-making forum for the people or the general public to answer to that? Yeah, so I'll take this to the legal context in which I work because it's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:10",
                    "end": "00:51:31",
                    "text": " Step one is transcribing what you're saying, right? So getting that accurately, which is already a hard problem because the language is, whether it's finance or law, it's technical, it's Indian languages. We don't have technology for those interfaces. So that's one. And the second is to build that intelligence layer on top of that material that leans and understands context, right? So for example,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:31",
                    "end": "00:51:55",
                    "text": " how to interpret a law, reading Section 2A with Section 10 of the Act and drawing the import of the law is a human analytical skill which AI still hasn't figured out. So I think the next step which you're talking about actually requires a lot of work. It requires experts to become data annotators. Today data annotators are largely those that are tagging speech output and text output and training models."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:55",
                    "end": "00:52:09",
                    "text": " Now the work is shifting to lawyers, financial experts, training these models so that the next layer of analysis comes in. And that's where the harder part is. I will, though, say that in the law, I am a little nervous about this being given out as legal advice."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:09",
                    "end": "00:52:39",
                    "text": " there are actually very serious risks or challenges around it. First, regulatory, the bar council, how do you allow a non-licensed lawyer to give legal advice and someone to act on the legal advice? And second, you know, you could be held for professional malpractice. What are the kind of safeguards that someone can have? So you want to make this accessible, but while within the import of the law and for finance, law, health care, the biggest challenge here will be how do regulations keep up? How does the regulatory framework enable these to provide"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:39",
                    "end": "00:53:09",
                    "text": " technical information that's otherwise inaccessible, but also in a way that's safe. And that's going to be a big question mark for the regulatory space going forward. So we're just short of time. So last question. Should I go ahead? Yeah, this question is to Prashanth and James. As philanthropists or funders of social impact, what are some key principles, pre-AI and post-AI, that you're looking in the post-AI environment that is guiding your philanthropy?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:12",
                    "end": "00:53:40",
                    "text": " Well, one of the things that, so I work in development broadly. And so the thing that happened contemporaneously with AI becoming very popular was America decided it didn't want to do foreign aid anymore. And so that, I think, has transformed the whole conversation around what social impact can mean, in particular, in contexts where public funds are kind of are limited."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:40",
                    "end": "00:54:02",
                    "text": " I think what it has done is it has changed how people think about, you know, how are we going to address problems like early childhood development, maternal health, moving really towards a focus on cost effectiveness and, you know, delivering effective state capability in ways that are technology driven."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:02",
                    "end": "00:54:30",
                    "text": " And yeah, ultimately much less expensive. So that kind of focus on reducing costs by leverage technology is what I see as the critical change. It's not so much that it didn't exist before. I think rocket learning are an illustrative example. They are working with government and they are able to deliver highly cost-effective interventions as a result of technology. But I just see that now expanding radically."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:31",
                    "end": "00:54:56",
                    "text": " No, I think I'll just say that the more hairy problems that we were never, as philanthropists, were comfortable taking on, I think we'll take on bigger challenges and problems. And I think that's the shift you'll see. So hopefully this diffusion of AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:56",
                    "end": "00:55:08",
                    "text": " in a cost-effective manner to be able to solve these hairy problems is what we'll want to champion and take forward. Well, thank you, everyone, and thank you to the panelists for a very engaging discussion."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Hindi"
},
{
    "_id": "69940f5a25c9f4e0d7e6fffd",
    "projectId": "187856",
    "projectName": "AI for the Last Mile_ Human-Centred Design for Bharat - English",
    "projectDescription": "AI for the Last Mile_ Human-Centred Design for Bharat - English",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T06:49:03.668Z",
    "projectUpdatedDate": "2026-02-17T06:49:03.668Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T08:08:20.428Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "AI for the Last Mile_ Human-Centred Design for Bharat.mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771310927614.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:30",
                    "text": " electronics and communications at the government of Telangana. At Telangana, she did something very interesting. She took the data that was in the government archives and made it available for innovators who actually work with that to build new solutions. So welcome. Next to her is Prashan Prakash. He's a founding partner of Excel, and he's also a board member of ACT Foundation. ACT is the one that's a knowledge partner for this."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:30",
                    "end": "00:00:58",
                    "text": " Prashanth has been involved from the very start. He's a pioneering venture capitalist who has helped shape the country's startup ecosystem. He was awarded the Pradham Sri in 2025 for his contribution to trade, industry and entrepreneurship. Welcome Prashanth. On my left, I have James Walsh. He's a senior behavioral scientist with the Agency Fund. He advances social innovation through strategic planning, cross-sector partnerships and resource mobilization. He focuses on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:58",
                    "end": "00:01:28",
                    "text": " building scalable solutions to complex global challenges, helping communities achieve sustainable lasting impact. He's also been involved with some of the work that we have done with one of our portfolio companies at ACT in the area of education. Thanks, James. And then we have the people who are actually doing some serious work on the ground. We have Ramakant. Ramakant is a co-founder of VISA. VISA is one of the leading healthcare bots for mental health."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:28",
                    "end": "00:01:55",
                    "text": " It's working across 100 countries with a billion conversations that are happening across multiple languages. A quick background, VISA used to be in English only. And during COVID, we worked with them to actually bring it to Hindi. And in that process, they made it multilingual. So that has really scaled the business. And last, I have Utkarsh Saxena."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:55",
                    "end": "00:02:27",
                    "text": " Utkarsh is the co-founder and CEO of Adalat AI. As the name suggests, you know what this is about. They work across India's justice and policy ecosystem, including at the Supreme Court of India, on court reform, marriage equality litigation, and reducing judicial backlogs. He holds an LLM from Harvard Law School and a master's from the Harvard Kennedy School. Welcome. So on that note, maybe I'll just give a little bit of a sort of set the stage for what this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:27",
                    "end": "00:02:57",
                    "text": " discussion is about and then we'll do a short video after that about one of our portfolio companies. So the question that is in front of us is as AI adoption accelerates globally, a critical question that we in India have to sort of think through is can AI help narrow our social and economic divides or will it deepen them? The reason why we're having this discussion is ACT as a philanthropic organization has been working very closely with various entrepreneurs"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:57",
                    "end": "00:03:27",
                    "text": " in this area of bringing technology-led innovation to either public education, public health, or the environment. All three areas are big problems. And to do it at scale, we believe that technology, and now AI in particular, has a big role to play. But what we also believe that building for Bharat is not necessarily just copying what is happening in Silicon Valley. And we have to think about it differently. So today's conversation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:27",
                    "end": "00:03:56",
                    "text": " is not what AI can do in theory, but what it takes to deliver real outcomes, not just innovation, but at the last mile in clinics, classrooms, courts, and communities. We'll focus on three things. What breaks when AI meets real users and real systems? You all had lunch outside. Did you see any AI working there? Right? That's the challenge. Second, how does trust, safety, and explainability"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:56",
                    "end": "00:04:27",
                    "text": " get designed from day one, and with what end user in mind and front and center? And lastly, what needs to change across government, capital, and builders for AI to scale responsibly? So those are the three questions that we'll deal with. I'll start with the first question, right? So, ma'am, I'll start with you. You know, you have been very involved as a... Sorry, I've been told, yes, the video. Can somebody play the video?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:27",
                    "end": "00:04:54",
                    "text": " Sorry."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:54",
                    "end": "00:05:38",
                    "text": " A lot of kids have learned, ma'am. My daughter has learned a lot. I think my knowledge has also increased. And the knowledge of the kids is also increasing. In the old days, they used to get a lot of help. In the old days, they used to learn a lot. And they didn't learn. So now the activity is starting through mobile apps. So the kids also get to learn a lot through this. So these are..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:38",
                    "end": "00:06:10",
                    "text": " Rocket learning and the team there has actually scaled very well. James has been very involved in that journey. And James, I'll come to you after I ask this question. So ma'am, from a government perspective, what is the role that you have seen AI play and where have you seen things work and not work? Hello? Yeah, yeah. So post-learn session is a bit difficult, I think. Right? Yeah."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:10",
                    "end": "00:06:34",
                    "text": " Coming to your question, Sandeep, like, currently I'm working, we have implemented a number of, I would say, I've qualified as pilots, you know, AI pilots in different sectors, ranging from healthcare to forestry to education to, you know, mental wellness tool. So while, you know, some of the pilots have been successful, some we couldn't scale it up."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:34",
                    "end": "00:06:59",
                    "text": " Honestly, many of them we couldn't scale it up. That's a different issue. But the underlying core problem that we faced across all these solutions was the availability of quality data. That was a single biggest roadblock for us, in fact, to either initiate the initially even to start up the projects and then going forward even to scale up."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:59",
                    "end": "00:07:27",
                    "text": " This we started way back in around 2021, building these AI solutions. At that point of time, now everybody's talking about AI, but it was too early for the government to even think about AI. So even the startup that worked, now it's a very mature startup in India, in the agriculture sector. Actually we worked with them. Even they were not sure what kind of data they need. So that was, they used to ask us data. We used to go to different departments, get it, give it to them. Then they used to process it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:27",
                    "end": "00:07:56",
                    "text": " But even getting the raw data used to take three to four months time for us. Then the startup used to process it. Another 30% of the time used to go. Finally, they used to come back and say, no, no, this is not the data set we needed. So that really took us six months to actually get the right quality data. So that was the biggest lesson for us. And based on those learnings, in Telangana, we built the agriculture data exchange platform initially. That made all the data related to agriculture."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:56",
                    "end": "00:08:12",
                    "text": " This we did two years back. It's now running up and running and startups and others are using data. Based on the success of that, we actually extended that from agriculture to other sectors as well. Just this July, I think, just before I left Telangana, they've launched the state as long as the..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:12",
                    "end": "00:08:37",
                    "text": " Telangana data exchange platform where data related to all the sectors is now made available. I heard the recent number is about 1100 data sets are made available for, I mean, related to different sectors. So that's from the learning, you know, that's what we built. Yeah. Well, that's great. If there are any founders, entrepreneurs that are sitting here, you know that Telangana is one place to go to go build your AI business."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:37",
                    "end": "00:09:06",
                    "text": " You know James, going to the video that we just saw with Rocket Learning and you know you've been a big supporter of the work that they're doing. What have you seen work on the ground, your thesis, your theory of change and supporting them? What has worked and what do you see going forward? What would work better? Thank you. So we have been working with Rocket Learning for about three years now or maybe four."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:06",
                    "end": "00:09:37",
                    "text": " Sorry, can everyone hear me? Yes, thanks. And so I think there's a couple of factors that have made rocket learning so successful. They now are working with hundreds of thousands of Angamwadi workers around India. They're touching with millions of parents. And so there's several million kids that are ultimately being impacted by the work that rocket do. I think one of the things that rocket have done that it's been so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:37",
                    "end": "00:09:39",
                    "text": " effective is that from the beginning"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:39",
                    "end": "00:10:08",
                    "text": " they took a product first approach. And so they have a suite of engineers that are really kind of driving forward the capabilities of what they do. If you're not familiar with rocket learning, they have a WhatsApp product where they communicate lesson plans to body workers and they provide advice to parents. And the fact that they have such a product led approach, which is quite unusual, actually, in the nonprofit sector. And what that has meant has been that they've been able to expand, solve problems, figure out and track users."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:08",
                    "end": "00:10:28",
                    "text": " engagement and really get, like, develop a solution that works for the folks that they're trying to support. But I think there's another thing that makes Rocket particularly unusual, which is that very early on, they also decided to embed quite rigorous research into the work that they've done."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:28",
                    "end": "00:10:57",
                    "text": " within a year or two, they began a randomized control trial with some of the top economists in the world and to test whether or not they were having the impact that they were hoping to on child development. And they found a bit midway through the course, they were having some impact, but it wasn't the impact that they were hoping for. And actually what they did in the response to that was that they started to make adaptations, they started to A-B test different variants of their solution and ultimately improve the product based on that feedback that they weren't achieving what they wanted to do. So I think research"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:57",
                    "end": "00:11:04",
                    "text": " is the kind of second really exciting thing that they've been able to embed. The fact that they're both product and research makes them"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:04",
                    "end": "00:11:33",
                    "text": " like relatively, kind of in a very small number of nonprofits advancing at the frontier. And I think moving forward, I think there are two real things that could make them even more impactful. The first, of course, is AI and thinking about how can you generate bespoke solutions that support that kind of very individual needs that an Angamwadi worker may need at that given point in time, be able to have a conversation with them, be able to help them, support them with their confidence or engage in the community in particular ways. These are often things that are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:33",
                    "end": "00:12:00",
                    "text": " quite difficult for anybody workers, and they often need a lot of support in customized ways. And I think the second thing that would really excite me, at least from Rocket's perspective, is this is truly one of the best nonprofits in the world. And I think they have been an amazing addition to the ecosystem in India. What's really exciting is could"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:00",
                    "end": "00:12:27",
                    "text": " large countries around the world benefit from these Indian innovations. So could you imagine these kind of supports in Brazil or in Nigeria and other large countries? I think that's the next frontier for some of these amazing innovations that have been developed in India and other countries to start bringing that skill set and that capability that is kind of indigenous to the Indian nonprofit tech sector to the world at large."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:28",
                    "end": "00:12:56",
                    "text": " You know, it's interesting that you say it's indigenous to the Indian nonprofit sector. I think Rocket is a very unique company in the sense that it really behaves like a company. I think Prashan, you know, you've been involved with ACT from day one. You've seen multiple opportunities to work with different founders, right? And obviously with Rocket as well. As you know, we are hearing James speak, it sounds like it's a VC company, right? Rocket is no different than any VC company."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:56",
                    "end": "00:13:28",
                    "text": " Would love to hear what you believe, you know what from everything that you've seen what has worked and where things haven't worked and You know how again as a VC as a capital provider across various areas you I know you're also big in climate and sustainability What are your thinking in AI AI's role what will work and what will not work? So said I think The pilot to scale Has been a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:28",
                    "end": "00:13:55",
                    "text": " has been a challenge if like in the rocket case, it has worked because they've day one thought about it like a product, like you were saying. So it has been engineered like a product and it has been built for scale. Now with AI, this starts getting slightly more, even more complex because AI systems, even at the pilot stage can be slightly compute intensive and the..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:55",
                    "end": "00:14:23",
                    "text": " Capital allocation becomes a problem. So it's a very early days in AI only products. What we are seeing is that unless it is something that is very monetizable, even for Bharat, even for Bharat, unless it's like test prep, very monetizable, different price point, even at a very different price point, you'll be able to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:23",
                    "end": "00:14:46",
                    "text": " kind of quickly do a pilot and see product market fit. But if it's not in those areas, then capital allocation becomes a problem. So that's one challenge that we see. The second challenge is going from pilot to scale. I think there has to be a lot more re-engineering for even more efficiency."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:46",
                    "end": "00:15:09",
                    "text": " with AI compute again. So whether it's bringing in edge compute related stuff, SLMs and your own proprietary LLM technologies. So I think it's not very straightforward from..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:09",
                    "end": "00:15:39",
                    "text": " I mean, if you built technologies in the previous era, I think server costs won't disproportionately increase when you went from pilot to scale. So that time we are starting to see is a problem. The third problem is, I think, very much what I was talking about, which is data. And the data environments in our country are extremely archaic and broken. So when you look at something like health care,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:39",
                    "end": "00:16:08",
                    "text": " So even if you do a pilot in a particular environment, how do you then actually integrate it into a clinical system and make it work across states, across boundaries becomes, you can make it work in Telangana at a pilot level, but then how do you take it to another state? So that's the third problem. And fourth, I think just the engineering that is required to make it easy to evaluate."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:08",
                    "end": "00:16:37",
                    "text": " monitor continuously. So it's not like your old products that you've built, right? So you've built the product and then you could have a next version three months from now. But AI products need continuous monitoring and evaluation. And I think that's one way you can differentiate a good AI team from a not so great AI team in terms of what kind of eval and monitoring mechanisms have they built from day one. I think Prashant, you pick up a very important point that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:37",
                    "end": "00:16:57",
                    "text": " you have to retrain in some ways the donors, because the donors need to understand how capital will get deployed. Yes. Because if you're saying that there's a much higher compute requirement upfront, and donors need to pay for that, and donors don't understand that at this point. Yes. Because as you scale, the cost will only go up."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:57",
                    "end": "00:17:26",
                    "text": " But I think that's where the ingenuity and the ability to build frugally as AI evolves for these impact sectors. I think teams have to, you need to be able to judge the capability of the teams to be able to do that because capital is not going to be available just like you have in these other sectors. You know, I think from Ramakant, your standpoint with VISA, the VISA was a for-profit from day one."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:26",
                    "end": "00:17:53",
                    "text": " but you were solving a problem that was a huge impact, right? How did you think about, and you were AI first, you know, before AI became big, right, in terms of how you built the bot, how did you think about this, you know, AI, for-profit, scaling, who would pay for it, how, you know, how did that all thinking evolve? So I think the answer, Sandeep, obviously is different from India versus probably outside India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:53",
                    "end": "00:18:17",
                    "text": " And we started life as wise as a product launched in 2017 on direct to consumer, almost as an experiment, personal experience with mental health. We said you need to have a mental health intervention which doesn't need you to step out of your home and go to a therapist. Can there be a chatbot? This was five years before chat GPT. So we were literally whistling in the wind."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:17",
                    "end": "00:18:44",
                    "text": " And there was nothing there. And most people used to look at us and saying, when we used to say, oh, we are building a chat bot for mental health. And they used to say, what? And now it's not what, it's how. So the question has changed because now it's become much more accepted and it's almost become, I think, the zeitgeist. I think monetization models have also evolved from 2017 to 2026 now. So nine years initially we tried monetizing in direct to consumer."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:44",
                    "end": "00:19:13",
                    "text": " COVID happened in 2019. Suddenly, I think the world woke up to mental health as an issue. And I remember the email we got from a clinician in the NHS, the National Health Service. That was 2019 saying, who are you guys? What have you invented? Have you any idea what you have done? Because wait lists for therapists in the NHS in the UK are about as much as 12 months long. So you have to literally wait for a year before you can see a therapist. So what do you do in the meantime? So she tried it, tried this with her patients."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:13",
                    "end": "00:19:40",
                    "text": " And she said, this is the only thing which is ticking on their phones and nothing else is. So you have done something special. Have you thought about clinical safety? That was the first time in 2019 we said, oh, we need to think about clinical safety when you're thinking mental health. And now, of course, that's the only thing we probably think about. So but at the same time, the ability of I think advanced public health systems to absorb AI to implement it and also to pay for it has evolved."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:40",
                    "end": "00:20:00",
                    "text": " I think the story became very different when we moved to India. So because the topic of this panel as well as AI for Bharat, so what broke or what changed? So we started life as a chatbot for mental health and then we came to Bharat and one chatbot didn't work, mental health didn't work."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:00",
                    "end": "00:20:24",
                    "text": " So you're like, what do you do now? Basically we wanted to implement this as a mental health intervention for young people saying that it's a very vulnerable time in their lives. There's huge amount of neuroplasticity. The intervention has a generational impact because it impacts families for many, many, especially if you're thinking about young girls. So which is the target group, which we wanted to go after."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:24",
                    "end": "00:20:37",
                    "text": " But then we also realized that you also have to go beyond the English language speaking, you know, all the people in this room. And if you wanted to go to say a municipal school in Raigad in Maharashtra, we are working with the Women and Child Development Ministry in Maharashtra right now."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:37",
                    "end": "00:21:05",
                    "text": " So we realized that a chatbot format doesn't work because one, the girl doesn't have access to a phone. She probably borrows her phone from her mother twice, maybe two hours a day. It's also very closely monitored. And the consumption pattern is very different. The user behavior is very different and nobody's willing to pay for it. So you cannot build a business model around here. So then obviously lots of bloody noses later. We, I think a chance to upon her, we created a product with said that this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:05",
                    "end": "00:21:25",
                    "text": " should not be a digital intervention. It has to be a physical intervention, not digital, which was combination of a physical plus digital saying that the actual intervention, which the child is receiving is a workbook, which looks like a normal scrapbook or a journal, which the say, imagine a 13 year old girl."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:25",
                    "end": "00:21:46",
                    "text": " which is getting access to a colorful book like a scrapbook given to her by her teacher, which has QR codes embedded inside it, which take the child, even if she's on a borrowed phone, they take her to a very private anonymous chatbot interaction on the web. So you're still delivering digital, but you're delivering it in this blended format."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:46",
                    "end": "00:22:15",
                    "text": " And the second thing is nobody's going to pay for a mental health intervention. So we said this actually has to become social emotional learning. So this is a skill building intervention and there are budgets for books. So schools pay for books. They don't pay for digital apps. So that's the direction we're going in saying that our intervention for mental health for India, for Bharat is actually education. So ma'am, I'll come to you before I'll go to Utkash later, because I think what Ramakant is outlining here."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:15",
                    "end": "00:22:44",
                    "text": " in terms of how AI will come to India, you mentioned this a little bit, but the government will have a big role to play because to the point that Prashanth made that the cost of rolling out AI interventions will be expensive and the government is already doing some work in this. But where do you see, like for example, if Ramakant were to come with Vaisa to Telangana or to come to Niti Aayog, what would your reaction be? How would you see that being taken by the government in India?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:44",
                    "end": "00:23:13",
                    "text": " I can actually share one example. We worked with one of the mental wellness, forgot the name actually in Telangana. So and government of India is also doing that currently subsidizing the compute. That's one way of, you know, helping the startups where the cost of compute is born by the government because at the end of the day, as someone said, you know, we don't look as a government, we don't look at profits, but we look at impact or, you know, solving a certain problem. So that's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:13",
                    "end": "00:23:42",
                    "text": " the main thing that the government looks at. So, the computer is one thing. And one of the biggest challenge start-ups face, you know, is access to, if you're, it's an agri-tech solution, access to farmers. Because at the ground level, even for the mental health solutions that we implemented for substance abuse, you know, I mean, in Telangana, many of the states, this is one of the biggest problem with children are actually getting addicted to these drugs. So, we actually implemented an AI solution."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:42",
                    "end": "00:24:10",
                    "text": " where which is administered for children below 18 years. So one of the biggest problem the startup faced was, well, the government also wanted to implement that. How do we reach out to the students who are actually out there who really need the most? That's where the entire government machinery can help the startup. It could be you may call it as a first customer or whatever it is, but that scale and size the government can really handle."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:10",
                    "end": "00:24:39",
                    "text": " Number one, infrastructure. Number two, you may call it customer base or, you know. Second, third, you know, at the ground level, there are certain nuances. You may have the best product in the world. We just rolled out. We actually had this experience in the agritech sector, you know. It was a marketing, you know, access to market solution, like it's a very good product. We actually rolled it out to the, in one of the districts in Telangana. So what happened was nobody was using it except very few farmers."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:39",
                    "end": "00:25:07",
                    "text": " we didn't understand why they you know they're getting access to market very good information that they can actually get price higher than what they're paying so then we did a you know the ground study and all our team went there and said finally what they found out is the farmers they rely on local suppliers for available to credit you know getting credit and all so they really I mean they don't see actually going beyond what is available because of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:07",
                    "end": "00:25:36",
                    "text": " those already relationships that they've built. So, these kind of nuances, which the startups may not know, you know, that's another thing that the government can help them with. Great. Thank you. I just want to add here, when I spoke about data before, we are working on a very interesting project in Odisha. In fact, I was talking to you, right? So, it's in the health sector where we're trying to actually understand the entire life cycle of data in the health sector because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:36",
                    "end": "00:26:03",
                    "text": " for any government, we don't have the right data for decision making right from the ASHA worker to even the chief secretary level. So there, the government actually wanted us to identify the prioritized list of challenges that they're facing in Odisha and come out with suitable solutions in health care that help ASHA workers to deliver better services. Just wanted to mention this. Great."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:03",
                    "end": "00:26:33",
                    "text": " Again, taking off on that point about data, Utkash, I think I would love if you could talk a little bit about what Adalat does, because you are fundamentally based on data, right? And you have taken on the responsibility of digitizing this and bringing it into AI. So we will talk a little bit about that. And what have been the challenges as you've gone down the journey? Yeah. So Adalat AI is a legal tech nonprofit that builds tech solutions for district courts in India to help them process their cases faster so that we can sunset that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:33",
                    "end": "00:26:57",
                    "text": " dialogue of Tariq per Tariq, which is so iconic of our courts. And the idea is to identify the manual and clerical processes that become pain points in the plumbing of a court. So we are the proud plumbers of the courts that are trying to unclog the pipes of all these processes and paperwork. I think in terms of the thinking and what worked and what started, it's very sector-specific."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:57",
                    "end": "00:27:25",
                    "text": " I'm quite envious of education and health care because there are all these incredible nonprofits that have done a lot of incredible work that's paved the way for impact. For us, we don't know other nonprofits that work directly with courts. That's a big blind spot. And so we had some responsibility to educate the sector on the role of civil society and doing this kind of work. So our ethos was identify painkillers and not multivitamins. A lot of solutions are multivitamins, the green vegetables that would be lovely to have."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:25",
                    "end": "00:27:53",
                    "text": " But I think we thought that let's identify the high fever points in the courts and go with a painkiller to develop that credibility that hey, civil society can actually help you. Give me an example of that. Exactly. So our first solution was stenography, which confused a lot of people. They were like, you're going to solve delays and backlogs by expediting stenography in courtrooms. It's actually a big pain point because in a cross examination in a courtroom, if I ask you a question and you answered it and Prashant is the judge,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:53",
                    "end": "00:28:15",
                    "text": " There are no stenographers in most of India's courts, so Prashanth will write the question and answer. Then Judge Prashanth will get transferred, and Judge Jamie will come in. And Judge Jamie will be like, I can't read Prashanth's handwriting. So judges have WhatsApp groups where they send each other images of depositions saying, hey, what have you written? And Judge Prashanth will say, it's been two years. I can't read my own handwriting."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:15",
                    "end": "00:28:39",
                    "text": " That's the kind of challenges the courts work with. Some states, which were innovative, came forward with a solution called the Fair Copy Department. And as the word suggests, there's a department which will take Judge Prashan's depositions every day and fair them out in writing by hand. And so we asked them, saying, why are you writing this? You can type it out. And they're like, we don't have Tamil keyboards. We don't have Hindi keyboards."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:39",
                    "end": "00:28:54",
                    "text": " And so we were like, this is clearly a high fever point. Judges are developing spondylitis. It's physical manual work. They will say, we're not judges. We are scribes in the system. And so we built a speech-to-text legal transcription tool powered by AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:54",
                    "end": "00:29:13",
                    "text": " Scaling was not by push, which is what often happens in the social impact space. It was pulling us into the system. We were getting calls from states every day saying, hey, we need this painkiller as well. And we were like, wait, we have to build capacity, get more compute, build more language models. This will take time. And I think once we established that successfully,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:13",
                    "end": "00:29:34",
                    "text": " the states then started coming to us for the multivitamins. There was a relationship of trust saying, hey, you actually helped me address a fever in the system. Can you now help me build paperless filings? Can you now help me build chatbots? Because we've proven ourselves with the painkiller. So I think that was our approach. And I think it's proven. It's worked well to some extent."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:34",
                    "end": "00:30:02",
                    "text": " In terms of challenges, as you said, we work in an extremely confidential setting. This is a courtroom. This is a judge dictating a judgment before the world finds out. Stock markets can move on that kind of information. And so security is the highest priority. And we have a three-pronged approach to security. One is we don't use any third-party APIs. So no OpenAI, Google, nothing. These are Adalat AI's models. We are an AI native company, which means a lot of compute."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:02",
                    "end": "00:30:30",
                    "text": " for both experimentation, for the fine tuning of the models, but also for the inferencing. The unit economics changes when the API call is not going to Google, but is coming to Adalat AI. And that's obviously a big challenge. The second is we have to think about encryption systems. This is extremely sensitive databases. And so we've kind of built a blockchain kind of decentralized encryption system where each judge gets an encryption key. And to decrypt the database, you need every judge's localized local machine encryption key."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:30",
                    "end": "00:30:54",
                    "text": " And so security is a big challenge. And the third is just the peculiar nature of the domain and the language we use. As per the doctrine of res ipsa loquitar, per order 39, rule 1 and 2, whatever that means, all conversational speech to text collapses in a courtroom because the language is so pedantic, so technical, that we are kind of."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:54",
                    "end": "00:31:19",
                    "text": " building our own speech to text models for our own languages. Law is its own language. One is Tamil, and then one is legal Tamil. And so I think all of these complicate the work that we do, but I think this is unique for sovereign AI. And so we are really looking forward to more support on compute, number one, that if someone asked me what's the one thing that you would like more from the system, it's compute. And I would say"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:19",
                    "end": "00:31:48",
                    "text": " To the C, I would add coordination in the compute. You know, I was just standing outside and I was talking to someone from an IIT, I won't name which one, and they were like, we're building this amazing chat bot that does this thing to demystify laws. And I was like, I know five other IITs and three other triple IITs that are doing the exact same thing. Why aren't the organizations talking to each other if compute is so precious? I'm not sure that there's as much thinking going into like using it in the most optimized and efficient way. So yeah, I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:48",
                    "end": "00:32:18",
                    "text": " We've now kind of got a foot in the door with the painkiller. The system is now opening the door for us to build more end-to-end systems, but for that to happen at scale, compute is pre-acquisite. Prashant, I'm going to come back to you and to you, Jimmy, as donors, funders in this. You've heard both, I think all three in some ways talk about compute being a big sort of bottleneck for impact, at least impact. You know, you're getting VCs to put money into"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:18",
                    "end": "00:32:51",
                    "text": " other companies but on the impact side how do you see your organizations you know act or in your case agency fund actually playing a role in in driving making that change you know either of you can go. So I think like Utkush was saying I think this requires a very sovereign approach right so it's not even when you look at how to optimize some of this compute so there's a company called E6 data for example right so you're aware of that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:51",
                    "end": "00:33:20",
                    "text": " Now they are doing a lot of this optimization for the larger companies, the Flipkarts, the Swiggies. I don't know what it would take to get some of that into a government stack, sovereign stack, which would be able to provide the same compute at a very differential cost. So it's not just compute. Can the government AI stack look at compute at a very differential, optimize it to a very differential cost? That's one."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:20",
                    "end": "00:33:50",
                    "text": " And I think over time, I would say, I think the solution is edge compute. I don't think there's, I mean, if this has to drop down to, you know, one 10th, kind of a price point. So, and we just launched something in Karnataka, the first AI kind of risk computer, right? Yeah, so that this is an AI edge computer where it runs on Chrome, the models run on Chrome."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:50",
                    "end": "00:34:10",
                    "text": " 15,000 rupees. So it's going to be first deployed in schools for the teachers to personalize curriculums for the classroom they're teaching. So you can't give one of these for every student, but can you give it to every teacher?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:10",
                    "end": "00:34:28",
                    "text": " so that the lessons are personalized at Microsoft. You can run like models on that. Yeah, models run on that. I mean, it's totally off the grid. You can run the model. So I'm seeing that come down to 5,000 rupees to 6,000 rupees, such edge AI compute."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:28",
                    "end": "00:34:58",
                    "text": " computers, right? And then I can see in a courtroom, it's going to be very, very practical to roll it out across the country. So I think there'll be innovation on the central AI stack, the API-driven stuff that's going to be, of course, made more optimal. But there's more category-specific compute solutions on the edge that will also hopefully play a role."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:59",
                    "end": "00:35:29",
                    "text": " Thanks, so that very much resonates. I think there's a need for technological innovation to come. In terms of how we, the agency funds, approach this, I don't think we're radically different to other funders in the philanthropic space. We see our role very early stage, and so where we pay attention is trying to help early stage non-profits."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:29",
                    "end": "00:35:55",
                    "text": " Identify a model a product that works and that might mean model evaluation to get the actual AI to be behaving as it should be, but all the way up to kind of getting user engagement effectively. This is something that is well understood in the private sector, but not always that kind of muscle doesn't exist in the nonprofit sector in the same way. And then ultimately helping nonprofits run randomized trials, impact evaluations to show impact."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:55",
                    "end": "00:36:15",
                    "text": " Because I think ultimately there we want to be able to work with nonprofits to help them demonstrate to the ultimate kind of payer, which is going to be government, that this is a cost effective, high impact solution. And I think that's that kind of is the second part of where we see the solution."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:15",
                    "end": "00:36:43",
                    "text": " From because we're relatively small funder to these big questions of large-scale fund a kind of large-scale costs Ultimately, like I would love to see a world where the user and the payer are much closer I don't think we want a world where like the Gates Foundation or the agency fund or whoever else are paying for indefinitely for things not only because of the kind of the sustainability question, but also because exactly to with crushes point about"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:43",
                    "end": "00:36:45",
                    "text": " You know pain points"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:45",
                    "end": "00:37:14",
                    "text": " You know, if the funder is the pair, the pain points that are going to be often solved are the funders pain points. But really, what you want is the nonprofit or the organization, the product to be solving the pain points off the government who are essentially acting on behalf of citizens. And so getting that much more close to citizens, I think, is ultimately better as well. So those are the two things. One is just helping nonprofits build products that work through evaluations, whether that's technical AI model evaluation or it's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:14",
                    "end": "00:37:38",
                    "text": " standard kind of evaluate, impact evaluation. And then the second is really helping find models where the user is the payer. And I loved hearing about the two stories that you guys had. I think that's a good either of you take on what he just said. I'd like to probably add a different dimension. One is definitely because we are talking about AI and the cost of compute."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:38",
                    "end": "00:37:54",
                    "text": " But I think the cost of implementation or the cost of actually rolling this out in real life is not just compute or my sense is compute is probably a very small fraction of the cost you would need to incur to actually make this rollout at scale in government."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:54",
                    "end": "00:38:22",
                    "text": " or in the public sphere in real life. And I'm talking about the transformation, I'm talking about the fact that you need to train people, you need to make sure that they're comfortable with whatever you're implementing. That's a non-trivial cost, not just in terms of energy and attention, but also in terms of money. And I had two or three different, just in terms of our work on the ground, both with in Maharashtra and now we are about to do something very significant in Meghalaya as well. And we're talking to the government of Punjab and we've just..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:22",
                    "end": "00:38:51",
                    "text": " We just won a $7 million five-year grant with the Wellcome Trust as a clinical trial to actually implement this in UP. So there are so multiple places where we are doing this. And I think we are very quickly realizing that AI is not just tech. It is tech plus people. And in the mental health space or in the emotional resilience and skill building space, you do need to, for example, gain the young girl's trust."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:51",
                    "end": "00:39:21",
                    "text": " And how do you do that? You need to think laterally or you need to think very non tech wise saying, go with the teacher and don't go out and say, hey, here is an app or here is a digital solution which you need. You need to get the parent on your side. So how do you do that? So you need to first get the government to say, look, we are backing this. And then we are now sending letters to each of the parents of all the children in school saying this is actually supported by the government of Maharashtra so that they allow their children to actually use the service."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:21",
                    "end": "00:39:40",
                    "text": " So these are non-trivial problems and if you don't think about them then the tech will just be a piece of tech lying on the shelf. That's a very good point that the systemic change actually ends up in many cases stopping things and that probably is true for you more than anybody else right? Oh yeah absolutely I think any impact has behavioral change behind it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:40",
                    "end": "00:40:09",
                    "text": " and convincing judges who are pretty set in their ways, who aren't usually told how to do their work to now be told how to change their workflows is very radical. And so it's a twin engine mission. I completely agree. Part of it is tech, and that's only a part of it. The other big part of it is programming to actually train and workshop with the audience and the population that needs to use the technology because otherwise the product is not going to see that kind of uptake. So we actually work and workshop with the judges."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:09",
                    "end": "00:40:33",
                    "text": " But then with the stenographers, the court clerk, the court record keeper, the whole team around the judge that operates in a courtroom so that the technology is internalized in their workflow and is actually implemented in court. And that requires sometimes bottom-up approaches where you can kind of go and work with courts and do all of it. But honestly, in our setting, because it's the law,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:33",
                    "end": "00:41:01",
                    "text": " top-down approaches also work very effectively. So we've seen very clear impact of things like mandates. Kerala became the first state in the world a few months ago to mandate us as an AI solution in every courtroom to transcribe court proceedings. And while we had seen very healthy uptake till then, the level up, I mean, after that has been absolutely transformational. Does government pay for that? Partly. Not entirely, but partly."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:01",
                    "end": "00:41:12",
                    "text": " Now it's become a precedent for other states. But I was going to come to the second point, which is that the end goal for us and many other organizations is government adoption. And today they don't have the compute to run these systems."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:12",
                    "end": "00:41:32",
                    "text": " and not just compute they don't have internal tech teams to actually productionize models and scale them and so we are filling that gap now but the end goal is at some point to migrate these solutions to state systems just like governments have very routine IT teams will they have routine AI teams in five years ten years that actually run these systems and so how do you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:32",
                    "end": "00:41:59",
                    "text": " build systems in a way from the outset that prepare you for that migration and future, I think keeping that direction in mind completely changes how you design and implement solutions and it changes the engineering. On that point, I think that's a very interesting point he brings up that the government IT teams effectively have to become government AI teams going forward, right? Are you, Niti Ayog, how are you driving this?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:59",
                    "end": "00:42:26",
                    "text": " I mean, basically there is a shift in the mindset of the CDAC or any of these organizations that have been building your issues and other applications in the past. Now I have to start thinking AI first. How are you seeing or how are you going to see something like Niti Aayog is at the center of this, right? And working with all the states, how do you see that driving that change from an IT first mindset to an AI first mindset?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:26",
                    "end": "00:42:55",
                    "text": " In fact, I mean, I work with Frontier Tech Hub in India. So, which actually looks at different technologies, how it's going to impact, you know, India for the next 10 years and we come out with the roadmap. So, one of the roadmap that we have recently launched a couple of months back, two months back is AI for jobs, you know, wherein we have identified actually different profiles of individuals, you know, what are the roles that are going to change, different roles that are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:55",
                    "end": "00:43:25",
                    "text": " new rules that are going to be created and what are the jobs that are going to be lost and we actually came out with certain recommendations to the government as to how the you know government should be prepared for it. I mean since it was part of the budget I'm not sure if I can speak it now speak now about it we actually recommended setting up a talent AI talent machine you know across the governments so but that recommendation they actually it came out in a different way which government is setting up a standing AI committee."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:25",
                    "end": "00:43:49",
                    "text": " they are going to and there's so many fragmented initiatives across the country you know in AI scaling like so how do you actually bring all those into one umbrella and create that AI talent pool for the government as well as for the private sector is what you were thinking it's that's a very good report that has been launched I think I think we need a AI just for coordinating government that I think is a starting Bharat we can start with that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:49",
                    "end": "00:44:07",
                    "text": " You know, what I'll do, thank you very much. I can keep asking questions, but I'm sure people over here have questions as well of the panel. We'll give some mics if anybody has any questions. Please, just if you have a specific question to a panel member, please ask them specifically."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:07",
                    "end": "00:44:30",
                    "text": " Yes. So when we are talking about education and a person who is working in education, I want you to understand from law to health care to the education system that we are talking about, any individual that is going to be like taking a job in future, how skills and capacities are being changed from 21st century skills to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:30",
                    "end": "00:44:53",
                    "text": " AI skills, what will be replaced by the skills that are required to for a person to let's say work with AI or what what are all sort of changes in curriculum are required per se to know where the students are going to head in future. So that is one question that I wanted to ask. Yeah."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:53",
                    "end": "00:45:23",
                    "text": " That's exactly what I said before, you know, we came out with a roadmap, which actually identifies all that, you know, and what level we need to actually introduce courses, you know, so that students right from certain level understand AI first, as you rightly said, you know, so, and government of India and the minister of education, they also actually set up NSDs national skill development, you know, centers where they're actually offering these courses. That's one aspect of second MSM is."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:23",
                    "end": "00:45:50",
                    "text": " That's one of the biggest sector in India, you know, they contribute almost more than 50% of the GDP, but they don't actually use technology and forget about AI. So that is one sector where the government is also focusing on to build skills so that their productivity increases. So one study that was done earlier long back, you know, said that why is that the MSMEs remain without, you know, improving their income and returns on."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:50",
                    "end": "00:46:18",
                    "text": " One of the biggest, you know, roadblock for them is they just don't use technology. Forget about AI. Even basic IT tools also they don't use. So, you need to start from there, you know, build the technology and then come to AI. So, but again, MSMEs are into different levels of maturity. Certain MSMEs, they do have technology. Then how do you introduce AI to them? That's a planned approach, strategic approach, you know, you need to actually adopt. So, right from schools to colleges to undergrads to..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:18",
                    "end": "00:46:48",
                    "text": " to fellowships there's a whole life cycle you know level of educational programs we are actually recommending. Hi I wanted to ask about the like how do you think the the the police or like the people who are protecting us should be trained in order to like find the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:48",
                    "end": "00:47:20",
                    "text": " upcoming uh state of the art uh like uh cyber hacks that are happening plus like how do you the my uh second query will be it should be in those regional languages and like how should like uh not just the training but uh uh the uh inference from from the infinite sides of things how should that be done yeah"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:20",
                    "end": "00:47:49",
                    "text": " No, I think, you know, this is something that recognizes is a, is a need. This issue of, you know, obviously there has been cyber hacks. And particularly those that have been targeted towards financial crimes and so on. And today I would say it's probably the number 1 of the 1st, top most things that the government is looking at. From a protection perspective, you're seeing the RBI is involved in this, the whole ministry is involved in this."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:49",
                    "end": "00:48:17",
                    "text": " The various banks, there's a lot more. It starts with awareness. Number one is, you know, in cyber, the likely victim is the one that has to be made aware that these things can happen. And there is a huge budget that is now has been created around awareness in particular. But beyond that, you know, 25,000 or something is given as insight, you know."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:17",
                    "end": "00:48:44",
                    "text": " I mean, people who lose money, they're going to government is going to pay something like that. So there is understanding of that. Now, the problem that you are highlighting is a global problem. It is not just an India-specific problem. And the solutions also have to be done at that level of scale. So I don't think this is something that is just a Bharat problem. The advantage that, and obviously you are aware of what Sarvam and others are doing,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:44",
                    "end": "00:49:14",
                    "text": " But the level of voice capabilities that are getting built in AI and solutions in India is actually a very big positive for us. Vaisa explained that. But one of the things that we are seeing is voice is going to become the primary means of communicating with technology in India. We're already seeing a lot of that happen. Adalat is a great example of all transcription. Everything is happening now. Voice makes technology accessible."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:14",
                    "end": "00:49:42",
                    "text": " And these problems require that accessibility, right? Somebody should feel comfortable being able to communicate and say, this is what has happened to me. So I think- And just to add, most of these police departments, they have cybersecurity cells within their organizations. And they're well-equipped, you know, both technically. They have the kind of expertise also built within the organizations that's being done."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:42",
                    "end": "00:50:00",
                    "text": " Hi, sir. My question is to Mr. Utkarsh. Sir, I was reading a newspaper report recently in which there have been two startups which have come up in which they have built an AI model which studies the material which is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:00",
                    "end": "00:50:25",
                    "text": " published by the companies which are listed on the stock market. So they do all those stuff to analyze. So they say there are 3000 to 5000 paper report that a single company submits to the stock exchange by which people can actually analyze and make the decision to kind of forecast what and when they do need to put in money when they do need to take it back. So I was"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:25",
                    "end": "00:50:44",
                    "text": " uh seeing a startup pitch also on a similar note in which these AI generated models are actually transcripting it making it short and making it worthwhile for the people to actually analyze and then make a decision but still people find it very what's the question"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:44",
                    "end": "00:51:10",
                    "text": " Yeah, but still people think that when they transcript all of this and make it in a smaller decision-making forum, do we still think that AI is equipped enough alone technologically to make it as a decision-making forum for the people or the general public to answer to that? Yeah, so I'll take this to the legal context in which I work because it's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:10",
                    "end": "00:51:31",
                    "text": " Step one is transcribing what you're saying, right? So getting that accurately, which is already a hard problem because the language is, whether it's finance or law, it's technical, it's Indian languages. We don't have technology for those interfaces. So that's one. And the second is to build that intelligence layer on top of that material that leans and understands context, right? So for example,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:31",
                    "end": "00:51:55",
                    "text": " how to interpret a law, reading Section 2A with Section 10 of the Act and drawing the import of the law is a human analytical skill which AI still hasn't figured out. So I think the next step which you're talking about actually requires a lot of work. It requires experts to become data annotators. Today data annotators are largely those that are tagging speech output and text output and training models."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:55",
                    "end": "00:52:09",
                    "text": " Now the work is shifting to lawyers, financial experts, training these models so that the next layer of analysis comes in. And that's where the harder part is. I will, though, say that in the law, I am a little nervous about this being given out as legal advice."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:09",
                    "end": "00:52:39",
                    "text": " there are actually very serious risks or challenges around it. First, regulatory, the bar council, how do you allow a non-licensed lawyer to give legal advice and someone to act on the legal advice? And second, you know, you could be held for professional malpractice. What are the kind of safeguards that someone can have? So you want to make this accessible, but while within the import of the law and for finance, law, health care, the biggest challenge here will be how do regulations keep up? How does the regulatory framework enable these to provide"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:39",
                    "end": "00:53:09",
                    "text": " technical information that's otherwise inaccessible, but also in a way that's safe. And that's going to be a big question mark for the regulatory space going forward. So we're just short of time. So last question. Should I go ahead? Yeah, this question is to Prashanth and James. As philanthropists or funders of social impact, what are some key principles, pre-AI and post-AI, that you're looking in the post-AI environment that is guiding your philanthropy?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:12",
                    "end": "00:53:40",
                    "text": " Well, one of the things that, so I work in development broadly. And so the thing that happened contemporaneously with AI becoming very popular was America decided it didn't want to do foreign aid anymore. And so that, I think, has transformed the whole conversation around what social impact can mean, in particular, in contexts where public funds are kind of are limited."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:40",
                    "end": "00:54:02",
                    "text": " I think what it has done is it has changed how people think about, you know, how are we going to address problems like early childhood development, maternal health, moving really towards a focus on cost effectiveness and, you know, delivering effective state capability in ways that are technology driven."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:02",
                    "end": "00:54:30",
                    "text": " And yeah, ultimately much less expensive. So that kind of focus on reducing costs by leverage technology is what I see as the critical change. It's not so much that it didn't exist before. I think rocket learning are an illustrative example. They are working with government and they are able to deliver highly cost-effective interventions as a result of technology. But I just see that now expanding radically."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:31",
                    "end": "00:54:56",
                    "text": " No, I think I'll just say that the more hairy problems that we were never, as philanthropists, were comfortable taking on, I think we'll take on bigger challenges and problems. And I think that's the shift you'll see. So hopefully this diffusion of AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:56",
                    "end": "00:55:08",
                    "text": " in a cost-effective manner to be able to solve these hairy problems is what we'll want to champion and take forward. Well, thank you, everyone, and thank you to the panelists for a very engaging discussion."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6994305025c9f4e0d7e70329",
    "projectId": "409518",
    "projectName": "Building Sustainable and Resilient AI Infrastructure",
    "projectDescription": "Building Sustainable and Resilient AI Infrastructure",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T09:09:35.559Z",
    "projectUpdatedDate": "2026-02-17T09:09:35.559Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-17T09:09:35.559Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "Building Sustainable and Resilient AI Infrastructure(MP3_160K).mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771319359451.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:25",
                    "text": " runs in basically two directions. One is that we are trying to create infrastructure so that we can use AI for the good that we want. And the second is that the same AI can help run our infrastructure in a more resilient and sustainable way. So there are these two sides to this, which is very interesting, which is probably not true for any other kind of a broad use case, if you think about it. So that is very interesting."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:25",
                    "end": "00:00:55",
                    "text": " And the second was that the design choices when we think about our infrastructure are shaped by what we want that infrastructure to deliver. So ultimately, when we think about the resilience and sustainability as the topics that we are discussing, they are also shaped by our constraints and the choices that we make. So with that kind of an overview, and now to build on some of these discussions forward,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:56",
                    "end": "00:01:24",
                    "text": " We have a very interesting set of speakers. So I'm going to start with you, Sharad, and you happen to be sitting right next to me also. And so Sharad runs SIFI, Infinite Space Limited. So those of you know SIFI, and those of you who are in my generation, SIFI is a old player in the Indian ecosystem. But when we are thinking about AI infrastructure, which is a new in thing, SIFI represents the data center layer in this."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:24",
                    "end": "00:01:50",
                    "text": " this conversation and in this infrastructure. So Sherb, let's take a step back and let's have an assessment of this infrastructure. I think it was you who was telling me or I think it was you only that our entire data center capacity is less than the size of the data center capacity of Chicago. And I looked up Chicago and it's true."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:50",
                    "end": "00:02:05",
                    "text": " And maybe it is mature and all of that, but I want you to tell us that when we look at in the global context where we are, what are some of the things that define the ecosystem as it stands?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:05",
                    "end": "00:02:35",
                    "text": " the shape and size of this ecosystem. Where do you think there are differences in terms of how this ecosystem is shaping up in India as to probably what you see outside? And also what are the areas where there are gaps in this ecosystem? So we'll go about in the future later, but give us a good sense of the present today. Yeah, let's just start with the history, that's right. In India, the data centers started taking a shape."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:35",
                    "end": "00:03:01",
                    "text": " after the cloud transformation of the cloud movement started. So the hyperscalers and the cloud service providers, they started entering into the country in 2015, 16. Before that, many of these startups that in India came on the back of the cloud services. They ended up using the infrastructure outside India. One of the..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:01",
                    "end": "00:03:22",
                    "text": " a large hyperscaler player on more than $200 million revenue from India in 2015 without having the data center presence in the country. From there, when it is started, it still has not reached to any stage where it could be the kind of capacity that the United States had."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:22",
                    "end": "00:03:50",
                    "text": " So for example, the United States at the end of 2024, my numbers could be wrong here and there, but somewhere between 20 to 25 gigawatt of the data center capacity versus India's just 1.3 gigawatt, which is today probably 1.5 gigawatt. So a lot definitely needs to happen to get where we should be. So talking about the future later, what is happening today?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:50",
                    "end": "00:04:06",
                    "text": " Out of that 1.5 gigawatt capacity that we have today in the data center infrastructure, close to 3 to 3.5 gigawatt capacity is under development. Now, underdevelopment capacity could come live, let's say, next two to three years."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:06",
                    "end": "00:04:33",
                    "text": " You would say that India will be three times the data center capacity of today in two to four years. But it still will not be enough what is required to be taken care of. And there is a lot more need to happen and a lot faster it needs to happen so that the time to market is shorter and shorter possible so that we could bring 10 gigawatt in the next four years, not just the three to four gigawatt."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:33",
                    "end": "00:04:50",
                    "text": " things are happening towards that. How is industry reacting to that? How is the infrastructure reacting to that? And how the whole ecosystem is building around that. So it starts with the, you could say, the capital. It feels like that there is no shortage of capital right now."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:50",
                    "end": "00:05:07",
                    "text": " This is the capital intensive business and this is the business where the returns could be made in the assured way with a very high credit worthy customers. Therefore the credit is not such a, the capital is not such a big issue. Second is."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:07",
                    "end": "00:05:24",
                    "text": " the government support and the policy and the regulatory framework. It does look like government is very, very supportive to the data center infrastructure. It's such a huge event being organized today in this week for AI, that AI event."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:24",
                    "end": "00:05:43",
                    "text": " will make no sense if we didn't have the data center infrastructure, the layer below what will run the AI. So government is supportive. Recent times, we would have seen Achish and I both have followed that how government had been consistently organizing the consultations with the industry."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:43",
                    "end": "00:06:10",
                    "text": " In fact, government has heard the hyperscalers also, and their demand has been met, it seems, in the latest union budget. And that will probably create a lot more demand. One of the big tax hurdles that they were facing, one of the big policy hurdles they were facing, it is now removed. So that does create a demand acceleration to bring the hyperscaler support also. Infrastructure is coming up very nicely."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:10",
                    "end": "00:06:21",
                    "text": " Unlike the US, where the time to market for energy currently right now is three to five years."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:21",
                    "end": "00:06:51",
                    "text": " the hyperscalers or the data center service providers, they cannot wait for utility to bring that energy so long. And therefore, they are doing a lot of behind the meter infrastructure of the power development. That scene doesn't exist in India. India has the surplus power capacity. And the transmission infrastructure is in such a way that it can bring power to whatever extent possible to the facility at the moment when it is required. So in a way, the ecosystem is fully"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:51",
                    "end": "00:07:20",
                    "text": " in place to really catapult us to the extent where we need to be. It's just that we all together need to execute it. And there is enough number of us who are making sure that we get to that capacity level as quickly as possible. All right, so Sharath, we heard from you, and it's very clear and loud in this room that as far as the data center layer is concerned, we are only dealing with the problem of plenty."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:20",
                    "end": "00:07:42",
                    "text": " and really it is for us as an industry to capture the market. And I also agree with you that from a government point of view, from the policy lens, because that's something that we've been working closely with the industry, that they've really tried to address whatever concerns that could be there in terms of making a lot of those commitments in investment actually happen."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:42",
                    "end": "00:08:12",
                    "text": " But before I move on, and I'll come to Rushali just in a minute, is that, tell me, when you talk about the ecosystem today, how would you rate the Indian data center ecosystem when you think in terms of things like resilience or things like incident response? So just a quick answer, what is your sense? See, we can measure it in terms of the outcomes. You can give me an example to explain."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:12",
                    "end": "00:08:39",
                    "text": " We measure it in terms of the outcomes and what is the expected outcome of an infrastructure. It should always be available. It should be resilient in the sense that redundancies should not be lost. There should be least amount of risk into the infrastructure. It should be secured physically and the software layer and the network layer and all layers. It should be safe."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:39",
                    "end": "00:09:07",
                    "text": " It's the electrons per square millimeter is very high in the data center. It's not such a safe place if it's not well taken care of. And then it should be working towards the sustainability that is the renewable energy consumption and then also towards the IE 100 or net zero goals. In all these aspects, India had been a zero."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:07",
                    "end": "00:09:21",
                    "text": " Zero availability incidents. We haven't seen any downtime of data centers in the country. We haven't seen any physical security incidents in the data center. Safety during weather."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:21",
                    "end": "00:09:45",
                    "text": " the construction and deployment or operation. We don't have any recorded safety incidents that have impacted the life. Obviously, there will be good saves and near misses at times, but not to the extent that there had been the losses. And then generally, it had been resilient. In fact, our customers..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:45",
                    "end": "00:10:06",
                    "text": " keep us on the toe when we lose the resiliency. So the example is that. If there is a redundancy in the infrastructure and the redundancy goes down, then not just us, but our customers also get nervous about that. And that's where we are. The country is full of engineering talent."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:06",
                    "end": "00:10:33",
                    "text": " who are designing and deploying and then operating these very, you could say, mission critical setups. And they are able to do it very nicely. All right, great. So I'm going to come back to you later. And I'm going to shift track a little bit. And Rushali, you are the global director for climate operations at Google. And we are so delighted to have you here. And what I want to do is to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:33",
                    "end": "00:10:56",
                    "text": " straight away jump into your role in terms of what you do and and and try and understand that you know and the topic is the sustainability aspect which also Sharad alluded to and by the way if any of the claims which he made if you as Google say no no no no we we think these are some of these things which are not really so true feel free to tell us but first on the sustainability"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:56",
                    "end": "00:11:12",
                    "text": " So, given the fact that in a lot of these things, when we think about from a business lens, cost, reliability, speed, these drive the entire infra ecosystem."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:12",
                    "end": "00:11:34",
                    "text": " And in some sense, one could probably argue that some of these might be at cross purposes sometimes or at friction sometimes when you think of sustainability goals. For example, we also think in terms of time to deploy and so on and so forth. So how do you?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:34",
                    "end": "00:11:58",
                    "text": " How does those decisions sit within an organization? And what is the journey that you have seen? And also maybe technology comes to rescue at times. And also help us understand in terms of geographies, because you operate at a global level, that do these decisions and these aspects materially change when you look at a particular kind of geographies? Because in India, south of equator, we are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:58",
                    "end": "00:12:27",
                    "text": " There is a heat. There is all of those kind of issues. So help us understand how does your role really define some of these things when you look at sustainability as an objective in itself. Thank you. That's a lot of questions. I'm going to try and remember as much as I can. Can everybody hear us? Thank you. I'm going to start by saying thank you. It's an interesting event to be at, having an AI Summit in India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:27",
                    "end": "00:12:57",
                    "text": " at the time when you're talking about so much of infrastructure expansion, and it's so linked to clean energy and that expansion. So it's a fascinating time, and I'm super privileged to be here. I also appreciate the fact that between the three main pillars and the seven chakras, you have all of the elements of sustainability, which is planet or whether it's resiliency or people. And I appreciate that because I think that's this kind of philosophy we have internally within Google. And I'm sure all the hyperscalers have as to how do you look at"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:57",
                    "end": "00:13:18",
                    "text": " How do you look at growth? And there was a time where sustainability used to be more like a CSR, more like a wrapper, more like building on things. And what I think has shifted in a good way is, how do you build all of these things into it versus bolting it on later on? And I'll walk you through a few examples. So my role at Google is to help Google decarbonize."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:18",
                    "end": "00:13:29",
                    "text": " across all its operations and its value chain. That's sort of at the top of it. But what that includes is looking at every single aspect that we have in Google, whether it's our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:29",
                    "end": "00:13:48",
                    "text": " you know, your foundations from your minerals, your materials that you get into your construction to how do you operate a data center to then how do you actually, you know, sort of generate anything on top of it, whether it's green software or the application. So how do we build sustainability or carbon or any of those metrics as we go along the process?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:48",
                    "end": "00:14:14",
                    "text": " And there are trade-offs. I don't think if anybody tells you it's a win-win, it's a good idea. But in general in life, I think we always have trade-offs. And what I believe is more authentic is appreciating those trade-offs and then working through it as you go through. And so a good example is I think what you said is right now we're at an interesting point where we have the technologies, but the speed at which we need AI deployed, the speed at which we need infrastructure built."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:14",
                    "end": "00:14:38",
                    "text": " And the speed at what it takes to normally deploy some of these technologies don't always go hand in hand. And you mentioned about in the US, the time for interconnection, the time to get grid energy is so much longer. And you're trying to figure out new models around it. So there is a trade-off between speed, technology, efficacy, and how do you do that? And where does innovation come into play?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:38",
                    "end": "00:15:03",
                    "text": " In some other ways, there are trade-offs. Within trade-offs, I would say there's more design considerations. So you talked about different geographies. Now, some are hotter than the others. Some are more water risk zones than the others. And so when we look at design for data centers, we look at some of those pieces. We use an air-cooled system. It's a water-intensive region. We use a water one where we have enough access to water. So those are actually design considerations. And I can tell you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:03",
                    "end": "00:15:26",
                    "text": " The hyperscalers have been doing this for quite a few years. So from a design perspective, I think we've had enough knowledge. The question is how fast and where you can customize it to that location. And that, I think, is the next piece that we're trying to get to. So the India data centers will be fantastic. We've got smaller pieces, but the scale at which we're looking at now means we look at these design considerations slightly differently."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:26",
                    "end": "00:15:50",
                    "text": " And I'll address one other thing. In general, I feel there's a lot of discussion around energy, power, and water, the bigger constraints that come out of the data centers, for the most part. So Google has so far done 22 gigawatts of clean energy across. And that's a huge thing. We started off with initiating and inventing PPAs before everybody was on board."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:50",
                    "end": "00:16:04",
                    "text": " And I say that because it's actually a good thing to have this growth and tech companies who are generally have, you know, are good at CapEx and have have good capital foundations and who are good at innovation and scaling."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:04",
                    "end": "00:16:34",
                    "text": " at the cutting edge of energy transition, because I think that momentum is helping everyone to move forward faster. You're seeing a lot more, you know, amongst all the hyperscalers, you're doing first of a kind projects, you're doing things which then moves the innovation faster, because you have someone taking the bet, you have someone, you know, who's investing the early pilot phases into it. And then you also have them as big off takers. Remember, in the energy market, you need big off takers, you need commitments to do that. I think those are the pieces that we need to do. So from a sustainability perspective, just rounding it up."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:34",
                    "end": "00:16:46",
                    "text": " It's it's in your design. It's in your procurement. It's in your, how you're thinking about data center, citing access to clean energy and not what you take from the grid, but what you can actually add and bring on as you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:46",
                    "end": "00:17:09",
                    "text": " All right, that's good to know, Rushali, and I think it is amazing to also hear those numbers in terms of 22 gigawatt of clean energy done. So I'll come back to you because I'm going to talk to you about your, some maybe about your India plans, and we'll put that in the context of sustainability when we come back. But Nicole, I want to now focus on your work at AWS."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:09",
                    "end": "00:17:38",
                    "text": " So Nicole Foster, she is the Director of Global AI and ML Policy at the AWS. So I want to shift focus on resilience, because I think at the end of it, when we think about AI infrastructure, the single most important thing would be trust. So when you think about AWS journey in terms of building out that kind of infrastructure, which kind of relays that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:38",
                    "end": "00:17:60",
                    "text": " So where, if you want to kind of reflect and don't go too back in the past, we want to keep it a little more current in terms of maybe last couple of years or so, that where do you think the industry has really moved forward in terms of genuine improvements to be able to deliver that kind of a trust for all the..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:60",
                    "end": "00:18:15",
                    "text": " all that sits on the other side of the infrastructure, which is the client and customers and users. Are there some examples and what is your assessment on where we are in terms of resilience and trust as an industry?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:15",
                    "end": "00:18:41",
                    "text": " If we think about trust, we also think a lot about security and controls as well, not just resiliency of the infrastructure. And Amazon, just to go back a little bit in terms of where Amazon started on some of these topics, it was really Amazon, the bookseller. If you think about the early days of the internet and think about how weird it would seem to use your credit card online."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:41",
                    "end": "00:19:05",
                    "text": " And so that was sort of the first initial hurdle that we had to overcome as a company to sort of start to kind of create that strong kind of e-commerce business was how do we make people feel secure about using their credit cards online? And then you build from there. So security is definitely built into everything we do. And to your point about not adding security on later or sustainability on later, it's the same with security. Security is..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:05",
                    "end": "00:19:34",
                    "text": " integrated into every part. And that's a huge part of, I think, of building the trust. And then the other aspect we think about are also the legal controls of all of that, but also the technical controls. And so that govern access as well as security. And so not only are our services now secure by design, but they're also sovereign by design. And when we think about that, the first thing that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:34",
                    "end": "00:20:00",
                    "text": " that customers decide when they open their AWS account is where they want their data to be. So where do you want your account to reside? And so whether you want that in India or Canada, where I live, that you are choosing where you are putting your data as a first decision. And then the other thing in terms of access and control is at AWS, we can't see any of our customers' data. So by design,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:00",
                    "end": "00:20:29",
                    "text": " it we have no access to our customer data and so that and then why do customers choose AWS or choose hyperscalers it's really because of that the access to security the reliability the availability and the resiliency of what they're able to access and build and not just build build sort of locally where they are but also that ability to scale globally is a huge part of what"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:29",
                    "end": "00:20:53",
                    "text": " our customers are looking for and to be able to do that in a consistent and resilient way. So, if I can maybe quickly follow up on that a little bit, right? So, I completely get that and I think it's tremendous because this is a bedrock in terms of taking the industry forward in terms of the overall, the growth that we are hoping to see over the next few years."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:53",
                    "end": "00:21:06",
                    "text": " So, but if you were to also highlight for us that what are these one or two areas where probably there is at industry level scope for improvement when we think about either security or resilience, which."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:06",
                    "end": "00:21:31",
                    "text": " And given technology is also disrupting on the other side, right? We talk about quantum, we talk about risk and from a technology side itself. So that kind of a lens and also you talked about customer choices. So given the fact that there are many, many kinds of use cases, so in terms of customer choices, what has come in new, which, you know, which,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:31",
                    "end": "00:21:57",
                    "text": " a CSP like yourself as a market is probably now addressing or providing, which wasn't a demand even maybe a couple of years back, if there is anything like that. Well, I think it's the one thing that is very true is that the innovation never stops and the sort of new issues that you're mitigating against don't stop, but also the ability to address those challenges is constantly evolving."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:57",
                    "end": "00:22:27",
                    "text": " And I think one of the other things that we're seeing is, yes, there are new threats arising because of AI or from quantum, but we're also deploying AI to address those threats as well. And so the more that you take people, and this is a very sophisticated audience, but the more that you take people out of security, the more secure it is. And so the ability to sort of automate who has access when they have access, but also being able to understand."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:27",
                    "end": "00:22:54",
                    "text": " and address threats in real time through machine learning has been one of the main innovations. But it never stops. All right. So on that note, we're going to shift a bit of a gear. So I think people who are familiar with data centers, people who are also familiar with the cloud service providers. But does anybody know what does it mean when we say voice infrastructure?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:54",
                    "end": "00:23:22",
                    "text": " Just to show fans, if anybody has a good understanding, what do we mean when we say voice infrastructure? Okay, all right. So I'll just take it to mean that this is an interesting conversation. So I'll bring in Ritu Mehrotra. Ritu is the founder of Shunya Labs. And I met Ritu only a few months back at another meeting and was so impressed and to hear their story."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:22",
                    "end": "00:23:48",
                    "text": " and that we decided at NASCOM that we must have her as part of this conversation. So, Ritu, will you tell us what exactly does it mean when you say voice infrastructure, and why is it important, and how important it is, and what is it that you are actually doing? Thanks, Ashish, and first of all, great to be here, privileged to be here today."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:48",
                    "end": "00:24:17",
                    "text": " So when I say voice infra what it means is at the upper end of the ecosystem when we say something is it can you understand when I say understand it's and it's very nuanced by the way right so when we are speaking we don't realize but we have these multiple different dialects different languages we speak with emotions the sentiments all of it kind of comes together so as easy as it sounds it's extremely nuanced so that's a first layer to it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:17",
                    "end": "00:24:44",
                    "text": " The second layer to it is where you understand. So context creation. So feel free to give examples. We'll understand much better. Yes. So after I explain, I am going to give example. So that's the second piece, where you understand what is said. And then the third piece is where you dish it out, which is called text to speech in our way. So picture this. I am somewhere in Bihar."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:44",
                    "end": "00:25:08",
                    "text": " And I speak Bhojpuri as a language. And I am trying to understand what my ration card balance looks like. That person may not know how to read or write. But are they able to say this in Bhojpuri?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:08",
                    "end": "00:25:34",
                    "text": " If you're looking at anybody who's doing it, and I'm sure he will concur with me as well, the way it works today is you speak in bhojpuri. First of all, understanding bhojpuri is not there today. We launched the first model there. Translating it to English and then translating it back. The whole three layers of translation, the context is gone."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:34",
                    "end": "00:25:48",
                    "text": " So that's where when we say that we're building voice infra, that's what it means. But it's not just understanding the voice and dishing the voice out. The important piece here is also the context."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:48",
                    "end": "00:26:16",
                    "text": " One other piece that we work, Ashish here, is on smaller models that can actually sit on people's phone, the CPU level architecture that people would want to use rather than running kind of GPUs as well. One thing I often say is that the kind of resources Wild West has in terms of compute, in terms of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:16",
                    "end": "00:26:39",
                    "text": " resources, capital data. Sometimes it's like telling people that, OK, I want to solve traffic problem, but hey, let's get everybody a helicopter. But that does not happen for 80% of the population globally, which is very India-like. And when I say which is very India-like, if you look at"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:39",
                    "end": "00:27:09",
                    "text": " large part of Europe which is where GDPR is like so prevalent and sovereignty is you know one of the key themes or if you look at Africa or some parts of India and some parts of Asia where price is a is a big deal and you know people are looking at or the linguistic You know diversity from all of that standpoint, you know, we don't speak just in English or a few top languages So that's where I mean, you know when I say voice that voice ability to be able to hear"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:09",
                    "end": "00:27:34",
                    "text": " understand, respond. That's what it means. So let me just quickly follow up. So given Indian context and given that for us dialect and language changes every so many kilometers, but infrastructure, when we say we always say it along with scale. So when you talk about your work as in building these models which do this voice infra,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:34",
                    "end": "00:27:50",
                    "text": " help us understand little more in terms of what are those use cases where this will be so valuable and beneficial at scale that they kind of solve a problem which nobody else will solve for us. So I'll give you example of Anganwadi worker."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:50",
                    "end": "00:28:19",
                    "text": " We have about 1.4 million Anganwadi centers and about 800,000 workers that are working there. So what this means is if I have to launch AI at scale, and if it has to be purely done on GPUs, you can just imagine 870 million conversations, the kind of GPU consumption I'm going to have, it's going to be humongous for something like that. And that's where you pitch in, wherein you say that, OK,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:19",
                    "end": "00:28:40",
                    "text": " Number one, can you make this at, I don't know, 1 20th of the cost? Can it actually understand the language of the person that a local Angarwadi worker is speaking in Marathi and being it not translated to English and then translated back in Marathi, but Marathi being spoken?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:40",
                    "end": "00:29:03",
                    "text": " fetching into the system and giving the response back so that you don't lose the context in translation. And when you have three different villages actually looking at this data, they are triangulating if there is a disease that could, early detection of a disease or something. So this is just one example that I gave you in the healthcare context, if you will."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:03",
                    "end": "00:29:33",
                    "text": " All right, all right. So we'll come back and we'll talk about the future also, Ritu, with you. So from a founder, we'll move to a professor. All right, so we have a professor here, Mr. Ravi Kiran. And so you are associate professor at the International Institute of Information Technology at Hyderabad. And at Bharat Jain, you are the core team member and principal investigator for vision projects. And as we know, Bharat Jain is one of the startups, the 12 startups"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:33",
                    "end": "00:30:03",
                    "text": " that has made that cut to be developing the language models to solve for India. So again, Professor Kern, delighted to have you on this panel. And so I think the first question, and I'm asking on behalf of maybe a larger audience, is that what is it that one thing or maybe two that an Indian model"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:03",
                    "end": "00:30:32",
                    "text": " can solve better or can solve, which we can't get from whatever is available elsewhere. So I know there is a capability question that even if it is available elsewhere, we should do it. We should. But that's not my question here. My question is, what is it that we are trying to solve which probably cannot be solved by models elsewhere? And it's more in the Bharat Jain context because we already heard Ritu. Thank you. Yeah. OK. Yeah."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:32",
                    "end": "00:30:60",
                    "text": " Yeah, so I'm really enjoying listening to all the conversation here, all the discussion here, because well, we operate mostly at the software level, and we sleep better at night knowing that the foundation is secure. We kind of take it for granted, but you know, when foundations fall apart, then we appreciate them. So for one, I'm really happy to be in a learning environment I have."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:60",
                    "end": "00:31:26",
                    "text": " picked up a lot of useful things, right? So Bharat Jain's primary motive is to build sovereign models and build applications based on that in India for India. And as part of this, it's important to understand that to answer your question, technically there is nothing that others cannot do."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:26",
                    "end": "00:31:48",
                    "text": " in terms of what we are doing right so given the same kind of conditions right it's almost like a lab experiment right like can you if you are given x y and z will the you know will the outcome of that process be the same so i think from uh technology point of view i don't think there is something fundamentally unique having said that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:48",
                    "end": "00:32:17",
                    "text": " I think there are certain aspects that I believe that we are uniquely positioned to solve. And I'll give you an example of that. For example, take public grievances. Ordinarily, citizens, when they have grievances, they take them to the law, like the law enforcement or the government. And it's unlikely that they would expect their problems to be solved by typing into ChatGBT."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:17",
                    "end": "00:32:42",
                    "text": " Well, I mean, they might get some ideas about how to go about it, but it's unlikely that they will be solved. Or Gemini, for that matter. On this panel, Gemini. On this panel, Gemini. Fair enough. So the government actually has a direct rate of public grievances, which is established for this purpose."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:42",
                    "end": "00:33:12",
                    "text": " What you have over there is a portal where citizens can upload their grievances in different formats. And when I mean different formats, it is truly different formats. People can record their grievance, people can upload documents related to their grievance. Now, think about handling the diversity and the different modalities. You're talking about speech, you're talking about text, documents and so on. Now, dealing with all of this,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:12",
                    "end": "00:33:29",
                    "text": " as a manual process is going to be totally not scalable and that is where i think we can step in and in fact we have stepped in right so this is i think a good example of something that this is a real real use case you're talking yeah"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:30",
                    "end": "00:33:51",
                    "text": " This is, I think it takes away the follow on question because I was going to ask you that, whom are you building for? So I think, and you're so right that while the underlying technology doesn't change, it is the context which really makes it very different. And given AI is all about data and the algo working together."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:51",
                    "end": "00:34:21",
                    "text": " We'll come back to it because it's very interesting. A lot of people want to know how soon, how fast, how good will some of this come out. And okay, all right. So with that, we have kind of one round. The intent of this round was to get us up to speed to where we are mostly in the world. What is really happening? What are we doing? Now in this next round, which is our last round maybe, we are going to focus on the future. We are going to look forward because this is a summit on AI impact."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:21",
                    "end": "00:34:32",
                    "text": " And I think we have established that what is happening is highly impactful. But what is that impact going to further translate into next 12 odd months or further is something that we'll all be very interested in understanding."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:32",
                    "end": "00:34:57",
                    "text": " So what I'm going to do is that I'm going to switch gears a little bit in terms of the format. I'm going to pull in my colleague Sudipto Banerjee, who is sitting in the audience with his nice pink tie. So Sudipto has been working with us a lot on the data center work, and he has a question for Sharad. So I'll not run with my question. I'll ask Sudipto to go ahead and ask what he wants to know. Yeah, hi, Sharad. So"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:57",
                    "end": "00:35:26",
                    "text": " We heard in the first round that the reality check of India's one point gigawatt and which is less than the total data center capacity of Chicago. And we also get to hear these projections of 10 gigawatt by 2030. So I mean, as Ashish said that we are expecting this sound to be a future forward looking. How realistic is this estimate? And also, if we dig deeper into this, if we bifurcate the demand right now,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:26",
                    "end": "00:35:54",
                    "text": " How much is the domestic demand? How much is actually driven by the hyperscalers? Because we get to hear when we interact that we are dependent on the demand that is created by the hyperscalers. So our demand market is actually being driven by mostly by the hyperscalers. And please correct me if I'm wrong. So if you can give also a mix of how that demand would actually look like. And one last, I mean, I'm getting opportunities. I'll ask one more question. That if one wish were to be granted,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:54",
                    "end": "00:36:20",
                    "text": " in terms of any policy change or any particular change in the Indian ecosystem, what would that be? And hypothetically, imagine that wish will get granted. So please be bold if one lever you had to pick up. So for data centers, the demand may be coming from the hyperscalers, but hyperscalers are not serving the world. Hyperscalers in India are serving India. So their India deployment is for India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:20",
                    "end": "00:36:50",
                    "text": " Whether the India deployment is for the world or not, that is we will watch. I'm hoping that they will serve world also from India and then we will get benefited by that. So it is India demand. So let's get this out that we are dependent on any kind of a foreign demand. If the foreign demand comes, then my 10 gigawatt could turn into 20. So let's not get there. Let's focus on this. Second, just the data pattern."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:50",
                    "end": "00:37:13",
                    "text": " Since COVID, India's data center infrastructure grew at a 28% CAGR from 0.4 gigawatt to 1.3 gigawatt by end of 2024. I have these exact numbers. That's why I'm counting end of 2024, not guessing any 2025 number yet. Even if I keep the same growth rate,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:13",
                    "end": "00:37:32",
                    "text": " and just put in a small acceleration to the fact that the cloud adaption in the country is still in high single digit or maybe very low double digit, no more than 10 or 11%. In comparison to the developed world where the cloud penetration is closer to 40%."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:32",
                    "end": "00:38:01",
                    "text": " Let's assume that only 40% of the IT can be cloudified. Remaining code will still have to be non-cloud. Then also, there is a four times of just the cloud demand to happen. And if I count that same 28% CAGR, and let's take it in the 7% acceleration on that, then 1.3 gigawatt becomes 6 gigawatt by the end of 2030. Did not count anything to do with artificial intelligence."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:01",
                    "end": "00:38:22",
                    "text": " did not count anything to do with what the data protection acts could come into the picture. And if you add to that, then someone, if says it's a 10 gigawatt, then it's entirely possible. So the demand-wise, there is a solid foundation to that number. That's where we could say that it will be there."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:22",
                    "end": "00:38:42",
                    "text": " And in recent times, what we have seen, there is a good amount of inferencing related workloads has started to come to India. And therefore, that 10 gigawatt is possible. That was the first question about the demand, whether 10 is real or not. The second one was?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:42",
                    "end": "00:39:04",
                    "text": " Okay, this is the domestic demand, but how do you also see the demand coming from Oversafe, the Indian data centers actually catering to the global... That's where I am fully relying on the latest reunion budgets clarification on the tax avoidance that has come up. Earlier, the what is called poem..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:04",
                    "end": "00:39:26",
                    "text": " place of effective management was a big concern with the global hyperscalers or the global players. Now that being removed by the 20 year tax holiday, I'm assuming that some of that will come to India and that should be able to bring to India. There are all the ingredients in place in India to be able to attract at least the regional demand."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:26",
                    "end": "00:39:45",
                    "text": " Power is available immediately. Talent is available immediately. Ecosystem is ready to build as much data center as possible. It's not just us, but then at least three big conglomerates have also talked about gear over capacity in the country, right?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:45",
                    "end": "00:40:12",
                    "text": " And with our experience, obviously, we will be the bigger one with them. So that way, there are all the tailwinds in place to host at least the regional data. The cable landing stations are getting ready in the eastern coast, which can easily serve at least the Southeast Asian demand. So yes, the foreign demand also should be able to be hosted in India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:12",
                    "end": "00:40:37",
                    "text": " One thing that I, you wanted my wish, right? Yes, yes, don't forget the wish, yes. The wish is very simple. True, true, true single window clearance. If today I have to go, and if any new data center project starts in the country, it's minimum four years from the operation. One year takes to get your head around the right land with the power."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:37",
                    "end": "00:41:06",
                    "text": " Another one year takes to get all the approvals in place. And then you take two years to deploy and commission. At least, yeah. So at least these two years can be made zero with the single window clearance and true single window clearance. OK, so the government is listening and your wish is granted. But Vrtika and Swapnil, you know this is an area that we are working for, so it would be good to double click on that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:06",
                    "end": "00:41:36",
                    "text": " Okay, all right. So I know that power and connectivity is also the base layer for you. And in that sense, from an India point of view, I think the story is let's build it and there is so much demand that that all effort will be well counted. So and I'll come back to Rishali again and then to Nicole. So when you look at sustainability as something that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:36",
                    "end": "00:41:38",
                    "text": " we discussed earlier."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:38",
                    "end": "00:42:07",
                    "text": " And often in tech industry what happens is what is science fiction today becomes science tomorrow. So in the sustainability kind of a space, what are some of the things that you are looking forward to in terms of breakthroughs, whether they are processes or whether they are technologies or what are the new developments which you hope to see in market or even experiment? And I'm going to also be a little more take an India lens because you announced this big investment which is coming."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:07",
                    "end": "00:42:34",
                    "text": " up in Vizag. So probably that will see some of those newer things in terms of sustainability play out. So help us understand in terms of future. And it's not just about Google. I think you can take a broader industry lens and give us a peek into what to expect. Yeah, I like the analogy of like what we see in movies and becomes real, which is good and bad both, right? It's a little bit scary too. So I'll just let me start with, you know,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:34",
                    "end": "00:42:54",
                    "text": " I look at the physical infrastructure side first from a sustainability perspective. I think from that, and if you go to, especially for India, so India's got tremendous solar capacity. We know that right now. It almost produces more than it can use during the daytime. And so I think the big thing for India from a sustainability point of view will be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:54",
                    "end": "00:43:22",
                    "text": " um sort of storage so either it is best sort of you know battery storage or long duration hydro how do you figure out a storage model that takes benefit of all the solar and wind i don't think generation is a problem right now as much as then making sure that it is 24 hours and then you bring that into a grid and make sure the grid is reliable i think that to me is a big piece and you know it to me sustainability is right now if you take from an energy perspective is unlocking clean energy at scale"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:22",
                    "end": "00:43:37",
                    "text": " And for everybody and then make it very affordable and to do those two things you have to have things such as I think batteries are very important one long duration storage is important and grid infrastructure. So the way I one of the things I loved I was here last year in Bikaner and I saw"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:37",
                    "end": "00:44:03",
                    "text": " in not just the miles and miles of solar panels, but the new stations that they are being built and the new grid infrastructure and the high-frequency grid, which then becomes an interconnected grid across India, which is fascinating. So I think that, going to your point, the faster you can scale that, I think is a big unlock. So that's the physical infrastructure side. From a data construction point of view, I think there's a lot under green steel and green cement that we can do in India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:03",
                    "end": "00:44:17",
                    "text": " A lot of the big players are in it, and there are many different ways. It could be from the innovation side of materials. It could be from a curation side. And the reason these are specific, because most of you know, like, from a construction perspective, cement is very local."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:17",
                    "end": "00:44:46",
                    "text": " You could have amazing technology breakthrough somewhere else, but you're not going to get the cement to a local place. And same with steel. We've got great steel plants in India. So how do you make that construction as green as possible is the next big one. And then the third is the application layer. What I like about what you were saying is you don't want to run large language models. You don't want to do this training, this inference. And then how can you be hyperlocal with your inference and running those smaller models? And I think we haven't tapped into that as much. And so the more efficiency we can bring in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:46",
                    "end": "00:45:11",
                    "text": " and the more innovation we can bring into that. And especially when you're looking at a billion consumers, as much as I love infrastructure and I'm responsible for it, I would love to have more easy applications and fine-tune deployment that you know where the demand is. You can have real-time demand response. You can figure out which models to use well and how to right-size it. And AI can come to rescue in that side also."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:11",
                    "end": "00:45:39",
                    "text": " Yes, I'm hoping, and we are hoping. We are on the demand response side, and load balancing, we are trying. There is, and even on the grid infrastructure and optimization, we are. But it's early stages, so. All right, good luck to that. So, Vatika, you had a question for Ritu as a follow on. And Vatika, she's part of the team. She's been working on a lot of these works with startups along with Supnik. So I'm going to request her to pose a question for you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:39",
                    "end": "00:46:09",
                    "text": " Thank you, Ashish. Ritu, first of all, thank you for educating all of us on voice infrastructure. It was really interesting. I just wanted to know, you know, looking ahead, maybe say the next two years, what, according to you, would be the most tangible change, you know, from actually this India-built, you know, frugal, accessible AI movement that we are all witnessing today? You know, from that perspective, if you could just pick one sector."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:09",
                    "end": "00:46:39",
                    "text": " where you feel we'll see the impact fastest, you know, be it education, banking, healthcare, I don't know, you pick it. And if you can tell us what is it that the users will see as a difference, you know, and if I can just extend the question a little more. If from the same sector, you could also tell us what is it today that is the most hard part where, you know, which is kind of holding you from seeing the impact that you would want to see. And what do you think will change"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:39",
                    "end": "00:47:04",
                    "text": " you know, in the next two years to make it accessible. It could be, you know, from junior labs perspective, it could be from, you know, other startups perspective, but please feel free to answer. Thank you. I'll start with the industry lens and then maybe I'll move towards, you know, where do I see? So the first thing I think as the industry we have to work on is trust."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:04",
                    "end": "00:47:25",
                    "text": " And that's the most important thing. So when I'm talking to any of the answering engines, Gemini or any other player, I think the first thing that we're thinking that, OK, where is the data going? What is it storing? So that's number one. Number two is its ability to contextualize me."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:25",
                    "end": "00:47:48",
                    "text": " And I'll give you an example, as I said a couple of minutes back as well, that let's say I'm in a rural town. I'm a woman who has to deliver a baby. Or I'm going to a local health facility. Its ability to understand my language, my problem related to my ecosystem."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:48",
                    "end": "00:48:16",
                    "text": " is available on my device and is kind of understanding that. So that's my level of, you know, that in the next couple of years, quarters, if we can get there, I think that is going to be great. So I think the distance that we have to cover is trust. Even in the companies today, when we deploy our voice solutions, right, the first thing that we hear is that the first time when you go in and your AI system doesn't understand what people are saying,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:16",
                    "end": "00:48:44",
                    "text": " The trust level goes down by 22%, and it happens one more time. It goes dramatically, and then people kind of lose interest. And that's why we are seeing that the abandonment of a large AI projects is kind of, it's interesting to see what the numbers look like today. So that's point number one. Point number two is, if we have to make it to scale, it has to be efficient. I'm not saying it has to be cheap. There's a difference between making it cheap and making it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:44",
                    "end": "00:49:12",
                    "text": " frugal, the word that you use. It has to be efficient. I'll give an example. One of the large companies has spent 680,000 hours to train a model that will dish out results. Whereas a company in India, and I know of many companies in India, and one of them, as I said, is sitting next to me. And Shunia, by the way, does the same thing. Can you do this in, let's say,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:12",
                    "end": "00:49:43",
                    "text": " 40 GPU hours right or a thousand voice samples. So efficiencies will have to kick in I think in different and sometimes you also say right that when you don't have enough resources then it becomes the mother of all invention. So I think that's where it will go in my mind. So that's the end state but the way to do it is going to be building efficient systems you know that that is going to be able to give you better tomorrow."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:43",
                    "end": "00:50:13",
                    "text": " Well, one of the biggest beneficiaries, having said that, obviously finance, right? It's another one. Banking. How do you take banking to every remote part of the country? And I will not say illiterate. I will say low literate, right? Well, we have the largest number of smartphones in the world, right? And there are people who can speak, but they can't read English. Can we actually build something for these people to be able to be a part of the ecosystem?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:13",
                    "end": "00:50:35",
                    "text": " All right, thanks. Nicole, I was holding a tough conversation for you, for a gap, because my conversation is going to shift back onto the larger discussion that is now so common as against maybe even three years back is on sovereignty."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:35",
                    "end": "00:50:52",
                    "text": " I mean, now a lot of the conversation is around delivering that as a feature of the infrastructure itself. So being a global corporation and looking at that entire space, I mean, if you help us understand going forward what."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:52",
                    "end": "00:51:20",
                    "text": " where do you think that entire kind of a piece is headed in terms of, in a very simple terms, where do you see, for example, this is going to be non-negotiable in terms of certain areas maybe, and in certain other areas where it might be a combination of factors where in some areas people will choose to do that, and what are those areas where one could still say that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:20",
                    "end": "00:51:50",
                    "text": " the end-to-end experience can be delivered without necessarily compromising in terms of the global nature of the entire delivery, while ensuring that the experiences in terms of control or access meets the sovereignty requirement, but doesn't necessarily need to compromise in terms of the way the physical architecting has to be done. So can you give us some sense of the future?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:50",
                    "end": "00:52:17",
                    "text": " So, at first, I just want to say that this is not just how this conversation is not just happening in sort of my jet lag brain is having trouble in India, but it's certainly happening all over the world in many countries. And it's very understandable as to why I live in Canada and we, you know, this is a very, very kind of active discussion where I live as well. I did."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:17",
                    "end": "00:52:24",
                    "text": " I think it's worth thinking about what it is that cloud brings and why people are still going to end up."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:24",
                    "end": "00:52:54",
                    "text": " really having to make difficult decisions just because of the opportunities of working with the cloud. Cloud is not a data center. It's a lot more than a data center. And cloud is, so first I just want to say we live in an extremely globally interconnected world. So the idea of kind of unweaving the interconnectedness is very difficult and probably not that desirable. But when we think about what cloud has done for artificial intelligence, cloud is really"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:54",
                    "end": "00:53:18",
                    "text": " this technology to be so accessible to people who are not PhDs in AI machine learning. And so cloud has really been an enabler of the accessibility of this technology. And like I said, cloud is not just a data center. We can break it down into sort of three basic levels of the technology stack. And it's important to think about the different layers because how we..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:18",
                    "end": "00:53:42",
                    "text": " how we mitigate some of the risks that we might see is gonna change in how we think about the different layers of the stack. So for example, at the foundational layer of the stack or the training level of the stack, that is really only, I mean, that is probably a little bit easier to build technically would be the."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:42",
                    "end": "00:54:10",
                    "text": " you know, if you just want to have a training data center. But there's very few people who can use it. You have to be really clever, like some of the people on this panel, but not all organizations have really sophisticated developers of technology. And so that's that while that layer is maybe easy to build, it's not the most accessible layer to use. And so the middle layer of the stack is the stack where"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:10",
                    "end": "00:54:24",
                    "text": " cloud providers have created all kinds of different tools and basically the Lego blocks to build AI applications. And they also make lots of models available. You can bring your own model."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:24",
                    "end": "00:54:45",
                    "text": " It gives you all of the security protections, all the responsible AI innovations. And so this is the layer for your economy that is really important. So if you're thinking about how organizations and how companies are going to sort of be able to use AI and deploy AI without having a PhD."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:45",
                    "end": "00:55:08",
                    "text": " like somebody who's really super expert at it, but someone who's got really strong IT skills, that middle layer is super important for your economy. And that's the innovation, the real innovation of AI that people are able to access in the cloud that is gonna be the game changer for our economies. And so that's the part that's hard to replicate."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:08",
                    "end": "00:55:16",
                    "text": " in just building a data center. And so that's why you're going to see different types of strategies or mitigations."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:16",
                    "end": "00:55:44",
                    "text": " When you're thinking about what is important for me to have control over, what is important for me to be able to govern both from a legal aspect or a technical control aspect. And it can be different things. It could be maybe you want to use open source models instead of first party models because those give you more portability. And there's a bunch of different ways that organizations are going to think about that. But there are going to be significant tradeoffs."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:44",
                    "end": "00:56:13",
                    "text": " that we have to make in order to access the best innovation or the most resilient infrastructure or the most secure capacity in order to kind of also manage those other concerns. But it is complicated, but I do think that there's still a ton of value that CSPs offer in an environment where organizations are very concerned about."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:13",
                    "end": "00:56:43",
                    "text": " about having that control that they want to have. Yeah, I think that beep is for us, maybe, but I'm not going to end without getting Professor Kiran for a last word on this and then we'll conclude. So I just want to come straight to the future again. And we know about the India AI mission, the fact that what the Indian startups are trying to do. So help us understand what will success really look like in the next four or five years."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:43",
                    "end": "00:57:07",
                    "text": " and try to explain it to us. I know the beep is going on, but will probably take two, three minutes for you to make us understand that. And also if you don't reach where we wish to, I'm sure we will, but what is that one factor that we would probably need to make sure that doesn't give us that place for any losses? So help us understand."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:07",
                    "end": "00:57:36",
                    "text": " So I think that's really a very nice question. This is a question that we ask ourselves all the time. I think all organizations do ask, what would success look like? You visualize it, you plan for it. So I think success, and when I say success, not just for solely Bharat Jain, but for all the initiatives which are part of this cohort, would be to have four"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:36",
                    "end": "00:58:05",
                    "text": " AI what UPI did for the country. So we want to have the UPI movement for India so that would be what I call the first level of success and then to have the Nokia level of success where there is a group of a very small group of people sitting somewhere in a very small country, but you have they are making a product which everybody"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:05",
                    "end": "00:58:33",
                    "text": " seems to use right and at some point that was certainly the case right everybody seemed to have a Nokia phone with them right and they were you know they had a certain resilience to them right everybody I think can attest to the resilience of that's true that's true I remember yeah like you know people claim to throw them from you know 10 stories and then pick them up and yeah works yeah too bad right so that would be I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:33",
                    "end": "00:59:00",
                    "text": " the second level of success. So these are what I call broad successes. I think the third would be contextual and hyperfocal level of success. And what I mean by that is success which reaches the last mile, right? So you can put things in context. And when that last mile is reached, then I think we have gotten there. Now for the failure, I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:00",
                    "end": "00:59:28",
                    "text": " Data is more siloed than we would like it to be. And if at all we fall short in reaching any of these levels, I think it is going to be the silos, be it at governance level or any other level, that is going to be a showstopper. And the second is talent. Because this is not a labor problem. You cannot throw people at it. So as an academic, one of the core duties and responsibilities for me is skilling."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:28",
                    "end": "00:59:44",
                    "text": " teaching as a very important part of what I do and so this cannot be done like any other teaching right like you are looking for a very small set of people who need to be able to do this right so this I think are some things to think about yeah"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:44",
                    "end": "01:00:13",
                    "text": " All right, great. So we are so completely out of time. I'm Ashish Agarwal, Head of Public Policy at NASSCOM. As NASSCOM, it has been a pleasure to be hosting this event. And thank you so much for bringing so much insight and value to this. I think it's the combination of you and each of you highlighting. And so thank you, Ravi. Thank you, Ritu. Thank you, Rushali. Thank you, Nicole. And thank you, Sharad, for this very insightful panel. And all the very best to you from all of us at NASSCOM."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:13",
                    "end": "01:00:20",
                    "text": " Thank you audience for your patience and I'm sure there will be time for questions after we are done with this. Thank you."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6994274a25c9f4e0d7e70232",
    "projectId": "444753",
    "projectName": "A Billion Voices One AI How Language Tech Transforms Nations AI Summit India 2026",
    "projectDescription": "A Billion Voices One AI How Language Tech Transforms Nations AI Summit India 2026",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T08:31:05.752Z",
    "projectUpdatedDate": "2026-02-17T08:31:05.752Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-17T08:31:05.752Z",
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "Content",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "Building Language AI at Scale _ Voice AI _ Global Collaboration _ India AI Impact Summit 2026(MP3_160K).mp3",
            "col3": " The last mile connectivity is concerned. That is a huge concern. And there is still thought processes that are going on. And I don't think there is a ready-made solution as of now. I must admit that. But you have raised a very important point. And for Asha's, we are trying to give them tablets. And we are holding hackathons. And we are holding competitions amongst youngsters, startups, who are trying to...  create models which will suit specifically, say, ASHA. Because these are very, very specific needs. So after the need assessment has been carried out, we are throwing open a competition. And we are asking the public and the youngsters, startups, to come with solutions. And then we will look at it. But not any one-size-fits-all. Thank you. Thanks a lot. So as an conclusion.  Professor, you said at the beginning you really wanted to hear the panelist. I'd like to bring back the panel. Thank you very much.  We are talking about public health experience.  That the word troubling is very important. That we need to be looking at things that include things like old development. And another point is being mentioned that development is not easy. I know that to go up and down might be a hard way, but we need to go up, up and down. And then we have to see that how it is going to work.  Do you know any?  Thank you.  Thank you.  There is no use.  of reinventing the wheel of speed and energy and existing the already existing technology. So that is why I'm actually happy.  They start using it and they work on it. So you know, we are working as we started it was about all about transcending the language barrier but as we move forward it is  now situation where we are bridging the digital divide bridging the literacy divide because we are using voice as a medium you know people are using for increasing productivity again because voice is there and then obviously the language barrier anyway is gone today we operate this for means we touch every citizen of the country through our partnerships like UIDI  We touch every village and panchayat through our partnership with Ministry of Panchayat-e-Raj and we do about 15 to 18 million inferences a day which is that means 15 to 18 million times our system is requested for a translation there and almost 100 million a week or 400 million a month.  And we are growing at a rate of approximately 15 to 20% month on month. That means there was a latent need for this kind of a system. And one of the reasons why we are able to pick up or work with most of the customers is our commitment to look at continuous improvement and look at how do we actually build this entire system together with the  stakeholders and the community, we feel that every end user is a contributor as well, he can. Things that's happening in India, so thank you Amitabh. I think Manmeet, you needed a second? Okay, so like they say, break your butt. We'll get to the next question that I had. So I think Amitabh, you talked a little bit about the voice AI technology being critical. So I'm going to go to Santosh a little bit in terms of  What do you see in terms of speech technology and its reach for truly population scale? What is it that you're seeing progress-wise? What are you seeing challenges-wise? Yeah. I think, first of all, so far, many people used to think that voice is just a convenience. We have built these great technology systems which are available to people, a lot of apps.  government, private apps have been built. They used to support English, and then we added languages. But still, we thought that adding a few more languages to apps may be sufficient. But how many people actually know? I mean, India still has 300 million people using feature phones. There are a much bigger number of the people who are digitally, like Amitabh was also saying, digitally not literally. They can forward things on WhatsApp. They can watch videos on YouTube.  But they can't really interact with these apps, right? Definitely ask, and I'm a Sindhi, if you ask me, do you know how to type in Sindhi in mobile phone? No, I don't. It's not, I don't even know that feature, how easy it is or not. So, and that is where the voice and language becomes very, very crucial, right? I mean, voice is probably the oldest technology that we have. I mean, voice is how people connect to each other in their own language, in their native language.  So it's for technology to be truly inclusive, it shouldn't be harder for people to adopt it. By making people learn new thing in typing or using smartphones and understand how these buttons operate, we are actually not necessarily including everyone. It's not just inconveniences and exclusion of people.  And same goes for AI also. I mean, globally, there are about 7,000 spoken languages. In India alone, we have about 100 plus spoken languages. But if you ask me how many of these languages are there on the AI map, I'm sure they are in hundreds, not even in thousands. So there are a lot of languages which are not in AI map. In last couple of years, I think, so obviously, we need to accelerate, not just in India and global south, I think.  in African nations everywhere to bring these languages onto AI map, which means the LLMs that we are building shouldn't just understand few languages. They should be able to understand any languages. There needs to be urgency of building these voice models which can understand not necessarily 22 official languages of India, but every other language of India. And India is a unique place, just like I think South, Global South also.  For every language, we have so many dialects. The same Hindi that we speak in Delhi is very different than in Lucknow, than in Bihar. So how can AI understand all these languages? Like Amitabh said, there's a lot of work already happening on language side, but I think there's an urgency and criticality to accelerate that work on the language side. And good news is that technology is evolving faster than we anticipated.  And it's moving very, very fast. Yeah, absolutely. Thank you, Santosh. Thank you. I'm going to go to Vijay a little bit in terms of... So obviously, language technology is supposed to bridge the divide amongst the haves and the have-nots. Do you see this happening in particular sectors, or is it something that you're seeing, you know, broadly across multiple sectors? Yeah. The Gates Foundation plays the role of a convener. So we've been looking at...  making lessons learned in India applied Africa the same going the other way. Some of the work we've done so far, yeah, some of the work we've been doing so far is in training data. It's more fundamental. It's not sector-specific work. It's been taking African languages and creating very high-quality training data so you can bring the word error rates down under 10%. Some of the languages we...  tested with 200 hours of high quality data, you could get it to that error rate. So our work has been primarily more fundamental, not sector-specific. That said, one of the initiatives that we supported, called your REMBO in Rwanda, is to make voice-based access to government services. There's 240 services that would normally be available to someone that's literate.  And by building voice-based capabilities, now it's available to a lot more. And the costs come down substantially. That's one, which is about government service delivery. The other one is, I would say, in health. It's another project in Rwanda called Horizon 1000. It's just getting started. We've kicked off some pilots. And that one's about primary health care. And that has a voice and chatbot component, too. So as Santosh was saying, it's no longer  we need to do voice. It's not a nice to have. It's a core offering. And this is how we deliver inclusive solutions. Thanks, Vijay. Thank you. Chinay, I'm going to come to you because obviously, like I mentioned earlier in your introduction, there's a lot of focus on research on your side as well. So I wanted to ask you, how do you see the translation from research to production systems? And what is your experience with that?  Thank you so much. So firstly, I'll start by saying it's good to be back in Delhi. I actually studied at JNU University a couple of years ago, so it feels like a full circle moment to come talk about languages here. So what we've seen, so as Masakane, what we exist as is a community of practitioners interested in research, and trying to then see how that research that we do, but not just...  research is where we started from. Now we are investing into the ecosystem looking at the creation of datasets and the governance around that data because that's quite critical as well as enabling the ecosystem through understanding research for innovation and benchmarking and then as well innovation itself. So particularly looking at from research to practical use one of the key things that we actually just closed as a request for proposals was trying to understand how we  benchmark this work and then how it can be applied. Because the biggest issue is that some of these innovations, as mentioned, you can't have a system that only represents the national 22 or national languages. People speak a whole lot more. So which means that when some of these models or technologies are introduced into different contexts, they don't work.  and particularly for global majority countries because the technologies we are often playing catch up or left behind in the innovation. So this is part of our purpose then of saying how do we evaluate the models that are out there, how do we create a systematic standard that everyone can then assist to see this is what works within a context of diversity and cultural nuance and difference. Because that I think is oftentimes what is the beautiful part  of global majority context was that, like what was being mentioned, Swahili spoken in Nairobi is not the same as Swahili spoken in Tanzania because they actually grow up learning to speak that. So that is the essence of this work when we think about from research to practice. And then also kind of like as Masakani, one of our core.  pieces of work is actually to connect with the tech entrepreneurs on the continent. A lot of innovation on the continent is actually driven by tech entrepreneurs in addition to the standard big tech companies. So the question really becomes, how do you ensure that what you're creating in terms of, is it high quality enough to be able to go to market? And also thinking about what is the business case that we often have to make with entrepreneurs. I was in a conversation with an entrepreneur who was building  a voice solution for lawyers, and they were like, I don't see the value of building for African languages. And this was someone from, I think they were from Kinga. So we were actually then having a conversation with them to actually show that the value of work, as particularly that it's already been done in India, for example, so citing Bashni, to actually then say there is a value in actually being proactive in ensuring that African languages are represented in the product you take to market.  the importance of connecting to the researchers so that the work is not done in silent efforts. Absolutely. That's an excellent perspective. Thank you. Amita, I'm going to come back to you with a question. There are two questions, actually, I have, because Bhashini has been doing a lot of work in this area.  Are you seeing particular sectors where the adoption is more? Is the impact more in certain sectors or domains like agriculture or education? Where are you seeing the proliferation of this technology? And the second thing is, this is an important question because of the experience you have, what remains stubbornly difficult? Okay, so what remains difficult is a...  very long list of things actually and although we have progressed a long way with obviously Santosh we are working on Mahavizkar which is actually giving you know voice advice to the farmers in Marathi the complete journey is voice enabled but then obviously the challenges remain the challenges remain in multiple areas  One is that you still do not have the digital dictionary of the country. There are a lot of words which are not known in the native language to the digital systems. It includes even the names of places because when we worked with Ministry of Panchayat, we got 270,000 panchayats onto the digital systems. But when we talked to Survey of India, they also do not have, they have, they say that there is about 16 to 18 lakh  places which they want to digitize into the system. Those are the number of places which are there, you know. Either it is a Kaspa or a Suba or whatever. It is basically not there in digital systems and that is something which we are working upon. So that's one area. Emotions is another area. And emotional rendering of voice is also changes the meaning. And there are some languages like Miso or  those are more or less phonetic languages. Same words mean multiple things spoken differently. So that's another area of challenge. So there are multiple issues which are there and those are to be sorted out over a period of time as technology progresses. But what we have, you know, we are quite encouraged to see that option. That's the brighter side of it because it's delivering value.  Right and social sector because of its very nature that it is You know the service delivery and governance can actually happen at the last mile is the biggest uptake for that So sometimes the systems are pushed that means it is from the governance and service delivery angle Sometimes the systems are actually adopted because it's giving certain action ables to the people as far as information is concerned to help the person so you know  Social sector happens to be agriculture, education, health. These are the sectors which actually need that because of their very nature. And also, you know, because there is information available which can be delivered. But obviously, you know, taking it to information delivery or service delivery or governance to actionable impact will take some time because we still need to be  getting very familiar with one of the very important things in voices about dialects. You know, as she said that every language can be same, but it can be spoken in different ways by different people. And sometimes it also differentiates between one person to another person. So that's another challenge which we have.  whatever results which we are seeing is exciting and it should prompt the researchers and all of us to do more work in that area and ensure that, you know, we are living in better tomorrow because when we had recently, you know, were in some of the, you know, district areas and AP, for example, in Copern or Arco Valley and we, you know, we were running these systems because people were speaking in English and  Only two people understand English in the entire district, the collector and his assistant. The rest of the whole people understand Telugu and the only way to survive there is to have a language bridge and fortunately the person whom we had sent, he was a Malayali, so he didn't have any of the background, so he was using the app to actually communicate. So it's important when we look at  When we, of course, sit at a metro, we find that it is a cosmopolitan. We switch over languages as soon as we see there is an objection or there is a non-understanding from other side. But 100 kilometers from here, the world is different. And there is a lot of divide. Absolutely. Thank you. Chinna, I'm going to come to you for a follow-up question. So Amit, I've talked about some of the challenges from an India standpoint. Africa has got infrastructure challenges too.  So can you kind of elaborate on what is it that you see from that standpoint? Thanks so much. So thinking about the challenges across the continent.  One of the key things will always come from an infrastructure of resourcing around compute. So do people have enough compute? And if they do have it, how well is it used? But in addition to that, there are legacy issues that are consistent when it comes to connectivity on the content anyway. So I think a point was earlier mentioned about people having access to feature phones. So there is still that reality that people are still connected to, you know, like.  feature phones or at least lower end smartphones because there's an affordability aspect to it. Which device can they afford? And we specifically talk about phones because people's first access to a digital device is a mobile phone. Not a computer, not a laptop, but the mobile phone that they have. And then secondly, you're also looking at issues around electricity.  So actual access to the power to power the models that you're going to be building, the technology you have. So that's a key issue in terms of then needing that infrastructural intervention beyond if you build the greatest data set and the greatest model. If there's no power to run it, there's no power to run it. And then also just really thinking about we've been having conversations about capacity. Can people actually?  Do we have the skill set on the continent? I think we do have it, but we're always also faced with the reality of brain drain and people traveling and leaving the continent. And then I think lastly, in terms of thinking about an infrastructure challenge, and again, I will take this to being about to market, is in the adoption of the services that are created in people's languages. And this is determined by, I think at a national level, if a country already has that type of policy in place.  So coming from South Africa, the government recognizes 11 arms.  11 languages across the country and they have launched digital services through digital public infrastructure work where they do recognize those languages. And the question becomes to what extent then are people creating tools and resources not just for information giving but also knowledge exchange. Because a lot of the times then that is the question of you can be told what to do but can you actually interact with the platform in your own chosen language. So that again is one of the  of connecting the infrastructure issues to the end use and to the market. So those are some of the challenges we face. But I think most critical, which I had almost forgotten, is also high quality data.  So we have seen earlier investments that were done, Gergates Foundation, GIZ, where then we have seen languages such as Kisohele and Kinyarwanda and Luganda move from being low resource languages to at least well-resourced languages. So there's quite a number of data sets you can find there. But there are over 2,000 languages on the continent.  And so then that's just like a drop in the ocean of what is there. So then there is that need of actually creating high quality data sets that allow for people to actually build on top of them to have that foundational basis. So those are some of the core issues that we face across the continent and they always have to be placed in context, dependent on what is the actual national drive for these interventions to take place.  Absolutely. Thank you, Chennai. Thank you. Now, Vijay, I'm going to come back to you. I think from a Gates Foundation perspective, obviously, you're looking globally. You're looking in Africa. You're looking at Southeast Asia. You're looking at India. You're looking at different parts of the world. How do you prioritize what you actually want to focus on? That's a hard question. I mean, there's 7,000 languages.  And I think over 80% are spoken by less than 100,000 people. So it is hard to go after all the languages. I think there's a breadth-first approach and a depth-first approach, depending on what you're pursuing. On the broader side is multilingual models that can address multiple languages in one shot. Those are the kinds of collecting training data. That's one approach where you address multiple languages in one shot.  On the depth first, it's more sector-specific use cases, like the integrated government services in Rwanda or the agriculture services that we work on in India, where you go deep and address them. So we are, as a convener, we're more use case driven. We work with what the needs are, and it's not a language-first approach, but more a use case-first approach. And the language comes if we're working in Nigeria.  you know, we would choose the languages that are most spoken in that area. If it's like an agriculture advisory for a government service, he would choose the most spoken language and target that person. Fair enough, fair enough. Santosh, I'm going to come back to you with a question.  for low resource languages in India. I know that you're working on some low resource languages currently, and you're trying to kind of create something that will happen for remote areas, remote languages, et cetera. Can you talk a little bit about that? Sure. So yeah, I think like everyone in the panel is saying, getting all the languages included is so critical for AI systems to work and to be inclusive for the last mile population.  And we recently, I mean, actually did a project along with the government of Maharashtra and Agri. It's called Mahavistar AI and one of the, I think, showcase project in the AI. The Mahavistar app already has been downloaded by about 25,000,000 downloads have happened. Some 15,000,000 AI queries have been done, some 5,000,000 AI interactions. And this is entirely built in Marathi for Marathi.  been a journey of last five, six months. Obviously, there are a lot of partners who came together to build it. But then we got a call from one of the district collectors. I mean, in fact, government marshal, Nandur Bar, is saying, in my district, there are three languages. Tribal languages are spoken. And Bheeli is among one of them. And can Mahavistar Agri app support Bheeli? He said, OK. It's not even one of the official languages. It's spoken by less than, I mean.  1 crore people across Gujarat, Maharashtra, Rajasthan. And just whatever support is needed, I'm going to bring it onto the table. But I want this language to be part of the Maharashtra Agri. And that's when the journey started. And there were partners like Aarya, I think Bhashni Amitabh's team, AI for Bharat. Everyone came together with a mission that can we bring the Pili language onto an AI map.  And it's been a journey of just last six to eight weeks, not months. With the support of the local administration, we got a lot of local community people who came together to what we call donate their data or their voices to be part of the community. It's not a trivial task, by the way, to bring a lowly resource language onto the AI map. You need to have a lot of data collection exercise.  And a data collection has to be in a certain way. You have to have a spoken speech, spontaneous speech. You need to have a translation. Bheeli also doesn't have an independent script, so it relies on the Devanagari script or the other scripts. So you needed a translation from Bheeli to Marathi or Bheeli to English. You needed somebody who could speak Bheeli, somebody who could read Bheeli, and then when translated, verified it. It's a hard job.  Right? And you require, typically it takes a year or longer for bringing a language to an AI map. But with the support of the local administration and the teams like Karya and Bhashini, we were able to do this job in less than six weeks. And what we are also saying is it's not just about one language bringing onto an AI map. Can we create a template that can be used?  for bringing all the other low resource language, not just for India, but even for Global South. I think we're talking about 11 languages, official languages in South Africa alone. So can this be a template to collect data, annotate data, which is the hardest part, by the way, build models, and then integrate into applications? It's not just if you build a model, the job doesn't end there.  you need to also be able to apply into a real use case. And what we are trying to do is to bring it into a Mahavistar Agri app so that people in Nandurbar district or in Gujarat, they could interact with this Agri bot and get their local farming advisor. Absolutely. Thank you, Santosh. Amita, one question for you. There's a lot of talk about data security and data privacy. What is your perspective on  how do we adhere to those norms that are nationally mandated and how do you see this going forward? Yeah this you know currently to of course you know we have a DPDP Act in India and we are supposed to abide by that and we are abiding by that but this will become a big challenge as we go by because as the systems will use voice and  voice being attached to a person and becoming a personal information, there would be a way we should be looking at to handle this particular situation. Now currently, of course, when we are looking at data contribution on voice, we are looking for a consent from the people who can actually say that this data can be used for training purposes. But when the users actually come to the system,  and they start using it and specifically for the system which are quote unquote free of charge, then there is obvious risk that the data is actually going into the system and is being used for the purpose which is not it's supposed to be. It becomes very also more difficult when you're looking at kind of health data, right? Health data is something which you are  giving means normal health data as well as voice when you are putting your voice into it means you are consenting to give the data to the doctor for a purpose which is treatment but should it be used for AI model is not for which you are giving the consent so now there's a second dimension the first dimension was that you're giving the data that is a consent but the purpose has also got to be defined and the purpose is  to train the AI model, perhaps a consent has to be taken for that also. Then when we are looking at the, perhaps the granularity will again come into picture because you want to perhaps give the data to do certain activities, but it should be a portion of data, not the complete data. So I think there would be complexities which will start coming in over a period of time.  And we would have to basically just as AI, you know, when the technology progresses, you don't have, you may perhaps will not be able to think of all the things. You know, there are known unknowns and there are unknown unknowns which will keep unfolding over a period of time. And we would have to handle them as they come by. But the thing which we have currently adopted is to ensure that we start  We keep ourselves on open source domain. So it helps in two things. It is we are socially auditable, so that's one good part. So somebody wants to raise an objection that can be taken up. Second is it also ensures that we compete with ourselves, because to stay relevant, you know, you already have put out what you know, so you have to stay relevant, so you have to do better than what you're doing. Otherwise, shelf life of any AI system is about six months or so.  Somewhere we have to find mechanisms which is easily monitorable or whatever. Otherwise, we are going to be in a situation where we are continuously looking, picking up nitty gritties and trying to solve them rather than looking at the technology development which perhaps would be far more has a very fast pace today. Absolutely, absolutely. Vijay, I'm going to come to you because we talked about this earlier. I think you had this whole  debate about whether we use the global models that AI companies offer versus sovereign models created in countries. Your thoughts on what is your viewpoint on that? I think I would point back to the actual use case. How quickly do you want to get it to market? What are the economics of it? And sovereignty, how important is that? So that would determine  whether you go build the model yourself, or you use one of the off-the-shelf ones. I think longer term building your own is the better way to go. May I piggyback on what Amitabhji said? One of the things in the health sector is that voice is always considered identifiable data, PHI. So federated approaches have been a solution that has been widely used.  where the model travels to the data and you learn from it, but the actual data never is made available to the users of the data. So I think there's some precedent in the health sector that can be broadly used for a waste-based solution. Fair enough. Chinay, I'm going to come back to you with one question.  So we talked about the differences between what India is doing, what Africa is doing, et cetera. How do we create a South-South dialogue to make this faster? How do we accelerate the process of benefiting wholesome across the sector? Yeah, absolutely. Thank you for that. So one of the core things we always talk about is actually breaking down the silos and the fragmentation.  So how we do this is, as we gather in this room, so at least, you know, people are connecting, people are learning about Masakane. Masakane actually means to build together, loosely translated from Esisulu, so we live that way. So we are often in conversations, actually, with partners from speaking to course, speaking to, I think, career, and then also facilitated by our funding partners.  But it's thinking about South-South collaborations from those who have created in Southeast Asia, from those who have created in India, from those who have created in Latin America, and finding the similarities in the issues that are there. So we often try to engage at...  firstly of that research level, so co-creation of knowledge, but also engaging in spaces of actually creating the solutions. So working in spaces where they're collaborators, the tech entrepreneurs, we bring them together. Quite recently, we supported a conference called African Machine Learning Days.  comes out of the students from Tunisia, based in Switzerland. So it actually did bring together, and it was held in South Africa. So there's already three different countries coming together. But in those spaces, what they managed to do is also bring in multi-stakeholders, even from the private sector, big tech, like when I say big tech, I mean Google, into the conversation.  What that actually created was an avenue for young people to have, because Africa is quite a youthful continent. So whenever I talk about collaboration, I'm starting from, how do I ensure that young people actually do have access to these spaces? And the critical questions they were asking was, if I build my model, can I actually then be able to make money out of it?  And so then to have someone in the room who can actually give an example of how they can build up that scaling for purposeful impact is quite important. So I think the South-South collaboration, and this is our call from Asakane, comes in thinking about what is the practicality of impact and how do we ensure that we do not repeat...  The same mistakes or the same areas that other people faced before, of course, they have to be nuanced into context. We're not all the same. But there are lessons that we can learn, particularly thinking about privacy and data protection. That's a significant issue everywhere. So as Vijay mentioned, we can look to health. That means that we don't have to recreate the will. We can actually look at what others have done. And that's only possible if people are also connecting with us as Masakane and our role in enabling the ecosystem.  Absolutely. And this summit is a good way to actually start that dialogue. Thanks. I've got very little time, so I'm going to take two questions. Yeah, go ahead. The lady first. Yeah.  Yeah, so anybody in particular? Maybe I'll take a stab at it. So I think one of the things that we found was, for example, Agri terms in Maharashtra were not some things that were the same across. So we actually have started using glossaries up front in terms of saying XYZ mean the same thing, kind of the framework. So that's one way to do this, to essentially translate it.  upfront before you actually feed it to a model. But I'll let any other person in the panel actually. So I think, I mean, this is a real problem and that having the right glossary, vertical glossary as well as the language glossary is not like Amitabh, he also said that this doesn't always exist, right, in terms of, and there's no easy answer, but I think the best possible way is to work with the local administration.  the department, if you're working with health care, then the local health care department would be the best place to create a glossary if it doesn't exist, or validate a glossary to say, yes, it makes sense if it is spoken like that. And then for that, also, you need a lot of linguistic experts, which, again, the department can bring in. They have access to these school teachers, professors, the linguistic experts.  The best way that we have found out working for us is to work with the local state governments and their corresponding departments. So one of the very high value, you know, challenge which is there, which has got a very high intrinsic value is land records. You know, it's an example where every state has its own terminology for the same thing. And there's no other way but to, you know, ensure that  these departments start preparing the glossaries. And fortunately for us that we have been able to create that momentum, the land records glossary has been prepared and it would soon get digitized so that you are in a position to actually, you know, be interoperable as far as land records are concerned. Same is the case with the food grains and other things. There are terminologies which are mismatching.  Yes, you know, there is an active work which is going on in the government and the departments to build those glossaries which will actually help us take out these anomalies. And if required, you know, the experts can sit and say that what should be the case, you know. We are doing that for...  Auditor's office, we're doing it for multiple places as of now. Of course, we are not doing it because we are not experts. We are experts in technology. But we have currently, I think that every state would have perhaps a million words which has to be digitized, right? And every line ministry would have approximately a million words which needs to be digitized. So it's a massive work actually. I'm gonna give this gentleman a chance because I think he's been very persistent, sir.  So one last question. We got your question, so let them answer. Let's chat after. It's probably better. I mean, you're asking about how startups compete with big tech. It's a classic problem that all startups face. How do you differentiate yourself? So that's what you need to focus on. So find your target audience. And can you build a killer solution that's better than the big tech?  Absolutely don't reinvent the wheel, but if you have something that is unique that you can add as a startup, go for it. So you know, going by what you talked about, accuracy per se. So I was in the other session where I was saying that we experimented with speech-to-speech translation in front of our minister as well as the Gujarat government officials.  The officials felt that it was 80% right, while as the minister said it was, you know, beyond his imagination. It was so good. Now the point which is there is not about whether it is accurate or not. Accuracy is all about whether the people are able to accept it or not. But the pertinent question which you have raised, which is basically, what do I do? It is to go back to the basics. And as we just said, you know,",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771317056563.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:28",
                    "text": " The last mile connectivity is concerned. That is a huge concern. And there is still thought processes that are going on. And I don't think there is a ready-made solution as of now. I must admit that. But you have raised a very important point. And for Asha's, we are trying to give them tablets. And we are holding hackathons. And we are holding competitions amongst youngsters, startups, who are trying to..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:28",
                    "end": "00:00:56",
                    "text": " create models which will suit specifically, say, ASHA. Because these are very, very specific needs. So after the need assessment has been carried out, we are throwing open a competition. And we are asking the public and the youngsters, startups, to come with solutions. And then we will look at it. But not any one-size-fits-all. Thank you. Thanks a lot. So as an conclusion."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:56",
                    "end": "00:01:28",
                    "text": " Professor, you said at the beginning you really wanted to hear the panelist. I'd like to bring back the panel. Thank you very much."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:28",
                    "end": "00:01:58",
                    "text": " We are talking about public health experience."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:58",
                    "end": "00:02:27",
                    "text": " That the word troubling is very important. That we need to be looking at things that include things like old development. And another point is being mentioned that development is not easy. I know that to go up and down might be a hard way, but we need to go up, up and down. And then we have to see that how it is going to work."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:27",
                    "end": "00:02:56",
                    "text": " Do you know any?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:56",
                    "end": "00:03:23",
                    "text": " Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:23",
                    "end": "00:03:48",
                    "text": " Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:48",
                    "end": "00:04:17",
                    "text": " There is no use."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:17",
                    "end": "00:04:45",
                    "text": " of reinventing the wheel of speed and energy and existing the already existing technology. So that is why I'm actually happy."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:45",
                    "end": "00:16:37",
                    "text": " They start using it and they work on it. So you know, we are working as we started it was about all about transcending the language barrier but as we move forward it is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:37",
                    "end": "00:17:05",
                    "text": " now situation where we are bridging the digital divide bridging the literacy divide because we are using voice as a medium you know people are using for increasing productivity again because voice is there and then obviously the language barrier anyway is gone today we operate this for means we touch every citizen of the country through our partnerships like UIDI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:05",
                    "end": "00:17:33",
                    "text": " We touch every village and panchayat through our partnership with Ministry of Panchayat-e-Raj and we do about 15 to 18 million inferences a day which is that means 15 to 18 million times our system is requested for a translation there and almost 100 million a week or 400 million a month."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:33",
                    "end": "00:18:01",
                    "text": " And we are growing at a rate of approximately 15 to 20% month on month. That means there was a latent need for this kind of a system. And one of the reasons why we are able to pick up or work with most of the customers is our commitment to look at continuous improvement and look at how do we actually build this entire system together with the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:01",
                    "end": "00:19:25",
                    "text": " stakeholders and the community, we feel that every end user is a contributor as well, he can. Things that's happening in India, so thank you Amitabh. I think Manmeet, you needed a second? Okay, so like they say, break your butt. We'll get to the next question that I had. So I think Amitabh, you talked a little bit about the voice AI technology being critical. So I'm going to go to Santosh a little bit in terms of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:25",
                    "end": "00:19:53",
                    "text": " What do you see in terms of speech technology and its reach for truly population scale? What is it that you're seeing progress-wise? What are you seeing challenges-wise? Yeah. I think, first of all, so far, many people used to think that voice is just a convenience. We have built these great technology systems which are available to people, a lot of apps."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:53",
                    "end": "00:20:22",
                    "text": " government, private apps have been built. They used to support English, and then we added languages. But still, we thought that adding a few more languages to apps may be sufficient. But how many people actually know? I mean, India still has 300 million people using feature phones. There are a much bigger number of the people who are digitally, like Amitabh was also saying, digitally not literally. They can forward things on WhatsApp. They can watch videos on YouTube."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:22",
                    "end": "00:20:50",
                    "text": " But they can't really interact with these apps, right? Definitely ask, and I'm a Sindhi, if you ask me, do you know how to type in Sindhi in mobile phone? No, I don't. It's not, I don't even know that feature, how easy it is or not. So, and that is where the voice and language becomes very, very crucial, right? I mean, voice is probably the oldest technology that we have. I mean, voice is how people connect to each other in their own language, in their native language."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:50",
                    "end": "00:21:13",
                    "text": " So it's for technology to be truly inclusive, it shouldn't be harder for people to adopt it. By making people learn new thing in typing or using smartphones and understand how these buttons operate, we are actually not necessarily including everyone. It's not just inconveniences and exclusion of people."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:13",
                    "end": "00:21:43",
                    "text": " And same goes for AI also. I mean, globally, there are about 7,000 spoken languages. In India alone, we have about 100 plus spoken languages. But if you ask me how many of these languages are there on the AI map, I'm sure they are in hundreds, not even in thousands. So there are a lot of languages which are not in AI map. In last couple of years, I think, so obviously, we need to accelerate, not just in India and global south, I think."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:43",
                    "end": "00:22:12",
                    "text": " in African nations everywhere to bring these languages onto AI map, which means the LLMs that we are building shouldn't just understand few languages. They should be able to understand any languages. There needs to be urgency of building these voice models which can understand not necessarily 22 official languages of India, but every other language of India. And India is a unique place, just like I think South, Global South also."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:12",
                    "end": "00:22:40",
                    "text": " For every language, we have so many dialects. The same Hindi that we speak in Delhi is very different than in Lucknow, than in Bihar. So how can AI understand all these languages? Like Amitabh said, there's a lot of work already happening on language side, but I think there's an urgency and criticality to accelerate that work on the language side. And good news is that technology is evolving faster than we anticipated."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:40",
                    "end": "00:23:11",
                    "text": " And it's moving very, very fast. Yeah, absolutely. Thank you, Santosh. Thank you. I'm going to go to Vijay a little bit in terms of... So obviously, language technology is supposed to bridge the divide amongst the haves and the have-nots. Do you see this happening in particular sectors, or is it something that you're seeing, you know, broadly across multiple sectors? Yeah. The Gates Foundation plays the role of a convener. So we've been looking at..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:12",
                    "end": "00:23:42",
                    "text": " making lessons learned in India applied Africa the same going the other way. Some of the work we've done so far, yeah, some of the work we've been doing so far is in training data. It's more fundamental. It's not sector-specific work. It's been taking African languages and creating very high-quality training data so you can bring the word error rates down under 10%. Some of the languages we..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:42",
                    "end": "00:24:09",
                    "text": " tested with 200 hours of high quality data, you could get it to that error rate. So our work has been primarily more fundamental, not sector-specific. That said, one of the initiatives that we supported, called your REMBO in Rwanda, is to make voice-based access to government services. There's 240 services that would normally be available to someone that's literate."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:09",
                    "end": "00:24:39",
                    "text": " And by building voice-based capabilities, now it's available to a lot more. And the costs come down substantially. That's one, which is about government service delivery. The other one is, I would say, in health. It's another project in Rwanda called Horizon 1000. It's just getting started. We've kicked off some pilots. And that one's about primary health care. And that has a voice and chatbot component, too. So as Santosh was saying, it's no longer"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:39",
                    "end": "00:25:08",
                    "text": " we need to do voice. It's not a nice to have. It's a core offering. And this is how we deliver inclusive solutions. Thanks, Vijay. Thank you. Chinay, I'm going to come to you because obviously, like I mentioned earlier in your introduction, there's a lot of focus on research on your side as well. So I wanted to ask you, how do you see the translation from research to production systems? And what is your experience with that?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:09",
                    "end": "00:25:34",
                    "text": " Thank you so much. So firstly, I'll start by saying it's good to be back in Delhi. I actually studied at JNU University a couple of years ago, so it feels like a full circle moment to come talk about languages here. So what we've seen, so as Masakane, what we exist as is a community of practitioners interested in research, and trying to then see how that research that we do, but not just..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:34",
                    "end": "00:26:03",
                    "text": " research is where we started from. Now we are investing into the ecosystem looking at the creation of datasets and the governance around that data because that's quite critical as well as enabling the ecosystem through understanding research for innovation and benchmarking and then as well innovation itself. So particularly looking at from research to practical use one of the key things that we actually just closed as a request for proposals was trying to understand how we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:03",
                    "end": "00:26:23",
                    "text": " benchmark this work and then how it can be applied. Because the biggest issue is that some of these innovations, as mentioned, you can't have a system that only represents the national 22 or national languages. People speak a whole lot more. So which means that when some of these models or technologies are introduced into different contexts, they don't work."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:24",
                    "end": "00:26:53",
                    "text": " and particularly for global majority countries because the technologies we are often playing catch up or left behind in the innovation. So this is part of our purpose then of saying how do we evaluate the models that are out there, how do we create a systematic standard that everyone can then assist to see this is what works within a context of diversity and cultural nuance and difference. Because that I think is oftentimes what is the beautiful part"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:53",
                    "end": "00:27:16",
                    "text": " of global majority context was that, like what was being mentioned, Swahili spoken in Nairobi is not the same as Swahili spoken in Tanzania because they actually grow up learning to speak that. So that is the essence of this work when we think about from research to practice. And then also kind of like as Masakani, one of our core."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:16",
                    "end": "00:27:46",
                    "text": " pieces of work is actually to connect with the tech entrepreneurs on the continent. A lot of innovation on the continent is actually driven by tech entrepreneurs in addition to the standard big tech companies. So the question really becomes, how do you ensure that what you're creating in terms of, is it high quality enough to be able to go to market? And also thinking about what is the business case that we often have to make with entrepreneurs. I was in a conversation with an entrepreneur who was building"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:46",
                    "end": "00:28:15",
                    "text": " a voice solution for lawyers, and they were like, I don't see the value of building for African languages. And this was someone from, I think they were from Kinga. So we were actually then having a conversation with them to actually show that the value of work, as particularly that it's already been done in India, for example, so citing Bashni, to actually then say there is a value in actually being proactive in ensuring that African languages are represented in the product you take to market."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:15",
                    "end": "00:28:33",
                    "text": " the importance of connecting to the researchers so that the work is not done in silent efforts. Absolutely. That's an excellent perspective. Thank you. Amita, I'm going to come back to you with a question. There are two questions, actually, I have, because Bhashini has been doing a lot of work in this area."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:33",
                    "end": "00:28:58",
                    "text": " Are you seeing particular sectors where the adoption is more? Is the impact more in certain sectors or domains like agriculture or education? Where are you seeing the proliferation of this technology? And the second thing is, this is an important question because of the experience you have, what remains stubbornly difficult? Okay, so what remains difficult is a..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:58",
                    "end": "00:29:27",
                    "text": " very long list of things actually and although we have progressed a long way with obviously Santosh we are working on Mahavizkar which is actually giving you know voice advice to the farmers in Marathi the complete journey is voice enabled but then obviously the challenges remain the challenges remain in multiple areas"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:27",
                    "end": "00:29:57",
                    "text": " One is that you still do not have the digital dictionary of the country. There are a lot of words which are not known in the native language to the digital systems. It includes even the names of places because when we worked with Ministry of Panchayat, we got 270,000 panchayats onto the digital systems. But when we talked to Survey of India, they also do not have, they have, they say that there is about 16 to 18 lakh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:57",
                    "end": "00:30:25",
                    "text": " places which they want to digitize into the system. Those are the number of places which are there, you know. Either it is a Kaspa or a Suba or whatever. It is basically not there in digital systems and that is something which we are working upon. So that's one area. Emotions is another area. And emotional rendering of voice is also changes the meaning. And there are some languages like Miso or"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:25",
                    "end": "00:30:54",
                    "text": " those are more or less phonetic languages. Same words mean multiple things spoken differently. So that's another area of challenge. So there are multiple issues which are there and those are to be sorted out over a period of time as technology progresses. But what we have, you know, we are quite encouraged to see that option. That's the brighter side of it because it's delivering value."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:54",
                    "end": "00:31:24",
                    "text": " Right and social sector because of its very nature that it is You know the service delivery and governance can actually happen at the last mile is the biggest uptake for that So sometimes the systems are pushed that means it is from the governance and service delivery angle Sometimes the systems are actually adopted because it's giving certain action ables to the people as far as information is concerned to help the person so you know"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:24",
                    "end": "00:31:54",
                    "text": " Social sector happens to be agriculture, education, health. These are the sectors which actually need that because of their very nature. And also, you know, because there is information available which can be delivered. But obviously, you know, taking it to information delivery or service delivery or governance to actionable impact will take some time because we still need to be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:54",
                    "end": "00:32:16",
                    "text": " getting very familiar with one of the very important things in voices about dialects. You know, as she said that every language can be same, but it can be spoken in different ways by different people. And sometimes it also differentiates between one person to another person. So that's another challenge which we have."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:16",
                    "end": "00:32:45",
                    "text": " whatever results which we are seeing is exciting and it should prompt the researchers and all of us to do more work in that area and ensure that, you know, we are living in better tomorrow because when we had recently, you know, were in some of the, you know, district areas and AP, for example, in Copern or Arco Valley and we, you know, we were running these systems because people were speaking in English and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:45",
                    "end": "00:33:13",
                    "text": " Only two people understand English in the entire district, the collector and his assistant. The rest of the whole people understand Telugu and the only way to survive there is to have a language bridge and fortunately the person whom we had sent, he was a Malayali, so he didn't have any of the background, so he was using the app to actually communicate. So it's important when we look at"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:13",
                    "end": "00:33:39",
                    "text": " When we, of course, sit at a metro, we find that it is a cosmopolitan. We switch over languages as soon as we see there is an objection or there is a non-understanding from other side. But 100 kilometers from here, the world is different. And there is a lot of divide. Absolutely. Thank you. Chinna, I'm going to come to you for a follow-up question. So Amit, I've talked about some of the challenges from an India standpoint. Africa has got infrastructure challenges too."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:39",
                    "end": "00:33:52",
                    "text": " So can you kind of elaborate on what is it that you see from that standpoint? Thanks so much. So thinking about the challenges across the continent."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:52",
                    "end": "00:34:18",
                    "text": " One of the key things will always come from an infrastructure of resourcing around compute. So do people have enough compute? And if they do have it, how well is it used? But in addition to that, there are legacy issues that are consistent when it comes to connectivity on the content anyway. So I think a point was earlier mentioned about people having access to feature phones. So there is still that reality that people are still connected to, you know, like."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:18",
                    "end": "00:34:39",
                    "text": " feature phones or at least lower end smartphones because there's an affordability aspect to it. Which device can they afford? And we specifically talk about phones because people's first access to a digital device is a mobile phone. Not a computer, not a laptop, but the mobile phone that they have. And then secondly, you're also looking at issues around electricity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:39",
                    "end": "00:35:04",
                    "text": " So actual access to the power to power the models that you're going to be building, the technology you have. So that's a key issue in terms of then needing that infrastructural intervention beyond if you build the greatest data set and the greatest model. If there's no power to run it, there's no power to run it. And then also just really thinking about we've been having conversations about capacity. Can people actually?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:04",
                    "end": "00:35:33",
                    "text": " Do we have the skill set on the continent? I think we do have it, but we're always also faced with the reality of brain drain and people traveling and leaving the continent. And then I think lastly, in terms of thinking about an infrastructure challenge, and again, I will take this to being about to market, is in the adoption of the services that are created in people's languages. And this is determined by, I think at a national level, if a country already has that type of policy in place."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:33",
                    "end": "00:35:38",
                    "text": " So coming from South Africa, the government recognizes 11 arms."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:38",
                    "end": "00:36:08",
                    "text": " 11 languages across the country and they have launched digital services through digital public infrastructure work where they do recognize those languages. And the question becomes to what extent then are people creating tools and resources not just for information giving but also knowledge exchange. Because a lot of the times then that is the question of you can be told what to do but can you actually interact with the platform in your own chosen language. So that again is one of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:08",
                    "end": "00:36:19",
                    "text": " of connecting the infrastructure issues to the end use and to the market. So those are some of the challenges we face. But I think most critical, which I had almost forgotten, is also high quality data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:19",
                    "end": "00:36:40",
                    "text": " So we have seen earlier investments that were done, Gergates Foundation, GIZ, where then we have seen languages such as Kisohele and Kinyarwanda and Luganda move from being low resource languages to at least well-resourced languages. So there's quite a number of data sets you can find there. But there are over 2,000 languages on the continent."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:40",
                    "end": "00:37:08",
                    "text": " And so then that's just like a drop in the ocean of what is there. So then there is that need of actually creating high quality data sets that allow for people to actually build on top of them to have that foundational basis. So those are some of the core issues that we face across the continent and they always have to be placed in context, dependent on what is the actual national drive for these interventions to take place."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:08",
                    "end": "00:37:36",
                    "text": " Absolutely. Thank you, Chennai. Thank you. Now, Vijay, I'm going to come back to you. I think from a Gates Foundation perspective, obviously, you're looking globally. You're looking in Africa. You're looking at Southeast Asia. You're looking at India. You're looking at different parts of the world. How do you prioritize what you actually want to focus on? That's a hard question. I mean, there's 7,000 languages."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:36",
                    "end": "00:38:04",
                    "text": " And I think over 80% are spoken by less than 100,000 people. So it is hard to go after all the languages. I think there's a breadth-first approach and a depth-first approach, depending on what you're pursuing. On the broader side is multilingual models that can address multiple languages in one shot. Those are the kinds of collecting training data. That's one approach where you address multiple languages in one shot."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:04",
                    "end": "00:38:33",
                    "text": " On the depth first, it's more sector-specific use cases, like the integrated government services in Rwanda or the agriculture services that we work on in India, where you go deep and address them. So we are, as a convener, we're more use case driven. We work with what the needs are, and it's not a language-first approach, but more a use case-first approach. And the language comes if we're working in Nigeria."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:33",
                    "end": "00:38:51",
                    "text": " you know, we would choose the languages that are most spoken in that area. If it's like an agriculture advisory for a government service, he would choose the most spoken language and target that person. Fair enough, fair enough. Santosh, I'm going to come back to you with a question."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:51",
                    "end": "00:39:21",
                    "text": " for low resource languages in India. I know that you're working on some low resource languages currently, and you're trying to kind of create something that will happen for remote areas, remote languages, et cetera. Can you talk a little bit about that? Sure. So yeah, I think like everyone in the panel is saying, getting all the languages included is so critical for AI systems to work and to be inclusive for the last mile population."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:21",
                    "end": "00:39:51",
                    "text": " And we recently, I mean, actually did a project along with the government of Maharashtra and Agri. It's called Mahavistar AI and one of the, I think, showcase project in the AI. The Mahavistar app already has been downloaded by about 25,000,000 downloads have happened. Some 15,000,000 AI queries have been done, some 5,000,000 AI interactions. And this is entirely built in Marathi for Marathi."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:51",
                    "end": "00:40:21",
                    "text": " been a journey of last five, six months. Obviously, there are a lot of partners who came together to build it. But then we got a call from one of the district collectors. I mean, in fact, government marshal, Nandur Bar, is saying, in my district, there are three languages. Tribal languages are spoken. And Bheeli is among one of them. And can Mahavistar Agri app support Bheeli? He said, OK. It's not even one of the official languages. It's spoken by less than, I mean."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:21",
                    "end": "00:40:48",
                    "text": " 1 crore people across Gujarat, Maharashtra, Rajasthan. And just whatever support is needed, I'm going to bring it onto the table. But I want this language to be part of the Maharashtra Agri. And that's when the journey started. And there were partners like Aarya, I think Bhashni Amitabh's team, AI for Bharat. Everyone came together with a mission that can we bring the Pili language onto an AI map."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:48",
                    "end": "00:41:12",
                    "text": " And it's been a journey of just last six to eight weeks, not months. With the support of the local administration, we got a lot of local community people who came together to what we call donate their data or their voices to be part of the community. It's not a trivial task, by the way, to bring a lowly resource language onto the AI map. You need to have a lot of data collection exercise."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:12",
                    "end": "00:41:40",
                    "text": " And a data collection has to be in a certain way. You have to have a spoken speech, spontaneous speech. You need to have a translation. Bheeli also doesn't have an independent script, so it relies on the Devanagari script or the other scripts. So you needed a translation from Bheeli to Marathi or Bheeli to English. You needed somebody who could speak Bheeli, somebody who could read Bheeli, and then when translated, verified it. It's a hard job."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:40",
                    "end": "00:42:07",
                    "text": " Right? And you require, typically it takes a year or longer for bringing a language to an AI map. But with the support of the local administration and the teams like Karya and Bhashini, we were able to do this job in less than six weeks. And what we are also saying is it's not just about one language bringing onto an AI map. Can we create a template that can be used?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:07",
                    "end": "00:42:32",
                    "text": " for bringing all the other low resource language, not just for India, but even for Global South. I think we're talking about 11 languages, official languages in South Africa alone. So can this be a template to collect data, annotate data, which is the hardest part, by the way, build models, and then integrate into applications? It's not just if you build a model, the job doesn't end there."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:32",
                    "end": "00:43:00",
                    "text": " you need to also be able to apply into a real use case. And what we are trying to do is to bring it into a Mahavistar Agri app so that people in Nandurbar district or in Gujarat, they could interact with this Agri bot and get their local farming advisor. Absolutely. Thank you, Santosh. Amita, one question for you. There's a lot of talk about data security and data privacy. What is your perspective on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:00",
                    "end": "00:43:29",
                    "text": " how do we adhere to those norms that are nationally mandated and how do you see this going forward? Yeah this you know currently to of course you know we have a DPDP Act in India and we are supposed to abide by that and we are abiding by that but this will become a big challenge as we go by because as the systems will use voice and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:29",
                    "end": "00:43:59",
                    "text": " voice being attached to a person and becoming a personal information, there would be a way we should be looking at to handle this particular situation. Now currently, of course, when we are looking at data contribution on voice, we are looking for a consent from the people who can actually say that this data can be used for training purposes. But when the users actually come to the system,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:59",
                    "end": "00:44:27",
                    "text": " and they start using it and specifically for the system which are quote unquote free of charge, then there is obvious risk that the data is actually going into the system and is being used for the purpose which is not it's supposed to be. It becomes very also more difficult when you're looking at kind of health data, right? Health data is something which you are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:27",
                    "end": "00:44:55",
                    "text": " giving means normal health data as well as voice when you are putting your voice into it means you are consenting to give the data to the doctor for a purpose which is treatment but should it be used for AI model is not for which you are giving the consent so now there's a second dimension the first dimension was that you're giving the data that is a consent but the purpose has also got to be defined and the purpose is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:55",
                    "end": "00:45:22",
                    "text": " to train the AI model, perhaps a consent has to be taken for that also. Then when we are looking at the, perhaps the granularity will again come into picture because you want to perhaps give the data to do certain activities, but it should be a portion of data, not the complete data. So I think there would be complexities which will start coming in over a period of time."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:22",
                    "end": "00:45:51",
                    "text": " And we would have to basically just as AI, you know, when the technology progresses, you don't have, you may perhaps will not be able to think of all the things. You know, there are known unknowns and there are unknown unknowns which will keep unfolding over a period of time. And we would have to handle them as they come by. But the thing which we have currently adopted is to ensure that we start"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:51",
                    "end": "00:46:20",
                    "text": " We keep ourselves on open source domain. So it helps in two things. It is we are socially auditable, so that's one good part. So somebody wants to raise an objection that can be taken up. Second is it also ensures that we compete with ourselves, because to stay relevant, you know, you already have put out what you know, so you have to stay relevant, so you have to do better than what you're doing. Otherwise, shelf life of any AI system is about six months or so."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:20",
                    "end": "00:46:49",
                    "text": " Somewhere we have to find mechanisms which is easily monitorable or whatever. Otherwise, we are going to be in a situation where we are continuously looking, picking up nitty gritties and trying to solve them rather than looking at the technology development which perhaps would be far more has a very fast pace today. Absolutely, absolutely. Vijay, I'm going to come to you because we talked about this earlier. I think you had this whole"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:49",
                    "end": "00:47:20",
                    "text": " debate about whether we use the global models that AI companies offer versus sovereign models created in countries. Your thoughts on what is your viewpoint on that? I think I would point back to the actual use case. How quickly do you want to get it to market? What are the economics of it? And sovereignty, how important is that? So that would determine"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:20",
                    "end": "00:47:47",
                    "text": " whether you go build the model yourself, or you use one of the off-the-shelf ones. I think longer term building your own is the better way to go. May I piggyback on what Amitabhji said? One of the things in the health sector is that voice is always considered identifiable data, PHI. So federated approaches have been a solution that has been widely used."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:47",
                    "end": "00:48:06",
                    "text": " where the model travels to the data and you learn from it, but the actual data never is made available to the users of the data. So I think there's some precedent in the health sector that can be broadly used for a waste-based solution. Fair enough. Chinay, I'm going to come back to you with one question."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:06",
                    "end": "00:48:30",
                    "text": " So we talked about the differences between what India is doing, what Africa is doing, et cetera. How do we create a South-South dialogue to make this faster? How do we accelerate the process of benefiting wholesome across the sector? Yeah, absolutely. Thank you for that. So one of the core things we always talk about is actually breaking down the silos and the fragmentation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:30",
                    "end": "00:48:56",
                    "text": " So how we do this is, as we gather in this room, so at least, you know, people are connecting, people are learning about Masakane. Masakane actually means to build together, loosely translated from Esisulu, so we live that way. So we are often in conversations, actually, with partners from speaking to course, speaking to, I think, career, and then also facilitated by our funding partners."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:56",
                    "end": "00:49:12",
                    "text": " But it's thinking about South-South collaborations from those who have created in Southeast Asia, from those who have created in India, from those who have created in Latin America, and finding the similarities in the issues that are there. So we often try to engage at..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:12",
                    "end": "00:49:33",
                    "text": " firstly of that research level, so co-creation of knowledge, but also engaging in spaces of actually creating the solutions. So working in spaces where they're collaborators, the tech entrepreneurs, we bring them together. Quite recently, we supported a conference called African Machine Learning Days."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:33",
                    "end": "00:49:56",
                    "text": " comes out of the students from Tunisia, based in Switzerland. So it actually did bring together, and it was held in South Africa. So there's already three different countries coming together. But in those spaces, what they managed to do is also bring in multi-stakeholders, even from the private sector, big tech, like when I say big tech, I mean Google, into the conversation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:56",
                    "end": "00:50:17",
                    "text": " What that actually created was an avenue for young people to have, because Africa is quite a youthful continent. So whenever I talk about collaboration, I'm starting from, how do I ensure that young people actually do have access to these spaces? And the critical questions they were asking was, if I build my model, can I actually then be able to make money out of it?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:17",
                    "end": "00:50:38",
                    "text": " And so then to have someone in the room who can actually give an example of how they can build up that scaling for purposeful impact is quite important. So I think the South-South collaboration, and this is our call from Asakane, comes in thinking about what is the practicality of impact and how do we ensure that we do not repeat..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:38",
                    "end": "00:51:07",
                    "text": " The same mistakes or the same areas that other people faced before, of course, they have to be nuanced into context. We're not all the same. But there are lessons that we can learn, particularly thinking about privacy and data protection. That's a significant issue everywhere. So as Vijay mentioned, we can look to health. That means that we don't have to recreate the will. We can actually look at what others have done. And that's only possible if people are also connecting with us as Masakane and our role in enabling the ecosystem."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:07",
                    "end": "00:52:32",
                    "text": " Absolutely. And this summit is a good way to actually start that dialogue. Thanks. I've got very little time, so I'm going to take two questions. Yeah, go ahead. The lady first. Yeah."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:33",
                    "end": "00:53:02",
                    "text": " Yeah, so anybody in particular? Maybe I'll take a stab at it. So I think one of the things that we found was, for example, Agri terms in Maharashtra were not some things that were the same across. So we actually have started using glossaries up front in terms of saying XYZ mean the same thing, kind of the framework. So that's one way to do this, to essentially translate it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:02",
                    "end": "00:53:29",
                    "text": " upfront before you actually feed it to a model. But I'll let any other person in the panel actually. So I think, I mean, this is a real problem and that having the right glossary, vertical glossary as well as the language glossary is not like Amitabh, he also said that this doesn't always exist, right, in terms of, and there's no easy answer, but I think the best possible way is to work with the local administration."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:29",
                    "end": "00:53:55",
                    "text": " the department, if you're working with health care, then the local health care department would be the best place to create a glossary if it doesn't exist, or validate a glossary to say, yes, it makes sense if it is spoken like that. And then for that, also, you need a lot of linguistic experts, which, again, the department can bring in. They have access to these school teachers, professors, the linguistic experts."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:55",
                    "end": "00:54:23",
                    "text": " The best way that we have found out working for us is to work with the local state governments and their corresponding departments. So one of the very high value, you know, challenge which is there, which has got a very high intrinsic value is land records. You know, it's an example where every state has its own terminology for the same thing. And there's no other way but to, you know, ensure that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:23",
                    "end": "00:54:51",
                    "text": " these departments start preparing the glossaries. And fortunately for us that we have been able to create that momentum, the land records glossary has been prepared and it would soon get digitized so that you are in a position to actually, you know, be interoperable as far as land records are concerned. Same is the case with the food grains and other things. There are terminologies which are mismatching."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:51",
                    "end": "00:55:10",
                    "text": " Yes, you know, there is an active work which is going on in the government and the departments to build those glossaries which will actually help us take out these anomalies. And if required, you know, the experts can sit and say that what should be the case, you know. We are doing that for..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:10",
                    "end": "00:55:39",
                    "text": " Auditor's office, we're doing it for multiple places as of now. Of course, we are not doing it because we are not experts. We are experts in technology. But we have currently, I think that every state would have perhaps a million words which has to be digitized, right? And every line ministry would have approximately a million words which needs to be digitized. So it's a massive work actually. I'm gonna give this gentleman a chance because I think he's been very persistent, sir."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:39",
                    "end": "00:57:17",
                    "text": " So one last question. We got your question, so let them answer. Let's chat after. It's probably better. I mean, you're asking about how startups compete with big tech. It's a classic problem that all startups face. How do you differentiate yourself? So that's what you need to focus on. So find your target audience. And can you build a killer solution that's better than the big tech?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:17",
                    "end": "00:57:50",
                    "text": " Absolutely don't reinvent the wheel, but if you have something that is unique that you can add as a startup, go for it. So you know, going by what you talked about, accuracy per se. So I was in the other session where I was saying that we experimented with speech-to-speech translation in front of our minister as well as the Gujarat government officials."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:50",
                    "end": "00:58:18",
                    "text": " The officials felt that it was 80% right, while as the minister said it was, you know, beyond his imagination. It was so good. Now the point which is there is not about whether it is accurate or not. Accuracy is all about whether the people are able to accept it or not. But the pertinent question which you have raised, which is basically, what do I do? It is to go back to the basics. And as we just said, you know,"
                }
            ]
        }
    ],
    "projectDeleted": false,
    "codeFrameColumn": [
        {
            "col1": "col1",
            "col2": "col3",
            "researchObjective": "",
            "filterColumn": [],
            "netSetting": true,
            "subNetSetting": true,
            "categorySetting": true,
            "sentimentSetting": true,
            "thinkingBudgetSetting": 2000,
            "questionType": "OPENEND",
            "otherBrand": false,
            "existingCodeFrame": [],
            "isList": false,
            "codeFrameWithDataList": [],
            "codeFrameList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 8715,
                        "CandidatesTokenCount": 3124,
                        "TotalTokenCount": 13136,
                        "thoughtsTokenCount": 1297,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Net": "Voice and Language AI Technology",
                                "Subnets": [
                                    {
                                        "Subnet": "Role of Voice AI",
                                        "Categories": [
                                            {
                                                "Category": "Critical for Digital Inclusion",
                                                "Summary": "Voice AI is seen as a fundamental tool for bringing unserved populations into the digital fold, enabling interaction for those traditionally excluded by digital interfaces."
                                            },
                                            {
                                                "Category": "Bridging Digital/Literacy Divide",
                                                "Summary": "Overcoming barriers for people who cannot type in their native languages or effectively interact with complex apps, providing access to information and services."
                                            },
                                            {
                                                "Category": "Not Just a Convenience",
                                                "Summary": "Voice technology has evolved from a 'nice to have' feature to a core offering essential for delivering inclusive solutions, especially in the Global South."
                                            }
                                        ],
                                        "Summary": "Voice AI is recognized as critical for achieving digital inclusion and overcoming literacy barriers, moving beyond being a mere convenience to an essential service for global populations."
                                    },
                                    {
                                        "Subnet": "Linguistic Diversity & AI",
                                        "Categories": [
                                            {
                                                "Category": "Low-Resource Language Integration",
                                                "Summary": "Efforts to bring a vast number of languages spoken by smaller populations (e.g., Bheeli, African languages) onto the AI map, which often lack existing digital resources."
                                            },
                                            {
                                                "Category": "Dialectal Variations",
                                                "Summary": "The challenge and necessity for AI models to understand and process numerous regional and local dialectal differences within a single language (e.g., Hindi in Delhi vs. Lucknow)."
                                            },
                                            {
                                                "Category": "Accelerating Language Work",
                                                "Summary": "The urgency and criticality of accelerating the development of voice models for diverse languages and dialects, not just official ones, to ensure comprehensive coverage."
                                            }
                                        ],
                                        "Summary": "The discussion addresses the significant challenge of integrating a multitude of spoken languages and dialects into AI systems to ensure true inclusivity, emphasizing the urgent need for focused development in this area."
                                    },
                                    {
                                        "Subnet": "Technology Evolution",
                                        "Categories": [
                                            {
                                                "Category": "Fast-Evolving Technology",
                                                "Summary": "Acknowledgement that AI technology is progressing rapidly, offering new capabilities and solutions that can address previously difficult linguistic and access challenges faster than anticipated."
                                            },
                                            {
                                                "Category": "Multilingual Models",
                                                "Summary": "The development of AI models capable of processing and understanding multiple languages simultaneously is key to broader application and efficiency."
                                            }
                                        ],
                                        "Summary": "The rapid advancements in AI and language models are viewed as a positive factor in addressing complex linguistic challenges, making the development of inclusive solutions more feasible and efficient."
                                    }
                                ],
                                "Summary": "This topic covers the inherent capabilities, evolving nature, and fundamental importance of voice and language AI technology in achieving widespread digital inclusion, particularly for diverse linguistic communities."
                            },
                            {
                                "Net": "Challenges in AI Adoption & Development",
                                "Subnets": [
                                    {
                                        "Subnet": "Infrastructure & Affordability",
                                        "Categories": [
                                            {
                                                "Category": "Compute & Connectivity Limitations",
                                                "Summary": "Shortages of computational resources, stable internet connectivity, and the high cost of data pose significant barriers to AI development and widespread adoption in many regions."
                                            },
                                            {
                                                "Category": "Electricity Access",
                                                "Summary": "Lack of consistent and reliable electricity supply is a fundamental challenge, impeding the deployment and operation of AI models and related technologies."
                                            },
                                            {
                                                "Category": "Device Affordability (Feature Phones)",
                                                "Summary": "The prevalence of feature phones and lower-end smartphones due to affordability limits user interaction with advanced apps, highlighting the need for voice-first solutions."
                                            }
                                        ],
                                        "Summary": "This subtopic explores the basic infrastructural deficits and economic barriers, such as limited compute power, inconsistent electricity, poor connectivity, and the high cost of devices, that hinder AI development and adoption."
                                    },
                                    {
                                        "Subnet": "Data & Linguistic Gaps",
                                        "Categories": [
                                            {
                                                "Category": "Lack of Digital Dictionaries & Glossaries",
                                                "Summary": "The absence of comprehensive, digitized dictionaries and standardized glossaries for local terminologies (e.g., land records, agri-terms) in native languages creates significant hurdles for AI systems."
                                            },
                                            {
                                                "Category": "High-Quality Training Data Shortage",
                                                "Summary": "The critical need for extensive, high-quality, and diverse training datasets for low-resource languages and dialects, a laborious and time-consuming process."
                                            },
                                            {
                                                "Category": "Complex Linguistic Nuances",
                                                "Summary": "Challenges in AI understanding phonetic languages, emotional rendering of voice, and how same words can mean different things when spoken differently, impacting accuracy."
                                            }
                                        ],
                                        "Summary": "This covers the technical challenges related to data scarcity and quality, including the absence of digital linguistic resources, insufficient training data for diverse languages, and the complexities of capturing subtle linguistic nuances."
                                    },
                                    {
                                        "Subnet": "Human Capital & Skill Gaps",
                                        "Categories": [
                                            {
                                                "Category": "Brain Drain",
                                                "Summary": "The challenge of retaining skilled talent within developing regions, as professionals often move abroad, impacting local capacity building for AI development."
                                            },
                                            {
                                                "Category": "Need for Linguistic Experts",
                                                "Summary": "The requirement for linguistic experts, local communities, and administrative support to collect, annotate, and validate data for diverse languages and dialects."
                                            }
                                        ],
                                        "Summary": "This section addresses the challenges related to human resources, including the loss of skilled professionals and the critical need for local linguistic expertise to develop and deploy AI solutions effectively."
                                    }
                                ],
                                "Summary": "This topic outlines the various significant hurdles faced in the widespread adoption, development, and effective implementation of AI and language technologies, encompassing infrastructure, data, and human resource limitations."
                            },
                            {
                                "Net": "Impact & Application of AI",
                                "Subnets": [
                                    {
                                        "Subnet": "Public Sector & Governance",
                                        "Categories": [
                                            {
                                                "Category": "Government Service Delivery",
                                                "Summary": "Voice-based AI is enabling access to numerous government services for non-literate populations, significantly reducing costs and increasing inclusivity (e.g., Irembo in Rwanda)."
                                            },
                                            {
                                                "Category": "Last Mile Connectivity in Governance",
                                                "Summary": "AI helps extend governance and service delivery to remote and underserved populations, bridging the urban-rural divide by providing information and actionable insights."
                                            }
                                        ],
                                        "Summary": "This subtopic highlights the transformative impact of AI in enhancing public sector efficiency and accessibility, particularly in delivering government services and governance to the last mile."
                                    },
                                    {
                                        "Subnet": "Key Social Sectors",
                                        "Categories": [
                                            {
                                                "Category": "Agriculture",
                                                "Summary": "AI-powered voice solutions provide farmers with critical advice and information in their local languages, increasing productivity and access to expert knowledge (e.g., Mahavistar AI)."
                                            },
                                            {
                                                "Category": "Health",
                                                "Summary": "Voice and chatbot components are being integrated into primary healthcare systems to provide accessible information and services, especially in rural areas (e.g., Horizon 1000 in Rwanda)."
                                            },
                                            {
                                                "Category": "Education",
                                                "Summary": "AI contributes to educational initiatives by transcending language barriers and providing information and learning opportunities in native languages."
                                            }
                                        ],
                                        "Summary": "Focuses on the significant impact of AI in vital social sectors like agriculture, health, and education, where it facilitates information dissemination, service delivery, and empowerment."
                                    }
                                ],
                                "Summary": "This topic details the tangible benefits and practical applications of voice and language AI technology across various sectors, highlighting its role in improving public services and empowering communities."
                            },
                            {
                                "Net": "Data & Resource Management",
                                "Subnets": [
                                    {
                                        "Subnet": "Data Security & Privacy",
                                        "Categories": [
                                            {
                                                "Category": "Compliance with DPDP Acts",
                                                "Summary": "Adherence to national data protection and privacy acts (e.g., India's DPDP Act) is crucial, especially as voice data becomes identifiable personal information."
                                            },
                                            {
                                                "Category": "Consent for Data Usage & Purpose",
                                                "Summary": "The challenge of obtaining explicit consent for both data contribution (voice donation) and the specific purpose of data use, particularly for training AI models and sensitive data like health records."
                                            },
                                            {
                                                "Category": "Federated Approaches",
                                                "Summary": "Utilizing federated learning where AI models travel to the data, ensuring data privacy as the actual data never leaves its secure environment (e.g., in healthcare)."
                                            }
                                        ],
                                        "Summary": "This subtopic addresses the critical concerns and complexities surrounding data security and privacy, emphasizing the need for robust frameworks and ethical practices in handling voice data for AI development."
                                    },
                                    {
                                        "Subnet": "Data Collection & Annotation",
                                        "Categories": [
                                            {
                                                "Category": "Community-Driven Data Collection",
                                                "Summary": "Engaging local communities to 'donate' their voices and data is a crucial, though non-trivial, process for building datasets for low-resource languages."
                                            },
                                            {
                                                "Category": "Structured Data Collection & Translation",
                                                "Summary": "The necessity for highly structured data collection exercises, including spontaneous speech, translation, and verification by linguistic experts, especially for languages without independent scripts."
                                            },
                                            {
                                                "Category": "Templates for Data Collection",
                                                "Summary": "Developing replicable templates or frameworks for data collection and annotation to accelerate the process for a multitude of low-resource languages across different regions."
                                            }
                                        ],
                                        "Summary": "Focuses on the labor-intensive yet essential processes of collecting and annotating high-quality data for AI models, particularly for under-resourced languages, often requiring community and expert involvement."
                                    }
                                ],
                                "Summary": "This topic encompasses the challenges and strategies related to managing data, including ensuring security and privacy, as well as the intricate processes involved in collecting, annotating, and curating high-quality datasets for AI."
                            },
                            {
                                "Net": "Ecosystem & Collaboration",
                                "Subnets": [
                                    {
                                        "Subnet": "Stakeholder Engagement",
                                        "Categories": [
                                            {
                                                "Category": "Government & Administration Support",
                                                "Summary": "The crucial role of local and national governments and administrations in facilitating data collection, providing linguistic expertise, and driving adoption of AI solutions."
                                            },
                                            {
                                                "Category": "Tech Entrepreneurs & Startups",
                                                "Summary": "Innovation is significantly driven by tech entrepreneurs, but they need support in benchmarking, market understanding, and building sustainable business cases for African languages."
                                            },
                                            {
                                                "Category": "Researchers & Academia",
                                                "Summary": "The importance of connecting researchers and academia with practitioners and entrepreneurs to ensure research translates into practical, market-ready solutions and avoids silent efforts."
                                            }
                                        ],
                                        "Summary": "This subtopic emphasizes the critical role of various stakeholders, including government, startups, and researchers, in fostering a collaborative environment for AI development and adoption."
                                    },
                                    {
                                        "Subnet": "South-South Collaboration",
                                        "Categories": [
                                            {
                                                "Category": "Breaking Down Silos & Fragmentation",
                                                "Summary": "The importance of fostering dialogue and collaboration between regions like India, Southeast Asia, Latin America, and Africa to share knowledge, experiences, and co-create solutions."
                                            },
                                            {
                                                "Category": "Learning from Shared Challenges",
                                                "Summary": "Identifying similarities in issues across global south contexts allows for learning from past mistakes and successes, particularly in areas like privacy, data protection, and model development."
                                            },
                                            {
                                                "Category": "Enabling Ecosystem & Youth Access",
                                                "Summary": "Collaborative efforts aim to empower young people and ensure they have access to spaces and resources to contribute to and benefit from AI innovation, considering Africa's youthful demography."
                                            }
                                        ],
                                        "Summary": "This section highlights the imperative for South-South collaboration to share knowledge, avoid redundant efforts, and accelerate the development and deployment of AI solutions across developing regions."
                                    }
                                ],
                                "Summary": "This topic explores the collaborative efforts and multi-stakeholder engagement essential for building a robust AI ecosystem, emphasizing the role of cross-regional partnerships and collective learning."
                            },
                            {
                                "Net": "Policy & Governance",
                                "Subnets": [
                                    {
                                        "Subnet": "National Language Policies",
                                        "Categories": [
                                            {
                                                "Category": "Policy-Driven Language Recognition",
                                                "Summary": "National policies that officially recognize and support multiple languages (e.g., South Africa's 11 official languages) provide a foundation for digital service delivery in native tongues."
                                            },
                                            {
                                                "Category": "Incentivizing Language Representation",
                                                "Summary": "Government-led initiatives and policies are crucial to encourage the creation of tools and resources for local languages, moving beyond mere information giving to knowledge exchange."
                                            }
                                        ],
                                        "Summary": "This subtopic focuses on the importance of national language policies and government initiatives in driving the recognition and integration of diverse languages into digital public infrastructure and AI applications."
                                    },
                                    {
                                        "Subnet": "Global vs. Sovereign AI Models",
                                        "Categories": [
                                            {
                                                "Category": "Use Case & Sovereignty as Determinants",
                                                "Summary": "The decision to build sovereign AI models versus using global, off-the-shelf models depends on factors like speed to market, economics, and the importance of national data sovereignty."
                                            },
                                            {
                                                "Category": "Longer-Term Sovereign Advantage",
                                                "Summary": "While quicker to market, relying on global models might not be ideal long-term; building sovereign models is seen as a better strategy for sustained control and localized relevance."
                                            }
                                        ],
                                        "Summary": "This addresses the strategic debate between developing country-specific (sovereign) AI models versus adopting global AI solutions, weighing factors like cost, speed, and national autonomy over data and technology."
                                    }
                                ],
                                "Summary": "This topic examines the strategic frameworks and regulatory considerations that influence the development and adoption of AI, including national language policies and the debate around sovereign AI models."
                            },
                            {
                                "Net": "996 Non specific TOPIC",
                                "Subnets": [
                                    {
                                        "Subnet": "Uncodable Responses",
                                        "Categories": [
                                            {
                                                "Category": "Irrelevant Content",
                                                "Summary": "Responses that are not directly relevant to the survey questions or the themes identified in the codeframe."
                                            },
                                            {
                                                "Category": "Needs Upcoding",
                                                "Summary": "Responses that are somewhat relevant but too broad or ambiguous to fit neatly into existing categories, requiring further analysis or refinement of the codeframe."
                                            }
                                        ],
                                        "Summary": "This subtopic captures responses that are generally irrelevant or difficult to categorize within the predefined themes, potentially requiring further review for upcoding."
                                    }
                                ],
                                "Summary": "This category is for responses that are not easily coded into the established themes, appearing irrelevant or requiring a higher level of coding abstraction."
                            },
                            {
                                "Net": "997 Don’t Know Can’t Say",
                                "Subnets": [
                                    {
                                        "Subnet": "Non-Committal Responses",
                                        "Categories": [
                                            {
                                                "Category": "Vague/Uncertain",
                                                "Summary": "Responses indicating uncertainty, lack of knowledge, or a general inability to provide a definitive answer."
                                            },
                                            {
                                                "Category": "Refusal to Answer",
                                                "Summary": "Explicit statements of not wanting to provide an answer or being unable to comment."
                                            }
                                        ],
                                        "Summary": "This subtopic groups responses that are non-committal, vague, or express uncertainty, indicating a lack of clear opinion or knowledge from the respondent."
                                    }
                                ],
                                "Summary": "This category captures responses where respondents explicitly state they do not know or cannot say, or provide vague and non-committal answers."
                            },
                            {
                                "Net": "998 Not Applicable",
                                "Subnets": [
                                    {
                                        "Subnet": "Out of Scope Responses",
                                        "Categories": [
                                            {
                                                "Category": "Irrelevant to Question/Project",
                                                "Summary": "Responses that are clearly outside the scope of the specific question asked or the overall project objectives."
                                            },
                                            {
                                                "Category": "Not Personal Experience",
                                                "Summary": "Responses indicating the question is not applicable to their personal situation or experience."
                                            }
                                        ],
                                        "Summary": "This subtopic includes responses that are deemed irrelevant or not applicable to the context of the question or the overall market research project."
                                    }
                                ],
                                "Summary": "This category is for responses that are irrelevant to the specific question or the broader project, indicating they are not applicable to the respondent's situation or the research context."
                            },
                            {
                                "Net": "999 None-Nothing",
                                "Subnets": [
                                    {
                                        "Subnet": "Gibberish/Meaningless",
                                        "Categories": [
                                            {
                                                "Category": "Non-Language Characters",
                                                "Summary": "Responses consisting of random characters, symbols, or keyboard mashing that do not form coherent words."
                                            },
                                            {
                                                "Category": "Meaningless Phrases",
                                                "Summary": "Responses that are grammatically incorrect, nonsensical, or convey no discernible meaning."
                                            }
                                        ],
                                        "Summary": "This subtopic covers responses that are essentially gibberish, meaningless, or composed of non-language characters, providing no valuable information."
                                    }
                                ],
                                "Summary": "This category is reserved for responses that are entirely nonsensical, include plain gibberish, or contain non-language characters, meaning they provide no valuable data."
                            }
                        ],
                        "Summary": "The panel discussion explores the critical role of voice and language AI technology in bridging digital and linguistic divides, particularly in India and African nations. Key discussion points include the importance of voice AI for digital inclusion, the challenges of integrating low-resource languages and dialects, infrastructural limitations, the necessity of high-quality data, and data privacy concerns. The conversation also highlights successful applications in sectors like agriculture and government services, the debate between global and sovereign AI models, and the crucial need for South-South collaboration to accelerate development and share lessons learned."
                    }
                }
            ],
            "onlyCodeFrame": true,
            "errorList": [],
            "codeFrameGenerateType": "CODE_FRAME_ONLY",
            "requestType": "CODEFRAME",
            "requestMode": "DEVELOPMENT",
            "sourceLanguage": "English",
            "targetLanguage": "English",
            "response": {
                "status": "OK",
                "data": {
                    "status": "OK",
                    "PromptTokenCount": 8715,
                    "CandidatesTokenCount": 3124,
                    "TotalTokenCount": 13136,
                    "thoughtsTokenCount": 1297,
                    "finishReason": "STOP",
                    "data": [
                        {
                            "Net": "Voice and Language AI Technology",
                            "Subnets": [
                                {
                                    "Subnet": "Role of Voice AI",
                                    "Categories": [
                                        {
                                            "Category": "Critical for Digital Inclusion",
                                            "Summary": "Voice AI is seen as a fundamental tool for bringing unserved populations into the digital fold, enabling interaction for those traditionally excluded by digital interfaces.",
                                            "code": 3
                                        },
                                        {
                                            "Category": "Bridging Digital/Literacy Divide",
                                            "Summary": "Overcoming barriers for people who cannot type in their native languages or effectively interact with complex apps, providing access to information and services.",
                                            "code": 4
                                        },
                                        {
                                            "Category": "Not Just a Convenience",
                                            "Summary": "Voice technology has evolved from a 'nice to have' feature to a core offering essential for delivering inclusive solutions, especially in the Global South.",
                                            "code": 5
                                        }
                                    ],
                                    "Summary": "Voice AI is recognized as critical for achieving digital inclusion and overcoming literacy barriers, moving beyond being a mere convenience to an essential service for global populations.",
                                    "code": 2
                                },
                                {
                                    "Subnet": "Linguistic Diversity & AI",
                                    "Categories": [
                                        {
                                            "Category": "Low-Resource Language Integration",
                                            "Summary": "Efforts to bring a vast number of languages spoken by smaller populations (e.g., Bheeli, African languages) onto the AI map, which often lack existing digital resources.",
                                            "code": 7
                                        },
                                        {
                                            "Category": "Dialectal Variations",
                                            "Summary": "The challenge and necessity for AI models to understand and process numerous regional and local dialectal differences within a single language (e.g., Hindi in Delhi vs. Lucknow).",
                                            "code": 8
                                        },
                                        {
                                            "Category": "Accelerating Language Work",
                                            "Summary": "The urgency and criticality of accelerating the development of voice models for diverse languages and dialects, not just official ones, to ensure comprehensive coverage.",
                                            "code": 9
                                        }
                                    ],
                                    "Summary": "The discussion addresses the significant challenge of integrating a multitude of spoken languages and dialects into AI systems to ensure true inclusivity, emphasizing the urgent need for focused development in this area.",
                                    "code": 6
                                },
                                {
                                    "Subnet": "Technology Evolution",
                                    "Categories": [
                                        {
                                            "Category": "Fast-Evolving Technology",
                                            "Summary": "Acknowledgement that AI technology is progressing rapidly, offering new capabilities and solutions that can address previously difficult linguistic and access challenges faster than anticipated.",
                                            "code": 11
                                        },
                                        {
                                            "Category": "Multilingual Models",
                                            "Summary": "The development of AI models capable of processing and understanding multiple languages simultaneously is key to broader application and efficiency.",
                                            "code": 12
                                        }
                                    ],
                                    "Summary": "The rapid advancements in AI and language models are viewed as a positive factor in addressing complex linguistic challenges, making the development of inclusive solutions more feasible and efficient.",
                                    "code": 10
                                }
                            ],
                            "Summary": "This topic covers the inherent capabilities, evolving nature, and fundamental importance of voice and language AI technology in achieving widespread digital inclusion, particularly for diverse linguistic communities.",
                            "code": 1
                        },
                        {
                            "Net": "Challenges in AI Adoption & Development",
                            "Subnets": [
                                {
                                    "Subnet": "Infrastructure & Affordability",
                                    "Categories": [
                                        {
                                            "Category": "Compute & Connectivity Limitations",
                                            "Summary": "Shortages of computational resources, stable internet connectivity, and the high cost of data pose significant barriers to AI development and widespread adoption in many regions.",
                                            "code": 15
                                        },
                                        {
                                            "Category": "Electricity Access",
                                            "Summary": "Lack of consistent and reliable electricity supply is a fundamental challenge, impeding the deployment and operation of AI models and related technologies.",
                                            "code": 16
                                        },
                                        {
                                            "Category": "Device Affordability (Feature Phones)",
                                            "Summary": "The prevalence of feature phones and lower-end smartphones due to affordability limits user interaction with advanced apps, highlighting the need for voice-first solutions.",
                                            "code": 17
                                        }
                                    ],
                                    "Summary": "This subtopic explores the basic infrastructural deficits and economic barriers, such as limited compute power, inconsistent electricity, poor connectivity, and the high cost of devices, that hinder AI development and adoption.",
                                    "code": 14
                                },
                                {
                                    "Subnet": "Data & Linguistic Gaps",
                                    "Categories": [
                                        {
                                            "Category": "Lack of Digital Dictionaries & Glossaries",
                                            "Summary": "The absence of comprehensive, digitized dictionaries and standardized glossaries for local terminologies (e.g., land records, agri-terms) in native languages creates significant hurdles for AI systems.",
                                            "code": 19
                                        },
                                        {
                                            "Category": "High-Quality Training Data Shortage",
                                            "Summary": "The critical need for extensive, high-quality, and diverse training datasets for low-resource languages and dialects, a laborious and time-consuming process.",
                                            "code": 20
                                        },
                                        {
                                            "Category": "Complex Linguistic Nuances",
                                            "Summary": "Challenges in AI understanding phonetic languages, emotional rendering of voice, and how same words can mean different things when spoken differently, impacting accuracy.",
                                            "code": 21
                                        }
                                    ],
                                    "Summary": "This covers the technical challenges related to data scarcity and quality, including the absence of digital linguistic resources, insufficient training data for diverse languages, and the complexities of capturing subtle linguistic nuances.",
                                    "code": 18
                                },
                                {
                                    "Subnet": "Human Capital & Skill Gaps",
                                    "Categories": [
                                        {
                                            "Category": "Brain Drain",
                                            "Summary": "The challenge of retaining skilled talent within developing regions, as professionals often move abroad, impacting local capacity building for AI development.",
                                            "code": 23
                                        },
                                        {
                                            "Category": "Need for Linguistic Experts",
                                            "Summary": "The requirement for linguistic experts, local communities, and administrative support to collect, annotate, and validate data for diverse languages and dialects.",
                                            "code": 24
                                        }
                                    ],
                                    "Summary": "This section addresses the challenges related to human resources, including the loss of skilled professionals and the critical need for local linguistic expertise to develop and deploy AI solutions effectively.",
                                    "code": 22
                                }
                            ],
                            "Summary": "This topic outlines the various significant hurdles faced in the widespread adoption, development, and effective implementation of AI and language technologies, encompassing infrastructure, data, and human resource limitations.",
                            "code": 13
                        },
                        {
                            "Net": "Impact & Application of AI",
                            "Subnets": [
                                {
                                    "Subnet": "Public Sector & Governance",
                                    "Categories": [
                                        {
                                            "Category": "Government Service Delivery",
                                            "Summary": "Voice-based AI is enabling access to numerous government services for non-literate populations, significantly reducing costs and increasing inclusivity (e.g., Irembo in Rwanda).",
                                            "code": 27
                                        },
                                        {
                                            "Category": "Last Mile Connectivity in Governance",
                                            "Summary": "AI helps extend governance and service delivery to remote and underserved populations, bridging the urban-rural divide by providing information and actionable insights.",
                                            "code": 28
                                        }
                                    ],
                                    "Summary": "This subtopic highlights the transformative impact of AI in enhancing public sector efficiency and accessibility, particularly in delivering government services and governance to the last mile.",
                                    "code": 26
                                },
                                {
                                    "Subnet": "Key Social Sectors",
                                    "Categories": [
                                        {
                                            "Category": "Agriculture",
                                            "Summary": "AI-powered voice solutions provide farmers with critical advice and information in their local languages, increasing productivity and access to expert knowledge (e.g., Mahavistar AI).",
                                            "code": 30
                                        },
                                        {
                                            "Category": "Health",
                                            "Summary": "Voice and chatbot components are being integrated into primary healthcare systems to provide accessible information and services, especially in rural areas (e.g., Horizon 1000 in Rwanda).",
                                            "code": 31
                                        },
                                        {
                                            "Category": "Education",
                                            "Summary": "AI contributes to educational initiatives by transcending language barriers and providing information and learning opportunities in native languages.",
                                            "code": 32
                                        }
                                    ],
                                    "Summary": "Focuses on the significant impact of AI in vital social sectors like agriculture, health, and education, where it facilitates information dissemination, service delivery, and empowerment.",
                                    "code": 29
                                }
                            ],
                            "Summary": "This topic details the tangible benefits and practical applications of voice and language AI technology across various sectors, highlighting its role in improving public services and empowering communities.",
                            "code": 25
                        },
                        {
                            "Net": "Data & Resource Management",
                            "Subnets": [
                                {
                                    "Subnet": "Data Security & Privacy",
                                    "Categories": [
                                        {
                                            "Category": "Compliance with DPDP Acts",
                                            "Summary": "Adherence to national data protection and privacy acts (e.g., India's DPDP Act) is crucial, especially as voice data becomes identifiable personal information.",
                                            "code": 35
                                        },
                                        {
                                            "Category": "Consent for Data Usage & Purpose",
                                            "Summary": "The challenge of obtaining explicit consent for both data contribution (voice donation) and the specific purpose of data use, particularly for training AI models and sensitive data like health records.",
                                            "code": 36
                                        },
                                        {
                                            "Category": "Federated Approaches",
                                            "Summary": "Utilizing federated learning where AI models travel to the data, ensuring data privacy as the actual data never leaves its secure environment (e.g., in healthcare).",
                                            "code": 37
                                        }
                                    ],
                                    "Summary": "This subtopic addresses the critical concerns and complexities surrounding data security and privacy, emphasizing the need for robust frameworks and ethical practices in handling voice data for AI development.",
                                    "code": 34
                                },
                                {
                                    "Subnet": "Data Collection & Annotation",
                                    "Categories": [
                                        {
                                            "Category": "Community-Driven Data Collection",
                                            "Summary": "Engaging local communities to 'donate' their voices and data is a crucial, though non-trivial, process for building datasets for low-resource languages.",
                                            "code": 39
                                        },
                                        {
                                            "Category": "Structured Data Collection & Translation",
                                            "Summary": "The necessity for highly structured data collection exercises, including spontaneous speech, translation, and verification by linguistic experts, especially for languages without independent scripts.",
                                            "code": 40
                                        },
                                        {
                                            "Category": "Templates for Data Collection",
                                            "Summary": "Developing replicable templates or frameworks for data collection and annotation to accelerate the process for a multitude of low-resource languages across different regions.",
                                            "code": 41
                                        }
                                    ],
                                    "Summary": "Focuses on the labor-intensive yet essential processes of collecting and annotating high-quality data for AI models, particularly for under-resourced languages, often requiring community and expert involvement.",
                                    "code": 38
                                }
                            ],
                            "Summary": "This topic encompasses the challenges and strategies related to managing data, including ensuring security and privacy, as well as the intricate processes involved in collecting, annotating, and curating high-quality datasets for AI.",
                            "code": 33
                        },
                        {
                            "Net": "Ecosystem & Collaboration",
                            "Subnets": [
                                {
                                    "Subnet": "Stakeholder Engagement",
                                    "Categories": [
                                        {
                                            "Category": "Government & Administration Support",
                                            "Summary": "The crucial role of local and national governments and administrations in facilitating data collection, providing linguistic expertise, and driving adoption of AI solutions.",
                                            "code": 44
                                        },
                                        {
                                            "Category": "Tech Entrepreneurs & Startups",
                                            "Summary": "Innovation is significantly driven by tech entrepreneurs, but they need support in benchmarking, market understanding, and building sustainable business cases for African languages.",
                                            "code": 45
                                        },
                                        {
                                            "Category": "Researchers & Academia",
                                            "Summary": "The importance of connecting researchers and academia with practitioners and entrepreneurs to ensure research translates into practical, market-ready solutions and avoids silent efforts.",
                                            "code": 46
                                        }
                                    ],
                                    "Summary": "This subtopic emphasizes the critical role of various stakeholders, including government, startups, and researchers, in fostering a collaborative environment for AI development and adoption.",
                                    "code": 43
                                },
                                {
                                    "Subnet": "South-South Collaboration",
                                    "Categories": [
                                        {
                                            "Category": "Breaking Down Silos & Fragmentation",
                                            "Summary": "The importance of fostering dialogue and collaboration between regions like India, Southeast Asia, Latin America, and Africa to share knowledge, experiences, and co-create solutions.",
                                            "code": 48
                                        },
                                        {
                                            "Category": "Learning from Shared Challenges",
                                            "Summary": "Identifying similarities in issues across global south contexts allows for learning from past mistakes and successes, particularly in areas like privacy, data protection, and model development.",
                                            "code": 49
                                        },
                                        {
                                            "Category": "Enabling Ecosystem & Youth Access",
                                            "Summary": "Collaborative efforts aim to empower young people and ensure they have access to spaces and resources to contribute to and benefit from AI innovation, considering Africa's youthful demography.",
                                            "code": 50
                                        }
                                    ],
                                    "Summary": "This section highlights the imperative for South-South collaboration to share knowledge, avoid redundant efforts, and accelerate the development and deployment of AI solutions across developing regions.",
                                    "code": 47
                                }
                            ],
                            "Summary": "This topic explores the collaborative efforts and multi-stakeholder engagement essential for building a robust AI ecosystem, emphasizing the role of cross-regional partnerships and collective learning.",
                            "code": 42
                        },
                        {
                            "Net": "Policy & Governance",
                            "Subnets": [
                                {
                                    "Subnet": "National Language Policies",
                                    "Categories": [
                                        {
                                            "Category": "Policy-Driven Language Recognition",
                                            "Summary": "National policies that officially recognize and support multiple languages (e.g., South Africa's 11 official languages) provide a foundation for digital service delivery in native tongues.",
                                            "code": 53
                                        },
                                        {
                                            "Category": "Incentivizing Language Representation",
                                            "Summary": "Government-led initiatives and policies are crucial to encourage the creation of tools and resources for local languages, moving beyond mere information giving to knowledge exchange.",
                                            "code": 54
                                        }
                                    ],
                                    "Summary": "This subtopic focuses on the importance of national language policies and government initiatives in driving the recognition and integration of diverse languages into digital public infrastructure and AI applications.",
                                    "code": 52
                                },
                                {
                                    "Subnet": "Global vs. Sovereign AI Models",
                                    "Categories": [
                                        {
                                            "Category": "Use Case & Sovereignty as Determinants",
                                            "Summary": "The decision to build sovereign AI models versus using global, off-the-shelf models depends on factors like speed to market, economics, and the importance of national data sovereignty.",
                                            "code": 56
                                        },
                                        {
                                            "Category": "Longer-Term Sovereign Advantage",
                                            "Summary": "While quicker to market, relying on global models might not be ideal long-term; building sovereign models is seen as a better strategy for sustained control and localized relevance.",
                                            "code": 57
                                        }
                                    ],
                                    "Summary": "This addresses the strategic debate between developing country-specific (sovereign) AI models versus adopting global AI solutions, weighing factors like cost, speed, and national autonomy over data and technology.",
                                    "code": 55
                                }
                            ],
                            "Summary": "This topic examines the strategic frameworks and regulatory considerations that influence the development and adoption of AI, including national language policies and the debate around sovereign AI models.",
                            "code": 51
                        }
                    ],
                    "Summary": "The panel discussion explores the critical role of voice and language AI technology in bridging digital and linguistic divides, particularly in India and African nations. Key discussion points include the importance of voice AI for digital inclusion, the challenges of integrating low-resource languages and dialects, infrastructural limitations, the necessity of high-quality data, and data privacy concerns. The conversation also highlights successful applications in sectors like agriculture and government services, the debate between global and sovereign AI models, and the crucial need for South-South collaboration to accelerate development and share lessons learned."
                }
            },
            "summaryList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 8376,
                        "CandidatesTokenCount": 3079,
                        "TotalTokenCount": 12919,
                        "thoughtsTokenCount": 1464,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Summary": "<h2>Comprehensive Synthesis Report on Voice and Language Technology for Inclusivity</h2><h3>I. Introduction and Overview of the Digital Divide</h3><p>The discussion commenced by highlighting the significant challenge of **last-mile connectivity** in achieving digital inclusivity. It was acknowledged that a ready-made solution for this complex issue is not yet available, prompting ongoing thought processes and innovative approaches. A key strategy for empowering frontline workers, such as ASHAs, involves providing them with tablets and fostering innovation through hackathons and competitions. These initiatives aim to encourage youngsters and startups to develop bespoke models that cater to specific needs, moving away from a 'one-size-fits-all' approach. The underlying philosophy emphasizes leveraging existing technology and energy rather than reinventing the wheel, focusing instead on its effective adoption and adaptation to bridge critical divides.</p><h3>II. Addressing the Digital Divide with Voice and Language Technology</h3><h4>A. Bridging Language and Literacy Barriers</h4><p>The panel extensively discussed how **voice technology** is proving critical in transcending language barriers and bridging the digital and literacy divides. By using voice as a medium, productivity is enhanced, especially for those who may be digitally literate (e.g., forwarding WhatsApp messages, watching YouTube videos) but lack the skills or comfort to interact with complex apps through typing. Voice technology is positioned as an inherent human connection method, making technology adoption intuitive and inclusive. Initiatives like Bhashini, through partnerships with UIDAI and the Ministry of Panchayat-e-Raj, now reach every citizen and village, processing 15-18 million daily inferences and showing 15-20% month-on-month growth, indicating a significant latent need. This growth is attributed to a commitment to continuous improvement and a collaborative approach where end-users are seen as contributors.</p><h4>B. The Crucial Role of Voice Technology for Inclusion</h4><p>Voice is no longer considered a mere convenience but a **core offering for inclusive solutions**. Many people still use feature phones (300 million in India), and a larger segment lacks the 'digital literacy' for app interaction, even in their native languages. Making people learn new interfaces (typing, smartphone operations) is inherently exclusionary. Voice technology ensures that digital solutions are accessible and easy to adopt, aligning with how people naturally communicate.</p><h4>C. AI and Language Representation</h4><p>A critical challenge is the vast disparity in **language representation on the AI map**. Globally, there are around 7,000 spoken languages, with over 100 in India alone, yet only hundreds are represented in AI models. There is an urgent need to accelerate the inclusion of these languages, particularly in the Global South and African nations. This involves building Large Language Models (LLMs) that understand not only official languages but also regional languages and their diverse dialects (e.g., Hindi variations across states). The good news is that technology is evolving rapidly, making this acceleration possible.</p><h4>D. Sectoral Impact and Use Cases</h4><p>Voice and language technology demonstrate significant impact across multiple sectors. The **social sector**, including agriculture, education, and health, shows the biggest uptake due to its inherent need for last-mile service delivery and governance. Examples include:</p><ul><li><b>Government Service Delivery:</b> Rwanda's "
                            },
                            {
                                "Summary": "Your REMBO"
                            },
                            {
                                "Summary": "initiative provides voice-based access to 240 government services, drastically reducing costs and expanding accessibility for non-literate populations."
                            },
                            {
                                "Summary": "</li><li><b>Healthcare:</b> Rwanda's "
                            },
                            {
                                "Summary": "Horizon 1000"
                            },
                            {
                                "Summary": "project in primary healthcare integrates voice and chatbot components. Federated approaches are being explored in health to ensure data privacy for voice, which is considered identifiable PHI."
                            },
                            {
                                "Summary": "</li><li><b>Agriculture:</b> India's "
                            },
                            {
                                "Summary": "Mahavistar AI"
                            },
                            {
                                "Summary": "app provides voice-based advice to farmers in Marathi, demonstrating significant downloads and AI interactions.</li></ul><p>While information delivery is happening, the transition to 'actionable impact' still requires familiarization, especially with dialectal nuances.</p><h3>III. Challenges and Obstacles in Language Technology Implementation</h3><h4>A. Data and Linguistic Gaps</h4><p>Several fundamental challenges persist:</p><ul><li><b>Digital Dictionary Absence:</b> A significant lack of digital dictionaries for native languages, including place names. In India, 1.6-1.8 million places need to be digitized.</li><li><b>Emotional Rendering:</b> AI struggles with the emotional rendering of voice, which can alter meaning, especially in phonetic languages like Miso.</li><li><b>Dialectal Variations:</b> Even within the same language, dialects vary significantly (e.g., Hindi in Delhi vs. Lucknow vs. Bihar), posing challenges for AI understanding.</li><li><b>High-Quality Data Scarcity:</b> Low-resource languages often lack the spontaneous speech, translations, and verified data necessary for training robust AI models.</li><li><b>Glossary Development:</b> The absence of standardized vertical and language glossaries (e.g., for agricultural terms or land records) hinders interoperability and accuracy. Every state and line ministry may have millions of unique words to digitize.</li></ul><h4>B. Infrastructure and Resource Constraints (Africa Specific, but Global South Relevance)</h4><p>In addition to linguistic challenges, infrastructure presents significant hurdles, particularly in Africa, which resonate across the Global South:</p><ul><li><b>Compute Resources:</b> Insufficient access to and efficient utilization of computational power.</li><li><b>Connectivity and Devices:</b> Legacy issues with connectivity, high prevalence of feature phones or lower-end smartphones due to affordability, and mobile phones often being the first digital device.</li><li><b>Electricity Access:</b> A fundamental barrier to running models and technologies.</li><li><b>Capacity Building & Brain Drain:</b> While skills exist, brain drain depletes local talent pools.</li><li><b>Policy and Government Recognition:</b> The extent to which national policies recognize and support digital services in indigenous languages (e.g., South Africa's 11 official languages) impacts adoption.</li><li><b>Knowledge Exchange vs. Information Giving:</b> Moving beyond simply providing information to enabling interactive knowledge exchange in local languages.</li></ul><h4>C. Translation from Research to Production</h4><p>The journey from research to practical application involves:</p><ul><li><b>Benchmarking and Evaluation:</b> Developing systematic standards to evaluate models for diversity and cultural nuance, ensuring they work effectively in global majority contexts where technologies often lag.</li><li><b>Ecosystem Development:</b> Creating datasets, establishing data governance, and fostering innovation through research.</li><li><b>Connecting with Tech Entrepreneurs:</b> Demonstrating the business value of integrating local languages into products to drive market adoption and innovation (e.g., citing Bhashini's success in India to African entrepreneurs).</li><li><b>Breaking Silos:</b> Fostering collaboration among researchers, developers, and entrepreneurs to avoid fragmented efforts.</li></ul><h3>IV. Strategies and Solutions for Accelerated Development</h3><h4>A. Prioritization and Approach</h4><p>Given the vast number of languages, a balanced approach is necessary:</p><ul><li><b>Breadth-first:</b> Developing multilingual models that can address multiple languages simultaneously.</li><li><b>Depth-first:</b> Focusing on sector-specific use cases (e.g., government services, agriculture) to go deep and address immediate needs.</li></ul><p>The overall strategy should be **use case-driven**, with language selection determined by the most spoken languages in a specific area relevant to the use case.</p><h4>B. Low-Resource Language Inclusion: The Bheeli Example</h4><p>A notable success story is the inclusion of the Bheeli language (spoken by ~10 million people across Gujarat, Maharashtra, Rajasthan) into the Mahavistar Agri app. This project, a collaboration between the Government of Maharashtra, Bhashini, AI for Bharat, and Karya, demonstrated:</p><ul><li><b>Community-Driven Data Collection:</b> Local administration and community members "
                            },
                            {
                                "Summary": "donated"
                            },
                            {
                                "Summary": "their voices and data."
                            },
                            {
                                "Summary": "</li><li><b>Rapid Development:</b> Despite challenges like Bheeli lacking an independent script and requiring complex data collection (spontaneous speech, translation to Marathi/English, verification), the language was brought onto the AI map in less than six weeks, a process that typically takes a year or more.</li><li><b>Scalable Template:</b> The project aims to create a template for rapidly including other low-resource languages, not just in India but across the Global South (e.g., for Africa's 2,000+ languages).</li></ul><h4>C. Data Security and Privacy</h4><p>Adherence to data protection norms (like India's DPDP Act) is crucial. Key considerations include:</p><ul><li><b>Consent:</b> Obtaining explicit consent for data contribution, especially for training AI models.</li><li><b>Purpose Definition:</b> Clearly defining the purpose for which data (e.g., health data, voice as personal information) is collected and used.</li><li><b>Granularity:</b> Allowing users to consent to specific portions or uses of their data.</li><li><b>Open-Source Approach:</b> Using open-source domains ensures social auditability and fosters continuous improvement, pushing developers to stay relevant.</li><li><b>Federated Learning:</b> For sensitive data like in health, federated approaches (where the model travels to the data, but data itself is not shared) are a proven solution to maintain privacy.</li></ul><h4>D. Global vs. Sovereign Models</h4><p>The choice between using global, off-the-shelf AI models versus building sovereign models depends on the use case, time-to-market, economics, and the importance of data sovereignty. While global models offer quick deployment, the long-term preference is to **build sovereign models** for greater control and tailored solutions.</p><h4>E. South-South Collaboration</h4><p>Accelerating progress requires breaking down silos and fostering South-South collaboration, particularly between India, Southeast Asia, Latin America, and Africa. Key aspects include:</p><ul><li><b>Co-creation of Knowledge:</b> Sharing research and learnings.</li><li><b>Multi-stakeholder Engagement:</b> Bringing together collaborators, tech entrepreneurs, and even big tech companies.</li><li><b>Youth Inclusion:</b> Ensuring young people have access to these spaces and understand the economic viability of local language solutions.</li><li><b>Learning from Shared Experiences:</b> Adapting lessons learned from challenges like privacy and data protection to avoid repeating mistakes, as championed by organizations like Masakane ("
                            },
                            {
                                "Summary": "to build together"
                            },
                            {
                                "Summary": "). This summit itself serves as a vital platform for dialogue."
                            },
                            {
                                "Summary": "</li></ul><h4>F. Glossary Development</h4><p>A critical, high-value challenge is the creation and digitization of **vertical and language glossaries**. This involves:</p><ul><li><b>Local Government Collaboration:</b> Working with local administrations and relevant departments (e.g., health, land records, agriculture) to prepare and validate glossaries.</li><li><b>Linguistic Expertise:</b> Engaging local linguistic experts, school teachers, and professors.</li><li><b>Massive Undertaking:</b> Recognizing that each state and line ministry may have millions of words requiring digitization to ensure interoperability and accuracy in digital systems. This proactive work is essential to overcome current terminological mismatches.</p><h3>V. Conclusions and Recommendations</h3><h4>A. Conclusions</h4><p>Voice AI and language technology are indispensable tools for achieving true digital inclusion, particularly for the vast populations in the Global South and remote areas that face significant language, literacy, and infrastructure barriers. While remarkable progress is being made through collaborative efforts and innovative approaches, persistent challenges related to data quality, linguistic diversity (especially dialects and low-resource languages), infrastructure, and regulatory frameworks (privacy) demand continuous attention. The ability to rapidly onboard new languages, as demonstrated by the Bheeli project, offers a template for broader impact. The focus is shifting from mere information delivery to enabling actionable impact through inclusive and culturally nuanced digital solutions.</p><h4>B. Actionable Recommendations</h4><p>Based on the findings, the following recommendations are proposed:</p><ul><li><b>Invest in Comprehensive Glossary Digitization:</b> Actively support and mandate state governments and line ministries to develop and digitize comprehensive vertical and language-specific glossaries. This will address the current lack of digital dictionaries and ensure interoperability across various government and social sector services.</li><li><b>Standardize Low-Resource Language Inclusion Templates:</b> Leverage successful initiatives like the Bheeli language project to create and disseminate standardized templates and best practices for rapid data collection, annotation, model building, and application integration for low-resource languages. This template should be adaptable for various Global South contexts.</li><li><b>Foster South-South Collaborative Ecosystems:</b> Establish and strengthen formal and informal platforms (e.g., summits, research grants, joint hackathons) for South-South collaboration. These initiatives should connect researchers, tech entrepreneurs, and policymakers from regions like India and Africa to share knowledge, co-create solutions, and learn from shared challenges (e.g., data privacy, infrastructure constraints).</li><li><b>Prioritize Use Case-Driven Language Development:</b> Continue to adopt a use case-first approach when prioritizing language development. This ensures that resources are directed towards languages and dialects that serve immediate and critical needs within specific sectors (e.g., agriculture, health, government services), maximizing practical impact.</li><li><b>Develop Robust Data Privacy and Security Frameworks:</b> Implement and continually refine data privacy and security frameworks, especially for voice and health data. This includes clear consent mechanisms, options for granular data sharing, and exploring advanced techniques like federated learning to protect personal identifiable information while still enabling AI model training. The adoption of open-source methodologies is recommended for transparency and auditability.</li><li><b>Address Infrastructure Gaps Strategically:</b> Collaborate with governments and private entities to address fundamental infrastructure challenges in underserved regions, including access to compute resources, affordable connectivity (beyond feature phones), and reliable electricity, which are prerequisites for widespread technology adoption.</li><li><b>Bridge Research to Market Gaps:</b> Create stronger linkages between academic research and commercial applications by supporting benchmarking initiatives and providing guidance to tech entrepreneurs on building viable business cases for local language solutions, ensuring that innovations are high-quality and market-ready.</li></ul>"
                            }
                        ]
                    }
                }
            ]
        }
    ],
    "headerCols": {
        "col1": "Part Id",
        "col2": "FileName",
        "col3": "Content",
        "col4": "Status"
    },
    "fileName": "",
    "projectType": "VIDEO",
    "sourceLanguage": "English",
    "targetLanguage": "English",
    "__v": 0
},

{
    "_id": "69940e7b25c9f4e0d7e6ffe7",
    "projectId": "783433",
    "projectName": "AI for Education Building Future-Ready Universities and Schools - Spanish ",
    "projectDescription": "AI for Education Building Future-Ready Universities and Schools",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T06:45:20.261Z",
    "projectUpdatedDate": "2026-02-17T06:45:20.261Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T06:45:20.261Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "AI for Education_ Building Future-Ready Universities and Schools.mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771310707292.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:27",
                    "text": " Quality is the best. Everything about Japan is the best. But as we see the societal issues that we have in Japan and the scale issues that we have in Japan and the technical issues that we have in Japan, it's become much, much clearer in all domains, construction especially, right? That yes, we have a lot of good and important knowledge, but in order to scale that globally and in order to continue growing, we now have to look outside of Japan."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:27",
                    "end": "00:00:54",
                    "text": " And so you see major enterprise companies in my industry everywhere, right? We have the top five general contractor companies in Japan, the mega super general contractor companies, acquiring companies all over the place, whether that be in India, whether that be in places like Singapore. You see all of the construction consultancies expanding their operations overseas and developing up similar to Fujitsu, like research centers and innovation centers outside of Japan as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:54",
                    "end": "00:01:33",
                    "text": " So I think that is a movement that we are seeing. I think that ultimately it becomes a two-way flow. And so we do need to import talent. We do need to make sure that Japan itself is shored up. But on an innovation scale, that connection continues to grow globally. And I think you'll see it continue to keep scaling into billions of dollars of investment. Yeah. Thank you. Okay. Okay. I think you strike the very important point."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:34",
                    "end": "00:02:04",
                    "text": " I think for Japanese, yeah, that is a very important point to look at our challenges. So I think that in Japan, the labor force, we are decreasing the labor force. So we need more talent from India to sustain our industry. So of course, robotics can help us, but I think the real human interface is still needed."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:04",
                    "end": "00:02:33",
                    "text": " I think in that sense, like I said, yeah, we can work together. So I think this is not straight question, but I think, yes, you pointed out a very important point. Thank you very much. I just have a comment. See, I mean, I represent this side of the equation. I've spent about six, seven years in Japan. There is still a very significant gap between"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:33",
                    "end": "00:03:06",
                    "text": " how we see ourselves and how Japan sees India in terms of policy, in terms of risk, in terms of a lot of other factors which are very cultural in nature. In the seven years post-COVID, due to significant challenges in supply chain, they are significantly improved."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:06",
                    "end": "00:03:36",
                    "text": " their sort of investment strategies with India as a focus. So if you look at the two very successful Japanese companies in India, of course Suzuki stands out. The second one is Daikin. Manufacture for the world with India as a base. These were the two standout companies. But you don't see too many other companies coming. It's not that they don't have problems."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:37",
                    "end": "00:04:03",
                    "text": " The fact of the matter is culturally I might be politically incorrect to make the statement here They are very comfortable working with Southeast Asian countries Okay, so we need to really look at how can we make India a Great destination for their investment not just as a market for also scaling up our manufacturing capability"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:03",
                    "end": "00:04:33",
                    "text": " Vietnam, Thailand, Indonesia, Malaysia are markets where Japanese companies outside of the West have significant exposure to. So that is the primary destination and those were colonies of Japan prior to World War II. Therefore, there is a lot of cultural integration between these two markets. So there is a little bit of history, so we need to unwind that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:33",
                    "end": "00:05:03",
                    "text": " I think our governments have done a fantastic job when Prime Minister visited Japan in August 2025 when they signed the next version of India-Japan journey. There are a lot of commitments made on both sides. Commitments alone do not help. I think commitments need to be followed by investment. Very recently I am seeing a lot of banks investing. You would have also heard"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:03",
                    "end": "00:05:31",
                    "text": " SMBC is invested in a bank in India. Mizzou has also acquired a bank in India and that is just the starting point. Once you get the banking ecosystem starting to invest in another country, then you will find capital flow. So I am very, very hopeful the next five years is possibly going to be the best time for the India-Japan relationship given the struggles that Japan is having with the other Southeast Asian partners."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:31",
                    "end": "00:05:60",
                    "text": " So keep this in mind, there is a history associated with it, and we need to work to unwind some of the things that have happened in the past, and that's possibly why there is still reluctance. Thank you very much. Time flies. Now we have already exceeded the end of the session. Thanks once again to the panelists and the audience. Actually, we have a small gift for the panelists. Dear panelists, could you stand up?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:01",
                    "end": "00:06:41",
                    "text": " And thank you so much for your kind cooperation. And please give them a lot of applause. Now, we would like to close the event. Thank you so much for the participation. And lastly, the company that participated, the panelists from Japan Today, are also exhibiting the Japan Pavilion as a whole for a team. A total seven companies, including those, will not present today. I'll showcase their work. So please feel free to visit if you're interested."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:41",
                    "end": "00:06:55",
                    "text": " This concludes the event, and you may exit at your convenience. Once again, thank you very much for participating today."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Spanish"
},
{
    "_id": "69940d3c25c9f4e0d7e6ffb7",
    "projectId": "927654",
    "projectName": "AI for Education Building Future-Ready Universities and Schools - Hindi",
    "projectDescription": "AI for Education Building Future-Ready Universities and Schools - Hindi",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T06:40:00.990Z",
    "projectUpdatedDate": "2026-02-17T06:40:00.990Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T06:40:00.990Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "AI for Education_ Building Future-Ready Universities and Schools.mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771310386302.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:27",
                    "text": " Quality is the best. Everything about Japan is the best. But as we see the societal issues that we have in Japan and the scale issues that we have in Japan and the technical issues that we have in Japan, it's become much, much clearer in all domains, construction especially, right? That yes, we have a lot of good and important knowledge, but in order to scale that globally and in order to continue growing, we now have to look outside of Japan."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:27",
                    "end": "00:00:54",
                    "text": " And so you see major enterprise companies in my industry everywhere, right? We have the top five general contractor companies in Japan, the mega super general contractor companies, acquiring companies all over the place, whether that be in India, whether that be in places like Singapore. You see all of the construction consultancies expanding their operations overseas and developing up similar to Fujitsu, like research centers and innovation centers outside of Japan as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:54",
                    "end": "00:01:33",
                    "text": " So I think that is a movement that we are seeing. I think that ultimately it becomes a two-way flow. And so we do need to import talent. We do need to make sure that Japan itself is shored up. But on an innovation scale, that connection continues to grow globally. And I think you'll see it continue to keep scaling into billions of dollars of investment. Yeah. Thank you. Okay. Okay. I think you strike the very important point."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:34",
                    "end": "00:02:04",
                    "text": " I think for Japanese, yeah, that is a very important point to look at our challenges. So I think that in Japan, the labor force, we are decreasing the labor force. So we need more talent from India to sustain our industry. So of course, robotics can help us, but I think the real human interface is still needed."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:04",
                    "end": "00:02:33",
                    "text": " I think in that sense, like I said, yeah, we can work together. So I think this is not straight question, but I think, yes, you pointed out a very important point. Thank you very much. I just have a comment. See, I mean, I represent this side of the equation. I've spent about six, seven years in Japan. There is still a very significant gap between"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:33",
                    "end": "00:03:06",
                    "text": " how we see ourselves and how Japan sees India in terms of policy, in terms of risk, in terms of a lot of other factors which are very cultural in nature. In the seven years post-COVID, due to significant challenges in supply chain, they are significantly improved."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:06",
                    "end": "00:03:36",
                    "text": " their sort of investment strategies with India as a focus. So if you look at the two very successful Japanese companies in India, of course Suzuki stands out. The second one is Daikin. Manufacture for the world with India as a base. These were the two standout companies. But you don't see too many other companies coming. It's not that they don't have problems."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:37",
                    "end": "00:04:03",
                    "text": " The fact of the matter is culturally I might be politically incorrect to make the statement here They are very comfortable working with Southeast Asian countries Okay, so we need to really look at how can we make India a Great destination for their investment not just as a market for also scaling up our manufacturing capability"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:03",
                    "end": "00:04:33",
                    "text": " Vietnam, Thailand, Indonesia, Malaysia are markets where Japanese companies outside of the West have significant exposure to. So that is the primary destination and those were colonies of Japan prior to World War II. Therefore, there is a lot of cultural integration between these two markets. So there is a little bit of history, so we need to unwind that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:33",
                    "end": "00:05:03",
                    "text": " I think our governments have done a fantastic job when Prime Minister visited Japan in August 2025 when they signed the next version of India-Japan journey. There are a lot of commitments made on both sides. Commitments alone do not help. I think commitments need to be followed by investment. Very recently I am seeing a lot of banks investing. You would have also heard"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:03",
                    "end": "00:05:31",
                    "text": " SMBC is invested in a bank in India. Mizzou has also acquired a bank in India and that is just the starting point. Once you get the banking ecosystem starting to invest in another country, then you will find capital flow. So I am very, very hopeful the next five years is possibly going to be the best time for the India-Japan relationship given the struggles that Japan is having with the other Southeast Asian partners."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:31",
                    "end": "00:05:60",
                    "text": " So keep this in mind, there is a history associated with it, and we need to work to unwind some of the things that have happened in the past, and that's possibly why there is still reluctance. Thank you very much. Time flies. Now we have already exceeded the end of the session. Thanks once again to the panelists and the audience. Actually, we have a small gift for the panelists. Dear panelists, could you stand up?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:01",
                    "end": "00:06:41",
                    "text": " And thank you so much for your kind cooperation. And please give them a lot of applause. Now, we would like to close the event. Thank you so much for the participation. And lastly, the company that participated, the panelists from Japan Today, are also exhibiting the Japan Pavilion as a whole for a team. A total seven companies, including those, will not present today. I'll showcase their work. So please feel free to visit if you're interested."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:41",
                    "end": "00:06:55",
                    "text": " This concludes the event, and you may exit at your convenience. Once again, thank you very much for participating today."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Hindi"
},
{
    "_id": "6994077c0d9c668b8501033c",
    "projectId": "362883",
    "projectName": "AI for Education Building Future-Ready Universities and Schools - English",
    "projectDescription": "AI for Education_ Building Future-Ready Universities and Schools - English 2026",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-17T06:15:29.791Z",
    "projectUpdatedDate": "2026-02-17T06:15:29.791Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-17T08:08:15.152Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "AI for Education_ Building Future-Ready Universities and Schools.mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771308917014.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:27",
                    "text": " Quality is the best. Everything about Japan is the best. But as we see the societal issues that we have in Japan and the scale issues that we have in Japan and the technical issues that we have in Japan, it's become much, much clearer in all domains, construction especially, right? That yes, we have a lot of good and important knowledge, but in order to scale that globally and in order to continue growing, we now have to look outside of Japan."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:27",
                    "end": "00:00:54",
                    "text": " And so you see major enterprise companies in my industry everywhere, right? We have the top five general contractor companies in Japan, the mega super general contractor companies, acquiring companies all over the place, whether that be in India, whether that be in places like Singapore. You see all of the construction consultancies expanding their operations overseas and developing up similar to Fujitsu, like research centers and innovation centers outside of Japan as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:54",
                    "end": "00:01:33",
                    "text": " So I think that is a movement that we are seeing. I think that ultimately it becomes a two-way flow. And so we do need to import talent. We do need to make sure that Japan itself is shored up. But on an innovation scale, that connection continues to grow globally. And I think you'll see it continue to keep scaling into billions of dollars of investment. Yeah. Thank you. Okay. Okay. I think you strike the very important point."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:34",
                    "end": "00:02:04",
                    "text": " I think for Japanese, yeah, that is a very important point to look at our challenges. So I think that in Japan, the labor force, we are decreasing the labor force. So we need more talent from India to sustain our industry. So of course, robotics can help us, but I think the real human interface is still needed."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:04",
                    "end": "00:02:33",
                    "text": " I think in that sense, like I said, yeah, we can work together. So I think this is not straight question, but I think, yes, you pointed out a very important point. Thank you very much. I just have a comment. See, I mean, I represent this side of the equation. I've spent about six, seven years in Japan. There is still a very significant gap between"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:33",
                    "end": "00:03:06",
                    "text": " how we see ourselves and how Japan sees India in terms of policy, in terms of risk, in terms of a lot of other factors which are very cultural in nature. In the seven years post-COVID, due to significant challenges in supply chain, they are significantly improved."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:06",
                    "end": "00:03:36",
                    "text": " their sort of investment strategies with India as a focus. So if you look at the two very successful Japanese companies in India, of course Suzuki stands out. The second one is Daikin. Manufacture for the world with India as a base. These were the two standout companies. But you don't see too many other companies coming. It's not that they don't have problems."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:37",
                    "end": "00:04:03",
                    "text": " The fact of the matter is culturally I might be politically incorrect to make the statement here They are very comfortable working with Southeast Asian countries Okay, so we need to really look at how can we make India a Great destination for their investment not just as a market for also scaling up our manufacturing capability"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:03",
                    "end": "00:04:33",
                    "text": " Vietnam, Thailand, Indonesia, Malaysia are markets where Japanese companies outside of the West have significant exposure to. So that is the primary destination and those were colonies of Japan prior to World War II. Therefore, there is a lot of cultural integration between these two markets. So there is a little bit of history, so we need to unwind that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:33",
                    "end": "00:05:03",
                    "text": " I think our governments have done a fantastic job when Prime Minister visited Japan in August 2025 when they signed the next version of India-Japan journey. There are a lot of commitments made on both sides. Commitments alone do not help. I think commitments need to be followed by investment. Very recently I am seeing a lot of banks investing. You would have also heard"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:03",
                    "end": "00:05:31",
                    "text": " SMBC is invested in a bank in India. Mizzou has also acquired a bank in India and that is just the starting point. Once you get the banking ecosystem starting to invest in another country, then you will find capital flow. So I am very, very hopeful the next five years is possibly going to be the best time for the India-Japan relationship given the struggles that Japan is having with the other Southeast Asian partners."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:31",
                    "end": "00:05:60",
                    "text": " So keep this in mind, there is a history associated with it, and we need to work to unwind some of the things that have happened in the past, and that's possibly why there is still reluctance. Thank you very much. Time flies. Now we have already exceeded the end of the session. Thanks once again to the panelists and the audience. Actually, we have a small gift for the panelists. Dear panelists, could you stand up?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:01",
                    "end": "00:06:41",
                    "text": " And thank you so much for your kind cooperation. And please give them a lot of applause. Now, we would like to close the event. Thank you so much for the participation. And lastly, the company that participated, the panelists from Japan Today, are also exhibiting the Japan Pavilion as a whole for a team. A total seven companies, including those, will not present today. I'll showcase their work. So please feel free to visit if you're interested."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:41",
                    "end": "00:06:55",
                    "text": " This concludes the event, and you may exit at your convenience. Once again, thank you very much for participating today."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
}, {
    "_id": "699307a14c5e8f0974c87779",
    "projectId": "563196",
    "projectName": "From AI User to Creator India's Next Innovation Leap ",
    "projectDescription": "From AI User to Creator India's Next Innovation Leap ",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-16T12:03:44.330Z",
    "projectUpdatedDate": "2026-02-16T12:03:44.330Z",
    "projectStatus": "Active",
    "companyId": "69803c46f08f26d9c13921b1",
    "projectCreatedBy": "69803c46f08f26d9c13921b2",
    "projectUpdatedBy": "69803c46f08f26d9c13921b2",
    "projectLastAccessDate": "2026-02-16T12:03:44.330Z",
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "Content",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "From AI User to Creator_ India_s Next Innovation Leap(MP3_128K).mp3",
            "col3": " It went. It went.  For what purpose?  I don't know.  This is the company that's working to present it today. So what do we have?  I have started five minutes for the next session.  Thank you so much for all your help. I work in the U.S. in Colorado.  No.  I'm going to get this.  Ambience is on  I've been told we'll start in two minutes.  Fantastic. It's a miracle you're doing it. That's pretty impressive.  Super excited to welcome everyone. It's a packed room, so very, very happy to get us started. So happy to see so many familiar faces, so many new faces. Welcome to the Asia Summit. It's an honor to host a session during the summit. My name is Manu Chopra, and I head from Karya.  We think deeply about how we use AI to bring dignified learning and learning opportunities. Over the last three years, we've been doing this work. We've been able to engage 240,000 people across every single state in our country. We've been able to engage with them to build foundational data sets.  Make AI models fair and inclusive for all of us. We've been thinking about the AI economy for the last three years as a computer science student from us on Earth. And we realized that the AI economy, the way it is set, the way our models are built, the way companies run them, isn't well-suited for people to look inside. We think deeply about the fact that AI today is simply often not useful enough for the people it's affecting the most.  And today I'm honored to join with two other people on the panel who have not only thought about it, but have actually led India to make better decisions in the AI space. I have Sunehna Ji with me from Microsoft Research. Sunehna and I have known each other for nine years, which is incredible. I also think Kalipa and Dr. Gandhi are amazing.  Dr. Ballier and, of course, two of the star organizers of the AI Summit, and we are so honored to have them here. Kalika, there's a seat for you over there. Of course. And very, very excited to have both Naina and Kalika for Microsoft Research. Many of you may already know this guy was incubated at Microsoft Research, so it's a nine-year partnership, the fruits of which you're going to see today through a project called Smiksha.  I often say India has the most progressive public policy on multilingual AI as a country which has over 19,000 unique dialects. And as Dr. Kalikabali often says, a language is just a dialect with an army, right? Which means that we have at least 19,000 unique subcultures that need to be respected in the AI ecosystem. We need to make sure that our models speak these languages. We need to make sure that the models, as much as possible,  are owned by our communities. And Dr. Nav is also here with us today talking about a very exciting thing we're working on together. So with that, we'll get started with the session. Thank you so much for being so patient. And we'll call on, it's going to be an interactive session, so please feel free to ask questions and we're excited to get started. Thank you. Good morning, everyone, and welcome to the AI Impact Summit. It's so exciting to see your full room here in the morning, and really excited to share some of the work that we've been doing over the last term.  almost a year or so with Karya but of course building on a lot of work that we've been doing at Microsoft Research over the last five years or so on multilingual and multicultural evaluation. So today we are actually showing you some of the work that we've done with Karya called Samiksha. So Samiksha is a Sanskrit word that means collective analysis or evaluation and actually Kalika was the one who came up with this word and so thank you for such a nice name for the project.  So here we're really looking at how communities can be part of the evaluation process itself. And if you've been following some of the AI news, then you know that evaluations have become very big over the last year or so, right? Everyone's talking about building benchmarks. Even in the previous session by Gates Foundation where they were looking at agriculture, they mentioned how important it is to have benchmarks and evaluate for impact and so on, right?  And so we're really looking at it from a community point of view, trying to see how we can involve more people in this evaluation process. So I'll give you a little bit of background about why this is so important to do today, and also describe our efforts in the last few months on this. So I'm not able to see the screen, so I'll turn a little bit. But as we all know, there's been a lot of progress made in the way AI systems work for different languages, cultures, et cetera. But there are still some gaps that remain.  So there's this article that came out a few months ago, which shows that even a basic task like translation sometimes can be challenging for some of these models.  So this is an example of a Punjabi paragraph that we also tried to translate from English to Punjabi and found lots of different errors in terms of how this translation task was done. And so even though we've made a lot of progress, there are still some fundamental problems with these models when it comes to language coverage. And there's a lot of interest in that topic in this summit. And I'm sure you'll hear lots of announcements about new models, better models, and so on.  So what really contributes to this gap? So we talk a lot about data, and rightly so. Data is one of the key factors that contributes to how well a model works in a particular language, culture, and so on. And we know that the representation of data is extremely skewed towards English and Western data sets. But that's not the only thing. Data is not the only reason.  why models don't do well on, say, the global south context of different languages. There are a lot of decisions that are taken during modeling. And also, evaluation plays a key role today in actually determining how well models work for us. And that's because evaluation sends a lot of signals that are used during the model building process. So to give you some examples.  Evaluations basically tell us whether our interventions are working the way they should or not. And this is really critical when we have rapidly evolving AI systems. And it's even more critical when we want to do this for the global south. Because a lot of the AI systems today come with very strong global north assumptions and priors. And so for us to be able to actually tell whether a technology is likely to work or not, we need to be able to do this evaluation well.  Also, since we have limited resources, we may want to use evaluations as a tool to prioritize where we want to invest those resources. And also, most importantly, evaluations create accountability. At least in the short term, they are a signal to show us whether something is working as intended or not. And in the long term, we need to follow that up with longer-term impact evaluations and so on.  So here in this talk, I'm going to be focusing mainly on model-level evaluation. But I also recognize that system evaluation, user evaluation, and impact evaluation is extremely critical to know whether these technologies will really be adopted and whether they will have the outcomes that we want them to have. So what makes AI evaluation, particularly GenAI or LLM evaluation, so challenging today?  As we all know, the outputs of these systems are extremely hard to understand. And so we don't really know what they will say given a response or given a question or a small tweak to the question. You can have a completely different response. And so this makes it really hard to build an evaluation data set with which you can be sure that given a question about a specific topic, you're going to get a certain kind of answer.  The other thing is that evaluation of the outputs of these GNI systems are highly dependent on how you pass, right? So we have hyperparameters that you need to prompt that you need to pass. And so a small begin that can actually tweak the output. And this creates a challenge while doing evaluations. If you want to actually compare multiple systems, they need to be done very carefully with the same setting, right? With the same prompt, same hyperparameters across different models in order to tell which model is better, for example, at a particular task.  The third thing is that a lot of the open-ended problems that we have today don't really have one correct answer. So we were talking about agriculture in the previous session and there, right?  a farmer advisory may not have a single right answer. And for some of these more open-ended, nuanced questions that people ask, like advice-seeking questions, which if you've seen some of the reports by the large model builders, a lot of people actually are talking to these systems asking for advice on their day-to-day lives. There's no one answer. And so actually doing the evaluation process, which usually depends on a ground truth to compare a model output, becomes very challenging in this kind of thing.  Building on top of that, the metrics that we used to use for AI model evaluation earlier, which were really thinking about especially ground truth or the output of classification systems, which were labels, those metrics don't really apply anymore to gene AI kind of evaluation, especially when you're doing these more nuanced metrics. And finally,  systems are getting more and more complex. So with agentic systems and chains of reasoning and so on, it becomes even more complex to design the right evaluation protocols in order to do this. And so because of all these challenges, we end up with what is known as an evaluation deployment gap, where even though the numbers look really good on paper, when you actually try to deploy them in the real world, they fail.  And so we really want to try and close this gap. We know that it can never be perfect, because there are always things that happen in the real world that we cannot anticipate or plan for with these benchmarks. But we want to close the gap as much as possible. And that's the goal of doing good evaluation. And again, the validity of our current evaluation methods has been shown to be less than ideal, has been challenged. So for example, if you looked at this study, this was a study, I think, by Oxford, which  which looked at 400 odd benchmarks that are used for AI model evaluation today and found that they were flawed in some way. Either the data was flawed, or the method that they were using, the metrics were flawed, or the interpretation was flawed. What they interpreted the outcome to be from the study itself was flawed. Similarly, AI agents have been found to cheat on evaluation. So this was a missed report that showed that some of these agents actually look up the answer on the web while the evaluation is going on.  And so they try to cut off the web access when they're doing evaluation and so on. It's a really interesting read. I highly recommend you check it out. Similarly, there have been critiques of benchmarks themselves where, again, as I said, the interpretation of a benchmark score is not valid because the benchmark is very constrained in some way. There's a very, very popular paradigm called the LLM as judge paradigm where an LLM or a JNI model itself does the evaluation for you.  It basically acts like a human. You can tell it what to look for in the output, and it does the evaluation. That has been shown to be neither valid nor reliable, and that is a very, very common way to do evaluation today, both in industry and academia. There's been lots of instances of benchmark hacking, where you can artificially boost your model scores on certain benchmarks. And of course, this was a very interesting paper from the Ohio Group, which showed that even very well-known leaderboards can also have their own issues.  Now, this is just generally about evaluation and how challenging it is. But when you look at the multilingual, multicultural context, which is the kind of context that we are interested in here, things become even more challenging. And there are three main factors for this. One of them is that coverage of many of these benchmarks that exist. First of all, most of them are only for English. There are a few that have been created for other languages as well. There are some Indian language benchmarks as well.  But we only barely scratched the surface. So if you look at some of the popular multi-equal benchmarks like MMLU Pro and so on, those also have only a few languages in them. And most of those languages tend to be the so-called high resource languages, the ones with larger amounts of data in them, usually Western languages, usually rich languages and so on. So there's a very big gap between what these benchmarks measure in terms of coverage and the 7,000 languages that we have in the world today.  The second major challenge is that of representativeness. So the benchmarks that do exist, the multi-equal ones, they are not always representative of how something will be used in that context. Sometimes they are simply translated from English benchmarks. So somebody creates a benchmark in English, and then if you want an indie version, you simply put it through a translator, or you get a human to do the translation. But that is not going to measure well.  how that system is actually going to be used in India, how that model is going to be used in India, how well it can work in the Indian scenario of culture, context, and so on. And this is actually a problem with most benchmarks today that look at multilingual evaluation. And this is something that we've really been trying to change over the last couple of years, that we need to build benchmarks separately for each language, culture, and context, if we really want to be able to tell how well these models will do on those languages, culture, and context.  So this is a change that we've really been pushing for, and I'm really happy to see that some of the newer benchmarks, some of which will be announced during this summit, those are actually taking this on and saying, yes, we are not going to do English-centric benchmarking anymore. Finally, the problem of trust and rigor in the evaluation process. So I mentioned small changes in hyperparameters, small changes in prompts, et cetera.  those can really change the results completely. So it's very important if we are trying to compare multiple models that we are fair across all those models and we keep the conditions the same across all of them. So this is something that we take very seriously. Same thing is the problem of contamination which is when a benchmark is put on the web and then it gets consumed into the training data of models. So if the model has already seen the benchmark before, so that's like you're appearing for an exam.  But you've already seen the question paper the previous day. And so it's not a fair evaluation anymore if the model has already seen that in its training data. This is a big problem for all benchmarks, even in English. But for multilingual, multicultural, given that we have so few benchmarks to begin with, and it's so hard to create good ones, losing them to this kind of contamination becomes really, really challenging. So we've been trying to address all of these challenges. So trying to improve coverage, trying to make benchmarks more representative.  and trying to ensure that there is trust and rigor in the benchmarking process over the last few months where we've been building Samiksha. And before Samiksha, I just want to show you another project that we did with Karya, which was last year, last to last year now, which is called Pariksha. And so that's, again, a word that means exam. And this was actually the very first time that we engaged Karya workers in doing the evaluation process.  So if you're familiar with the kind of data work that the Karya workers do, now evaluation is also something that the Karya workers are able to do. And this was our very first attempt at doing evaluation with the communities in Karya. This was what we released last year as a Kariksha benchmark, where we did an evaluation across 10 different languages, Indian languages, 30 different models at that time. And we showed scores both by human evaluators from Karya as well as automated methods.  So we are also working on automated methods in order to try and make them closer to human evaluation as well, because there is also a need for some automation over here. So with that, I want to introduce Samik Shah, which is what we are all here for. And this is the new benchmark that we've created for Indian languages, contexts, and cultures.  completely from the ground up with community input. I'll be showing you how we constructed the benchmark. And this is being released on the CARIA website. Today, the leaderboard is there. There's a 37-page report if anyone wants to really dig deep in and look at a lot of the technical details. There are lots of interesting examples as well in the report, so I highly encourage you to take a look if you're interested. We are not releasing this benchmark, so the data points we are not going to release because we want to avoid contamination. So until we are done with all the evaluations of all the models that we want to evaluate.  we're not going to be releasing it as open source. So this is what Samiksha is. It covers four domains, 11 Indian languages, so 10 plus Indian English. We benchmarked 17 different models that we had access to so far, and we will do more as more models get released in the next WD. The benchmark comprises of 23,000 data points, so this is one of the largest multi-level benchmarks that's available now.  We did 150,000 human evaluations with Karya. So we broke our record for Pariksha, which was 90,000 evaluations. So 150K evaluations have been done by community members from Karya for this benchmark. And in order to build good automated methods, we also did more than a billion automated evaluations using the LMS judge paradigm that I mentioned earlier. So this is really like a very, very large scale evaluation. So we've gone for breadth in terms of coverage.  We did 11 languages. We want to go to 22 languages and beyond in the next iteration. But we've also done a lot of depth. So we've really analyzed all of the examples. We've done lots of statistical tests. You can find all of that in the report. We've also done a lot of qualitative analysis to look at each data point and see.  why it was rated a certain way, what kind of score it got, what the problem was with the answer, and so on. So we hope that that will really be useful to model developers to try and figure out why these models are doing well or not so well in specific languages and settings. So this is how we built the benchmark. So there were three different phases of our benchmark, and the idea here was really to involve the community in multiple ways in order to construct the benchmark.  The very first way that we reached out to communities is through civil society organizations. So in order to determine what goes into the benchmark, we actually leaned on the knowledge and experience of CSOs that work in specific domains. Some of them already have AI-based solutions in these languages and markets. Some of them are about to deploy. And so they have a really good idea of what kinds of questions people are asking AI monitors today.  And our benchmark has four domains in it, and we are going to expand on the domains in the future as well. Right now, we're focusing on health care, education, legal, and finance. And these are not questions that, say, a lawyer would be asking an AI system, but these are questions from the community. So somebody who has, say, a legal dispute with somebody, like their neighbor or something, is asking a question to an AI model and advice-seeking question, those are the kinds of questions that we are capturing in Samiksha.  So we engaged with 14 different civil society organizations in order to construct this benchmark. And the languages I already mentioned, we had 11 languages in our benchmark. We have 23,000 data points. And we generated response for all these 23,000 data points from 17 different models, both Indic and global. So you can take a look at the leaderboards and so on in the paper. And as I mentioned, we did both.  human evaluations with Karya, but we also do these LLM as judge automated evaluations as well, plus a lot of qualitative analysis and expert evaluation as well. All of those details are then included. So just to flash the domains again, education, finance, health care, and leadership. And here, the critical part which differentiates this benchmark from some of the other benchmarks is that we are not just testing for the accuracy of the answer.  We are also testing for other domains like cultural relevance, whether the user perceives these as being trustworthy or not, what the output quality is in general. And of course, we're also testing for whether the language is correct or not, whether it sounds natural to a native speaker or not when they look at the response from the LLM. So the key findings from Saviksha are that most of the models actually perform well on  both content quality as well as language quality. So the models that we have today, the global models, the frontier models, they're actually doing reasonably well with some exceptions on producing good language, producing coherent language, and also generally answering the question fine. There's nothing wrong with the answer. But the differentiator we find in our evaluations is the cultural relevance. So we actually find that some models do much better at being locally culturally relevant, whereas some of them don't.  And so this we can see very clearly with our evaluation effort. The second thing that we found is that among the domains that we tested, health care proved to be the most challenging domain. And so in fact, education was a very easy domain. Most of the models were able to do well. When I say education, I mean people asking for career advice and so on. Should I do this course? Should I take up this job, et cetera? Those kinds of questions are able to do well. But health care is something that models still struggle with quite a bit.  Yes, so in finance the questions and I'll show you some examples of questions So finance the questions are like should I do this kind? Should I you know invest in this or should I take this loan and things like so these are everyday questions that people would ask advice seeking questions that they would ask and You know the kind of model errors that we find we still find that there are some models that You know that use awkward languages that almost look translated they they translate frozen expressions from English You know these are the things that traders typically rate  low in our benchmark. And finally, as I mentioned, we also did this massive LLM judge or automated evaluation to see whether they agree with human evaluation or not. And our findings mirror the findings that we've had in the last couple of years on this topic is that just using LLM judges or automated methods, not a good idea at all. They diverge a lot from human preferences. But what we've done is we've actually used the human evaluation data from  from all of the CARIA evaluations. And we've used that to actually make the automated methods better. And we find significant improvements in this. So this is definitely a direction that can be explored further, where looking at how we can improve some of these automated methods by using the data that we collect from communities. It's still not a replacement for doing human evaluation. But yes, when we are doing these quick iterations, model building, et cetera, these automated techniques can provide useful signals as well.  So I'm going to end with a few examples. And here, the example is in the Indian language. Then there's a translation. And then at the bottom, it's basically either a high-scoring response or a low-scoring response with a reason why. So let me just read it out. So the first one is about how a student's abilities can be identified and how clear guidance can help. So this is a generic question about education.  And the interesting thing is that the answer that did poorly over here was not really suited to the Indian context. So the answer talked about psychometric assessments, job shadowing, industry visits, internships, and so on. And so the person who evaluated who was also a member of the community did not connect with this answer because it didn't seem relevant to them. So these are the kinds of differences we see between models that do well and not.  And similarly, on the right, if you look at the question is about how government schools are different from private schools and how to decide whether a child should go to the government school or not. And here, a high-scoring response actually managed to compare government and private schools on different dimensions and also suggested some actionable advice. So go and visit it and so on.  to try and assess the quality. So here, one more thing that we're really looking for is how trustworthy are some of these answers, right? Is somebody actually going to take an action based on the response that they give or not, right? So these are the kinds of responses that we got, which the users rated as the answer. This is a good response. I would actually trust this kind of response. Couple of other examples here. So on the left, we see a question in Megali. And one of the interesting things we found was that  Many of the Bengali responses actually assume that we are in the Bangladesh context. And to be fair to the models, we don't specify what country we're in. So we don't expect that the model can tell that this is from an Indian context and not from a Bangladeshi context. So this is something to keep in mind that probably next time we should do this. We should prompt the model with the location as well for these kinds of answers. But what was interesting is that  In some cases, the model assumed that it was a Bangladeshi context. But in other cases, it actually fixes the two contexts. And so it's referencing the description from both India and Bangladesh in the same answer, which causes a lot of confusion. So this is a low-scoring answer for us.  And here, there's another question about buying houses. And here, the high-scoring answer basically talked about all sorts of risks and concerns, including fraud and illegal transactions and so on, and also referenced India-specific processes which the human evaluator found very valuable. So this is really what we're testing for in Samiksha. Are these responses something that community members can connect with?  And to do this, Karya, of course, provides them a lot of training. You can go into more detail during the discussion. But we're really trying to empower our community to decide themselves whether something works for them or not. So they decide what the benchmark should have. They create the actual benchmark data points themselves in each language, in each culture, domain, separately. And then they also do the evaluations as well. So they're really determining whether India's models work for them or not.  So I'll end here with the link to the page that has the report as well as the leaderboards for you to look at. Thank you so much and I'm happy to take questions now or later.  Yes, but let me introduce one more concept and then we'll take a couple of questions. Thank you so much. So, thinking about redesigning AI economy, the work that Selena just presented on Saniksha, is so important. Because unless we center our communities in the way we evaluate AI models, we will never know what works for us. And so often in the tech sector, the way we talk about our communities can be slightly critical.  I remember one and a half, two years ago, we had to convince our clients that our communities can do a complex evaluation. As of that thing, we've convinced it. I think what the mixture shows is the communities we serve at Taria are not just excellent beneficiaries of AI models. They're not just passive beneficiaries of AI revolution. They're active builders. They're active evaluators. They have to be centered to make sure these models work. So Naina, you talked about one of the reasons these biases exist.  And one of the things that the Roshni team has done at Government of India is completely change who gets to contribute to the AI world. It's an absolute honor to have Amitabh Sir with us today. For people in this sector, I think Amitabh Sir is the biggest champion. I think so much of the work we do at Taria would not be possible without his support and his championship. So with no further ado, Amitabh Sir, I would love for you to talk about the second thing we're launching today, which is Samudar.  So actually, given that you're so unique in the room, I would love you to talk about what are the goals behind Bhashini, what separates Bhashini from anything else in the world, and why are you excited about what we're doing with some of that. So yeah, you know, taking the cue from what you have talked about, we're getting into the background. You know, we started as Bhashini, which is Basha Interface for India, about three and a half years back.  and we had five basic problems to solve. Obviously over a period of time, the technology has changed and obviously many things have happened. But currently, the building of those models, what was the challenge we had, was that we had about six to seven languages occupying  the most of the digital data which was available, while as rest of the languages did not have any digital data. And even the six to seven languages which were there, which had digital data, they were approximately, you know, about 2% or 3%, you know, anybody can guess, that amount of digital data which was available and of the entire digital data, you know, 3% of the entire digital data.  which is available on the internet. These languages are occupying such a low space. And when we were trying to work this out, we obvious, hence it was becoming difficult because how will you build the model if there is no digital data? Because AI models are built on digital data. So we had a unique exercise which we carried out to collect the data by brute force. We collected digital data through brute force.  We had a, you know, application which had a certain set of images, places, food, items, etc. which was taken onto the field. And there were about 200 plus people who were hired to let people speak about the picture which was available on the mobile.  and they speak, they transcribe, they translate and they move all of that which is there to build the digital data. With that we built the Morpus which was barely sufficient to build the models and obviously we could have waited for very long to actually see that these models have enough data to become 100% accurate and then we moved on to the field.  But we decided that we should actually look at how do we go to the field so that we kick off the improvement cycle which is so essential. So we built some of the use cases which, you know, narrow use cases which, you know, started benefiting people. You know, we are handled releases today. The top eight use cases which we are working upon. We are, you know, having a use case which is...  touching the panchayats today, 270,000 panchayats, where the idea was that we should bring the transparency in the way the meetings are organized in the panchayats. The panchayats essentially work on local languages and the minutes are actually created in English because the central ministry understands more English than any other place.  there was an intervention by the secretary or such a who will create that particular meetings and then obviously there are multiple interventions which will actually bring about lot of their own thought process, their own way of thinking rather than bringing the people's requirement into minutes. So now the minutes are actually recorded and they are sent to  The second use case which we are in the midst of launching  There are two of them. One of them is about agriculture in Maharashtra, where the farmer gets the advisory on his own language through a voice journey. So it's a chatbot which you can ask, the farmer can ask, what do I do about this? And the answer through a voice journey comes out that these are the things which need to be done.  Now that system when we were implementing, the first system when we were implementing about Sabha Saar, we started looking at again from a perspective of what is the language coming up and the schools which are provided on what are the feedback etc. which is there and where are we not able to capture dialects, where are we not able to capture the words which are specific to that particular place. But in the case of Mahavistar, we had a challenge which was  much different, which was basically saying that, you know, the agriculture taxonomy and agriculture nomenclatures were not available in the model because the model hadn't seen it and it's approximately 70,000 numbers and they were not pronounced correctly when it was basically, you know, talked about or the answers which were made.  So we had to go into a specific drive of involving the agriculture students to actually collect those utterances to train them on so that things start moving fine in that direction. So that's another intervention which we worked upon. The third intervention was, of course, where we touch every citizen across the country is UIDN, where the place names are translated and other names are translated.  There there is a unique challenge of number of places which are available. That is the digital dictionary of the place which is available in the system is actually very limited. When we digitized the panchayat, we had 270,000 panchayat names brought to the digital system. Now we have a second challenge is that the  actual number of names of places which the Survey of India has is approximately 16 lakhs to 18 lakhs and currently those are not in the digital sector and so we have to bring that back and hence we are working with Survey of India and then again those places have to be on the local dialect because those places are known on the local dialect to where the people eat there.  And the idea of creating that particular database is to, perhaps a person go there, that's what Survey of India would be doing, is a person go there, this geofence, and then we ask someone from that particular location to actually speak. And that gets translated, you know, through automatic speech recognition, that gets translated and put into the database. So, just to, why I gave this example is basically...  reflect upon that if we are wanting to build a large system, the value chain starts at the last mile. And the value chain of AI, which is basically related to data collection, then annotation and labeling, then it goes to model building, then you are vetting.  and then you are using it in application and then coming back to a feedback loop. So this is the AI value chain and then you have the community which is basically involved or the user who is involved in the work. If we start doing that evaluation sitting at, you know, any of our metros, the evaluation would be very different, right? So we will be starting to discuss whether biology should be represented as chiefs of three in Hindi.  or, you know, biology should be biology there. So, you know, that is the debate which we will have if we are doing that in Delhi. If, you know, I was basically doing a speech-to-speech demonstration in Gujarat and I had the privilege of having Honorable Minister of Mighty there listening to that, and then obviously there were other officers there. I wouldn't like to name the sake of this thing. The officers had an impression that...  This was about 80 to 85 percent correct. Whereas the Honorable Minister said that it's so pure that I have started learning. So I'm basically again trying to say that the perception of people would differ when you are trying to evaluate that. Some people think that, oh, the machine is doing so much. Let's go with it. Some people will say that, no, it's not still at that level. So don't go with it.  But at the end of this that's what I'm taking the community feedback. I'm not The officers said it is 80% the other people said the minister said that it's too good to be true. That's the way it is Now again, the fact comes across that when you are doing all of this and deploying all these use cases and working on this It is at the end of the day  The system is as good as the users need it, the people say. And if this people, it is not necessary that we should be looking if some of our people went to Akaru in Andhra Pradesh and to demonstrate the product. And actually, Honorable CM wanted us to do that. Because there are only two people who understand English there, the collector and his secretary. Rest, all are taken.  We sent a person who knew Madhyalam and some amount of Hindi and English. His survival was actually the app which we had built, the Vajni app. Otherwise, he wouldn't have been able to do anything. So if you go 50 to 100 kilometers away from a metro, you start actually seeing what is the language divide. And that is where the evaluation of the system matters. So the work which we are doing with Korea and we are trying to do this, which they have done a good job when we were  we started all of this was about creating those communities at the grassroots level to actually see how, you know, they are there. Often, you know, we are working with a situation where we are currently looking at how do we actually, you know, build a data working, data workers as such. That's basically the data collection people from whom we can collect data. Should we pay? Should we not pay? How do we pay?  et cetera are the things which are there. But first of all, we need to build a community which is actually going to aggregate all of this who are willing to act as a data worker across the value chain of what we are trying to do. You know, the language AI value chain. And that is where we have started an initiative called community so that we are at least able to...  recognize that these people are there they are looking at it we still do not have a name for them we are basically using the name which is combination of phases for this but still there is a reason for us to believe that these people would be the people who are there often i say that you know number one is that whatever in ai is very has a self-life of six months you can keep talking about  better today and better tomorrow. It's about six months, right? Because in six months somebody else will collect the data or build the things which are much better than this. So your objective is not to actually chase that. Your objective is how do I satisfy the customer for which the use case has been developed. So, you know, obviously there is a backend technology. There is an improvement which is required in the backend technology, but using that is not going to help us.  But if you build a community which is basically going to do that, we perhaps will be unleashing, over a period of time, a marketplace or an economy which is actually going to be, you know, self-sustaining in terms of what AI can do and how AI can be better. So that's the core point which I have. Fantastic. And I think, thank you so much. We are super excited about Samudai, which we are launching with Bhashini. Samudai, again, is Sanskrit for community.  We love Sanskrit words, let's start with this. You've seen the pattern here, with both Savitra and Samudaya. We're also launching Sammati at 4.13, so this is three for three today. But very, very excited about Samudaya. Sir talked about a project we wrote with government of Maharashtra, and that to me is the best example of how do we center our communities in the AI magnitude, right? Because in number one, tribal district in Maharashtra, the data collection for the local model is being done in the district.  by the communities. The model building, the technology building is being done in the district. And everything will be owned by the district and by the communities. And the evaluation will also be done by the same communities, like we did with Sameeksha. So we think about the whole AI value chain.  The goal to redesign the AI economy is to centralize our communities throughout the world. I would love to bring in Fezl. Fezl is one of the rock stars at Collective Intelligence Project CIP, cares deeply about evals, cares deeply about doing this globally, founding partner at Samiksha. Fezl, could you tell us about how you plan to take rubrics like Samiksha to other countries and how do we bring  community benchmarks from India to looking this up. Like Manu said, my name is Faisal. I'm the Head of World Partnerships at the Collective Intelligence Project. And a lot of the work that this fantastic team has done at Persimiksha is setting the precedent for a lot of the work we want to do in a lot of other countries. So the idea behind this really is the community-driven pipeline. The idea of community-driven AI governance, I think, is something very popular in a lot of different countries.  So one is how you communicate with representatives of particular domains, particular communities. Two is how you get both experts and non-experts involved in the validation of them. And then three is how do you keep this as a loop? How do you make it replicable? How do you make it scalable? And I think everything that's been done with Sumiksha is an example of that. So one big question that we're always trying to ask is...  What do AI evaluations capture right now? And what do community-driven pipelines offer that those do not? So we talked about depth, which is something that evaluations are always good at. But what community-driven pipelines do is capture breadth, contextuality, and variability. And that is just more efficient, not just morally superior, but just generally more efficient than other types of pipelines.  So what we're trying to do with the Collective Intelligence Project is make evaluations in general more accessible for everyone. We've set up platforms like WeVal, which you can go to at weval.com, in which teachers in Uganda, in doctors in Memphis, they can go on and create their own AI evaluations. They can actually take criteria from their own experience and make it something rigorous, make it something credible that AI labs and policymakers can use.  So just to end things here, everything that's happening with Simiksha is the precedent for community-driven governance that we want to see in a lot of other places. How do you build communities of practice when you're holding AI out? Thank you. Thank you so much for that, Faisal. I think, as we've seen with Samudai today, I wasn't able to get the internet functioning to get the Samudai demo on. But you can actually go to partially in Samudai.  It's D-H-A-S-H-I-N-I-S-A-M-U-D-M-I-E. And check it out. And if you're sticking around for the whole summit, you can come to the Bhashni group. And we're actually demoing some of the material. On 19th, we actually have a panel with Amitabh Sir and Mithali Ji, who's the district of the Nandabhar district. And from her perspective, you'll actually get to learn about what does community-centered EI mean. For her, I also want to call out Shannon Farley, who's in the room.  I think many of us know and love Shannon, but Shannon runs Fast Forward, one of the most prolific AI for Good investors in the world. Shannon, I want to quickly bring you in. Could you talk to us about, well, first of all, the first checks into Tarya, into CIB, into Dura Health, into Rocket Learning, so many organizations that have been highlighted at this summit today. When you think about redesigning the AI, you will immediately think about AI for Good. Are there some ideas that you find interesting that are coming from India, and then again should go to Dura?  Thank you so much for having me. Hi, I'm Shannon Farley. I run Fast Forward, which is an accelerator for tech nonprofits. We've had over 100 organizations go through our accelerator, so many Indian organizations over the last few years. And I think it's in part because of this connection to community. We're seeing Indian tech nonprofits leapfrog in AI. There's a deal.  There's an infrastructure here that has been here for a long time. There's lots of technical talent. And there is a deep need to integrate different voices into the models because there are so many people you're serving. It's really a model for what we want to see in the rest of the world. The other piece that's important is these organizations need money to operate. And India has a homegrown philanthropy that is making a huge difference.  Organizations like The Nudge, we partner with them. Organizations like ACT Grants, we partner with them. Have been deceived and follow on funding to these organizations. And when that happens, they can scale. So one of the things I hope comes out of this summit and the incredible work of CARIA and CIP and Rocket Learning and Nora Health is that in other regions of the world they get inspired by both local talent  great entrepreneurship and philanthropy to see what they can build. We're starting to see it in Colombia with the Velas Reyes Foundation. That has spurred an ecosystem of tech nonprofits, and we would love to see it in other parts of the world. So we thank you for your models, both AI and human, and we're excited to meet you. Thank you.  That was transcription, that was translation.  How do communities in the country find digital work opportunities directly from government and non-government sources?  who have done that work before being directly engaged. This one? We're very, very excited about launching both Samsung and Samsung XI. We have five minutes for questions. Thank you. I just have one quick question. So just I have a little bit of background.  I think I'm using the same people to do the evaluations.  And you prove that has any kind of impact on the person. Because the first time they're doing it as if they would be using that language themselves, they might not see that pattern in the end, right? Maybe the world looks at that and changes. I might not be the right person to do the evaluation for you of the world.  Do we get your thing ready?  Hello? Honey, how about that? Ha ha, session is over. Kahane?  Where exactly? What's so many people?  Fantastic. Yeah, so to answer that question, because it's an easier one, is to work with healthcare professionals, right? I mean, that's clear ones who know the domain, and they will be able to answer all those questions. So again, working with communities where it is actually healthcare. To answer your question, I think that's a very interesting one. So one thing is, of course, that Karya does extensive training for the workers to...  tell them what this evaluation actually means. For all of the different metrics and projects I mentioned, what the meaning is of a good language for a person, what trust means, what cultural relevance means. So there's a lot of examples given, a lot of training given in their own language, by language coordinators. So there's a lot of work there. And this is also done in partnership with the CSOs. So the CSOs have an input of how the data workers get trained for both.  create the data points, but also to evaluate them. So we do a lot of work on that. But having said that, there is variation also in these evaluation responses. So for some of the data points, we actually got two different evaluators to the same data point. And we measured, again, the agreement between the annotators and so on. So all those details are there in your report. Thank you so much. We'll add that, unfortunately. But we'll all be outside to answer the questions. I will end the conflict game. If you want to speak to the car yet.  We have 28 other panels we're doing across the next four days. So please come find us. If you are very interested in working with me, I'll send you a list of the panels that you can come to. We also have three booths. We're doing an incredible work outside India, across Kenya, South Africa, Ethiopia, the US.  Indonesia, so please check out the MasterCard group. We did a fantastic group with Bhasini, which we're very, very happy to do. We're showcasing our work on Samudaya. We actually try out Samudaya and try out all the ways we can do this. And finally, we have an exciting group with an India AI mission. We're showcasing, and that's right next door at call number 14. We're showcasing and releasing our datasets today. It's very, very exciting. Thank you so much for this session. And thank you so much to my terrible panelists.  You have the copy of the file, right? If someone asks for it from the Karya team, you don't have it. Which team? The organizing team. Okay, if someone asks for it from the organizing team, do you want to give it or not? Do you want to give it or not? No, I don't want to give it.  Thank you.  Good morning everyone.  So the next session is going to start. So I request everyone to please clear the stage and allow us to start the session on time.  This is different.  What's up?  So request all the speakers for the Preda Sapien event to be in the front row please. Good morning everyone. On behalf of Preda Sapien, I am happy to welcome you all for the event.  Uh...",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771243400240.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:02",
                    "end": "00:00:59",
                    "text": " It went. It went."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:01",
                    "end": "00:01:23",
                    "text": " For what purpose?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:23",
                    "end": "00:01:52",
                    "text": " I don't know."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:52",
                    "end": "00:02:05",
                    "text": " This is the company that's working to present it today. So what do we have?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:05",
                    "end": "00:02:28",
                    "text": " I have started five minutes for the next session."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:28",
                    "end": "00:02:58",
                    "text": " Thank you so much for all your help. I work in the U.S. in Colorado."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:58",
                    "end": "00:03:21",
                    "text": " No."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:21",
                    "end": "00:03:51",
                    "text": " I'm going to get this."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:51",
                    "end": "00:04:22",
                    "text": " Ambience is on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:22",
                    "end": "00:05:02",
                    "text": " I've been told we'll start in two minutes."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:22",
                    "end": "00:05:57",
                    "text": " Fantastic. It's a miracle you're doing it. That's pretty impressive."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:57",
                    "end": "00:06:12",
                    "text": " Super excited to welcome everyone. It's a packed room, so very, very happy to get us started. So happy to see so many familiar faces, so many new faces. Welcome to the Asia Summit. It's an honor to host a session during the summit. My name is Manu Chopra, and I head from Karya."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:12",
                    "end": "00:06:37",
                    "text": " We think deeply about how we use AI to bring dignified learning and learning opportunities. Over the last three years, we've been doing this work. We've been able to engage 240,000 people across every single state in our country. We've been able to engage with them to build foundational data sets."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:37",
                    "end": "00:07:07",
                    "text": " Make AI models fair and inclusive for all of us. We've been thinking about the AI economy for the last three years as a computer science student from us on Earth. And we realized that the AI economy, the way it is set, the way our models are built, the way companies run them, isn't well-suited for people to look inside. We think deeply about the fact that AI today is simply often not useful enough for the people it's affecting the most."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:07",
                    "end": "00:07:32",
                    "text": " And today I'm honored to join with two other people on the panel who have not only thought about it, but have actually led India to make better decisions in the AI space. I have Sunehna Ji with me from Microsoft Research. Sunehna and I have known each other for nine years, which is incredible. I also think Kalipa and Dr. Gandhi are amazing."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:32",
                    "end": "00:08:01",
                    "text": " Dr. Ballier and, of course, two of the star organizers of the AI Summit, and we are so honored to have them here. Kalika, there's a seat for you over there. Of course. And very, very excited to have both Naina and Kalika for Microsoft Research. Many of you may already know this guy was incubated at Microsoft Research, so it's a nine-year partnership, the fruits of which you're going to see today through a project called Smiksha."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:01",
                    "end": "00:08:26",
                    "text": " I often say India has the most progressive public policy on multilingual AI as a country which has over 19,000 unique dialects. And as Dr. Kalikabali often says, a language is just a dialect with an army, right? Which means that we have at least 19,000 unique subcultures that need to be respected in the AI ecosystem. We need to make sure that our models speak these languages. We need to make sure that the models, as much as possible,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:26",
                    "end": "00:09:14",
                    "text": " are owned by our communities. And Dr. Nav is also here with us today talking about a very exciting thing we're working on together. So with that, we'll get started with the session. Thank you so much for being so patient. And we'll call on, it's going to be an interactive session, so please feel free to ask questions and we're excited to get started. Thank you. Good morning, everyone, and welcome to the AI Impact Summit. It's so exciting to see your full room here in the morning, and really excited to share some of the work that we've been doing over the last term."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:14",
                    "end": "00:09:42",
                    "text": " almost a year or so with Karya but of course building on a lot of work that we've been doing at Microsoft Research over the last five years or so on multilingual and multicultural evaluation. So today we are actually showing you some of the work that we've done with Karya called Samiksha. So Samiksha is a Sanskrit word that means collective analysis or evaluation and actually Kalika was the one who came up with this word and so thank you for such a nice name for the project."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:42",
                    "end": "00:10:11",
                    "text": " So here we're really looking at how communities can be part of the evaluation process itself. And if you've been following some of the AI news, then you know that evaluations have become very big over the last year or so, right? Everyone's talking about building benchmarks. Even in the previous session by Gates Foundation where they were looking at agriculture, they mentioned how important it is to have benchmarks and evaluate for impact and so on, right?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:11",
                    "end": "00:10:41",
                    "text": " And so we're really looking at it from a community point of view, trying to see how we can involve more people in this evaluation process. So I'll give you a little bit of background about why this is so important to do today, and also describe our efforts in the last few months on this. So I'm not able to see the screen, so I'll turn a little bit. But as we all know, there's been a lot of progress made in the way AI systems work for different languages, cultures, et cetera. But there are still some gaps that remain."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:41",
                    "end": "00:10:50",
                    "text": " So there's this article that came out a few months ago, which shows that even a basic task like translation sometimes can be challenging for some of these models."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:50",
                    "end": "00:11:16",
                    "text": " So this is an example of a Punjabi paragraph that we also tried to translate from English to Punjabi and found lots of different errors in terms of how this translation task was done. And so even though we've made a lot of progress, there are still some fundamental problems with these models when it comes to language coverage. And there's a lot of interest in that topic in this summit. And I'm sure you'll hear lots of announcements about new models, better models, and so on."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:16",
                    "end": "00:11:40",
                    "text": " So what really contributes to this gap? So we talk a lot about data, and rightly so. Data is one of the key factors that contributes to how well a model works in a particular language, culture, and so on. And we know that the representation of data is extremely skewed towards English and Western data sets. But that's not the only thing. Data is not the only reason."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:40",
                    "end": "00:12:02",
                    "text": " why models don't do well on, say, the global south context of different languages. There are a lot of decisions that are taken during modeling. And also, evaluation plays a key role today in actually determining how well models work for us. And that's because evaluation sends a lot of signals that are used during the model building process. So to give you some examples."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:03",
                    "end": "00:12:28",
                    "text": " Evaluations basically tell us whether our interventions are working the way they should or not. And this is really critical when we have rapidly evolving AI systems. And it's even more critical when we want to do this for the global south. Because a lot of the AI systems today come with very strong global north assumptions and priors. And so for us to be able to actually tell whether a technology is likely to work or not, we need to be able to do this evaluation well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:28",
                    "end": "00:12:51",
                    "text": " Also, since we have limited resources, we may want to use evaluations as a tool to prioritize where we want to invest those resources. And also, most importantly, evaluations create accountability. At least in the short term, they are a signal to show us whether something is working as intended or not. And in the long term, we need to follow that up with longer-term impact evaluations and so on."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:51",
                    "end": "00:13:14",
                    "text": " So here in this talk, I'm going to be focusing mainly on model-level evaluation. But I also recognize that system evaluation, user evaluation, and impact evaluation is extremely critical to know whether these technologies will really be adopted and whether they will have the outcomes that we want them to have. So what makes AI evaluation, particularly GenAI or LLM evaluation, so challenging today?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:14",
                    "end": "00:13:38",
                    "text": " As we all know, the outputs of these systems are extremely hard to understand. And so we don't really know what they will say given a response or given a question or a small tweak to the question. You can have a completely different response. And so this makes it really hard to build an evaluation data set with which you can be sure that given a question about a specific topic, you're going to get a certain kind of answer."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:38",
                    "end": "00:14:07",
                    "text": " The other thing is that evaluation of the outputs of these GNI systems are highly dependent on how you pass, right? So we have hyperparameters that you need to prompt that you need to pass. And so a small begin that can actually tweak the output. And this creates a challenge while doing evaluations. If you want to actually compare multiple systems, they need to be done very carefully with the same setting, right? With the same prompt, same hyperparameters across different models in order to tell which model is better, for example, at a particular task."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:08",
                    "end": "00:14:19",
                    "text": " The third thing is that a lot of the open-ended problems that we have today don't really have one correct answer. So we were talking about agriculture in the previous session and there, right?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:19",
                    "end": "00:14:47",
                    "text": " a farmer advisory may not have a single right answer. And for some of these more open-ended, nuanced questions that people ask, like advice-seeking questions, which if you've seen some of the reports by the large model builders, a lot of people actually are talking to these systems asking for advice on their day-to-day lives. There's no one answer. And so actually doing the evaluation process, which usually depends on a ground truth to compare a model output, becomes very challenging in this kind of thing."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:47",
                    "end": "00:15:08",
                    "text": " Building on top of that, the metrics that we used to use for AI model evaluation earlier, which were really thinking about especially ground truth or the output of classification systems, which were labels, those metrics don't really apply anymore to gene AI kind of evaluation, especially when you're doing these more nuanced metrics. And finally,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:08",
                    "end": "00:15:32",
                    "text": " systems are getting more and more complex. So with agentic systems and chains of reasoning and so on, it becomes even more complex to design the right evaluation protocols in order to do this. And so because of all these challenges, we end up with what is known as an evaluation deployment gap, where even though the numbers look really good on paper, when you actually try to deploy them in the real world, they fail."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:32",
                    "end": "00:16:03",
                    "text": " And so we really want to try and close this gap. We know that it can never be perfect, because there are always things that happen in the real world that we cannot anticipate or plan for with these benchmarks. But we want to close the gap as much as possible. And that's the goal of doing good evaluation. And again, the validity of our current evaluation methods has been shown to be less than ideal, has been challenged. So for example, if you looked at this study, this was a study, I think, by Oxford, which"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:03",
                    "end": "00:16:32",
                    "text": " which looked at 400 odd benchmarks that are used for AI model evaluation today and found that they were flawed in some way. Either the data was flawed, or the method that they were using, the metrics were flawed, or the interpretation was flawed. What they interpreted the outcome to be from the study itself was flawed. Similarly, AI agents have been found to cheat on evaluation. So this was a missed report that showed that some of these agents actually look up the answer on the web while the evaluation is going on."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:32",
                    "end": "00:17:00",
                    "text": " And so they try to cut off the web access when they're doing evaluation and so on. It's a really interesting read. I highly recommend you check it out. Similarly, there have been critiques of benchmarks themselves where, again, as I said, the interpretation of a benchmark score is not valid because the benchmark is very constrained in some way. There's a very, very popular paradigm called the LLM as judge paradigm where an LLM or a JNI model itself does the evaluation for you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:00",
                    "end": "00:17:31",
                    "text": " It basically acts like a human. You can tell it what to look for in the output, and it does the evaluation. That has been shown to be neither valid nor reliable, and that is a very, very common way to do evaluation today, both in industry and academia. There's been lots of instances of benchmark hacking, where you can artificially boost your model scores on certain benchmarks. And of course, this was a very interesting paper from the Ohio Group, which showed that even very well-known leaderboards can also have their own issues."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:32",
                    "end": "00:18:01",
                    "text": " Now, this is just generally about evaluation and how challenging it is. But when you look at the multilingual, multicultural context, which is the kind of context that we are interested in here, things become even more challenging. And there are three main factors for this. One of them is that coverage of many of these benchmarks that exist. First of all, most of them are only for English. There are a few that have been created for other languages as well. There are some Indian language benchmarks as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:01",
                    "end": "00:18:30",
                    "text": " But we only barely scratched the surface. So if you look at some of the popular multi-equal benchmarks like MMLU Pro and so on, those also have only a few languages in them. And most of those languages tend to be the so-called high resource languages, the ones with larger amounts of data in them, usually Western languages, usually rich languages and so on. So there's a very big gap between what these benchmarks measure in terms of coverage and the 7,000 languages that we have in the world today."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:31",
                    "end": "00:18:56",
                    "text": " The second major challenge is that of representativeness. So the benchmarks that do exist, the multi-equal ones, they are not always representative of how something will be used in that context. Sometimes they are simply translated from English benchmarks. So somebody creates a benchmark in English, and then if you want an indie version, you simply put it through a translator, or you get a human to do the translation. But that is not going to measure well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:56",
                    "end": "00:19:24",
                    "text": " how that system is actually going to be used in India, how that model is going to be used in India, how well it can work in the Indian scenario of culture, context, and so on. And this is actually a problem with most benchmarks today that look at multilingual evaluation. And this is something that we've really been trying to change over the last couple of years, that we need to build benchmarks separately for each language, culture, and context, if we really want to be able to tell how well these models will do on those languages, culture, and context."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:24",
                    "end": "00:19:48",
                    "text": " So this is a change that we've really been pushing for, and I'm really happy to see that some of the newer benchmarks, some of which will be announced during this summit, those are actually taking this on and saying, yes, we are not going to do English-centric benchmarking anymore. Finally, the problem of trust and rigor in the evaluation process. So I mentioned small changes in hyperparameters, small changes in prompts, et cetera."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:48",
                    "end": "00:20:16",
                    "text": " those can really change the results completely. So it's very important if we are trying to compare multiple models that we are fair across all those models and we keep the conditions the same across all of them. So this is something that we take very seriously. Same thing is the problem of contamination which is when a benchmark is put on the web and then it gets consumed into the training data of models. So if the model has already seen the benchmark before, so that's like you're appearing for an exam."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:16",
                    "end": "00:20:45",
                    "text": " But you've already seen the question paper the previous day. And so it's not a fair evaluation anymore if the model has already seen that in its training data. This is a big problem for all benchmarks, even in English. But for multilingual, multicultural, given that we have so few benchmarks to begin with, and it's so hard to create good ones, losing them to this kind of contamination becomes really, really challenging. So we've been trying to address all of these challenges. So trying to improve coverage, trying to make benchmarks more representative."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:45",
                    "end": "00:21:10",
                    "text": " and trying to ensure that there is trust and rigor in the benchmarking process over the last few months where we've been building Samiksha. And before Samiksha, I just want to show you another project that we did with Karya, which was last year, last to last year now, which is called Pariksha. And so that's, again, a word that means exam. And this was actually the very first time that we engaged Karya workers in doing the evaluation process."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:10",
                    "end": "00:21:40",
                    "text": " So if you're familiar with the kind of data work that the Karya workers do, now evaluation is also something that the Karya workers are able to do. And this was our very first attempt at doing evaluation with the communities in Karya. This was what we released last year as a Kariksha benchmark, where we did an evaluation across 10 different languages, Indian languages, 30 different models at that time. And we showed scores both by human evaluators from Karya as well as automated methods."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:40",
                    "end": "00:22:01",
                    "text": " So we are also working on automated methods in order to try and make them closer to human evaluation as well, because there is also a need for some automation over here. So with that, I want to introduce Samik Shah, which is what we are all here for. And this is the new benchmark that we've created for Indian languages, contexts, and cultures."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:01",
                    "end": "00:22:31",
                    "text": " completely from the ground up with community input. I'll be showing you how we constructed the benchmark. And this is being released on the CARIA website. Today, the leaderboard is there. There's a 37-page report if anyone wants to really dig deep in and look at a lot of the technical details. There are lots of interesting examples as well in the report, so I highly encourage you to take a look if you're interested. We are not releasing this benchmark, so the data points we are not going to release because we want to avoid contamination. So until we are done with all the evaluations of all the models that we want to evaluate."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:31",
                    "end": "00:23:01",
                    "text": " we're not going to be releasing it as open source. So this is what Samiksha is. It covers four domains, 11 Indian languages, so 10 plus Indian English. We benchmarked 17 different models that we had access to so far, and we will do more as more models get released in the next WD. The benchmark comprises of 23,000 data points, so this is one of the largest multi-level benchmarks that's available now."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:01",
                    "end": "00:23:31",
                    "text": " We did 150,000 human evaluations with Karya. So we broke our record for Pariksha, which was 90,000 evaluations. So 150K evaluations have been done by community members from Karya for this benchmark. And in order to build good automated methods, we also did more than a billion automated evaluations using the LMS judge paradigm that I mentioned earlier. So this is really like a very, very large scale evaluation. So we've gone for breadth in terms of coverage."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:31",
                    "end": "00:23:49",
                    "text": " We did 11 languages. We want to go to 22 languages and beyond in the next iteration. But we've also done a lot of depth. So we've really analyzed all of the examples. We've done lots of statistical tests. You can find all of that in the report. We've also done a lot of qualitative analysis to look at each data point and see."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:49",
                    "end": "00:24:15",
                    "text": " why it was rated a certain way, what kind of score it got, what the problem was with the answer, and so on. So we hope that that will really be useful to model developers to try and figure out why these models are doing well or not so well in specific languages and settings. So this is how we built the benchmark. So there were three different phases of our benchmark, and the idea here was really to involve the community in multiple ways in order to construct the benchmark."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:16",
                    "end": "00:24:42",
                    "text": " The very first way that we reached out to communities is through civil society organizations. So in order to determine what goes into the benchmark, we actually leaned on the knowledge and experience of CSOs that work in specific domains. Some of them already have AI-based solutions in these languages and markets. Some of them are about to deploy. And so they have a really good idea of what kinds of questions people are asking AI monitors today."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:42",
                    "end": "00:25:10",
                    "text": " And our benchmark has four domains in it, and we are going to expand on the domains in the future as well. Right now, we're focusing on health care, education, legal, and finance. And these are not questions that, say, a lawyer would be asking an AI system, but these are questions from the community. So somebody who has, say, a legal dispute with somebody, like their neighbor or something, is asking a question to an AI model and advice-seeking question, those are the kinds of questions that we are capturing in Samiksha."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:10",
                    "end": "00:25:36",
                    "text": " So we engaged with 14 different civil society organizations in order to construct this benchmark. And the languages I already mentioned, we had 11 languages in our benchmark. We have 23,000 data points. And we generated response for all these 23,000 data points from 17 different models, both Indic and global. So you can take a look at the leaderboards and so on in the paper. And as I mentioned, we did both."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:36",
                    "end": "00:26:03",
                    "text": " human evaluations with Karya, but we also do these LLM as judge automated evaluations as well, plus a lot of qualitative analysis and expert evaluation as well. All of those details are then included. So just to flash the domains again, education, finance, health care, and leadership. And here, the critical part which differentiates this benchmark from some of the other benchmarks is that we are not just testing for the accuracy of the answer."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:03",
                    "end": "00:26:29",
                    "text": " We are also testing for other domains like cultural relevance, whether the user perceives these as being trustworthy or not, what the output quality is in general. And of course, we're also testing for whether the language is correct or not, whether it sounds natural to a native speaker or not when they look at the response from the LLM. So the key findings from Saviksha are that most of the models actually perform well on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:29",
                    "end": "00:26:58",
                    "text": " both content quality as well as language quality. So the models that we have today, the global models, the frontier models, they're actually doing reasonably well with some exceptions on producing good language, producing coherent language, and also generally answering the question fine. There's nothing wrong with the answer. But the differentiator we find in our evaluations is the cultural relevance. So we actually find that some models do much better at being locally culturally relevant, whereas some of them don't."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:58",
                    "end": "00:27:28",
                    "text": " And so this we can see very clearly with our evaluation effort. The second thing that we found is that among the domains that we tested, health care proved to be the most challenging domain. And so in fact, education was a very easy domain. Most of the models were able to do well. When I say education, I mean people asking for career advice and so on. Should I do this course? Should I take up this job, et cetera? Those kinds of questions are able to do well. But health care is something that models still struggle with quite a bit."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:30",
                    "end": "00:28:01",
                    "text": " Yes, so in finance the questions and I'll show you some examples of questions So finance the questions are like should I do this kind? Should I you know invest in this or should I take this loan and things like so these are everyday questions that people would ask advice seeking questions that they would ask and You know the kind of model errors that we find we still find that there are some models that You know that use awkward languages that almost look translated they they translate frozen expressions from English You know these are the things that traders typically rate"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:01",
                    "end": "00:28:30",
                    "text": " low in our benchmark. And finally, as I mentioned, we also did this massive LLM judge or automated evaluation to see whether they agree with human evaluation or not. And our findings mirror the findings that we've had in the last couple of years on this topic is that just using LLM judges or automated methods, not a good idea at all. They diverge a lot from human preferences. But what we've done is we've actually used the human evaluation data from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:30",
                    "end": "00:28:58",
                    "text": " from all of the CARIA evaluations. And we've used that to actually make the automated methods better. And we find significant improvements in this. So this is definitely a direction that can be explored further, where looking at how we can improve some of these automated methods by using the data that we collect from communities. It's still not a replacement for doing human evaluation. But yes, when we are doing these quick iterations, model building, et cetera, these automated techniques can provide useful signals as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:58",
                    "end": "00:29:24",
                    "text": " So I'm going to end with a few examples. And here, the example is in the Indian language. Then there's a translation. And then at the bottom, it's basically either a high-scoring response or a low-scoring response with a reason why. So let me just read it out. So the first one is about how a student's abilities can be identified and how clear guidance can help. So this is a generic question about education."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:24",
                    "end": "00:29:52",
                    "text": " And the interesting thing is that the answer that did poorly over here was not really suited to the Indian context. So the answer talked about psychometric assessments, job shadowing, industry visits, internships, and so on. And so the person who evaluated who was also a member of the community did not connect with this answer because it didn't seem relevant to them. So these are the kinds of differences we see between models that do well and not."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:52",
                    "end": "00:30:17",
                    "text": " And similarly, on the right, if you look at the question is about how government schools are different from private schools and how to decide whether a child should go to the government school or not. And here, a high-scoring response actually managed to compare government and private schools on different dimensions and also suggested some actionable advice. So go and visit it and so on."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:17",
                    "end": "00:30:46",
                    "text": " to try and assess the quality. So here, one more thing that we're really looking for is how trustworthy are some of these answers, right? Is somebody actually going to take an action based on the response that they give or not, right? So these are the kinds of responses that we got, which the users rated as the answer. This is a good response. I would actually trust this kind of response. Couple of other examples here. So on the left, we see a question in Megali. And one of the interesting things we found was that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:46",
                    "end": "00:31:12",
                    "text": " Many of the Bengali responses actually assume that we are in the Bangladesh context. And to be fair to the models, we don't specify what country we're in. So we don't expect that the model can tell that this is from an Indian context and not from a Bangladeshi context. So this is something to keep in mind that probably next time we should do this. We should prompt the model with the location as well for these kinds of answers. But what was interesting is that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:12",
                    "end": "00:31:26",
                    "text": " In some cases, the model assumed that it was a Bangladeshi context. But in other cases, it actually fixes the two contexts. And so it's referencing the description from both India and Bangladesh in the same answer, which causes a lot of confusion. So this is a low-scoring answer for us."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:26",
                    "end": "00:31:56",
                    "text": " And here, there's another question about buying houses. And here, the high-scoring answer basically talked about all sorts of risks and concerns, including fraud and illegal transactions and so on, and also referenced India-specific processes which the human evaluator found very valuable. So this is really what we're testing for in Samiksha. Are these responses something that community members can connect with?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:56",
                    "end": "00:32:24",
                    "text": " And to do this, Karya, of course, provides them a lot of training. You can go into more detail during the discussion. But we're really trying to empower our community to decide themselves whether something works for them or not. So they decide what the benchmark should have. They create the actual benchmark data points themselves in each language, in each culture, domain, separately. And then they also do the evaluations as well. So they're really determining whether India's models work for them or not."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:24",
                    "end": "00:33:22",
                    "text": " So I'll end here with the link to the page that has the report as well as the leaderboards for you to look at. Thank you so much and I'm happy to take questions now or later."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:22",
                    "end": "00:34:07",
                    "text": " Yes, but let me introduce one more concept and then we'll take a couple of questions. Thank you so much. So, thinking about redesigning AI economy, the work that Selena just presented on Saniksha, is so important. Because unless we center our communities in the way we evaluate AI models, we will never know what works for us. And so often in the tech sector, the way we talk about our communities can be slightly critical."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:07",
                    "end": "00:34:34",
                    "text": " I remember one and a half, two years ago, we had to convince our clients that our communities can do a complex evaluation. As of that thing, we've convinced it. I think what the mixture shows is the communities we serve at Taria are not just excellent beneficiaries of AI models. They're not just passive beneficiaries of AI revolution. They're active builders. They're active evaluators. They have to be centered to make sure these models work. So Naina, you talked about one of the reasons these biases exist."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:34",
                    "end": "00:35:01",
                    "text": " And one of the things that the Roshni team has done at Government of India is completely change who gets to contribute to the AI world. It's an absolute honor to have Amitabh Sir with us today. For people in this sector, I think Amitabh Sir is the biggest champion. I think so much of the work we do at Taria would not be possible without his support and his championship. So with no further ado, Amitabh Sir, I would love for you to talk about the second thing we're launching today, which is Samudar."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:01",
                    "end": "00:35:35",
                    "text": " So actually, given that you're so unique in the room, I would love you to talk about what are the goals behind Bhashini, what separates Bhashini from anything else in the world, and why are you excited about what we're doing with some of that. So yeah, you know, taking the cue from what you have talked about, we're getting into the background. You know, we started as Bhashini, which is Basha Interface for India, about three and a half years back."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:36",
                    "end": "00:36:15",
                    "text": " and we had five basic problems to solve. Obviously over a period of time, the technology has changed and obviously many things have happened. But currently, the building of those models, what was the challenge we had, was that we had about six to seven languages occupying"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:15",
                    "end": "00:36:44",
                    "text": " the most of the digital data which was available, while as rest of the languages did not have any digital data. And even the six to seven languages which were there, which had digital data, they were approximately, you know, about 2% or 3%, you know, anybody can guess, that amount of digital data which was available and of the entire digital data, you know, 3% of the entire digital data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:44",
                    "end": "00:37:13",
                    "text": " which is available on the internet. These languages are occupying such a low space. And when we were trying to work this out, we obvious, hence it was becoming difficult because how will you build the model if there is no digital data? Because AI models are built on digital data. So we had a unique exercise which we carried out to collect the data by brute force. We collected digital data through brute force."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:13",
                    "end": "00:37:37",
                    "text": " We had a, you know, application which had a certain set of images, places, food, items, etc. which was taken onto the field. And there were about 200 plus people who were hired to let people speak about the picture which was available on the mobile."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:37",
                    "end": "00:38:04",
                    "text": " and they speak, they transcribe, they translate and they move all of that which is there to build the digital data. With that we built the Morpus which was barely sufficient to build the models and obviously we could have waited for very long to actually see that these models have enough data to become 100% accurate and then we moved on to the field."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:04",
                    "end": "00:38:33",
                    "text": " But we decided that we should actually look at how do we go to the field so that we kick off the improvement cycle which is so essential. So we built some of the use cases which, you know, narrow use cases which, you know, started benefiting people. You know, we are handled releases today. The top eight use cases which we are working upon. We are, you know, having a use case which is..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:33",
                    "end": "00:39:01",
                    "text": " touching the panchayats today, 270,000 panchayats, where the idea was that we should bring the transparency in the way the meetings are organized in the panchayats. The panchayats essentially work on local languages and the minutes are actually created in English because the central ministry understands more English than any other place."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:01",
                    "end": "00:39:30",
                    "text": " there was an intervention by the secretary or such a who will create that particular meetings and then obviously there are multiple interventions which will actually bring about lot of their own thought process, their own way of thinking rather than bringing the people's requirement into minutes. So now the minutes are actually recorded and they are sent to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:30",
                    "end": "00:39:58",
                    "text": " The second use case which we are in the midst of launching"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:59",
                    "end": "00:40:25",
                    "text": " There are two of them. One of them is about agriculture in Maharashtra, where the farmer gets the advisory on his own language through a voice journey. So it's a chatbot which you can ask, the farmer can ask, what do I do about this? And the answer through a voice journey comes out that these are the things which need to be done."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:25",
                    "end": "00:40:54",
                    "text": " Now that system when we were implementing, the first system when we were implementing about Sabha Saar, we started looking at again from a perspective of what is the language coming up and the schools which are provided on what are the feedback etc. which is there and where are we not able to capture dialects, where are we not able to capture the words which are specific to that particular place. But in the case of Mahavistar, we had a challenge which was"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:54",
                    "end": "00:41:21",
                    "text": " much different, which was basically saying that, you know, the agriculture taxonomy and agriculture nomenclatures were not available in the model because the model hadn't seen it and it's approximately 70,000 numbers and they were not pronounced correctly when it was basically, you know, talked about or the answers which were made."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:21",
                    "end": "00:41:49",
                    "text": " So we had to go into a specific drive of involving the agriculture students to actually collect those utterances to train them on so that things start moving fine in that direction. So that's another intervention which we worked upon. The third intervention was, of course, where we touch every citizen across the country is UIDN, where the place names are translated and other names are translated."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:49",
                    "end": "00:42:17",
                    "text": " There there is a unique challenge of number of places which are available. That is the digital dictionary of the place which is available in the system is actually very limited. When we digitized the panchayat, we had 270,000 panchayat names brought to the digital system. Now we have a second challenge is that the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:17",
                    "end": "00:42:44",
                    "text": " actual number of names of places which the Survey of India has is approximately 16 lakhs to 18 lakhs and currently those are not in the digital sector and so we have to bring that back and hence we are working with Survey of India and then again those places have to be on the local dialect because those places are known on the local dialect to where the people eat there."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:44",
                    "end": "00:43:14",
                    "text": " And the idea of creating that particular database is to, perhaps a person go there, that's what Survey of India would be doing, is a person go there, this geofence, and then we ask someone from that particular location to actually speak. And that gets translated, you know, through automatic speech recognition, that gets translated and put into the database. So, just to, why I gave this example is basically..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:14",
                    "end": "00:43:39",
                    "text": " reflect upon that if we are wanting to build a large system, the value chain starts at the last mile. And the value chain of AI, which is basically related to data collection, then annotation and labeling, then it goes to model building, then you are vetting."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:39",
                    "end": "00:44:08",
                    "text": " and then you are using it in application and then coming back to a feedback loop. So this is the AI value chain and then you have the community which is basically involved or the user who is involved in the work. If we start doing that evaluation sitting at, you know, any of our metros, the evaluation would be very different, right? So we will be starting to discuss whether biology should be represented as chiefs of three in Hindi."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:08",
                    "end": "00:44:37",
                    "text": " or, you know, biology should be biology there. So, you know, that is the debate which we will have if we are doing that in Delhi. If, you know, I was basically doing a speech-to-speech demonstration in Gujarat and I had the privilege of having Honorable Minister of Mighty there listening to that, and then obviously there were other officers there. I wouldn't like to name the sake of this thing. The officers had an impression that..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:37",
                    "end": "00:45:07",
                    "text": " This was about 80 to 85 percent correct. Whereas the Honorable Minister said that it's so pure that I have started learning. So I'm basically again trying to say that the perception of people would differ when you are trying to evaluate that. Some people think that, oh, the machine is doing so much. Let's go with it. Some people will say that, no, it's not still at that level. So don't go with it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:07",
                    "end": "00:45:35",
                    "text": " But at the end of this that's what I'm taking the community feedback. I'm not The officers said it is 80% the other people said the minister said that it's too good to be true. That's the way it is Now again, the fact comes across that when you are doing all of this and deploying all these use cases and working on this It is at the end of the day"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:35",
                    "end": "00:46:04",
                    "text": " The system is as good as the users need it, the people say. And if this people, it is not necessary that we should be looking if some of our people went to Akaru in Andhra Pradesh and to demonstrate the product. And actually, Honorable CM wanted us to do that. Because there are only two people who understand English there, the collector and his secretary. Rest, all are taken."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:04",
                    "end": "00:46:34",
                    "text": " We sent a person who knew Madhyalam and some amount of Hindi and English. His survival was actually the app which we had built, the Vajni app. Otherwise, he wouldn't have been able to do anything. So if you go 50 to 100 kilometers away from a metro, you start actually seeing what is the language divide. And that is where the evaluation of the system matters. So the work which we are doing with Korea and we are trying to do this, which they have done a good job when we were"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:34",
                    "end": "00:47:04",
                    "text": " we started all of this was about creating those communities at the grassroots level to actually see how, you know, they are there. Often, you know, we are working with a situation where we are currently looking at how do we actually, you know, build a data working, data workers as such. That's basically the data collection people from whom we can collect data. Should we pay? Should we not pay? How do we pay?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:04",
                    "end": "00:47:32",
                    "text": " et cetera are the things which are there. But first of all, we need to build a community which is actually going to aggregate all of this who are willing to act as a data worker across the value chain of what we are trying to do. You know, the language AI value chain. And that is where we have started an initiative called community so that we are at least able to..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:32",
                    "end": "00:48:01",
                    "text": " recognize that these people are there they are looking at it we still do not have a name for them we are basically using the name which is combination of phases for this but still there is a reason for us to believe that these people would be the people who are there often i say that you know number one is that whatever in ai is very has a self-life of six months you can keep talking about"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:01",
                    "end": "00:48:29",
                    "text": " better today and better tomorrow. It's about six months, right? Because in six months somebody else will collect the data or build the things which are much better than this. So your objective is not to actually chase that. Your objective is how do I satisfy the customer for which the use case has been developed. So, you know, obviously there is a backend technology. There is an improvement which is required in the backend technology, but using that is not going to help us."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:29",
                    "end": "00:49:01",
                    "text": " But if you build a community which is basically going to do that, we perhaps will be unleashing, over a period of time, a marketplace or an economy which is actually going to be, you know, self-sustaining in terms of what AI can do and how AI can be better. So that's the core point which I have. Fantastic. And I think, thank you so much. We are super excited about Samudai, which we are launching with Bhashini. Samudai, again, is Sanskrit for community."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:01",
                    "end": "00:49:30",
                    "text": " We love Sanskrit words, let's start with this. You've seen the pattern here, with both Savitra and Samudaya. We're also launching Sammati at 4.13, so this is three for three today. But very, very excited about Samudaya. Sir talked about a project we wrote with government of Maharashtra, and that to me is the best example of how do we center our communities in the AI magnitude, right? Because in number one, tribal district in Maharashtra, the data collection for the local model is being done in the district."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:30",
                    "end": "00:49:49",
                    "text": " by the communities. The model building, the technology building is being done in the district. And everything will be owned by the district and by the communities. And the evaluation will also be done by the same communities, like we did with Sameeksha. So we think about the whole AI value chain."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:49",
                    "end": "00:50:19",
                    "text": " The goal to redesign the AI economy is to centralize our communities throughout the world. I would love to bring in Fezl. Fezl is one of the rock stars at Collective Intelligence Project CIP, cares deeply about evals, cares deeply about doing this globally, founding partner at Samiksha. Fezl, could you tell us about how you plan to take rubrics like Samiksha to other countries and how do we bring"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:19",
                    "end": "00:50:58",
                    "text": " community benchmarks from India to looking this up. Like Manu said, my name is Faisal. I'm the Head of World Partnerships at the Collective Intelligence Project. And a lot of the work that this fantastic team has done at Persimiksha is setting the precedent for a lot of the work we want to do in a lot of other countries. So the idea behind this really is the community-driven pipeline. The idea of community-driven AI governance, I think, is something very popular in a lot of different countries."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:58",
                    "end": "00:51:22",
                    "text": " So one is how you communicate with representatives of particular domains, particular communities. Two is how you get both experts and non-experts involved in the validation of them. And then three is how do you keep this as a loop? How do you make it replicable? How do you make it scalable? And I think everything that's been done with Sumiksha is an example of that. So one big question that we're always trying to ask is..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:22",
                    "end": "00:51:47",
                    "text": " What do AI evaluations capture right now? And what do community-driven pipelines offer that those do not? So we talked about depth, which is something that evaluations are always good at. But what community-driven pipelines do is capture breadth, contextuality, and variability. And that is just more efficient, not just morally superior, but just generally more efficient than other types of pipelines."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:47",
                    "end": "00:52:17",
                    "text": " So what we're trying to do with the Collective Intelligence Project is make evaluations in general more accessible for everyone. We've set up platforms like WeVal, which you can go to at weval.com, in which teachers in Uganda, in doctors in Memphis, they can go on and create their own AI evaluations. They can actually take criteria from their own experience and make it something rigorous, make it something credible that AI labs and policymakers can use."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:17",
                    "end": "00:52:43",
                    "text": " So just to end things here, everything that's happening with Simiksha is the precedent for community-driven governance that we want to see in a lot of other places. How do you build communities of practice when you're holding AI out? Thank you. Thank you so much for that, Faisal. I think, as we've seen with Samudai today, I wasn't able to get the internet functioning to get the Samudai demo on. But you can actually go to partially in Samudai."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:43",
                    "end": "00:53:11",
                    "text": " It's D-H-A-S-H-I-N-I-S-A-M-U-D-M-I-E. And check it out. And if you're sticking around for the whole summit, you can come to the Bhashni group. And we're actually demoing some of the material. On 19th, we actually have a panel with Amitabh Sir and Mithali Ji, who's the district of the Nandabhar district. And from her perspective, you'll actually get to learn about what does community-centered EI mean. For her, I also want to call out Shannon Farley, who's in the room."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:11",
                    "end": "00:53:40",
                    "text": " I think many of us know and love Shannon, but Shannon runs Fast Forward, one of the most prolific AI for Good investors in the world. Shannon, I want to quickly bring you in. Could you talk to us about, well, first of all, the first checks into Tarya, into CIB, into Dura Health, into Rocket Learning, so many organizations that have been highlighted at this summit today. When you think about redesigning the AI, you will immediately think about AI for Good. Are there some ideas that you find interesting that are coming from India, and then again should go to Dura?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:43",
                    "end": "00:54:11",
                    "text": " Thank you so much for having me. Hi, I'm Shannon Farley. I run Fast Forward, which is an accelerator for tech nonprofits. We've had over 100 organizations go through our accelerator, so many Indian organizations over the last few years. And I think it's in part because of this connection to community. We're seeing Indian tech nonprofits leapfrog in AI. There's a deal."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:11",
                    "end": "00:54:40",
                    "text": " There's an infrastructure here that has been here for a long time. There's lots of technical talent. And there is a deep need to integrate different voices into the models because there are so many people you're serving. It's really a model for what we want to see in the rest of the world. The other piece that's important is these organizations need money to operate. And India has a homegrown philanthropy that is making a huge difference."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:40",
                    "end": "00:55:09",
                    "text": " Organizations like The Nudge, we partner with them. Organizations like ACT Grants, we partner with them. Have been deceived and follow on funding to these organizations. And when that happens, they can scale. So one of the things I hope comes out of this summit and the incredible work of CARIA and CIP and Rocket Learning and Nora Health is that in other regions of the world they get inspired by both local talent"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:09",
                    "end": "00:55:40",
                    "text": " great entrepreneurship and philanthropy to see what they can build. We're starting to see it in Colombia with the Velas Reyes Foundation. That has spurred an ecosystem of tech nonprofits, and we would love to see it in other parts of the world. So we thank you for your models, both AI and human, and we're excited to meet you. Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:40",
                    "end": "00:56:02",
                    "text": " That was transcription, that was translation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:02",
                    "end": "00:56:23",
                    "text": " How do communities in the country find digital work opportunities directly from government and non-government sources?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:23",
                    "end": "00:56:54",
                    "text": " who have done that work before being directly engaged. This one? We're very, very excited about launching both Samsung and Samsung XI. We have five minutes for questions. Thank you. I just have one quick question. So just I have a little bit of background."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:54",
                    "end": "00:57:01",
                    "text": " I think I'm using the same people to do the evaluations."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:01",
                    "end": "00:57:26",
                    "text": " And you prove that has any kind of impact on the person. Because the first time they're doing it as if they would be using that language themselves, they might not see that pattern in the end, right? Maybe the world looks at that and changes. I might not be the right person to do the evaluation for you of the world."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:26",
                    "end": "00:57:47",
                    "text": " Do we get your thing ready?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:47",
                    "end": "00:57:59",
                    "text": " Hello? Honey, how about that? Ha ha, session is over. Kahane?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:59",
                    "end": "00:58:28",
                    "text": " Where exactly? What's so many people?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:28",
                    "end": "00:58:50",
                    "text": " Fantastic. Yeah, so to answer that question, because it's an easier one, is to work with healthcare professionals, right? I mean, that's clear ones who know the domain, and they will be able to answer all those questions. So again, working with communities where it is actually healthcare. To answer your question, I think that's a very interesting one. So one thing is, of course, that Karya does extensive training for the workers to..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:58:50",
                    "end": "00:59:15",
                    "text": " tell them what this evaluation actually means. For all of the different metrics and projects I mentioned, what the meaning is of a good language for a person, what trust means, what cultural relevance means. So there's a lot of examples given, a lot of training given in their own language, by language coordinators. So there's a lot of work there. And this is also done in partnership with the CSOs. So the CSOs have an input of how the data workers get trained for both."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:15",
                    "end": "00:59:45",
                    "text": " create the data points, but also to evaluate them. So we do a lot of work on that. But having said that, there is variation also in these evaluation responses. So for some of the data points, we actually got two different evaluators to the same data point. And we measured, again, the agreement between the annotators and so on. So all those details are there in your report. Thank you so much. We'll add that, unfortunately. But we'll all be outside to answer the questions. I will end the conflict game. If you want to speak to the car yet."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:59:45",
                    "end": "01:00:10",
                    "text": " We have 28 other panels we're doing across the next four days. So please come find us. If you are very interested in working with me, I'll send you a list of the panels that you can come to. We also have three booths. We're doing an incredible work outside India, across Kenya, South Africa, Ethiopia, the US."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:10",
                    "end": "01:00:39",
                    "text": " Indonesia, so please check out the MasterCard group. We did a fantastic group with Bhasini, which we're very, very happy to do. We're showcasing our work on Samudaya. We actually try out Samudaya and try out all the ways we can do this. And finally, we have an exciting group with an India AI mission. We're showcasing, and that's right next door at call number 14. We're showcasing and releasing our datasets today. It's very, very exciting. Thank you so much for this session. And thank you so much to my terrible panelists."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:00:39",
                    "end": "01:01:11",
                    "text": " You have the copy of the file, right? If someone asks for it from the Karya team, you don't have it. Which team? The organizing team. Okay, if someone asks for it from the organizing team, do you want to give it or not? Do you want to give it or not? No, I don't want to give it."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:01:11",
                    "end": "01:01:21",
                    "text": " Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:01:21",
                    "end": "01:01:50",
                    "text": " Good morning everyone."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:01:50",
                    "end": "01:02:13",
                    "text": " So the next session is going to start. So I request everyone to please clear the stage and allow us to start the session on time."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:02:13",
                    "end": "01:02:42",
                    "text": " This is different."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:02:42",
                    "end": "01:02:59",
                    "text": " What's up?"
                },
                {
                    "speaker": "Unknown",
                    "start": "01:02:59",
                    "end": "01:03:29",
                    "text": " So request all the speakers for the Preda Sapien event to be in the front row please. Good morning everyone. On behalf of Preda Sapien, I am happy to welcome you all for the event."
                },
                {
                    "speaker": "Unknown",
                    "start": "01:03:29",
                    "end": "01:03:30",
                    "text": " Uh..."
                }
            ]
        }
    ],
    "projectDeleted": false,
    "codeFrameColumn": [
        {
            "col1": "col1",
            "col2": "col3",
            "researchObjective": "",
            "filterColumn": [],
            "netSetting": true,
            "subNetSetting": true,
            "categorySetting": true,
            "sentimentSetting": true,
            "thinkingBudgetSetting": 2000,
            "questionType": "OPENEND",
            "otherBrand": false,
            "existingCodeFrame": [],
            "isList": false,
            "codeFrameWithDataList": [],
            "codeFrameList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 11730,
                        "CandidatesTokenCount": 2943,
                        "TotalTokenCount": 16028,
                        "thoughtsTokenCount": 1355,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Net": "AI Model Evaluation & Benchmarking",
                                "Subnets": [
                                    {
                                        "Subnet": "Challenges in AI Evaluation",
                                        "Categories": [
                                            {
                                                "Category": "Output Complexity & Variability",
                                                "Summary": "AI outputs are hard to understand and can vary significantly with minor input changes."
                                            },
                                            {
                                                "Category": "Lack of Ground Truth/Metrics",
                                                "Summary": "Open-ended AI problems often lack a single correct answer, rendering traditional evaluation metrics ineffective."
                                            },
                                            {
                                                "Category": "Evaluation Deployment Gap",
                                                "Summary": "AI models perform well in controlled benchmarks but frequently fail in real-world deployment scenarios."
                                            }
                                        ],
                                        "Summary": "Discusses the inherent difficulties in evaluating modern AI systems, especially large language models and generative AI."
                                    },
                                    {
                                        "Subnet": "Samiksha: Community-Driven Evaluation",
                                        "Categories": [
                                            {
                                                "Category": "Construction Methodology",
                                                "Summary": "A benchmark built with input from civil society organizations and community members across 11 Indian languages, comprising 23,000 data points."
                                            },
                                            {
                                                "Category": "Evaluation Metrics & Findings",
                                                "Summary": "Evaluates AI models for accuracy, cultural relevance, trustworthiness, and language quality, finding healthcare to be the most challenging domain."
                                            },
                                            {
                                                "Category": "Human vs. Automated Evaluation",
                                                "Summary": "Human evaluations are preferred over LLM-as-judge methods, but human data can significantly improve automated techniques."
                                            }
                                        ],
                                        "Summary": "Details the Samiksha project, a comprehensive, community-driven benchmark for Indian languages, contexts, and cultures."
                                    },
                                    {
                                        "Subnet": "Benchmarking for Diverse Languages",
                                        "Categories": [
                                            {
                                                "Category": "Limited Language Coverage",
                                                "Summary": "Existing benchmarks are heavily skewed towards high-resource, Western languages, failing to cover the vast majority of global languages."
                                            },
                                            {
                                                "Category": "Non-Representative Benchmarks",
                                                "Summary": "Many multilingual benchmarks are merely translations of English ones, lacking cultural and contextual relevance for local use."
                                            },
                                            {
                                                "Category": "Trust, Rigor & Contamination Issues",
                                                "Summary": "Concerns about fair comparison across models due to hyperparameter sensitivity, and benchmarks being 'contaminated' by inclusion in training data."
                                            }
                                        ],
                                        "Summary": "Addresses the specific issues of coverage, representativeness, and rigor in benchmarks designed for multilingual and multicultural contexts."
                                    }
                                ],
                                "Summary": "This topic covers the methodologies, inherent challenges, and specific initiatives (like Samiksha) related to assessing the performance and impact of AI models, particularly in diverse linguistic and cultural settings."
                            },
                            {
                                "Net": "Community-Centric AI Development",
                                "Subnets": [
                                    {
                                        "Subnet": "Role of Community in AI Value Chain",
                                        "Categories": [
                                            {
                                                "Category": "Data Collection & Annotation",
                                                "Summary": "Community workers are actively involved in collecting and annotating local language data, often through 'brute force' methods (e.g., Bhashini)."
                                            },
                                            {
                                                "Category": "Model Building & Training Input",
                                                "Summary": "Local communities, including students, contribute specific domain-related data (e.g., agriculture taxonomy) to train AI models."
                                            },
                                            {
                                                "Category": "Evaluation & Feedback Loop",
                                                "Summary": "Communities are empowered to evaluate AI models based on local criteria and provide essential feedback for continuous improvement."
                                            }
                                        ],
                                        "Summary": "Highlights how communities are actively involved across the AI value chain, from data creation to evaluation, not just as passive beneficiaries."
                                    },
                                    {
                                        "Subnet": "Samudai Initiative",
                                        "Categories": [
                                            {
                                                "Category": "Goals of Samudai",
                                                "Summary": "Aimed at building a community of data workers at the grassroots level to aggregate and contribute to the AI value chain."
                                            },
                                            {
                                                "Category": "Self-Sustaining AI Economy",
                                                "Summary": "Envisions unleashing a marketplace and economy where community involvement leads to continuous AI improvement and sustainability."
                                            },
                                            {
                                                "Category": "Local Ownership & Empowerment",
                                                "Summary": "Emphasizes that local communities should own the data, models, and evaluation processes, ensuring relevance and control."
                                            }
                                        ],
                                        "Summary": "Introduces the Samudai initiative by Bhashini, focusing on building and empowering communities for data work and a localized AI economy."
                                    },
                                    {
                                        "Subnet": "Community Engagement Strategies",
                                        "Categories": [
                                            {
                                                "Category": "Civil Society Organization (CSO) Partnership",
                                                "Summary": "Engaging CSOs to leverage their domain expertise and connections for benchmark construction and data worker training."
                                            },
                                            {
                                                "Category": "Localized Training & Guidance",
                                                "Summary": "Providing extensive training to community workers in their native languages on evaluation metrics like trust and cultural relevance."
                                            },
                                            {
                                                "Category": "Addressing Language & Dialect Divide",
                                                "Summary": "Strategies to bridge the gap in digital data for local languages and dialects, ensuring AI models cater to subcultures."
                                            }
                                        ],
                                        "Summary": "Describes the methods and partnerships used to effectively engage and train communities for their involvement in AI development and evaluation."
                                    }
                                ],
                                "Summary": "Focuses on the philosophy and practical implementation of involving local communities at every stage of the AI value chain, from data creation to model building and evaluation."
                            },
                            {
                                "Net": "Challenges & Gaps in AI for Diverse Contexts",
                                "Subnets": [
                                    {
                                        "Subnet": "Linguistic & Cultural Bias",
                                        "Categories": [
                                            {
                                                "Category": "Global North Assumptions",
                                                "Summary": "Many current AI systems are built with strong assumptions and priors from Global North contexts, leading to irrelevance elsewhere."
                                            },
                                            {
                                                "Category": "Lack of Cultural Relevance",
                                                "Summary": "AI responses often lack local cultural context, making them untrustworthy or unhelpful to diverse users (e.g., Bangladeshi vs. Indian Bengali context)."
                                            },
                                            {
                                                "Category": "Dialect & Subculture Neglect",
                                                "Summary": "AI models often fail to account for the vast number of unique dialects and subcultures, particularly prominent in countries like India."
                                            }
                                        ],
                                        "Summary": "Examines how AI systems exhibit biases rooted in their training data and design, making them less effective or relevant for diverse global populations."
                                    },
                                    {
                                        "Subnet": "Data Scarcity & Representation",
                                        "Categories": [
                                            {
                                                "Category": "Skewed Data Distribution",
                                                "Summary": "Digital data is overwhelmingly skewed towards English and Western datasets, leaving many languages with insufficient training material."
                                            },
                                            {
                                                "Category": "Need for Localized Data Collection",
                                                "Summary": "Highlights the necessity of active, grassroots data collection in specific local contexts to build relevant AI models."
                                            },
                                            {
                                                "Category": "Domain-Specific Taxonomy Gaps",
                                                "Summary": "Lack of digital data for specialized vocabularies (e.g., agriculture taxonomy) in local languages poses significant challenges for AI accuracy."
                                            }
                                        ],
                                        "Summary": "Addresses the critical issue of insufficient and imbalanced data, which hinders the development of accurate and inclusive AI for non-English and specific cultural contexts."
                                    },
                                    {
                                        "Subnet": "Limitations of Current AI Paradigms",
                                        "Categories": [
                                            {
                                                "Category": "LLM as Judge Inadequacy",
                                                "Summary": "The common practice of using LLMs to evaluate other AI models has been shown to be neither valid nor reliable."
                                            },
                                            {
                                                "Category": "Model Complexity & Debugging",
                                                "Summary": "Increasing complexity of generative AI systems and agentic chains makes designing effective evaluation protocols and understanding model failures more difficult."
                                            },
                                            {
                                                "Category": "Benchmark Hacking & Flaws",
                                                "Summary": "The prevalence of benchmark hacking and inherent flaws in existing benchmarks (e.g., data quality, metrics, interpretation) undermines their validity."
                                            }
                                        ],
                                        "Summary": "Discusses the inherent weaknesses and limitations of existing AI evaluation methods and current model paradigms when applied to complex, diverse scenarios."
                                    }
                                ],
                                "Summary": "Discusses fundamental problems such as language coverage, cultural relevance, data scarcity, and the 'evaluation deployment gap' that hinder AI effectiveness, especially for non-English and Global South contexts."
                            },
                            {
                                "Net": "AI for Social Impact & Inclusivity",
                                "Subnets": [
                                    {
                                        "Subnet": "Application in Public Services",
                                        "Categories": [
                                            {
                                                "Category": "Panchayat Meeting Translation",
                                                "Summary": "AI is used to translate local language panchayat meeting minutes into English to improve transparency and accessibility for central ministries."
                                            },
                                            {
                                                "Category": "Agricultural Advisory Systems",
                                                "Summary": "Farmers receive voice-based AI advisories in their local languages, addressing specific agricultural queries and challenges."
                                            },
                                            {
                                                "Category": "Place Name Digitization (UIDN)",
                                                "Summary": "Efforts to digitize and translate place names in local dialects for national identification systems, working with Survey of India."
                                            }
                                        ],
                                        "Summary": "Showcases how AI is being deployed to enhance various public services, making them more accessible and relevant to citizens in local languages."
                                    },
                                    {
                                        "Subnet": "Fostering Dignified Learning & Opportunities",
                                        "Categories": [
                                            {
                                                "Category": "Fair & Inclusive AI Models",
                                                "Summary": "Using AI to build foundational datasets that ensure AI models are fair and inclusive for all, particularly in diverse linguistic landscapes."
                                            },
                                            {
                                                "Category": "Community-Owned AI Solutions",
                                                "Summary": "The goal of developing AI models that are owned and controlled by the communities they serve, promoting self-reliance and relevance."
                                            },
                                            {
                                                "Category": "Bridging the Language Divide",
                                                "Summary": "Leveraging AI to overcome language barriers, ensuring technology is useful for populations beyond urban metros, where language divides are stark."
                                            }
                                        ],
                                        "Summary": "Focuses on how AI can be leveraged to create equitable access to learning and opportunities, ensuring technology serves all segments of society."
                                    },
                                    {
                                        "Subnet": "AI for Good Initiatives",
                                        "Categories": [
                                            {
                                                "Category": "Promoting AI for Good Investments",
                                                "Summary": "Organizations like Fast Forward invest in tech nonprofits leveraging AI for social impact, inspired by India's community-centric approach."
                                            },
                                            {
                                                "Category": "Local Entrepreneurship & Philanthropy",
                                                "Summary": "Highlighting the role of homegrown philanthropy and local entrepreneurship in scaling AI for good initiatives in India and inspiring other regions."
                                            },
                                            {
                                                "Category": "Global Partnerships for Inclusive AI",
                                                "Summary": "Collaborations (e.g., Collective Intelligence Project) to scale community-driven AI governance models from India to other countries globally."
                                            }
                                        ],
                                        "Summary": "Discusses the broader movement and investment in AI applications that address social challenges and promote inclusivity globally."
                                    }
                                ],
                                "Summary": "Focuses on how AI can be made useful, fair, and inclusive for people, particularly in India, with examples in agriculture, education, healthcare, and legal domains, and the broader AI for Good movement."
                            },
                            {
                                "Net": "AI Ecosystem & Policy in India",
                                "Subnets": [
                                    {
                                        "Subnet": "Bhashini: Government AI Initiative",
                                        "Categories": [
                                            {
                                                "Category": "Addressing Data Scarcity",
                                                "Summary": "Bhashini's mission to overcome the lack of digital data for Indian languages, especially beyond the top 6-7, using brute-force collection methods."
                                            },
                                            {
                                                "Category": "Building Language Models for India",
                                                "Summary": "Focus on creating models specifically for Indian languages, recognizing the unique linguistic diversity (19,000+ dialects)."
                                            },
                                            {
                                                "Category": "Kicking Off Improvement Cycles",
                                                "Summary": "Deploying use cases early to gather field feedback and initiate continuous improvement cycles for AI models."
                                            }
                                        ],
                                        "Summary": "Highlights the Government of India's Bhashini initiative, its goals, and strategies for building an inclusive AI ecosystem for Indian languages."
                                    },
                                    {
                                        "Subnet": "Ecosystem Stakeholders & Collaboration",
                                        "Categories": [
                                            {
                                                "Category": "Role of Microsoft Research & Karya",
                                                "Summary": "Partnership in developing community-driven evaluation benchmarks (Samiksha) and engaging local workers for AI projects."
                                            },
                                            {
                                                "Category": "Government & Academia Involvement",
                                                "Summary": "The collaboration of government bodies (Bhashini, Survey of India) and research institutions in shaping AI policy and development."
                                            },
                                            {
                                                "Category": "International Partnerships",
                                                "Summary": "Collaborations with global organizations (e.g., Collective Intelligence Project, Fast Forward) to scale Indian AI models and best practices internationally."
                                            }
                                        ],
                                        "Summary": "Discusses the various entities, including government, industry, academia, and non-profits, that are collaborating to build India's AI ecosystem."
                                    },
                                    {
                                        "Subnet": "Funding & Support for AI Development",
                                        "Categories": [
                                            {
                                                "Category": "Homegrown Philanthropy",
                                                "Summary": "The significant role of Indian philanthropy (e.g., The Nudge, ACT Grants) in providing follow-on funding to scale tech nonprofits."
                                            },
                                            {
                                                "Category": "Accelerator & Investor Support",
                                                "Summary": "Organizations like Fast Forward provide early-stage checks and acceleration to tech nonprofits, including many Indian organizations."
                                            },
                                            {
                                                "Category": "Marketplace & Economy for Data Workers",
                                                "Summary": "The concept of building a self-sustaining economy and marketplace around data workers and community contributors to the AI value chain."
                                            }
                                        ],
                                        "Summary": "Examines the financial and institutional support mechanisms driving AI development, particularly for social good, within India and for Indian organizations."
                                    }
                                ],
                                "Summary": "Highlights initiatives like Bhashini, the role of government, local entrepreneurship, and philanthropy in building a robust and inclusive AI ecosystem in India."
                            },
                            {
                                "Net": "996 Non specific TOPIC",
                                "Subnets": [
                                    {
                                        "Subnet": "Non specific SubTopic",
                                        "Categories": [
                                            {
                                                "Category": "Non specific Category",
                                                "Summary": "Responses that are irrelevant to the codeframe's themes but are not gibberish or non-committal."
                                            }
                                        ],
                                        "Summary": "This category contains responses that are irrelevant to the established themes but do not fit other exclusion criteria."
                                    }
                                ],
                                "Summary": "This topic will contain all those responses which are irrelevant, but could not be coded in the themes within the codeframe easily."
                            },
                            {
                                "Net": "997 Don’t Know Can’t Say",
                                "Subnets": [
                                    {
                                        "Subnet": "Non-committal/Vague",
                                        "Categories": [
                                            {
                                                "Category": "Uncertain/Indecisive",
                                                "Summary": "Responses indicating uncertainty, lack of knowledge, or an unwillingness to provide a definitive answer."
                                            }
                                        ],
                                        "Summary": "This category contains responses that are non-committal or vague, such as 'Don't know' or 'Can't say'."
                                    }
                                ],
                                "Summary": "This topic will contain all those responses which are non-committal and vague."
                            },
                            {
                                "Net": "998 Not Applicable",
                                "Subnets": [
                                    {
                                        "Subnet": "Irrelevant/Out of Scope",
                                        "Categories": [
                                            {
                                                "Category": "Not Applicable to Question",
                                                "Summary": "Responses that are completely irrelevant to the survey question or the overall project context."
                                            }
                                        ],
                                        "Summary": "This category contains responses that are irrelevant or not applicable to the specific question asked or the project at hand."
                                    }
                                ],
                                "Summary": "This topic will contain all those responses which are irrelevant, not applicable to the question."
                            },
                            {
                                "Net": "999 None-Nothing",
                                "Subnets": [
                                    {
                                        "Subnet": "Gibberish/Meaningless",
                                        "Categories": [
                                            {
                                                "Category": "Non-Language Characters/Empty",
                                                "Summary": "Responses that consist of plain gibberish, non-language characters, or convey no meaningful content."
                                            }
                                        ],
                                        "Summary": "This category contains responses that are plain gibberish, non-language characters, or mean nothing."
                                    }
                                ],
                                "Summary": "This topic will contain all those responses which are plain gibberish non-language characters or mean nothing."
                            }
                        ],
                        "Summary": "The data is a transcription of a session at an AI Impact Summit, primarily focusing on the critical need for community-centric approaches in AI development and evaluation, especially for multilingual and multicultural contexts in India. Speakers highlight projects like Samiksha (community-driven AI evaluation) and Samudai (community for data collection), emphasizing challenges such as limited language coverage, cultural irrelevance in existing AI models, and the 'evaluation deployment gap.' The discussion underscores the importance of local context, trust, and inclusivity in AI to ensure its effectiveness and fairness for diverse communities, promoting AI for social good through initiatives like Bhashini and collaborations with local organizations and philanthropy."
                    }
                }
            ],
            "onlyCodeFrame": true,
            "errorList": [],
            "codeFrameGenerateType": "CODE_FRAME_ONLY",
            "requestType": "CODEFRAME",
            "requestMode": "DEVELOPMENT",
            "sourceLanguage": "English",
            "targetLanguage": "English",
            "response": {
                "status": "OK",
                "data": {
                    "status": "OK",
                    "PromptTokenCount": 11730,
                    "CandidatesTokenCount": 2943,
                    "TotalTokenCount": 16028,
                    "thoughtsTokenCount": 1355,
                    "finishReason": "STOP",
                    "data": [
                        {
                            "Net": "AI Model Evaluation & Benchmarking",
                            "Subnets": [
                                {
                                    "Subnet": "Challenges in AI Evaluation",
                                    "Categories": [
                                        {
                                            "Category": "Output Complexity & Variability",
                                            "Summary": "AI outputs are hard to understand and can vary significantly with minor input changes.",
                                            "code": 3
                                        },
                                        {
                                            "Category": "Lack of Ground Truth/Metrics",
                                            "Summary": "Open-ended AI problems often lack a single correct answer, rendering traditional evaluation metrics ineffective.",
                                            "code": 4
                                        },
                                        {
                                            "Category": "Evaluation Deployment Gap",
                                            "Summary": "AI models perform well in controlled benchmarks but frequently fail in real-world deployment scenarios.",
                                            "code": 5
                                        }
                                    ],
                                    "Summary": "Discusses the inherent difficulties in evaluating modern AI systems, especially large language models and generative AI.",
                                    "code": 2
                                },
                                {
                                    "Subnet": "Samiksha: Community-Driven Evaluation",
                                    "Categories": [
                                        {
                                            "Category": "Construction Methodology",
                                            "Summary": "A benchmark built with input from civil society organizations and community members across 11 Indian languages, comprising 23,000 data points.",
                                            "code": 7
                                        },
                                        {
                                            "Category": "Evaluation Metrics & Findings",
                                            "Summary": "Evaluates AI models for accuracy, cultural relevance, trustworthiness, and language quality, finding healthcare to be the most challenging domain.",
                                            "code": 8
                                        },
                                        {
                                            "Category": "Human vs. Automated Evaluation",
                                            "Summary": "Human evaluations are preferred over LLM-as-judge methods, but human data can significantly improve automated techniques.",
                                            "code": 9
                                        }
                                    ],
                                    "Summary": "Details the Samiksha project, a comprehensive, community-driven benchmark for Indian languages, contexts, and cultures.",
                                    "code": 6
                                },
                                {
                                    "Subnet": "Benchmarking for Diverse Languages",
                                    "Categories": [
                                        {
                                            "Category": "Limited Language Coverage",
                                            "Summary": "Existing benchmarks are heavily skewed towards high-resource, Western languages, failing to cover the vast majority of global languages.",
                                            "code": 11
                                        },
                                        {
                                            "Category": "Non-Representative Benchmarks",
                                            "Summary": "Many multilingual benchmarks are merely translations of English ones, lacking cultural and contextual relevance for local use.",
                                            "code": 12
                                        },
                                        {
                                            "Category": "Trust, Rigor & Contamination Issues",
                                            "Summary": "Concerns about fair comparison across models due to hyperparameter sensitivity, and benchmarks being 'contaminated' by inclusion in training data.",
                                            "code": 13
                                        }
                                    ],
                                    "Summary": "Addresses the specific issues of coverage, representativeness, and rigor in benchmarks designed for multilingual and multicultural contexts.",
                                    "code": 10
                                }
                            ],
                            "Summary": "This topic covers the methodologies, inherent challenges, and specific initiatives (like Samiksha) related to assessing the performance and impact of AI models, particularly in diverse linguistic and cultural settings.",
                            "code": 1
                        },
                        {
                            "Net": "Community-Centric AI Development",
                            "Subnets": [
                                {
                                    "Subnet": "Role of Community in AI Value Chain",
                                    "Categories": [
                                        {
                                            "Category": "Data Collection & Annotation",
                                            "Summary": "Community workers are actively involved in collecting and annotating local language data, often through 'brute force' methods (e.g., Bhashini).",
                                            "code": 16
                                        },
                                        {
                                            "Category": "Model Building & Training Input",
                                            "Summary": "Local communities, including students, contribute specific domain-related data (e.g., agriculture taxonomy) to train AI models.",
                                            "code": 17
                                        },
                                        {
                                            "Category": "Evaluation & Feedback Loop",
                                            "Summary": "Communities are empowered to evaluate AI models based on local criteria and provide essential feedback for continuous improvement.",
                                            "code": 18
                                        }
                                    ],
                                    "Summary": "Highlights how communities are actively involved across the AI value chain, from data creation to evaluation, not just as passive beneficiaries.",
                                    "code": 15
                                },
                                {
                                    "Subnet": "Samudai Initiative",
                                    "Categories": [
                                        {
                                            "Category": "Goals of Samudai",
                                            "Summary": "Aimed at building a community of data workers at the grassroots level to aggregate and contribute to the AI value chain.",
                                            "code": 20
                                        },
                                        {
                                            "Category": "Self-Sustaining AI Economy",
                                            "Summary": "Envisions unleashing a marketplace and economy where community involvement leads to continuous AI improvement and sustainability.",
                                            "code": 21
                                        },
                                        {
                                            "Category": "Local Ownership & Empowerment",
                                            "Summary": "Emphasizes that local communities should own the data, models, and evaluation processes, ensuring relevance and control.",
                                            "code": 22
                                        }
                                    ],
                                    "Summary": "Introduces the Samudai initiative by Bhashini, focusing on building and empowering communities for data work and a localized AI economy.",
                                    "code": 19
                                },
                                {
                                    "Subnet": "Community Engagement Strategies",
                                    "Categories": [
                                        {
                                            "Category": "Civil Society Organization (CSO) Partnership",
                                            "Summary": "Engaging CSOs to leverage their domain expertise and connections for benchmark construction and data worker training.",
                                            "code": 24
                                        },
                                        {
                                            "Category": "Localized Training & Guidance",
                                            "Summary": "Providing extensive training to community workers in their native languages on evaluation metrics like trust and cultural relevance.",
                                            "code": 25
                                        },
                                        {
                                            "Category": "Addressing Language & Dialect Divide",
                                            "Summary": "Strategies to bridge the gap in digital data for local languages and dialects, ensuring AI models cater to subcultures.",
                                            "code": 26
                                        }
                                    ],
                                    "Summary": "Describes the methods and partnerships used to effectively engage and train communities for their involvement in AI development and evaluation.",
                                    "code": 23
                                }
                            ],
                            "Summary": "Focuses on the philosophy and practical implementation of involving local communities at every stage of the AI value chain, from data creation to model building and evaluation.",
                            "code": 14
                        },
                        {
                            "Net": "Challenges & Gaps in AI for Diverse Contexts",
                            "Subnets": [
                                {
                                    "Subnet": "Linguistic & Cultural Bias",
                                    "Categories": [
                                        {
                                            "Category": "Global North Assumptions",
                                            "Summary": "Many current AI systems are built with strong assumptions and priors from Global North contexts, leading to irrelevance elsewhere.",
                                            "code": 29
                                        },
                                        {
                                            "Category": "Lack of Cultural Relevance",
                                            "Summary": "AI responses often lack local cultural context, making them untrustworthy or unhelpful to diverse users (e.g., Bangladeshi vs. Indian Bengali context).",
                                            "code": 30
                                        },
                                        {
                                            "Category": "Dialect & Subculture Neglect",
                                            "Summary": "AI models often fail to account for the vast number of unique dialects and subcultures, particularly prominent in countries like India.",
                                            "code": 31
                                        }
                                    ],
                                    "Summary": "Examines how AI systems exhibit biases rooted in their training data and design, making them less effective or relevant for diverse global populations.",
                                    "code": 28
                                },
                                {
                                    "Subnet": "Data Scarcity & Representation",
                                    "Categories": [
                                        {
                                            "Category": "Skewed Data Distribution",
                                            "Summary": "Digital data is overwhelmingly skewed towards English and Western datasets, leaving many languages with insufficient training material.",
                                            "code": 33
                                        },
                                        {
                                            "Category": "Need for Localized Data Collection",
                                            "Summary": "Highlights the necessity of active, grassroots data collection in specific local contexts to build relevant AI models.",
                                            "code": 34
                                        },
                                        {
                                            "Category": "Domain-Specific Taxonomy Gaps",
                                            "Summary": "Lack of digital data for specialized vocabularies (e.g., agriculture taxonomy) in local languages poses significant challenges for AI accuracy.",
                                            "code": 35
                                        }
                                    ],
                                    "Summary": "Addresses the critical issue of insufficient and imbalanced data, which hinders the development of accurate and inclusive AI for non-English and specific cultural contexts.",
                                    "code": 32
                                },
                                {
                                    "Subnet": "Limitations of Current AI Paradigms",
                                    "Categories": [
                                        {
                                            "Category": "LLM as Judge Inadequacy",
                                            "Summary": "The common practice of using LLMs to evaluate other AI models has been shown to be neither valid nor reliable.",
                                            "code": 37
                                        },
                                        {
                                            "Category": "Model Complexity & Debugging",
                                            "Summary": "Increasing complexity of generative AI systems and agentic chains makes designing effective evaluation protocols and understanding model failures more difficult.",
                                            "code": 38
                                        },
                                        {
                                            "Category": "Benchmark Hacking & Flaws",
                                            "Summary": "The prevalence of benchmark hacking and inherent flaws in existing benchmarks (e.g., data quality, metrics, interpretation) undermines their validity.",
                                            "code": 39
                                        }
                                    ],
                                    "Summary": "Discusses the inherent weaknesses and limitations of existing AI evaluation methods and current model paradigms when applied to complex, diverse scenarios.",
                                    "code": 36
                                }
                            ],
                            "Summary": "Discusses fundamental problems such as language coverage, cultural relevance, data scarcity, and the 'evaluation deployment gap' that hinder AI effectiveness, especially for non-English and Global South contexts.",
                            "code": 27
                        },
                        {
                            "Net": "AI for Social Impact & Inclusivity",
                            "Subnets": [
                                {
                                    "Subnet": "Application in Public Services",
                                    "Categories": [
                                        {
                                            "Category": "Panchayat Meeting Translation",
                                            "Summary": "AI is used to translate local language panchayat meeting minutes into English to improve transparency and accessibility for central ministries.",
                                            "code": 42
                                        },
                                        {
                                            "Category": "Agricultural Advisory Systems",
                                            "Summary": "Farmers receive voice-based AI advisories in their local languages, addressing specific agricultural queries and challenges.",
                                            "code": 43
                                        },
                                        {
                                            "Category": "Place Name Digitization (UIDN)",
                                            "Summary": "Efforts to digitize and translate place names in local dialects for national identification systems, working with Survey of India.",
                                            "code": 44
                                        }
                                    ],
                                    "Summary": "Showcases how AI is being deployed to enhance various public services, making them more accessible and relevant to citizens in local languages.",
                                    "code": 41
                                },
                                {
                                    "Subnet": "Fostering Dignified Learning & Opportunities",
                                    "Categories": [
                                        {
                                            "Category": "Fair & Inclusive AI Models",
                                            "Summary": "Using AI to build foundational datasets that ensure AI models are fair and inclusive for all, particularly in diverse linguistic landscapes.",
                                            "code": 46
                                        },
                                        {
                                            "Category": "Community-Owned AI Solutions",
                                            "Summary": "The goal of developing AI models that are owned and controlled by the communities they serve, promoting self-reliance and relevance.",
                                            "code": 47
                                        },
                                        {
                                            "Category": "Bridging the Language Divide",
                                            "Summary": "Leveraging AI to overcome language barriers, ensuring technology is useful for populations beyond urban metros, where language divides are stark.",
                                            "code": 48
                                        }
                                    ],
                                    "Summary": "Focuses on how AI can be leveraged to create equitable access to learning and opportunities, ensuring technology serves all segments of society.",
                                    "code": 45
                                },
                                {
                                    "Subnet": "AI for Good Initiatives",
                                    "Categories": [
                                        {
                                            "Category": "Promoting AI for Good Investments",
                                            "Summary": "Organizations like Fast Forward invest in tech nonprofits leveraging AI for social impact, inspired by India's community-centric approach.",
                                            "code": 50
                                        },
                                        {
                                            "Category": "Local Entrepreneurship & Philanthropy",
                                            "Summary": "Highlighting the role of homegrown philanthropy and local entrepreneurship in scaling AI for good initiatives in India and inspiring other regions.",
                                            "code": 51
                                        },
                                        {
                                            "Category": "Global Partnerships for Inclusive AI",
                                            "Summary": "Collaborations (e.g., Collective Intelligence Project) to scale community-driven AI governance models from India to other countries globally.",
                                            "code": 52
                                        }
                                    ],
                                    "Summary": "Discusses the broader movement and investment in AI applications that address social challenges and promote inclusivity globally.",
                                    "code": 49
                                }
                            ],
                            "Summary": "Focuses on how AI can be made useful, fair, and inclusive for people, particularly in India, with examples in agriculture, education, healthcare, and legal domains, and the broader AI for Good movement.",
                            "code": 40
                        },
                        {
                            "Net": "AI Ecosystem & Policy in India",
                            "Subnets": [
                                {
                                    "Subnet": "Bhashini: Government AI Initiative",
                                    "Categories": [
                                        {
                                            "Category": "Addressing Data Scarcity",
                                            "Summary": "Bhashini's mission to overcome the lack of digital data for Indian languages, especially beyond the top 6-7, using brute-force collection methods.",
                                            "code": 55
                                        },
                                        {
                                            "Category": "Building Language Models for India",
                                            "Summary": "Focus on creating models specifically for Indian languages, recognizing the unique linguistic diversity (19,000+ dialects).",
                                            "code": 56
                                        },
                                        {
                                            "Category": "Kicking Off Improvement Cycles",
                                            "Summary": "Deploying use cases early to gather field feedback and initiate continuous improvement cycles for AI models.",
                                            "code": 57
                                        }
                                    ],
                                    "Summary": "Highlights the Government of India's Bhashini initiative, its goals, and strategies for building an inclusive AI ecosystem for Indian languages.",
                                    "code": 54
                                },
                                {
                                    "Subnet": "Ecosystem Stakeholders & Collaboration",
                                    "Categories": [
                                        {
                                            "Category": "Role of Microsoft Research & Karya",
                                            "Summary": "Partnership in developing community-driven evaluation benchmarks (Samiksha) and engaging local workers for AI projects.",
                                            "code": 59
                                        },
                                        {
                                            "Category": "Government & Academia Involvement",
                                            "Summary": "The collaboration of government bodies (Bhashini, Survey of India) and research institutions in shaping AI policy and development.",
                                            "code": 60
                                        },
                                        {
                                            "Category": "International Partnerships",
                                            "Summary": "Collaborations with global organizations (e.g., Collective Intelligence Project, Fast Forward) to scale Indian AI models and best practices internationally.",
                                            "code": 61
                                        }
                                    ],
                                    "Summary": "Discusses the various entities, including government, industry, academia, and non-profits, that are collaborating to build India's AI ecosystem.",
                                    "code": 58
                                },
                                {
                                    "Subnet": "Funding & Support for AI Development",
                                    "Categories": [
                                        {
                                            "Category": "Homegrown Philanthropy",
                                            "Summary": "The significant role of Indian philanthropy (e.g., The Nudge, ACT Grants) in providing follow-on funding to scale tech nonprofits.",
                                            "code": 63
                                        },
                                        {
                                            "Category": "Accelerator & Investor Support",
                                            "Summary": "Organizations like Fast Forward provide early-stage checks and acceleration to tech nonprofits, including many Indian organizations.",
                                            "code": 64
                                        },
                                        {
                                            "Category": "Marketplace & Economy for Data Workers",
                                            "Summary": "The concept of building a self-sustaining economy and marketplace around data workers and community contributors to the AI value chain.",
                                            "code": 65
                                        }
                                    ],
                                    "Summary": "Examines the financial and institutional support mechanisms driving AI development, particularly for social good, within India and for Indian organizations.",
                                    "code": 62
                                }
                            ],
                            "Summary": "Highlights initiatives like Bhashini, the role of government, local entrepreneurship, and philanthropy in building a robust and inclusive AI ecosystem in India.",
                            "code": 53
                        }
                    ],
                    "Summary": "The data is a transcription of a session at an AI Impact Summit, primarily focusing on the critical need for community-centric approaches in AI development and evaluation, especially for multilingual and multicultural contexts in India. Speakers highlight projects like Samiksha (community-driven AI evaluation) and Samudai (community for data collection), emphasizing challenges such as limited language coverage, cultural irrelevance in existing AI models, and the 'evaluation deployment gap.' The discussion underscores the importance of local context, trust, and inclusivity in AI to ensure its effectiveness and fairness for diverse communities, promoting AI for social good through initiatives like Bhashini and collaborations with local organizations and philanthropy."
                }
            },
            "summaryList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 11391,
                        "CandidatesTokenCount": 2206,
                        "TotalTokenCount": 14688,
                        "thoughtsTokenCount": 1091,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Summary": "<html><body><h2>Comprehensive Synthesis Report: Redesigning the AI Economy for Inclusivity and Community Ownership</h2><h3>Introduction: Setting the Stage for an Inclusive AI Summit</h3><p>The Asia Summit commenced with a vibrant welcome, highlighting a shared vision for an AI economy that is both equitable and impactful. Manu Chopra, representing Karya, set the tone by underscoring Karya's three-year mission to leverage AI for dignified learning and opportunities. This initiative has successfully engaged 240,000 individuals across India, focusing on building foundational datasets and ensuring AI models are fair and inclusive for all. A central premise introduced was the current AI economy's design, which, with its existing model structures and operational practices, often fails to adequately serve the needs and realities of the people it affects most. The session's core objective was to explore how India, a nation characterized by over 19,000 unique dialects and subcultures, can lead in making AI decisions that respect this rich diversity, ensuring models speak local languages and are genuinely owned by communities. This vision is a result of a nine-year partnership between Karya and Microsoft Research, with 'Samiksha' as a flagship project embodying this collaborative spirit.</p><h3>Section 1: The Samiksha Project – Community-Centered AI Evaluation</h3><h4>1.1 Context and Rationale: Why Community Evaluation Matters</h4><p>Sunehna Ji from Microsoft Research elaborated on the 'Samiksha' project, a Sanskrit term meaning 'collective analysis or evaluation,' which was aptly named by Kalika. Samiksha stems from nearly a year of dedicated work with Karya, building upon Microsoft Research's five-year endeavor in multilingual and multicultural AI evaluation. The project's fundamental aim is to integrate communities directly into the AI evaluation process, recognizing the critical need for robust benchmarks to assess AI's real-world impact—a sentiment echoed by entities like the Gates Foundation. This community-centric approach is particularly vital for the Global South, where AI systems often carry inherent assumptions rooted in Global North contexts, potentially diminishing their effectiveness and relevance locally.</p><h4>1.2 Persistent Gaps and Contributing Factors in AI Systems</h4><p>Despite significant advancements in AI systems across diverse languages and cultures, notable gaps persist. Even basic tasks, such as English-to-Punjabi translation, demonstrate fundamental errors, highlighting an ongoing struggle with language coverage in many models. The primary factors contributing to these deficiencies are multifaceted: while data skew, heavily biased towards English and Western datasets, is a significant contributor, it is not the sole cause. Decisions made during the model building process and the current evaluation methodologies also play a pivotal role. Effective evaluation is crucial as it provides essential signals during model development, indicating whether interventions are working as intended, especially for rapidly evolving AI systems designed for the Global South. Furthermore, evaluation serves as a tool for prioritizing resource allocation and, most importantly, fosters accountability, offering both short-term performance signals and informing long-term impact assessments.</p><h4>1.3 Unique Challenges in Generative AI (GenAI) and Large Language Model (LLM) Evaluation</h4><p>Evaluating modern GenAI and LLM systems presents distinct challenges:</p><ul><li><b>Output Complexity:</b> Their outputs are inherently difficult to predict and understand, with minor adjustments in input often leading to drastically different responses.</li><li><b>Hyperparameter and Prompt Sensitivity:</b> Model performance is highly dependent on specific hyperparameters and prompts, demanding rigorous control and consistency when comparing different systems.</li><li><b>Open-ended Problems:</b> Many real-world applications (e.g., farmer advisories, personal advice) lack a single 'correct' answer, rendering traditional ground-truth-based evaluation metrics inadequate.</li><li><b>Inapplicable Traditional Metrics:</b> Older AI evaluation metrics, designed for classification systems with clear labels, fail to capture the nuanced performance of GenAI.</li><li><b>System Complexity:</b> The emergence of agentic systems and complex chains of reasoning further complicates the design of effective evaluation protocols.</li></ul><p>These challenges collectively lead to an 'evaluation deployment gap,' where models perform well in controlled benchmarks but falter in real-world deployment. Bridging this gap is Samiksha's core objective.</p><h4>1.4 Validity Concerns and Flaws in Current Evaluation Methods</h4><p>The validity of existing AI evaluation methods has been widely questioned:</p><ul><li><b>Flawed Benchmarks:</b> An Oxford study highlighted that over 400 AI benchmarks were flawed in terms of data, methodology, metrics, or interpretation.</li><li><b>AI Agent Cheating:</b> Reports from NIST revealed instances of AI agents 'cheating' by accessing web resources during evaluation.</li><li><b>Constrained Benchmarks:</b> Many benchmarks are too narrow, leading to invalid interpretations of model scores.</li><li><b>'LLM as Judge' Paradigm:</b> A popular method where an LLM evaluates another LLM's output has been proven to be neither valid nor reliable.</li><li><b>Benchmark Hacking and Leaderboard Issues:</b> Instances of artificially boosting model scores and inherent issues in widely recognized leaderboards undermine trust.</li></ul><h4>1.5 Enhanced Challenges in Multilingual and Multicultural Contexts</h4><p>These general evaluation issues are compounded in multilingual and multicultural settings:</p><ul><li><b>Limited Coverage:</b> Existing benchmarks, primarily English-centric, offer sparse coverage for the 7,000+ global languages, with a bias towards 'high-resource' Western languages.</li><li><b>Lack of Representativeness:</b> Many multilingual benchmarks are mere translations of English ones, failing to capture the unique cultural, contextual, and usage nuances of local languages (e.g., Indian contexts). There is a critical need to build separate, context-specific benchmarks for each language and culture.</li><li><b>Trust and Rigor Issues:</b> Minor changes in evaluation parameters (hyperparameters, prompts) can drastically alter results, demanding stringent fairness and consistency for reliable comparisons. The problem of 'contamination,' where models inadvertently train on benchmark data, further compromises evaluation integrity, particularly for scarce multilingual benchmarks.</li></ul><h4>1.6 The Samiksha Solution: A Ground-Up Community-Driven Benchmark</h4><p>To address these challenges, Samiksha was developed as a robust, community-input-driven benchmark for Indian languages, contexts, and cultures. It is being released on the Karya website with a leaderboard and a detailed 37-page technical report. To prevent contamination, the benchmark data points will not be openly released until all planned evaluations are complete.</p><ul><li><b>Scope:</b> Covers 4 domains (healthcare, education, legal, finance), 11 Indian languages (including Indian English), and has benchmarked 17 Indic and global models to date, with plans for more.</li><li><b>Scale:</b> Comprises 23,000 data points, making it one of the largest multilingual benchmarks. It involved 150,000 human evaluations by Karya community members (surpassing the previous record of 90,000 for 'Pariksha') and over a billion automated evaluations using LLM-as-judge paradigms.</li><li><b>Depth and Breadth:</b> Offers extensive language coverage and deep qualitative analysis, including statistical tests and detailed examination of individual data points to understand rating rationales and model deficiencies.</li></ul><p>The benchmark construction involved three phases:</p><ol><li><b>Community Engagement:</b> Collaborating with 14 Civil Society Organizations (CSOs) to identify relevant community questions within the chosen domains, ensuring the benchmark reflects real-world advice-seeking scenarios rather than expert-level queries.</li><li><b>Response Generation:</b> Generating responses for all 23,000 data points from 17 different models.</li><li><b>Evaluation:</b> Conducting both human and automated evaluations, complemented by qualitative analysis and expert review.</li></ol><p>Samiksha evaluates models not just on accuracy but also on crucial aspects like cultural relevance, perceived trustworthiness, general output quality, and linguistic naturalness for native speakers.</p><h4>1.7 Key Findings from Samiksha</h4><ul><li><b>Content and Language Quality:</b> Most global and frontier models generally perform well on content quality (coherence, answering the question) and language quality, with some exceptions.</li><li><b>Cultural Relevance as Differentiator:</b> The primary differentiator in model performance is cultural relevance; some models excel at providing locally relevant responses, while others do not.</li><li><b>Challenging Domains:</b> Healthcare proved to be the most challenging domain for models, while education (e.g., career advice) was relatively easier. Finance questions involved everyday advice-seeking on investments and loans.</li><li><b>Model Errors:</b> Common errors include awkward language and literal translations of frozen English expressions, which human evaluators rate poorly.</li><li><b>Automated vs. Human Evaluation:</b> LLM-as-judge automated methods diverge significantly from human preferences. However, using human evaluation data from Karya significantly improves these automated methods, suggesting a hybrid approach can provide useful signals for rapid iteration, though not a replacement for human judgment.</li></ul><p>Examples provided illustrated these findings: a low-scoring education response suggested psychometric assessments and job shadowing, irrelevant to the Indian context, while a high-scoring one offered culturally appropriate, actionable advice. Similarly, some Bengali responses incorrectly assumed a Bangladesh context, causing confusion, whereas high-scoring housing advice included India-specific processes and risk warnings. Karya empowers communities by training them to define benchmark content, create data points in their own language and culture, and perform evaluations, effectively giving them agency to determine model suitability for India.</p><h3>Section 2: The Samudai Initiative – Redefining the AI Economy</h3><h4>2.1 The Vision for a Redesigned AI Economy</h4><p>Manu Chopra emphasized that centering communities in AI model evaluation is paramount for determining what truly works for them. He highlighted that Karya's communities are not just passive beneficiaries but active builders and evaluators of AI, challenging previous notions about their capacity for complex evaluation. This underscores the need to redesign the AI economy to centralize communities throughout the entire AI value chain.</p><h4>2.2 Bhashini and Samudai: Government Initiatives for Language AI</h4><p>Amitabh Sir introduced 'Bhashini' (Bhasha Interface for India), a government initiative launched three and a half years prior, aimed at solving fundamental problems in language AI. A critical challenge was the scarcity of digital data for most Indian languages, with only 6-7 languages having minimal (2-3% of total internet data) digital presence. Bhashini addressed this through a "
                            }
                        ]
                    }
                }
            ]
        }
    ],
    "headerCols": {
        "col1": "Part Id",
        "col2": "FileName",
        "col3": "Content",
        "col4": "Status"
    },
    "fileName": "",
    "projectType": "VIDEO",
    "sourceLanguage": "English",
    "targetLanguage": "English",
    "__v": 0
},
{
    "_id": "69930ad34c5e8f0974c877a9",
    "projectId": "814788",
    "projectName": "National Launch Event: AI Readiness Assessment Methodology (RAM) Report India  AI India Summit 2026",
    "projectDescription": "National Launch Event: AI Readiness Assessment Methodology (RAM) Report India  AI India Summit 2026",
     "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-16T12:17:22.031Z",
    "projectUpdatedDate": "2026-02-16T12:17:22.031Z",
    "projectStatus": "Active",
    "companyId": "69803c46f08f26d9c13921b1",
    "projectCreatedBy": "69803c46f08f26d9c13921b2",
    "projectUpdatedBy": "69803c46f08f26d9c13921b2",
    "projectLastAccessDate": "2026-02-16T12:17:22.031Z",
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "Content",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "India is Ready for AI_ Launch of the National AI Readiness Assessment Report(MP3_128K).mp3",
            "col3": " Readiness assessment that we are launching here today. This RAM is not merely a technical order, it's a diagnostic tool designed to ensure that the impact we discuss this week is inclusive, ethical and sustainable. By linking these readiness findings to the summit's efforts, we aim to transition from high-level ambition to a grounded, actionable roadmap for the nation. So as we heard the reporters...  The outcome of sustained efforts over the last 18 months with many stakeholders. We have five regional consultations across Delhi, Bangalore, Hyderabad and Guwahati. More than 600 stakeholders, including government, start-ups, research, civil society, helped to build a grounded picture of where India stands today and where it might want to go next. I would like also to thank the India A.I. Mission and METI.  in particular for their partnership on the RAM initiative. In particular, Secretary Krishnan and Mr. Adi Singh's team have been incredibly supportive throughout this endeavour, and we are excited about our joint collaboration to advance the ethics of AI in India. Of course, in a country as vast and complex as India, inclusivity in AI is foundational.  With 1.4 billion people, thousands of languages, and deep regional and socioeconomic diversity, the question is not whether AI will shape our daily life, but how evenly or unevenly it might do so if we are not deliberate about it. AI is already mediating access to welfare, employment, education, healthcare, justice. In this context, the design choices behind these systems have the power to determine who benefits and who is left behind.  When an algorithm determines who is shortlisted for a job, approved for a loan, or prioritized for public services, its choice carries real human consequences. And behind every data set is, of course, a citizen. Behind every algorithm, a life shaped by its outcome. So UNESCO's recommendation on the ethics of AI matters because it takes a clear baseline for how AI should be used, and in ways people can trust.  And the RAM helps map what's already in place and what might need strengthening. So while the detailed findings will be shortly presented by my colleague, let me just offer a quick sense of what the RAM tells us about India's overall picture. And I must say we've done this RAM now I think in 70 countries, more than 70 countries around the world, so it's giving us a collective data set. And India really does have a strong space for responsible AI.  The country accounts for about 16% of the world's AI talent. There is real momentum on multilingual AI and digital public services through efforts expanding Indian language data sets and models. And the innovation system is vibrant and supported by strong digital foundations. At the same time, there are areas that could improve according to the assessment. Access to an ethical use and sharing of data could be strengthened.  Environmental sustainability needs to be built more directly into AI infrastructure planning, and ethics is there but could be more consistently reflected in education and training, including regional language offerings. So building on all of these findings, the report sets out a few priority actions to address the above gaps and translate readiness into responsible adoption across sectors.  The India AI mission alongside key institutions such as the Office of the Principal Scientific Advisor play key roles in this exercise. So India's AI future will be shaped not just by what is built but how it is built, who it serves and what safeguards are embedded from the start. We should aim for AI that listens before it decides. AI that performs well not only in demos but in the lived complexity of real communities.  And herein lies the real challenge. The RAM report offers one practical path, one of many. It reflects what data the stakeholders are telling us at the moment, and it can translate into actions that the various stakeholders, government, industry, civil society can take forward together. So I'd like to once again thank the Government of India and all our partners for their unwavering commitment to this journey. And we look forward to continued collaboration on what will define India's leadership.  Thank you, Mr. Curtis, for setting the context so thoughtfully. I think what you said about being deliberate in our approach really resonates with some of the findings that we've got here. It now gives me great pleasure to invite Secretary Krishnan, Ministry of Electronics and Information Technology, Government of India, to deliver the opening remarks. Secretary Krishnan, I don't need to say this to this room, has been instrumental in steering India's digital public infrastructure and AI governance vision. So thank you for your presence.  Thank you very much and thanks to UNESCO, the agency which has worked on this readiness assessment report. And thank you to the principal scientific advisor for being present here today. And more than anything else, thanks to all of you who are participating as delegates as part of the India AI Impact Summit.  which I think has gotten off to a really, let's say, a roaring start. And all of you are here, and there's been extensive participation, a great deal of enthusiasm, a great deal of energy everywhere that we go. And it's wonderful to see this. This presumably also reflects part of our readiness to end A as AI and to sort of, at least.  the curiosity, the need to find out what it is all about, and the way that we can actually adapt it to what India's needs are. I think that is truly what the response to this event reflects. Of course, as was pointed out, this is an effort of over 15 months. I recall I was at an event about a year and a half ago at Piyamaskar.  to sort of discuss on how this will go forward and I'm very glad that today we are seeing the final outcome of the readiness framework. It would undoubtedly help to inform what we in India are attempting to do in the AI systems. Fortunately, the way we have designed the India-AI mission is flexible and we can sort of accommodate any mid-course corrections and changes that we need to do.  And the vision of the AI mission, apart from of course the AI summit itself, was fundamentally to make sure that we, without, I mean, given the constraints of how much public resources would be actually expended on AI, what is it that we can do to create the environment? So whether it is the approach to creating the AI compute, whether it is the approach to skilling, whether it is the approach to creating the models.  or whether it was the approach to create on the databases. It was very much grounded in something we wanted to do with partnerships. The government is not the only agency which is working in many of these spaces. We have a number of partners, a number of stakeholders who are working with the government to enable this. In the way that we are going out the AI compute, we are doing it through a mechanism whereby the private sector steps in to actually build it out.  and re-enable access to those who need it the most by subsidizing the access. Now this is an approach which would possibly fit many other countries who don't want to invest billions and trillions in actually building out the infrastructure at one go and are not sort of put off the idea of trying to access AI because by saying that it just costs too much and we can't actually do it.  I think there is an alternate approach, which is what India is now championing, to see the access to some of the infrastructure of the areas. Likewise, I think the skilling part of it is a very important part of the story, and I was very happy to note the comments that have been made and recommendations which have been made in the report on the skilling. We have to address it at every level, in existing organizations and institutions, how we still move forward.  for those who are still in schools and colleges in terms of what backgrounds they will need to move forward in this space. And more than anything else, I think the readiness of organizations and institutions to actually enterprise us to adopt that. Because ultimately, unless we have applications and unless we have use cases converted into actual solutions.  your acceptance of AI into enterprises is just not going to be there. Unless that acceptance takes place, you don't see the real world in that. Ultimately, what's the point of building a model? What's the point of building a huge computer? What's the point of setting up huge databases if it doesn't translate into something which is a practical utility in the real world? And that is the ultimate goal which we are attempting to reach. And in some ways,  I think this report to some extent validates the approach we've taken and actually gives us much more of an impetus to pursue the line that we've taken to sort of redouble our efforts on some of the specific parts of it and it's like a report card on some of the stuff that we've done and so thank you very much for sharing this with us it would be a very useful  commentary on what has happened and the way that we push this forward. And the work that you have done here, I'm sure, is something which can be translated to other parts of the world, to other countries in the Global South, for it to be really useful as to what the playbook which needs to be developed, how they can actually work on it in the future. So thank you again, Mr. Kerkus and the entire UNESCO team for everything you've just done. Thank you.  Thank you for your remarks and for your openness to receive inputs along the journey. Thank you so much. Next, we are honored to have with us today Dr. Ajay Kumar Sood, Principal Scientific Advisor to the Government of India to deliver the keynote address. Dr. Sood's leadership in advancing science, innovation and evidence-based policy continues to shape India's strategic direction in emerging technologies.  Thank you very much. Mr. Tim Curtis, Mr. Krishnan who was here two seconds back. All the AI enthusiasts in the room. You can see this room is full of energy. You can see what is happening outside. It really tells that AI is here to make an impact in our lives.  We are delighted to be here on the launch of this India AI Readiness Assessment Methodology report. And it has been a privilege for me and my office, Office of Principal Scientific Advisor, to be associated over the last two years or two and a half years very intimately with this AI scene in the country, helping to shape it.  because this was discussed in the Prime Minister's Science Technology Innovation Advisory Council which I chaired and it was very clear that we need a mission mode to really see the benefits of AI to the society and I'm so glad that we can now see that it is bearing some fruits and much much more has yet to come.  So we know that AI brings enormous opportunity. It's a no-brainer to say this at this time. But we also know that it also carries profound ethical, social, environmental, and human rights implications. And its inherent opportunities come with responsibilities. And that's why UNESCO's Readiness AI Assessment matters a lot.  And this is the right time for it. So the real question before us is not simply how fast we can adopt AI, but how thoughtfully we can shape it. So this is something for all of us to see. First, how do we ensure that AI truly serves people guided by a clear do no harm principle? And the second,  How do we make it fair, trustworthy, inclusive and align with constitutional values, human dignity and fundamental freedoms? So these are the two basic questions which all of us need to answer. So presently at this point, that is why this assessment methodology RAM is a good word from semiconductor technology.  It will be very catchy. I can assure you when we say ram people will be confused Is it nesco ram or is it Intel ram? So there will be a confusion, but it's good confusion For this report. It's very relevant at this time Because it offers a structured way for countries to reflect on their preparedness across legal social economic scientific  educational and technological dimensions. So many dimensions we have to see when we look at the assessment of this AI. So it also operationalizes global ethical principles into concrete actionable policy guidance without hampering innovation. So at today's launch, I thought time will be very brief in seven minutes which have been given to highlight three core priorities.  for advancing ethical AI so let me just say that these are not absolutely new which you have heard at this thing but let me say in my own words what it means because it is never harmful to say the good thing twice so I'm sure you will appreciate because it reinforces our commitment to this first AI ethics cannot be an afterthought  within the AI system from beginning. So that has to be understood. We have recently come out, perhaps most of you know, with a techno-legal framework from our office, namely strengthening AI governance through techno-legal framework. And we believe that the grounding principles of ethical AI, namely respect for human rights,  fairness, transparency, explainability, privacy, safety and accountability must be embedded across the entire AI life cycle right from data governance, model designs, training inference to final deployment with human oversight. So human has to be in the loop if we don't want catastrophe. So this is something  which we have articulated in our AI governance framework which was released in November 2025 and more recently in the techno-legal framework which came out a few days back. So as we scale AI across sectors, these principles must be reflected in standards, procurement frameworks, impact assessments,  and institutional mechanisms in all of them we need to implement this and this aligns obviously with UNESCO's call for right based policy measures and legal norms that keep pace with technological change and that is very essential because public trust depends on this and trust is the key thing which you have to realize that the technology doesn't get killed.  but it gets deployed. So trust is the key issue which has happened with other technologies which did not take off because of the trust. So this is something all of us should be very, very keenly aware of. Second, inclusions must remain central to India's AI journey. AI should work for all Indians across regions.  languages, genders, abilities, and socio-economic backgrounds. As was mentioned, we have a huge diversity in all of this. So it's an absolutely good platform and a sandbox to really see what works and what doesn't work. And AI should help bridge divides and not widen them. This is something is a good matrix to see when we see the deployment of AI.  and it must be accessible in locally relevant culturally sensitive format in an equitable manner for all. So please note culturally sensitive is very important which can be missed if you are blindly following AI developed somewhere else. So this has to be borne in mind because this will not come just by saying it unless you have the right data which is culturally sensitive it will not happen.  So capacity building, especially for states, small enterprise startups and underserved communities is equally important. And these insights will be crucial as we move from assessment to action. And the third point, the last point, collaboration is our biggest trend. So no single stakeholders can shape the future of AI alone. And we need government, we need industry, we need academia, civil society, research communities.  All of us have important roles to play. And this collaborative approach is what the India AI mission has taken, particularly its safe and trusted AI pillar and India's AI governance guidelines. So the convergence between UNESCO's global ethical framework and India's national priority gives us a robust foundation.  for leadership and not just adoption of AI. So the recommendation in this report, I'm very sure, will offer a practical roadmap strengthening institutional coordination, enhancing regulatory and ethical oversight, investing in talent and research, promoting data, media, and information literacy, and deepening public engagement on AI, on all these factors.  I hope this report will cover it, which it does. I didn't have a chance to look at all the things, but I'm sure it will cover them. So at this juncture, it's a pleasure to acknowledge my appreciation on record for UNESCO, the India AI Mission and MIT, which is really doing a phenomenal job.  uh mighty leadership with minister rational very very intimately engaged with all the stakeholders with our office all of us are in this together uh the ikegai law which is a part of this and most importantly all the stakeholders who contributed their perspective and expertise so thank you very much the next phase obviously is the implementation phase and i'm sure  This is the next phase of the outcome of assessment matrices and I am sure we will meet one year from now to see how the outcome of this assessment is done and that what will be the proof of the pudding. So I look forward to that and I would only say this is a wonderful moment. Let us seize this moment.  or to really see that AI makes a difference to the society, to the last mile and we hope that inclusive AI will be known and not an exception. Thank you so very much.  And I think the question everyone has on their minds right now is what does the report really say? And to walk us through the key findings and recommendations emerging from this national assessment, it is my pleasure to invite Ms. Yumsong Kim, Program Specialist and Chief of Section, Social and Human Sciences, UNESCO Regional Office for South Asia. Ms. Kim has closely read the RAM process in India and will present the report's key insights and policy recommendations.  Hello, everyone. It's amazing to see a full room. And I'm very excited to present the findings of the RAM in India. You must now be sick of hearing the RAM and wondering what is this? What is the RAM? So it starts off by UNESCO's interpretation of AI readiness. When UNESCO looks at AI readiness, it goes beyond infrastructure. We interpret readiness in terms of governance.  in terms of institutional frameworks, in terms of human capital readiness. So that is what we have tried to capture in the India AI RAM Report. As the structure of the India AI RAM Report is divided into two sections, we start off with the diagnosis, which has all the data that we have collected quantitatively through the indicators across 200 data points.  of the five dimensions the principal scientific advisor has mentioned, as well as all of the qualitative data that we collected through five stakeholder consultations across the country.  which we thought was important in a country as diverse as India. And then the second part is culminating all of these different inputs into the policy recommendations. We have eight policy recommendations, and each of the recommendations are linked with a concrete roadmap that is realistic, actionable, and also builds on the existing AI infrastructure and ecosystem in India already. So as you can see, we've conducted the secondary research as well as the  Tim mentioned in his speech as well that we were able to reach inputs from more than 600 stakeholders across the country and we were very very happy that we did it in cities like Bangalore and Hyderabad which has a very mature AI ecosystem but also in cities like Guwahati in the northeast where we got very different inputs and insights where regional diversity was an important aspect that we have reflected in the RAM report.  I'll get on with the task of summarizing a 120-page document into a 10-minute presentation, but let's see how well I can do. So the first dimension is what we call the legal and regulatory dimension. If you look at India, it's very much a hybrid AI governance framework, combining cross-sectoral laws, sector-specific standards, and voluntary guidelines, led by the ministry, METI, the PSA's office, NITIO,  and many other government bodies. It also has quite a strong cybersecurity readiness, as you can see in ITU's Global Cybersecurity Index. And just to let you know, the report is being launched today, so you can see all of these data points within the report. And it also has a very strong online participation readiness. It still has quite an evolving data ecosystem, reflected in a moderate open data performance.  And then the stakeholders are telling us that they support the government's principle-based approach and cautioned against overly prescriptive regulations, which I think the secretary also mentioned today. There was a strong call for a gap analysis in the existing laws to identify what the existing laws cover and what are the existing gaps that probably future research can address.  And definitely the experts called for a stronger state and center alignment in terms of AI initiatives. Now we go on to the social and cultural dimension, which I think is quite a unique aspect of the UNESCO RAM. We make sure that this is a part of readiness when we talk about AI readiness because we believe that there is no one size fits all kind of prescription.  There's definitely a strong momentum in India for multilingual AI as we can see in Bosni and different initiatives digital access to AI continues to expand but there's still a bit of a large issue in terms of the gender  gap especially in internet access. There's room for improvement in STEM inclusion especially for young women and girls and that's a very important part that we try to reiterate in the recommendations because without the STEM pool it's also a narrow pool going into the AI talent pool. There's room for improvement especially in terms of sustainability and the environmental impact assessments which is a strong recommendation from the report.  The experts continue to say that socially beneficial AI projects were receiving insufficient funding and what they mean by socially beneficial AI projects is things on sustainability on multilingual languages. And that's where we already quite see the government directing towards and making sure that local language data is collected in AI.  Again, the experts also echoed the importance of environmental impacts and making sure that is considered in the future development of AI of India. In terms of the scientific and educational dimension, there's a very, very strong innovation momentum in this country. We found that over 86,000 AI patents were filed between 2010 and 2025, and that was over 25% of all of the...  tech patents in this period, which is absolutely enormous. There's an expanding AI research ecosystem led a lot by the center of excellences led by the government and a very robust national AI skill scaling initiative as you can see here. There is definitely room for improvement. First of all, a stronger mainstreaming of AI ethics into the education system. When a lot of the initiatives talked about AI literacy,  We found that AI ethics wasn't necessarily always included. So that was something that we found was scope for language. And that made it difficult for some of these students to access so much of the opportunities that was out there in India. The stakeholders also corroborated a lot of these findings, such as AI skilling still being remained very urban centric and very English. So what about the students and informal workers who are not in the urban areas?  nor is very fluent with English. So that was another very important finding in the educational dimension. For the economic dimension, I hope I'm on the right page, there is a very, very strong global positioning of India. It hosts 16% of the world's AI talent, which is huge. And it is one of the fastest growing AI talents across the world. It has a very vibrant industry ecosystem.  with 2,000 plus companies only in Bangalore. So across the countries there's way more, which is a very, very positive sign. But again, in terms of distribution, outside of the urban tech hubs, there's still a lot of room for improvement. Access to financing for a lot of the startups beyond Bangalore, beyond Hyderabad, beyond Ali, will still continue to be posted as an issue that can be resolved.  Stakeholders also ask for targeted incentives to encourage firms to be more ethical in their approaches to inclusive AI development. A lot of the stakeholders are saying ethical AI is great.  But if you don't give financial incentives or any kind of incentives for the firms to follow, it's much more difficult for these firms to consider the ethical dimensions. The informal sector still largely remains outside of the AI ecosystem or of all the discussion we have here. And we must recognize that India still has quite a huge large informal economy sector that should be considered.  In the technical and infrastructural dimension, there was a strong national focus on building accessible, affordable AI infrastructure. But in terms of improvement, things like electricity in certain areas where there's not constant electricity.  could be a problem in accessing a lot of these infrastructure, AI infrastructure. We definitely welcome, stakeholders welcome the, in the AI missions, GPU procurement initiatives. But again, without sustainable or continued electricity, some of these issues were pointed out as there are room for improvement.  Now let's go on to the recommendations. The way we decided or came up with these recommendations are definitely based on the 200 data points that we collected through secondary research. It's corroborated with the stakeholder consultations that we conducted and we tried to also understand the existing policy landscape so that we don't recommend something that doesn't align with the government's initiative. We also wanted to make sure that they are realistic and actionable and you'll see here that each of the recommendations  have concrete roadmaps. So the first recommendation is to ensure informed AI governance across sectors through comprehensive evidence-based risk and legal review. What do we mean by that? In the roadmap, we suggest three things. Number one, let's do an AI risk mapping. The AI Safety Institute, which exists under the India AI Mission, can lead with academia, industry...  second recommendation is to create operational guidance for ethical AI. When we talk about ethical AI, we have a lot of principles, we have a lot of frameworks, but can we make something that's much more relevant and actionable to day-to-day lives of policymakers and developers? That's why we recommend  that the AI Safety Institute creates, first of all, an ethical AI repository with examples and case studies for people to really understand what specific use cases in ethical AI looks like, also have a voluntary measure, especially with industry partners to implement them, and experiment with regulatory sandboxes to test some of these responsible AI approaches, especially in high impact sectors like healthcare, agriculture,  and pharmaceutical areas. The third recommendation, which came out quite a lot, I must say, in the stakeholder consultations, is to strengthen the center and state coordination for AI governance and implementation. There are already emerging actions from the government, but dedicated center-state coordination mechanism  could help make sure that the AI policy and planning at the state level are informed from the voices and opinions of all of the different states in India. And also to develop modular policy implementation toolkits for states to make it much more easier to action on these policies. Next is leveraging the AI Cosh platform, which already exists and has huge potential in terms of data in this country.  a lot of efforts already underway and we recommend to strengthen some of these existing offices such as the AI data office, such as the data creation units in different parts of the government to make sure that AI COSCH can fulfill its full potential and we go a step ahead and also ask the private sector to contribute to AI COSCH so that we can make sure that this is very very useful for those inside and outside of the government.  Next recommendation is about trust. How do we increase public trust of AI governance? We recommend that the AI Safety Institute can guide some of these initiatives, for example, have a nationwide study that can inform some of the campaigns and literacy programs, and also have model AI procurement guidelines that ensures and gives that trust to people that when AI is being procured, it is being done in an ethical way.  Recommend six is all about skilling and workforce, and we have two parts to it, but I'll briefly go through it. There was a strong, strong call for making sure that India's workforce is AI future ready through capacity building, embedding ethics, and supporting inclusive transitions. There was a lot of anxiety and concerns around especially mid...  career workers on re-skilling and up-skilling, it's not as easy as it sounds, to suddenly embed all these new skills. So that was a strong call from a lot of the experts we talked to, and we feel like the Skills Development Ministry, the AI Safety Institute, the Ministry of Education can definitely play a role in pacifying some of these anxieties through platforms, through integrated up-skilling initiatives.  And yes, this was another very important point. AI capacity building for public officials was called for again and again all across the government at center and state level, how important it is for public officials to truly understand, at least have basic AI literacy in order to implement some of these public initiatives, but also policies.  The seventh recommendation is to ensure environmental sustainability considerations into AI infrastructure all across based on comprehensive studies and making sure that environmental impact assessments is truly meaningful rather than just tick in the box. Last but not least, we talked about diversity in the AI workforce. We had another session that documented really the...  the data on how women are not being as integrated into the AI ecosystem as we think it is. And there's definitely scope for improvement, especially going beyond the urban centers of India. So this takes me to the end of the presentation. The report will be available on UNESCO's homepage from today. So I hope you can all have a look and hope it's useful to the AI ecosystem in India. Thank you so much.  Thank you, Yunsong. I know that was a very difficult task to condense so much into such little time. I'm sure people have questions for you, but they can read it online, like you said. We'll now quickly move to the final segment of today's program, something I'm very excited about, a fireside conversation on global perspectives in ethical AI governance. It's my pleasure to invite on stage Dr. Mario Grazia. She's the...  Chief of Executive Office, Social and Human Sciences, UNESCO, and has been very closely involved with the RAM in India. And Mr. Vilas Dhar, President of the Patrick J. McGovern Foundation. The McGovern Foundation has been supporting the RAM exercises in India and globally. This conversation will reflect on the global implications of ethical AI frameworks and the role of multilateral collaboration in using and shaping trusted AI ecosystems. Over to you. There we go.  So it's really my pleasure and honor to be here with Dylan. Actually, there are three words in my mind that always remind me when I see Dylan. And these are the President and Trustee of the Montgomery Foundation. The first one is insightful. There hasn't been one single instance in which I was talking to Dylan, and then I thought, oh, I haven't thought about that. And this doesn't happen at all. So that's already one thing really that's always...  Help me, it's a great pleasure to be with you. The other one is that he's a visionary. He has been seeing this transformation, he has been seeing the challenges coming with it, and the opportunities, of course, much before others, and has taken a really active role in that, in steering it towards the right direction. He's committed, always. The energy you put in that, the kind of the determination, but also the unusual way of thinking on how to address the problem is something that always  You are striking me all the time by the time I've been talking to you. So I won't have easy questions for you today because we really want to get the best out of your brain. So the first one is, as the song was brilliantly showing us, this report is very rich and it shows that there is a lot of strength in India, but there are also real complexities. And I mean, scale is not the only one. So...  You have supported UNESCO and a lot of your work anyway looks at what are the factors that might be the ecosystem that might be conducive to AI that is ethical and how to make the best out of AI that really goes for people. In your mind, is there one example or one thing that India could adopt, replicate, tailor in order really to set up already let's say the transformation is very visible and it's very much there?  But what would be perhaps something, an addendum or an approach that you would suggest that India takes? Well, thank you so much for that very kind introduction. Too kind. You're supposed to set a low standard so that I can exceed it, not set a very high standard. I want to thank you for this invitation to UNESCO. And I also want to acknowledge Secretary Krishnan, who I know had to go, but also you, Dr. Soot. I want to say one scientist to another. I think it gives the world a great deal of confidence and comfort to know that  You really have the kind of scientific background that's coupled with a moral and a conscience driven approach. Your techno-legal frameworks, as you know, have been picked up in countries around the world already. I want to thank you for that leadership. Your question, I think, is wonderful, right? But I'm actually going to flip it for you. And I'm sorry you know me well. Rather than talk about what India should adopt, I actually first want to talk about what the rest of the world should adopt from India. Because I think this is not a story where we're saying, what do we start here and figure out who else in the world is better? Well, your readiness to present that shows that India is actually far ahead of us.  I think one of the things we should look to is a re-conceptualization of what AI governance is actually about. When you go to many places in the world, AI governance is seen as a way of limiting risk. Risk for institutions, liability for people, minimizing models. And in doing so, you miss the amazing opportunity that's possible. When you think of AI governance as a way to incentivize a positive vision for the possible, you do that through a set of interventions that go beyond minimizing models.  that go beyond restricting the worst possible outcome. You instead think about how you incentivize and name and elaborate a whole of society approach to AI. One that centers the idea of participatory mechanism. One that allows for people to engage in, as you have done with 600 consultations across the country, to let people not merely be the subjects of an AI governance regime, but rather to be the architect. We've seen that happen in India in the DPI stack, where we saw in digital public infrastructure the idea that  While there were guidelines and guardrails that were put forward, much of the innovation actually came from a combination of private and public sectors, who came together not merely to operate inside of a constraint, but to test it out. To say, how can we consistently deliver services to those who are left out of the market? And yet, in doing so, also build a framework that can then be picked up, abstracted, and shared with the world. So this is my answer to you. I would say what the world needs to learn is a governance principle that doesn't start from  heavy-handed restrictions, but rather from the light touch that's been espoused here in the country, but allows for the incredible innovation of 1.4 billion people to save the world's vision of what to happen. Absolutely. And actually, one strong component of that is really the skills, the endowment, the human capital, the humanity of this country. In your mind, what was, because we hear way too often, and the report also mentions that, in relation to the skill, the AI skills, and I've always been saying, to me, it's more skills for the digital era.  What do the set of skills for the digital era look like for you? And what is the role you see for awareness? Because I think a lot of the action, a lot of the directions that the development of it depend also on what we are aware of or might not be aware of. Look, I think the first part of the story was we have to build technical skills and scale across the population. We very quickly realized that doesn't make sense. We don't need 1.4 billion AI scientists.  We barely need a few hundred thousand. What we need instead, though, is a population that understands the opportunities that's available. In the past few months, I've had the opportunity to travel across India and across the world, as you know, visiting those top tier institutions, which are known for their engineering prowess, and having great conversations with students and faculty. But you know where I find inspiration is when you go to those other institutions. They're often termed tier two or tier three. I reject the premise. But here in India, institutions like PSG.  like Maika, like Amrita University, places where students come together in an environment where they have access to the underlying ecosystem of AI. But then they bring their lived experience from communities across the country. They may or may not be able to sit down and understand exactly how to build the next frontier model, but they do know that their parents are struggling with drought. They know that agricultural productivity is at risk. They know that the electricity grid is not functioning the way it should.  They know that disasters and crisis from mudslides all the way through to macro level events affect them. And in that moment, something magical happens. Because the skilling here is no longer about whether you can describe about your model. So you can take the tools that are available and build real solutions. Solutions that operate at low cost, sometimes in low power and low connectivity environments, that deliver real value for people. Somebody agrees with me. I like it.  India will go from about 1.4 gigawatts in data center capacity to projected 9 to 10 gigawatts in the next century.  And when you look at that and you understand it, on one side there is a wonderful storyline of companies investing and building out the maps. On the other is the communities where those data centers live, the environmental impact on the sustainability issues that will arise. Now, the conversation that I'm asking you to have is not to name those risks because we already know them, but rather to think about what a uniquely Indian model might be to address them. One where you center economic sustainability as a frontline matter, not as something you deal with as a factor.  thinking about mandatory disclosure instead of voluntary definition of the impact on water and local power plants, of building mechanisms that say fresh water cannot be used for cooling of data centers unless there's enough in the community already to go around, because we know there are commercially viable alternatives for air cooling or other things. And finally, to note that even in India, and especially in India, the fair tariff on renewable energy are actually lower than where natural gas is. What would happen if we were to mandate that in order to put a data center in a community?  You not only have to commit to the increasing purchase of renewable energy, but commit to building that capacity. This is the opportunity here in India, that as you begin this incredible ramp up, can you use those techno-legal frameworks to actually build a new mechanism and show the world what sustainable data center faces. Absolutely, and this would also speak to the issue of inequality, which was raised actually during the presentation. So, suppose you can decide tomorrow, put one thing in motion, and you have unlimited resources.  What would you do? What would be the first thing that you're going to do? You know what? It's the easiest question in the world because what I would do is ask communities across the country what they prioritize in the ring as their primary internet. But once you've asked them, huh? No, that's great. But I mean, I don't want to skip over it, right? I don't want to be flip about it because one of the things is when we name participatory AI,  We often skip over the fact that that requires participation. But once you ask them, I think there are a couple of things that I would advise. The first is really investing in this idea of translational AI. Take what models enable and allow, and invest meaningfully in building the architecture that brings it to communities in the form of apps and services, in the form of insight and wisdom that actually lets us change these things. I think we could look at priority sectors, and we've done that already. I think in the report you identified them.  but focus also on where you see innovation centers that are organically emerging across the country and invest in them deeply. So I think those of you in the room that can do something tomorrow might want to stop thinking about this. Now one last question because time's up and I see it very clearly unfortunately. But there is one thing we were talking before in the other panel about the different trends that AI is taking and actually we were discussing also about agents which are you know perhaps now the kind of developments that  excites many, but frightens others. And the question is, well, they've been with us now for some time, but we haven't seen the scale of adoption that we would expect. In your mind, given that you've been analyzing this thing, what is the big topic? And should we start thinking about dynamics like the ones we have seen, for instance, with the Internet of Things and the kind of failing to develop a being adopted at the pace we would expect?  It's a great question. And this is a really nice story. I'll be very quick about it. We saw the first generation of agents to do some of the most boring things you can imagine. Companies that build agents to increase profitability by redeploying your CRM, by working in your workspace. And can anybody in this room really say they were excited about that? I don't think so. But we saw a new phenomenon happen just in the last few weeks. We saw hobbyists come forward with an open platform for building agents. And it took the world by storm. Why? Because we allowed people to be creative in their use.  We let them bring agents into their lives and solve their problems. And with that, we saw hundreds of thousands of people deploy agents in scale. We even saw all kinds of weird effects like agents talking to each other online. And even most interestingly, now a service that lets agents autonomously go and contract a human being to go and do a task for them in the real world. This is the future, right? Agentic AI is not a corporate term, and it cannot become one. It has to be a way of extending human agency in ways that the powerless have never had.  a way of democratizing access to the ecosystem of both economic success, but also environmental, political, and economic success. Join me in thanking Vila D'Arcy. I think we've been cut off. Thank you, everybody. We've been cut off.",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771244221309.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:29",
                    "text": " Readiness assessment that we are launching here today. This RAM is not merely a technical order, it's a diagnostic tool designed to ensure that the impact we discuss this week is inclusive, ethical and sustainable. By linking these readiness findings to the summit's efforts, we aim to transition from high-level ambition to a grounded, actionable roadmap for the nation. So as we heard the reporters..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:29",
                    "end": "00:00:57",
                    "text": " The outcome of sustained efforts over the last 18 months with many stakeholders. We have five regional consultations across Delhi, Bangalore, Hyderabad and Guwahati. More than 600 stakeholders, including government, start-ups, research, civil society, helped to build a grounded picture of where India stands today and where it might want to go next. I would like also to thank the India A.I. Mission and METI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:57",
                    "end": "00:01:21",
                    "text": " in particular for their partnership on the RAM initiative. In particular, Secretary Krishnan and Mr. Adi Singh's team have been incredibly supportive throughout this endeavour, and we are excited about our joint collaboration to advance the ethics of AI in India. Of course, in a country as vast and complex as India, inclusivity in AI is foundational."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:21",
                    "end": "00:01:50",
                    "text": " With 1.4 billion people, thousands of languages, and deep regional and socioeconomic diversity, the question is not whether AI will shape our daily life, but how evenly or unevenly it might do so if we are not deliberate about it. AI is already mediating access to welfare, employment, education, healthcare, justice. In this context, the design choices behind these systems have the power to determine who benefits and who is left behind."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:50",
                    "end": "00:02:19",
                    "text": " When an algorithm determines who is shortlisted for a job, approved for a loan, or prioritized for public services, its choice carries real human consequences. And behind every data set is, of course, a citizen. Behind every algorithm, a life shaped by its outcome. So UNESCO's recommendation on the ethics of AI matters because it takes a clear baseline for how AI should be used, and in ways people can trust."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:19",
                    "end": "00:02:48",
                    "text": " And the RAM helps map what's already in place and what might need strengthening. So while the detailed findings will be shortly presented by my colleague, let me just offer a quick sense of what the RAM tells us about India's overall picture. And I must say we've done this RAM now I think in 70 countries, more than 70 countries around the world, so it's giving us a collective data set. And India really does have a strong space for responsible AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:48",
                    "end": "00:03:15",
                    "text": " The country accounts for about 16% of the world's AI talent. There is real momentum on multilingual AI and digital public services through efforts expanding Indian language data sets and models. And the innovation system is vibrant and supported by strong digital foundations. At the same time, there are areas that could improve according to the assessment. Access to an ethical use and sharing of data could be strengthened."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:15",
                    "end": "00:03:43",
                    "text": " Environmental sustainability needs to be built more directly into AI infrastructure planning, and ethics is there but could be more consistently reflected in education and training, including regional language offerings. So building on all of these findings, the report sets out a few priority actions to address the above gaps and translate readiness into responsible adoption across sectors."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:43",
                    "end": "00:04:14",
                    "text": " The India AI mission alongside key institutions such as the Office of the Principal Scientific Advisor play key roles in this exercise. So India's AI future will be shaped not just by what is built but how it is built, who it serves and what safeguards are embedded from the start. We should aim for AI that listens before it decides. AI that performs well not only in demos but in the lived complexity of real communities."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:14",
                    "end": "00:04:45",
                    "text": " And herein lies the real challenge. The RAM report offers one practical path, one of many. It reflects what data the stakeholders are telling us at the moment, and it can translate into actions that the various stakeholders, government, industry, civil society can take forward together. So I'd like to once again thank the Government of India and all our partners for their unwavering commitment to this journey. And we look forward to continued collaboration on what will define India's leadership."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:45",
                    "end": "00:05:20",
                    "text": " Thank you, Mr. Curtis, for setting the context so thoughtfully. I think what you said about being deliberate in our approach really resonates with some of the findings that we've got here. It now gives me great pleasure to invite Secretary Krishnan, Ministry of Electronics and Information Technology, Government of India, to deliver the opening remarks. Secretary Krishnan, I don't need to say this to this room, has been instrumental in steering India's digital public infrastructure and AI governance vision. So thank you for your presence."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:20",
                    "end": "00:05:49",
                    "text": " Thank you very much and thanks to UNESCO, the agency which has worked on this readiness assessment report. And thank you to the principal scientific advisor for being present here today. And more than anything else, thanks to all of you who are participating as delegates as part of the India AI Impact Summit."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:49",
                    "end": "00:06:18",
                    "text": " which I think has gotten off to a really, let's say, a roaring start. And all of you are here, and there's been extensive participation, a great deal of enthusiasm, a great deal of energy everywhere that we go. And it's wonderful to see this. This presumably also reflects part of our readiness to end A as AI and to sort of, at least."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:18",
                    "end": "00:06:46",
                    "text": " the curiosity, the need to find out what it is all about, and the way that we can actually adapt it to what India's needs are. I think that is truly what the response to this event reflects. Of course, as was pointed out, this is an effort of over 15 months. I recall I was at an event about a year and a half ago at Piyamaskar."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:46",
                    "end": "00:07:13",
                    "text": " to sort of discuss on how this will go forward and I'm very glad that today we are seeing the final outcome of the readiness framework. It would undoubtedly help to inform what we in India are attempting to do in the AI systems. Fortunately, the way we have designed the India-AI mission is flexible and we can sort of accommodate any mid-course corrections and changes that we need to do."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:13",
                    "end": "00:07:41",
                    "text": " And the vision of the AI mission, apart from of course the AI summit itself, was fundamentally to make sure that we, without, I mean, given the constraints of how much public resources would be actually expended on AI, what is it that we can do to create the environment? So whether it is the approach to creating the AI compute, whether it is the approach to skilling, whether it is the approach to creating the models."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:41",
                    "end": "00:08:10",
                    "text": " or whether it was the approach to create on the databases. It was very much grounded in something we wanted to do with partnerships. The government is not the only agency which is working in many of these spaces. We have a number of partners, a number of stakeholders who are working with the government to enable this. In the way that we are going out the AI compute, we are doing it through a mechanism whereby the private sector steps in to actually build it out."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:10",
                    "end": "00:08:35",
                    "text": " and re-enable access to those who need it the most by subsidizing the access. Now this is an approach which would possibly fit many other countries who don't want to invest billions and trillions in actually building out the infrastructure at one go and are not sort of put off the idea of trying to access AI because by saying that it just costs too much and we can't actually do it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:35",
                    "end": "00:09:04",
                    "text": " I think there is an alternate approach, which is what India is now championing, to see the access to some of the infrastructure of the areas. Likewise, I think the skilling part of it is a very important part of the story, and I was very happy to note the comments that have been made and recommendations which have been made in the report on the skilling. We have to address it at every level, in existing organizations and institutions, how we still move forward."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:04",
                    "end": "00:09:31",
                    "text": " for those who are still in schools and colleges in terms of what backgrounds they will need to move forward in this space. And more than anything else, I think the readiness of organizations and institutions to actually enterprise us to adopt that. Because ultimately, unless we have applications and unless we have use cases converted into actual solutions."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:31",
                    "end": "00:09:59",
                    "text": " your acceptance of AI into enterprises is just not going to be there. Unless that acceptance takes place, you don't see the real world in that. Ultimately, what's the point of building a model? What's the point of building a huge computer? What's the point of setting up huge databases if it doesn't translate into something which is a practical utility in the real world? And that is the ultimate goal which we are attempting to reach. And in some ways,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:59",
                    "end": "00:10:28",
                    "text": " I think this report to some extent validates the approach we've taken and actually gives us much more of an impetus to pursue the line that we've taken to sort of redouble our efforts on some of the specific parts of it and it's like a report card on some of the stuff that we've done and so thank you very much for sharing this with us it would be a very useful"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:28",
                    "end": "00:10:54",
                    "text": " commentary on what has happened and the way that we push this forward. And the work that you have done here, I'm sure, is something which can be translated to other parts of the world, to other countries in the Global South, for it to be really useful as to what the playbook which needs to be developed, how they can actually work on it in the future. So thank you again, Mr. Kerkus and the entire UNESCO team for everything you've just done. Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:01",
                    "end": "00:11:29",
                    "text": " Thank you for your remarks and for your openness to receive inputs along the journey. Thank you so much. Next, we are honored to have with us today Dr. Ajay Kumar Sood, Principal Scientific Advisor to the Government of India to deliver the keynote address. Dr. Sood's leadership in advancing science, innovation and evidence-based policy continues to shape India's strategic direction in emerging technologies."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:29",
                    "end": "00:12:03",
                    "text": " Thank you very much. Mr. Tim Curtis, Mr. Krishnan who was here two seconds back. All the AI enthusiasts in the room. You can see this room is full of energy. You can see what is happening outside. It really tells that AI is here to make an impact in our lives."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:03",
                    "end": "00:12:32",
                    "text": " We are delighted to be here on the launch of this India AI Readiness Assessment Methodology report. And it has been a privilege for me and my office, Office of Principal Scientific Advisor, to be associated over the last two years or two and a half years very intimately with this AI scene in the country, helping to shape it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:32",
                    "end": "00:12:54",
                    "text": " because this was discussed in the Prime Minister's Science Technology Innovation Advisory Council which I chaired and it was very clear that we need a mission mode to really see the benefits of AI to the society and I'm so glad that we can now see that it is bearing some fruits and much much more has yet to come."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:54",
                    "end": "00:13:24",
                    "text": " So we know that AI brings enormous opportunity. It's a no-brainer to say this at this time. But we also know that it also carries profound ethical, social, environmental, and human rights implications. And its inherent opportunities come with responsibilities. And that's why UNESCO's Readiness AI Assessment matters a lot."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:24",
                    "end": "00:13:54",
                    "text": " And this is the right time for it. So the real question before us is not simply how fast we can adopt AI, but how thoughtfully we can shape it. So this is something for all of us to see. First, how do we ensure that AI truly serves people guided by a clear do no harm principle? And the second,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:54",
                    "end": "00:14:22",
                    "text": " How do we make it fair, trustworthy, inclusive and align with constitutional values, human dignity and fundamental freedoms? So these are the two basic questions which all of us need to answer. So presently at this point, that is why this assessment methodology RAM is a good word from semiconductor technology."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:22",
                    "end": "00:14:51",
                    "text": " It will be very catchy. I can assure you when we say ram people will be confused Is it nesco ram or is it Intel ram? So there will be a confusion, but it's good confusion For this report. It's very relevant at this time Because it offers a structured way for countries to reflect on their preparedness across legal social economic scientific"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:51",
                    "end": "00:15:21",
                    "text": " educational and technological dimensions. So many dimensions we have to see when we look at the assessment of this AI. So it also operationalizes global ethical principles into concrete actionable policy guidance without hampering innovation. So at today's launch, I thought time will be very brief in seven minutes which have been given to highlight three core priorities."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:21",
                    "end": "00:15:48",
                    "text": " for advancing ethical AI so let me just say that these are not absolutely new which you have heard at this thing but let me say in my own words what it means because it is never harmful to say the good thing twice so I'm sure you will appreciate because it reinforces our commitment to this first AI ethics cannot be an afterthought"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:50",
                    "end": "00:16:18",
                    "text": " within the AI system from beginning. So that has to be understood. We have recently come out, perhaps most of you know, with a techno-legal framework from our office, namely strengthening AI governance through techno-legal framework. And we believe that the grounding principles of ethical AI, namely respect for human rights,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:18",
                    "end": "00:16:45",
                    "text": " fairness, transparency, explainability, privacy, safety and accountability must be embedded across the entire AI life cycle right from data governance, model designs, training inference to final deployment with human oversight. So human has to be in the loop if we don't want catastrophe. So this is something"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:45",
                    "end": "00:17:11",
                    "text": " which we have articulated in our AI governance framework which was released in November 2025 and more recently in the techno-legal framework which came out a few days back. So as we scale AI across sectors, these principles must be reflected in standards, procurement frameworks, impact assessments,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:11",
                    "end": "00:17:39",
                    "text": " and institutional mechanisms in all of them we need to implement this and this aligns obviously with UNESCO's call for right based policy measures and legal norms that keep pace with technological change and that is very essential because public trust depends on this and trust is the key thing which you have to realize that the technology doesn't get killed."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:39",
                    "end": "00:18:02",
                    "text": " but it gets deployed. So trust is the key issue which has happened with other technologies which did not take off because of the trust. So this is something all of us should be very, very keenly aware of. Second, inclusions must remain central to India's AI journey. AI should work for all Indians across regions."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:02",
                    "end": "00:18:30",
                    "text": " languages, genders, abilities, and socio-economic backgrounds. As was mentioned, we have a huge diversity in all of this. So it's an absolutely good platform and a sandbox to really see what works and what doesn't work. And AI should help bridge divides and not widen them. This is something is a good matrix to see when we see the deployment of AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:30",
                    "end": "00:18:59",
                    "text": " and it must be accessible in locally relevant culturally sensitive format in an equitable manner for all. So please note culturally sensitive is very important which can be missed if you are blindly following AI developed somewhere else. So this has to be borne in mind because this will not come just by saying it unless you have the right data which is culturally sensitive it will not happen."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:59",
                    "end": "00:19:29",
                    "text": " So capacity building, especially for states, small enterprise startups and underserved communities is equally important. And these insights will be crucial as we move from assessment to action. And the third point, the last point, collaboration is our biggest trend. So no single stakeholders can shape the future of AI alone. And we need government, we need industry, we need academia, civil society, research communities."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:29",
                    "end": "00:19:56",
                    "text": " All of us have important roles to play. And this collaborative approach is what the India AI mission has taken, particularly its safe and trusted AI pillar and India's AI governance guidelines. So the convergence between UNESCO's global ethical framework and India's national priority gives us a robust foundation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:56",
                    "end": "00:20:26",
                    "text": " for leadership and not just adoption of AI. So the recommendation in this report, I'm very sure, will offer a practical roadmap strengthening institutional coordination, enhancing regulatory and ethical oversight, investing in talent and research, promoting data, media, and information literacy, and deepening public engagement on AI, on all these factors."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:26",
                    "end": "00:20:52",
                    "text": " I hope this report will cover it, which it does. I didn't have a chance to look at all the things, but I'm sure it will cover them. So at this juncture, it's a pleasure to acknowledge my appreciation on record for UNESCO, the India AI Mission and MIT, which is really doing a phenomenal job."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:52",
                    "end": "00:21:21",
                    "text": " uh mighty leadership with minister rational very very intimately engaged with all the stakeholders with our office all of us are in this together uh the ikegai law which is a part of this and most importantly all the stakeholders who contributed their perspective and expertise so thank you very much the next phase obviously is the implementation phase and i'm sure"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:21",
                    "end": "00:21:50",
                    "text": " This is the next phase of the outcome of assessment matrices and I am sure we will meet one year from now to see how the outcome of this assessment is done and that what will be the proof of the pudding. So I look forward to that and I would only say this is a wonderful moment. Let us seize this moment."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:50",
                    "end": "00:22:19",
                    "text": " or to really see that AI makes a difference to the society, to the last mile and we hope that inclusive AI will be known and not an exception. Thank you so very much."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:19",
                    "end": "00:22:49",
                    "text": " And I think the question everyone has on their minds right now is what does the report really say? And to walk us through the key findings and recommendations emerging from this national assessment, it is my pleasure to invite Ms. Yumsong Kim, Program Specialist and Chief of Section, Social and Human Sciences, UNESCO Regional Office for South Asia. Ms. Kim has closely read the RAM process in India and will present the report's key insights and policy recommendations."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:49",
                    "end": "00:23:14",
                    "text": " Hello, everyone. It's amazing to see a full room. And I'm very excited to present the findings of the RAM in India. You must now be sick of hearing the RAM and wondering what is this? What is the RAM? So it starts off by UNESCO's interpretation of AI readiness. When UNESCO looks at AI readiness, it goes beyond infrastructure. We interpret readiness in terms of governance."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:14",
                    "end": "00:23:40",
                    "text": " in terms of institutional frameworks, in terms of human capital readiness. So that is what we have tried to capture in the India AI RAM Report. As the structure of the India AI RAM Report is divided into two sections, we start off with the diagnosis, which has all the data that we have collected quantitatively through the indicators across 200 data points."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:40",
                    "end": "00:23:52",
                    "text": " of the five dimensions the principal scientific advisor has mentioned, as well as all of the qualitative data that we collected through five stakeholder consultations across the country."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:52",
                    "end": "00:24:21",
                    "text": " which we thought was important in a country as diverse as India. And then the second part is culminating all of these different inputs into the policy recommendations. We have eight policy recommendations, and each of the recommendations are linked with a concrete roadmap that is realistic, actionable, and also builds on the existing AI infrastructure and ecosystem in India already. So as you can see, we've conducted the secondary research as well as the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:22",
                    "end": "00:24:50",
                    "text": " Tim mentioned in his speech as well that we were able to reach inputs from more than 600 stakeholders across the country and we were very very happy that we did it in cities like Bangalore and Hyderabad which has a very mature AI ecosystem but also in cities like Guwahati in the northeast where we got very different inputs and insights where regional diversity was an important aspect that we have reflected in the RAM report."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:50",
                    "end": "00:25:19",
                    "text": " I'll get on with the task of summarizing a 120-page document into a 10-minute presentation, but let's see how well I can do. So the first dimension is what we call the legal and regulatory dimension. If you look at India, it's very much a hybrid AI governance framework, combining cross-sectoral laws, sector-specific standards, and voluntary guidelines, led by the ministry, METI, the PSA's office, NITIO,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:19",
                    "end": "00:25:44",
                    "text": " and many other government bodies. It also has quite a strong cybersecurity readiness, as you can see in ITU's Global Cybersecurity Index. And just to let you know, the report is being launched today, so you can see all of these data points within the report. And it also has a very strong online participation readiness. It still has quite an evolving data ecosystem, reflected in a moderate open data performance."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:44",
                    "end": "00:26:13",
                    "text": " And then the stakeholders are telling us that they support the government's principle-based approach and cautioned against overly prescriptive regulations, which I think the secretary also mentioned today. There was a strong call for a gap analysis in the existing laws to identify what the existing laws cover and what are the existing gaps that probably future research can address."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:13",
                    "end": "00:26:36",
                    "text": " And definitely the experts called for a stronger state and center alignment in terms of AI initiatives. Now we go on to the social and cultural dimension, which I think is quite a unique aspect of the UNESCO RAM. We make sure that this is a part of readiness when we talk about AI readiness because we believe that there is no one size fits all kind of prescription."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:37",
                    "end": "00:26:56",
                    "text": " There's definitely a strong momentum in India for multilingual AI as we can see in Bosni and different initiatives digital access to AI continues to expand but there's still a bit of a large issue in terms of the gender"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:56",
                    "end": "00:27:23",
                    "text": " gap especially in internet access. There's room for improvement in STEM inclusion especially for young women and girls and that's a very important part that we try to reiterate in the recommendations because without the STEM pool it's also a narrow pool going into the AI talent pool. There's room for improvement especially in terms of sustainability and the environmental impact assessments which is a strong recommendation from the report."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:23",
                    "end": "00:27:44",
                    "text": " The experts continue to say that socially beneficial AI projects were receiving insufficient funding and what they mean by socially beneficial AI projects is things on sustainability on multilingual languages. And that's where we already quite see the government directing towards and making sure that local language data is collected in AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:44",
                    "end": "00:28:13",
                    "text": " Again, the experts also echoed the importance of environmental impacts and making sure that is considered in the future development of AI of India. In terms of the scientific and educational dimension, there's a very, very strong innovation momentum in this country. We found that over 86,000 AI patents were filed between 2010 and 2025, and that was over 25% of all of the..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:13",
                    "end": "00:28:40",
                    "text": " tech patents in this period, which is absolutely enormous. There's an expanding AI research ecosystem led a lot by the center of excellences led by the government and a very robust national AI skill scaling initiative as you can see here. There is definitely room for improvement. First of all, a stronger mainstreaming of AI ethics into the education system. When a lot of the initiatives talked about AI literacy,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:40",
                    "end": "00:29:18",
                    "text": " We found that AI ethics wasn't necessarily always included. So that was something that we found was scope for language. And that made it difficult for some of these students to access so much of the opportunities that was out there in India. The stakeholders also corroborated a lot of these findings, such as AI skilling still being remained very urban centric and very English. So what about the students and informal workers who are not in the urban areas?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:18",
                    "end": "00:29:48",
                    "text": " nor is very fluent with English. So that was another very important finding in the educational dimension. For the economic dimension, I hope I'm on the right page, there is a very, very strong global positioning of India. It hosts 16% of the world's AI talent, which is huge. And it is one of the fastest growing AI talents across the world. It has a very vibrant industry ecosystem."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:48",
                    "end": "00:30:14",
                    "text": " with 2,000 plus companies only in Bangalore. So across the countries there's way more, which is a very, very positive sign. But again, in terms of distribution, outside of the urban tech hubs, there's still a lot of room for improvement. Access to financing for a lot of the startups beyond Bangalore, beyond Hyderabad, beyond Ali, will still continue to be posted as an issue that can be resolved."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:14",
                    "end": "00:30:27",
                    "text": " Stakeholders also ask for targeted incentives to encourage firms to be more ethical in their approaches to inclusive AI development. A lot of the stakeholders are saying ethical AI is great."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:27",
                    "end": "00:30:51",
                    "text": " But if you don't give financial incentives or any kind of incentives for the firms to follow, it's much more difficult for these firms to consider the ethical dimensions. The informal sector still largely remains outside of the AI ecosystem or of all the discussion we have here. And we must recognize that India still has quite a huge large informal economy sector that should be considered."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:51",
                    "end": "00:31:08",
                    "text": " In the technical and infrastructural dimension, there was a strong national focus on building accessible, affordable AI infrastructure. But in terms of improvement, things like electricity in certain areas where there's not constant electricity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:08",
                    "end": "00:31:30",
                    "text": " could be a problem in accessing a lot of these infrastructure, AI infrastructure. We definitely welcome, stakeholders welcome the, in the AI missions, GPU procurement initiatives. But again, without sustainable or continued electricity, some of these issues were pointed out as there are room for improvement."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:30",
                    "end": "00:32:00",
                    "text": " Now let's go on to the recommendations. The way we decided or came up with these recommendations are definitely based on the 200 data points that we collected through secondary research. It's corroborated with the stakeholder consultations that we conducted and we tried to also understand the existing policy landscape so that we don't recommend something that doesn't align with the government's initiative. We also wanted to make sure that they are realistic and actionable and you'll see here that each of the recommendations"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:00",
                    "end": "00:32:25",
                    "text": " have concrete roadmaps. So the first recommendation is to ensure informed AI governance across sectors through comprehensive evidence-based risk and legal review. What do we mean by that? In the roadmap, we suggest three things. Number one, let's do an AI risk mapping. The AI Safety Institute, which exists under the India AI Mission, can lead with academia, industry..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:47",
                    "end": "00:33:06",
                    "text": " second recommendation is to create operational guidance for ethical AI. When we talk about ethical AI, we have a lot of principles, we have a lot of frameworks, but can we make something that's much more relevant and actionable to day-to-day lives of policymakers and developers? That's why we recommend"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:06",
                    "end": "00:33:36",
                    "text": " that the AI Safety Institute creates, first of all, an ethical AI repository with examples and case studies for people to really understand what specific use cases in ethical AI looks like, also have a voluntary measure, especially with industry partners to implement them, and experiment with regulatory sandboxes to test some of these responsible AI approaches, especially in high impact sectors like healthcare, agriculture,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:36",
                    "end": "00:33:57",
                    "text": " and pharmaceutical areas. The third recommendation, which came out quite a lot, I must say, in the stakeholder consultations, is to strengthen the center and state coordination for AI governance and implementation. There are already emerging actions from the government, but dedicated center-state coordination mechanism"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:57",
                    "end": "00:34:26",
                    "text": " could help make sure that the AI policy and planning at the state level are informed from the voices and opinions of all of the different states in India. And also to develop modular policy implementation toolkits for states to make it much more easier to action on these policies. Next is leveraging the AI Cosh platform, which already exists and has huge potential in terms of data in this country."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:26",
                    "end": "00:34:53",
                    "text": " a lot of efforts already underway and we recommend to strengthen some of these existing offices such as the AI data office, such as the data creation units in different parts of the government to make sure that AI COSCH can fulfill its full potential and we go a step ahead and also ask the private sector to contribute to AI COSCH so that we can make sure that this is very very useful for those inside and outside of the government."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:53",
                    "end": "00:35:21",
                    "text": " Next recommendation is about trust. How do we increase public trust of AI governance? We recommend that the AI Safety Institute can guide some of these initiatives, for example, have a nationwide study that can inform some of the campaigns and literacy programs, and also have model AI procurement guidelines that ensures and gives that trust to people that when AI is being procured, it is being done in an ethical way."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:21",
                    "end": "00:35:43",
                    "text": " Recommend six is all about skilling and workforce, and we have two parts to it, but I'll briefly go through it. There was a strong, strong call for making sure that India's workforce is AI future ready through capacity building, embedding ethics, and supporting inclusive transitions. There was a lot of anxiety and concerns around especially mid..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:43",
                    "end": "00:36:10",
                    "text": " career workers on re-skilling and up-skilling, it's not as easy as it sounds, to suddenly embed all these new skills. So that was a strong call from a lot of the experts we talked to, and we feel like the Skills Development Ministry, the AI Safety Institute, the Ministry of Education can definitely play a role in pacifying some of these anxieties through platforms, through integrated up-skilling initiatives."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:10",
                    "end": "00:36:32",
                    "text": " And yes, this was another very important point. AI capacity building for public officials was called for again and again all across the government at center and state level, how important it is for public officials to truly understand, at least have basic AI literacy in order to implement some of these public initiatives, but also policies."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:33",
                    "end": "00:36:58",
                    "text": " The seventh recommendation is to ensure environmental sustainability considerations into AI infrastructure all across based on comprehensive studies and making sure that environmental impact assessments is truly meaningful rather than just tick in the box. Last but not least, we talked about diversity in the AI workforce. We had another session that documented really the..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:58",
                    "end": "00:37:23",
                    "text": " the data on how women are not being as integrated into the AI ecosystem as we think it is. And there's definitely scope for improvement, especially going beyond the urban centers of India. So this takes me to the end of the presentation. The report will be available on UNESCO's homepage from today. So I hope you can all have a look and hope it's useful to the AI ecosystem in India. Thank you so much."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:28",
                    "end": "00:37:52",
                    "text": " Thank you, Yunsong. I know that was a very difficult task to condense so much into such little time. I'm sure people have questions for you, but they can read it online, like you said. We'll now quickly move to the final segment of today's program, something I'm very excited about, a fireside conversation on global perspectives in ethical AI governance. It's my pleasure to invite on stage Dr. Mario Grazia. She's the..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:52",
                    "end": "00:38:35",
                    "text": " Chief of Executive Office, Social and Human Sciences, UNESCO, and has been very closely involved with the RAM in India. And Mr. Vilas Dhar, President of the Patrick J. McGovern Foundation. The McGovern Foundation has been supporting the RAM exercises in India and globally. This conversation will reflect on the global implications of ethical AI frameworks and the role of multilateral collaboration in using and shaping trusted AI ecosystems. Over to you. There we go."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:35",
                    "end": "00:39:05",
                    "text": " So it's really my pleasure and honor to be here with Dylan. Actually, there are three words in my mind that always remind me when I see Dylan. And these are the President and Trustee of the Montgomery Foundation. The first one is insightful. There hasn't been one single instance in which I was talking to Dylan, and then I thought, oh, I haven't thought about that. And this doesn't happen at all. So that's already one thing really that's always..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:05",
                    "end": "00:39:36",
                    "text": " Help me, it's a great pleasure to be with you. The other one is that he's a visionary. He has been seeing this transformation, he has been seeing the challenges coming with it, and the opportunities, of course, much before others, and has taken a really active role in that, in steering it towards the right direction. He's committed, always. The energy you put in that, the kind of the determination, but also the unusual way of thinking on how to address the problem is something that always"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:36",
                    "end": "00:40:02",
                    "text": " You are striking me all the time by the time I've been talking to you. So I won't have easy questions for you today because we really want to get the best out of your brain. So the first one is, as the song was brilliantly showing us, this report is very rich and it shows that there is a lot of strength in India, but there are also real complexities. And I mean, scale is not the only one. So..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:02",
                    "end": "00:40:33",
                    "text": " You have supported UNESCO and a lot of your work anyway looks at what are the factors that might be the ecosystem that might be conducive to AI that is ethical and how to make the best out of AI that really goes for people. In your mind, is there one example or one thing that India could adopt, replicate, tailor in order really to set up already let's say the transformation is very visible and it's very much there?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:33",
                    "end": "00:41:02",
                    "text": " But what would be perhaps something, an addendum or an approach that you would suggest that India takes? Well, thank you so much for that very kind introduction. Too kind. You're supposed to set a low standard so that I can exceed it, not set a very high standard. I want to thank you for this invitation to UNESCO. And I also want to acknowledge Secretary Krishnan, who I know had to go, but also you, Dr. Soot. I want to say one scientist to another. I think it gives the world a great deal of confidence and comfort to know that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:02",
                    "end": "00:41:32",
                    "text": " You really have the kind of scientific background that's coupled with a moral and a conscience driven approach. Your techno-legal frameworks, as you know, have been picked up in countries around the world already. I want to thank you for that leadership. Your question, I think, is wonderful, right? But I'm actually going to flip it for you. And I'm sorry you know me well. Rather than talk about what India should adopt, I actually first want to talk about what the rest of the world should adopt from India. Because I think this is not a story where we're saying, what do we start here and figure out who else in the world is better? Well, your readiness to present that shows that India is actually far ahead of us."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:32",
                    "end": "00:42:01",
                    "text": " I think one of the things we should look to is a re-conceptualization of what AI governance is actually about. When you go to many places in the world, AI governance is seen as a way of limiting risk. Risk for institutions, liability for people, minimizing models. And in doing so, you miss the amazing opportunity that's possible. When you think of AI governance as a way to incentivize a positive vision for the possible, you do that through a set of interventions that go beyond minimizing models."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:01",
                    "end": "00:42:30",
                    "text": " that go beyond restricting the worst possible outcome. You instead think about how you incentivize and name and elaborate a whole of society approach to AI. One that centers the idea of participatory mechanism. One that allows for people to engage in, as you have done with 600 consultations across the country, to let people not merely be the subjects of an AI governance regime, but rather to be the architect. We've seen that happen in India in the DPI stack, where we saw in digital public infrastructure the idea that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:30",
                    "end": "00:42:60",
                    "text": " While there were guidelines and guardrails that were put forward, much of the innovation actually came from a combination of private and public sectors, who came together not merely to operate inside of a constraint, but to test it out. To say, how can we consistently deliver services to those who are left out of the market? And yet, in doing so, also build a framework that can then be picked up, abstracted, and shared with the world. So this is my answer to you. I would say what the world needs to learn is a governance principle that doesn't start from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:60",
                    "end": "00:43:31",
                    "text": " heavy-handed restrictions, but rather from the light touch that's been espoused here in the country, but allows for the incredible innovation of 1.4 billion people to save the world's vision of what to happen. Absolutely. And actually, one strong component of that is really the skills, the endowment, the human capital, the humanity of this country. In your mind, what was, because we hear way too often, and the report also mentions that, in relation to the skill, the AI skills, and I've always been saying, to me, it's more skills for the digital era."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:31",
                    "end": "00:44:01",
                    "text": " What do the set of skills for the digital era look like for you? And what is the role you see for awareness? Because I think a lot of the action, a lot of the directions that the development of it depend also on what we are aware of or might not be aware of. Look, I think the first part of the story was we have to build technical skills and scale across the population. We very quickly realized that doesn't make sense. We don't need 1.4 billion AI scientists."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:01",
                    "end": "00:44:31",
                    "text": " We barely need a few hundred thousand. What we need instead, though, is a population that understands the opportunities that's available. In the past few months, I've had the opportunity to travel across India and across the world, as you know, visiting those top tier institutions, which are known for their engineering prowess, and having great conversations with students and faculty. But you know where I find inspiration is when you go to those other institutions. They're often termed tier two or tier three. I reject the premise. But here in India, institutions like PSG."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:31",
                    "end": "00:44:56",
                    "text": " like Maika, like Amrita University, places where students come together in an environment where they have access to the underlying ecosystem of AI. But then they bring their lived experience from communities across the country. They may or may not be able to sit down and understand exactly how to build the next frontier model, but they do know that their parents are struggling with drought. They know that agricultural productivity is at risk. They know that the electricity grid is not functioning the way it should."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:56",
                    "end": "00:45:28",
                    "text": " They know that disasters and crisis from mudslides all the way through to macro level events affect them. And in that moment, something magical happens. Because the skilling here is no longer about whether you can describe about your model. So you can take the tools that are available and build real solutions. Solutions that operate at low cost, sometimes in low power and low connectivity environments, that deliver real value for people. Somebody agrees with me. I like it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:28",
                    "end": "00:45:55",
                    "text": " India will go from about 1.4 gigawatts in data center capacity to projected 9 to 10 gigawatts in the next century."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:55",
                    "end": "00:46:22",
                    "text": " And when you look at that and you understand it, on one side there is a wonderful storyline of companies investing and building out the maps. On the other is the communities where those data centers live, the environmental impact on the sustainability issues that will arise. Now, the conversation that I'm asking you to have is not to name those risks because we already know them, but rather to think about what a uniquely Indian model might be to address them. One where you center economic sustainability as a frontline matter, not as something you deal with as a factor."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:22",
                    "end": "00:46:51",
                    "text": " thinking about mandatory disclosure instead of voluntary definition of the impact on water and local power plants, of building mechanisms that say fresh water cannot be used for cooling of data centers unless there's enough in the community already to go around, because we know there are commercially viable alternatives for air cooling or other things. And finally, to note that even in India, and especially in India, the fair tariff on renewable energy are actually lower than where natural gas is. What would happen if we were to mandate that in order to put a data center in a community?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:51",
                    "end": "00:47:21",
                    "text": " You not only have to commit to the increasing purchase of renewable energy, but commit to building that capacity. This is the opportunity here in India, that as you begin this incredible ramp up, can you use those techno-legal frameworks to actually build a new mechanism and show the world what sustainable data center faces. Absolutely, and this would also speak to the issue of inequality, which was raised actually during the presentation. So, suppose you can decide tomorrow, put one thing in motion, and you have unlimited resources."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:21",
                    "end": "00:47:49",
                    "text": " What would you do? What would be the first thing that you're going to do? You know what? It's the easiest question in the world because what I would do is ask communities across the country what they prioritize in the ring as their primary internet. But once you've asked them, huh? No, that's great. But I mean, I don't want to skip over it, right? I don't want to be flip about it because one of the things is when we name participatory AI,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:49",
                    "end": "00:48:16",
                    "text": " We often skip over the fact that that requires participation. But once you ask them, I think there are a couple of things that I would advise. The first is really investing in this idea of translational AI. Take what models enable and allow, and invest meaningfully in building the architecture that brings it to communities in the form of apps and services, in the form of insight and wisdom that actually lets us change these things. I think we could look at priority sectors, and we've done that already. I think in the report you identified them."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:16",
                    "end": "00:48:45",
                    "text": " but focus also on where you see innovation centers that are organically emerging across the country and invest in them deeply. So I think those of you in the room that can do something tomorrow might want to stop thinking about this. Now one last question because time's up and I see it very clearly unfortunately. But there is one thing we were talking before in the other panel about the different trends that AI is taking and actually we were discussing also about agents which are you know perhaps now the kind of developments that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:45",
                    "end": "00:49:13",
                    "text": " excites many, but frightens others. And the question is, well, they've been with us now for some time, but we haven't seen the scale of adoption that we would expect. In your mind, given that you've been analyzing this thing, what is the big topic? And should we start thinking about dynamics like the ones we have seen, for instance, with the Internet of Things and the kind of failing to develop a being adopted at the pace we would expect?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:14",
                    "end": "00:49:43",
                    "text": " It's a great question. And this is a really nice story. I'll be very quick about it. We saw the first generation of agents to do some of the most boring things you can imagine. Companies that build agents to increase profitability by redeploying your CRM, by working in your workspace. And can anybody in this room really say they were excited about that? I don't think so. But we saw a new phenomenon happen just in the last few weeks. We saw hobbyists come forward with an open platform for building agents. And it took the world by storm. Why? Because we allowed people to be creative in their use."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:43",
                    "end": "00:50:12",
                    "text": " We let them bring agents into their lives and solve their problems. And with that, we saw hundreds of thousands of people deploy agents in scale. We even saw all kinds of weird effects like agents talking to each other online. And even most interestingly, now a service that lets agents autonomously go and contract a human being to go and do a task for them in the real world. This is the future, right? Agentic AI is not a corporate term, and it cannot become one. It has to be a way of extending human agency in ways that the powerless have never had."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:12",
                    "end": "00:50:51",
                    "text": " a way of democratizing access to the ecosystem of both economic success, but also environmental, political, and economic success. Join me in thanking Vila D'Arcy. I think we've been cut off. Thank you, everybody. We've been cut off."
                }
            ]
        }
    ],
    "projectDeleted": false,
    "codeFrameColumn": [
        {
            "col1": "col1",
            "col2": "col3",
            "researchObjective": "",
            "filterColumn": [],
            "netSetting": true,
            "subNetSetting": true,
            "categorySetting": true,
            "sentimentSetting": true,
            "thinkingBudgetSetting": 2000,
            "questionType": "OPENEND",
            "otherBrand": false,
            "existingCodeFrame": [],
            "isList": false,
            "codeFrameWithDataList": [],
            "codeFrameList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 9849,
                        "CandidatesTokenCount": 3219,
                        "TotalTokenCount": 14234,
                        "thoughtsTokenCount": 1166,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Net": "AI Readiness & Assessment Framework",
                                "Subnets": [
                                    {
                                        "Subnet": "RAM Report & Methodology",
                                        "Categories": [
                                            {
                                                "Category": "UNESCO's RAM Interpretation",
                                                "Summary": "Readiness interpreted beyond infrastructure to include governance, institutional frameworks, and human capital."
                                            },
                                            {
                                                "Category": "Report Structure & Data Collection",
                                                "Summary": "Divided into diagnosis (quantitative/qualitative data from 200 data points, 600+ stakeholders) and policy recommendations."
                                            },
                                            {
                                                "Category": "Scope & Dimensions",
                                                "Summary": "Covers legal, social, cultural, scientific, educational, economic, technical, and infrastructural aspects of AI readiness."
                                            }
                                        ],
                                        "Summary": "Overview of the UNESCO AI Readiness Assessment Methodology (RAM) report, its definition of readiness, and the comprehensive approach to data collection and analysis in India."
                                    },
                                    {
                                        "Subnet": "Purpose & Significance of RAM",
                                        "Categories": [
                                            {
                                                "Category": "Diagnostic & Informative Tool",
                                                "Summary": "Designed to ensure inclusive, ethical, and sustainable AI impact, providing a grounded picture of India's current standing."
                                            },
                                            {
                                                "Category": "Actionable Roadmap for India",
                                                "Summary": "Aims to transition from high-level ambitions to concrete, actionable roadmaps for national AI adoption and strategy."
                                            },
                                            {
                                                "Category": "Global Context & Learnings",
                                                "Summary": "India's RAM contributes to a collective dataset from over 70 countries, offering insights that can be translated globally, especially for the Global South."
                                            }
                                        ],
                                        "Summary": "The overarching goals and importance of the RAM report in shaping India's AI future and contributing to global understanding of ethical AI adoption."
                                    }
                                ],
                                "Summary": "This topic covers the foundational aspects of the AI Readiness Assessment Methodology (RAM) report, including its interpretation of AI readiness, the comprehensive approach taken in India, and its strategic importance as a diagnostic tool and actionable roadmap for national and global AI development."
                            },
                            {
                                "Net": "India's AI Landscape: Strengths & Challenges",
                                "Subnets": [
                                    {
                                        "Subnet": "Current Strengths & Momentum",
                                        "Categories": [
                                            {
                                                "Category": "Strong AI Talent Pool",
                                                "Summary": "India accounts for about 16% of the world's AI talent, showing significant growth and innovation in the sector."
                                            },
                                            {
                                                "Category": "Multilingual AI & Digital Public Services",
                                                "Summary": "Real momentum in expanding Indian language datasets and models, supporting digital public services like 'Bosni'."
                                            },
                                            {
                                                "Category": "Vibrant Innovation Ecosystem",
                                                "Summary": "Robust innovation system, supported by strong digital foundations, and expanding AI research led by Centers of Excellence."
                                            }
                                        ],
                                        "Summary": "Highlights India's existing advantages, significant achievements, and positive momentum in the field of Artificial Intelligence."
                                    },
                                    {
                                        "Subnet": "Identified Gaps & Weaknesses",
                                        "Categories": [
                                            {
                                                "Category": "Ethical Data Use & Sharing",
                                                "Summary": "Access to and ethical use and sharing of data needs strengthening, along with an evolving data ecosystem."
                                            },
                                            {
                                                "Category": "Environmental Sustainability",
                                                "Summary": "Needs to be built more directly into AI infrastructure planning and impact assessments."
                                            },
                                            {
                                                "Category": "AI Ethics in Education & Training",
                                                "Summary": "Ethics are present but could be more consistently reflected in education and training, including regional language offerings."
                                            }
                                        ],
                                        "Summary": "Areas where India currently faces limitations or requires significant improvement to achieve its AI goals."
                                    },
                                    {
                                        "Subnet": "Socio-Cultural Context & Inclusivity",
                                        "Categories": [
                                            {
                                                "Category": "Diversity & Inclusion Mandate",
                                                "Summary": "Acknowledges India's vast diversity (1.4 billion people, thousands of languages, regional/socioeconomic differences) as foundational for AI inclusivity."
                                            },
                                            {
                                                "Category": "Bridging Digital & Gender Divides",
                                                "Summary": "Challenges exist in gender gaps in internet access and STEM inclusion for young women/girls, which AI should help bridge, not widen."
                                            },
                                            {
                                                "Category": "Culturally Sensitive AI",
                                                "Summary": "Emphasizes the need for AI to be accessible in locally relevant, culturally sensitive formats to truly serve all Indians."
                                            }
                                        ],
                                        "Summary": "Focuses on the unique social and cultural considerations influencing AI development and adoption in India, particularly regarding inclusivity and accessibility."
                                    }
                                ],
                                "Summary": "This topic describes the current state of AI in India, detailing its significant strengths such as a large talent pool and innovation, alongside critical areas needing improvement like ethical data management and environmental sustainability. It also emphasizes the crucial role of India's unique socio-cultural diversity in shaping inclusive AI development."
                            },
                            {
                                "Net": "Ethical AI Governance & Principles",
                                "Subnets": [
                                    {
                                        "Subnet": "Governance Principles & Frameworks",
                                        "Categories": [
                                            {
                                                "Category": "Principle-Based Approach",
                                                "Summary": "India supports a principle-based AI governance framework, cautioning against overly prescriptive regulations to foster innovation."
                                            },
                                            {
                                                "Category": "Techno-Legal Frameworks",
                                                "Summary": "Development of frameworks like 'Strengthening AI Governance through Techno-Legal Frameworks' embedding ethical principles from the start."
                                            },
                                            {
                                                "Category": "Embedding Ethics in AI Lifecycle",
                                                "Summary": "Ethical principles (human rights, fairness, transparency, privacy, accountability) must be integrated across the entire AI lifecycle, from data to deployment."
                                            }
                                        ],
                                        "Summary": "Examines the foundational principles and regulatory approaches guiding AI development and deployment in India, emphasizing a human-centric and ethical stance."
                                    },
                                    {
                                        "Subnet": "Trust, Transparency & Accountability",
                                        "Categories": [
                                            {
                                                "Category": "Public Trust as Key",
                                                "Summary": "Public trust is essential for AI technology adoption; without it, technology may not get deployed effectively."
                                            },
                                            {
                                                "Category": "Human Oversight & Consequences",
                                                "Summary": "Human oversight in AI is crucial to prevent catastrophes, recognizing that algorithmic choices carry real human consequences."
                                            },
                                            {
                                                "Category": "Risk & Legal Reviews",
                                                "Summary": "Recommendation for comprehensive evidence-based risk mapping and legal reviews to ensure informed AI governance across sectors."
                                            }
                                        ],
                                        "Summary": "Discusses the importance of building public confidence, ensuring clarity in AI operations, and establishing mechanisms for responsibility."
                                    },
                                    {
                                        "Subnet": "Human-Centric AI",
                                        "Categories": [
                                            {
                                                "Category": "'Do No Harm' Principle",
                                                "Summary": "AI must serve people, guided by a clear 'do no harm' principle to prevent negative societal impacts."
                                            },
                                            {
                                                "Category": "Fair, Trustworthy & Inclusive AI",
                                                "Summary": "Shaping AI to be fair, trustworthy, inclusive, and aligned with constitutional values, human dignity, and fundamental freedoms."
                                            },
                                            {
                                                "Category": "AI that Listens & Performs",
                                                "Summary": "Aim for AI that listens before it decides, performing well in the lived complexity of real communities, not just demos."
                                            }
                                        ],
                                        "Summary": "Focuses on the imperative of designing and deploying AI systems that prioritize human well-being, rights, and societal benefit."
                                    }
                                ],
                                "Summary": "This topic delves into the core aspects of ethical AI governance, highlighting the principles and frameworks India is adopting to ensure AI development is human-centric, trustworthy, and aligns with societal values. It covers the need for ethical integration throughout the AI lifecycle, the importance of public trust, and the commitment to a 'do no harm' approach."
                            },
                            {
                                "Net": "AI Development & Adoption Strategy",
                                "Subnets": [
                                    {
                                        "Subnet": "Skilling & Capacity Building",
                                        "Categories": [
                                            {
                                                "Category": "AI Future-Ready Workforce",
                                                "Summary": "Strong call for making India's workforce AI future-ready through capacity building, ethics integration, and inclusive transitions."
                                            },
                                            {
                                                "Category": "Addressing Skilling Gaps (Urban/English)",
                                                "Summary": "AI skilling remains urban-centric and English-focused; need to extend to non-urban areas and informal workers through regional language offerings."
                                            },
                                            {
                                                "Category": "Public Officials' AI Literacy",
                                                "Summary": "Repeated calls for AI capacity building for public officials at all levels to effectively implement AI initiatives and policies."
                                            }
                                        ],
                                        "Summary": "Details the efforts and recommendations related to educating, training, and preparing the workforce and public officials for the AI era, with a focus on inclusivity."
                                    },
                                    {
                                        "Subnet": "Infrastructure & Ecosystem Development",
                                        "Categories": [
                                            {
                                                "Category": "Accessible & Affordable AI Infrastructure",
                                                "Summary": "Strong national focus on building accessible and affordable AI infrastructure, including GPU procurement initiatives."
                                            },
                                            {
                                                "Category": "Environmental Impact in Infrastructure",
                                                "Summary": "Environmental sustainability needs to be built more directly into AI infrastructure planning, considering issues like electricity reliability and water usage."
                                            },
                                            {
                                                "Category": "Vibrant Industry Ecosystem",
                                                "Summary": "India hosts a vibrant industry ecosystem with 2,000+ AI companies, particularly in tech hubs like Bangalore and Hyderabad."
                                            }
                                        ],
                                        "Summary": "Outlines the strategic development of the physical and digital foundations necessary for widespread AI adoption, including addressing sustainability concerns."
                                    },
                                    {
                                        "Subnet": "Funding & Incentives",
                                        "Categories": [
                                            {
                                                "Category": "Socially Beneficial AI Funding",
                                                "Summary": "Socially beneficial AI projects (e.g., sustainability, multilingual) were receiving insufficient funding, despite government's direction."
                                            },
                                            {
                                                "Category": "Targeted Incentives for Ethical AI",
                                                "Summary": "Stakeholders ask for targeted incentives to encourage firms to adopt more ethical approaches to inclusive AI development."
                                            },
                                            {
                                                "Category": "Subsidized Access to AI Compute",
                                                "Summary": "India is championing an alternate approach to AI compute, with the private sector building infrastructure and subsidizing access for those most in need."
                                            }
                                        ],
                                        "Summary": "Covers financial mechanisms, investment strategies, and incentives designed to promote ethical and inclusive AI development and adoption."
                                    },
                                    {
                                        "Subnet": "Application & Use Cases",
                                        "Categories": [
                                            {
                                                "Category": "Enterprise & Real-World Utility",
                                                "Summary": "Emphasis on converting use cases into actual solutions for enterprise acceptance and practical utility in the real world."
                                            },
                                            {
                                                "Category": "Focus on Priority Sectors",
                                                "Summary": "Experimentation with regulatory sandboxes in high-impact sectors like healthcare, agriculture, and pharmaceuticals."
                                            },
                                            {
                                                "Category": "Translational AI Investment",
                                                "Summary": "Investing meaningfully in building architecture that brings AI models to communities in the form of apps, services, and insights."
                                            }
                                        ],
                                        "Summary": "Focuses on the practical implementation and deployment of AI solutions, emphasizing real-world impact and utility across various sectors and communities."
                                    }
                                ],
                                "Summary": "This topic outlines the practical strategies and approaches for fostering AI development and adoption in India, encompassing efforts in skilling, infrastructure build-out, funding mechanisms, and the crucial translation of AI models into tangible, real-world applications for societal benefit."
                            },
                            {
                                "Net": "Collaboration & Global Leadership",
                                "Subnets": [
                                    {
                                        "Subnet": "Stakeholder Collaboration",
                                        "Categories": [
                                            {
                                                "Category": "Whole-of-Society Approach",
                                                "Summary": "No single stakeholder can shape AI alone; government, industry, academia, civil society, and research communities all have vital roles."
                                            },
                                            {
                                                "Category": "India AI Mission & Partners",
                                                "Summary": "The India AI Mission emphasizes partnerships with diverse stakeholders to create an enabling environment for AI."
                                            },
                                            {
                                                "Category": "Center-State Coordination",
                                                "Summary": "Strengthening coordination between central and state governments for AI governance and implementation, with modular policy toolkits."
                                            }
                                        ],
                                        "Summary": "Highlights the importance of multi-stakeholder engagement and partnerships in steering India's AI journey."
                                    },
                                    {
                                        "Subnet": "India's Global Role & Learnings",
                                        "Categories": [
                                            {
                                                "Category": "Leadership vs. Adoption",
                                                "Summary": "Convergence of UNESCO's global ethical framework and India's national priorities provides a robust foundation for leadership, not just adoption, of AI."
                                            },
                                            {
                                                "Category": "Playbook for Global South",
                                                "Summary": "India's approach and the RAM report's findings can serve as a useful playbook for other countries, especially in the Global South."
                                            },
                                            {
                                                "Category": "Re-conceptualizing AI Governance Globally",
                                                "Summary": "India's model of AI governance, focusing on incentivizing positive vision and participatory mechanisms, offers a valuable lesson for the rest of the world."
                                            }
                                        ],
                                        "Summary": "Focuses on India's potential to influence global AI discourse and offer valuable models for ethical and inclusive AI development."
                                    },
                                    {
                                        "Subnet": "Future Outlook & Implementation",
                                        "Categories": [
                                            {
                                                "Category": "Assessment to Action",
                                                "Summary": "The transition from assessment findings to concrete actions and implementation is the crucial next phase."
                                            },
                                            {
                                                "Category": "Long-Term Vision",
                                                "Summary": "Looking forward to reviewing the outcomes of the assessment in the future, aiming for inclusive AI to be the norm."
                                            },
                                            {
                                                "Category": "Seizing the Moment",
                                                "Summary": "Encouragement to seize the current momentum to ensure AI makes a significant difference to society, especially at the last mile."
                                            }
                                        ],
                                        "Summary": "Addresses the forward-looking aspects, emphasizing the transition from planning to execution and the long-term vision for AI's societal impact."
                                    }
                                ],
                                "Summary": "This topic emphasizes the collaborative spirit driving India's AI initiatives, recognizing the need for multi-stakeholder engagement. It highlights India's aspiration and potential to lead global ethical AI development, serving as a model, particularly for the Global South. The focus is also on the crucial next steps of implementing recommendations and realizing the long-term vision for AI's societal impact."
                            },
                            {
                                "Net": "996 Non specific TOPIC",
                                "Subnets": [
                                    {
                                        "Subnet": "Uncategorized/Irrelevant Responses",
                                        "Categories": [
                                            {
                                                "Category": "Miscellaneous Feedback",
                                                "Summary": "Responses that are not directly alignable with defined topics and subtopics but provide some form of feedback or comment."
                                            },
                                            {
                                                "Category": "General Observations",
                                                "Summary": "Broad statements or observations about the AI landscape that lack specificity for other categories."
                                            }
                                        ],
                                        "Summary": "Responses that are somewhat relevant to the general theme but cannot be specifically placed into the main hierarchical codeframe and may require further review."
                                    }
                                ],
                                "Summary": "This topic captures responses that are broadly related to the subject matter but do not fit neatly into the established hierarchical codeframe, indicating a need for potential up-coding or further analysis."
                            },
                            {
                                "Net": "997 Don’t Know Can’t Say",
                                "Subnets": [
                                    {
                                        "Subnet": "Non-Committal Responses",
                                        "Categories": [
                                            {
                                                "Category": "Vague or Indecisive",
                                                "Summary": "Responses that express uncertainty, indecision, or a lack of clear opinion on the subject."
                                            },
                                            {
                                                "Category": "Lack of Information/Awareness",
                                                "Summary": "Respondents explicitly stating they do not have enough information or understanding to provide a specific answer."
                                            }
                                        ],
                                        "Summary": "This subtopic captures responses where individuals express a lack of knowledge, uncertainty, or an inability to provide a definitive answer."
                                    }
                                ],
                                "Summary": "This topic includes responses that are non-committal, vague, or indicate that the respondent does not know or cannot provide a specific answer."
                            },
                            {
                                "Net": "998 Not Applicable",
                                "Subnets": [
                                    {
                                        "Subnet": "Responses Irrelevant to Question",
                                        "Categories": [
                                            {
                                                "Category": "Off-Topic Comments",
                                                "Summary": "Responses that are clearly unrelated to the specific question asked or the overall project context."
                                            },
                                            {
                                                "Category": "Misinterpretation of Question",
                                                "Summary": "Responses indicating the respondent misunderstood the question and provided an answer for something else."
                                            }
                                        ],
                                        "Summary": "This subtopic captures responses that are clearly irrelevant or not applicable to the question or project at hand."
                                    }
                                ],
                                "Summary": "This topic is for responses that are entirely irrelevant or not applicable to the question or the market research project."
                            },
                            {
                                "Net": "999 None-Nothing",
                                "Subnets": [
                                    {
                                        "Subnet": "Gibberish/Meaningless Responses",
                                        "Categories": [
                                            {
                                                "Category": "Non-Language Characters",
                                                "Summary": "Responses consisting of random characters, symbols, or numbers that do not form intelligible words or sentences."
                                            },
                                            {
                                                "Category": "Empty/Null Responses",
                                                "Summary": "Responses that are blank, contain only filler words without substance, or clearly indicate 'nothing' or 'none'."
                                            }
                                        ],
                                        "Summary": "This subtopic captures responses that are unintelligible, nonsensical, or convey no meaningful information."
                                    }
                                ],
                                "Summary": "This topic is for responses that are plain gibberish, non-language characters, or contain no meaningful content."
                            }
                        ],
                        "Summary": "The provided text is a comprehensive discussion about the launch and findings of UNESCO's AI Readiness Assessment Methodology (RAM) report for India. It highlights India's progress and unique approach to AI adoption, focusing on ethical, inclusive, and sustainable development. Speakers from UNESCO, the Government of India, and the Patrick J. McGovern Foundation detail the report's diagnostic insights across legal, social, scientific, educational, economic, and technical dimensions. Key strengths include India's AI talent pool, multilingual AI efforts, and digital public services. Areas for improvement noted are ethical data use, environmental sustainability in AI infrastructure, mainstreaming AI ethics in education, and addressing urban-centric skilling gaps. The discussion emphasizes India's 'light-touch' governance approach, participatory mechanisms, and its potential to set a global example, especially for the Global South, through collaborative efforts and a focus on real-world, community-centric applications of AI."
                    }
                }
            ],
            "onlyCodeFrame": true,
            "errorList": [],
            "codeFrameGenerateType": "CODE_FRAME_ONLY",
            "requestType": "CODEFRAME",
            "requestMode": "DEVELOPMENT",
            "sourceLanguage": "English",
            "targetLanguage": "English",
            "response": {
                "status": "OK",
                "data": {
                    "status": "OK",
                    "PromptTokenCount": 9849,
                    "CandidatesTokenCount": 3219,
                    "TotalTokenCount": 14234,
                    "thoughtsTokenCount": 1166,
                    "finishReason": "STOP",
                    "data": [
                        {
                            "Net": "AI Readiness & Assessment Framework",
                            "Subnets": [
                                {
                                    "Subnet": "RAM Report & Methodology",
                                    "Categories": [
                                        {
                                            "Category": "UNESCO's RAM Interpretation",
                                            "Summary": "Readiness interpreted beyond infrastructure to include governance, institutional frameworks, and human capital.",
                                            "code": 3
                                        },
                                        {
                                            "Category": "Report Structure & Data Collection",
                                            "Summary": "Divided into diagnosis (quantitative/qualitative data from 200 data points, 600+ stakeholders) and policy recommendations.",
                                            "code": 4
                                        },
                                        {
                                            "Category": "Scope & Dimensions",
                                            "Summary": "Covers legal, social, cultural, scientific, educational, economic, technical, and infrastructural aspects of AI readiness.",
                                            "code": 5
                                        }
                                    ],
                                    "Summary": "Overview of the UNESCO AI Readiness Assessment Methodology (RAM) report, its definition of readiness, and the comprehensive approach to data collection and analysis in India.",
                                    "code": 2
                                },
                                {
                                    "Subnet": "Purpose & Significance of RAM",
                                    "Categories": [
                                        {
                                            "Category": "Diagnostic & Informative Tool",
                                            "Summary": "Designed to ensure inclusive, ethical, and sustainable AI impact, providing a grounded picture of India's current standing.",
                                            "code": 7
                                        },
                                        {
                                            "Category": "Actionable Roadmap for India",
                                            "Summary": "Aims to transition from high-level ambitions to concrete, actionable roadmaps for national AI adoption and strategy.",
                                            "code": 8
                                        },
                                        {
                                            "Category": "Global Context & Learnings",
                                            "Summary": "India's RAM contributes to a collective dataset from over 70 countries, offering insights that can be translated globally, especially for the Global South.",
                                            "code": 9
                                        }
                                    ],
                                    "Summary": "The overarching goals and importance of the RAM report in shaping India's AI future and contributing to global understanding of ethical AI adoption.",
                                    "code": 6
                                }
                            ],
                            "Summary": "This topic covers the foundational aspects of the AI Readiness Assessment Methodology (RAM) report, including its interpretation of AI readiness, the comprehensive approach taken in India, and its strategic importance as a diagnostic tool and actionable roadmap for national and global AI development.",
                            "code": 1
                        },
                        {
                            "Net": "India's AI Landscape: Strengths & Challenges",
                            "Subnets": [
                                {
                                    "Subnet": "Current Strengths & Momentum",
                                    "Categories": [
                                        {
                                            "Category": "Strong AI Talent Pool",
                                            "Summary": "India accounts for about 16% of the world's AI talent, showing significant growth and innovation in the sector.",
                                            "code": 12
                                        },
                                        {
                                            "Category": "Multilingual AI & Digital Public Services",
                                            "Summary": "Real momentum in expanding Indian language datasets and models, supporting digital public services like 'Bosni'.",
                                            "code": 13
                                        },
                                        {
                                            "Category": "Vibrant Innovation Ecosystem",
                                            "Summary": "Robust innovation system, supported by strong digital foundations, and expanding AI research led by Centers of Excellence.",
                                            "code": 14
                                        }
                                    ],
                                    "Summary": "Highlights India's existing advantages, significant achievements, and positive momentum in the field of Artificial Intelligence.",
                                    "code": 11
                                },
                                {
                                    "Subnet": "Identified Gaps & Weaknesses",
                                    "Categories": [
                                        {
                                            "Category": "Ethical Data Use & Sharing",
                                            "Summary": "Access to and ethical use and sharing of data needs strengthening, along with an evolving data ecosystem.",
                                            "code": 16
                                        },
                                        {
                                            "Category": "Environmental Sustainability",
                                            "Summary": "Needs to be built more directly into AI infrastructure planning and impact assessments.",
                                            "code": 17
                                        },
                                        {
                                            "Category": "AI Ethics in Education & Training",
                                            "Summary": "Ethics are present but could be more consistently reflected in education and training, including regional language offerings.",
                                            "code": 18
                                        }
                                    ],
                                    "Summary": "Areas where India currently faces limitations or requires significant improvement to achieve its AI goals.",
                                    "code": 15
                                },
                                {
                                    "Subnet": "Socio-Cultural Context & Inclusivity",
                                    "Categories": [
                                        {
                                            "Category": "Diversity & Inclusion Mandate",
                                            "Summary": "Acknowledges India's vast diversity (1.4 billion people, thousands of languages, regional/socioeconomic differences) as foundational for AI inclusivity.",
                                            "code": 20
                                        },
                                        {
                                            "Category": "Bridging Digital & Gender Divides",
                                            "Summary": "Challenges exist in gender gaps in internet access and STEM inclusion for young women/girls, which AI should help bridge, not widen.",
                                            "code": 21
                                        },
                                        {
                                            "Category": "Culturally Sensitive AI",
                                            "Summary": "Emphasizes the need for AI to be accessible in locally relevant, culturally sensitive formats to truly serve all Indians.",
                                            "code": 22
                                        }
                                    ],
                                    "Summary": "Focuses on the unique social and cultural considerations influencing AI development and adoption in India, particularly regarding inclusivity and accessibility.",
                                    "code": 19
                                }
                            ],
                            "Summary": "This topic describes the current state of AI in India, detailing its significant strengths such as a large talent pool and innovation, alongside critical areas needing improvement like ethical data management and environmental sustainability. It also emphasizes the crucial role of India's unique socio-cultural diversity in shaping inclusive AI development.",
                            "code": 10
                        },
                        {
                            "Net": "Ethical AI Governance & Principles",
                            "Subnets": [
                                {
                                    "Subnet": "Governance Principles & Frameworks",
                                    "Categories": [
                                        {
                                            "Category": "Principle-Based Approach",
                                            "Summary": "India supports a principle-based AI governance framework, cautioning against overly prescriptive regulations to foster innovation.",
                                            "code": 25
                                        },
                                        {
                                            "Category": "Techno-Legal Frameworks",
                                            "Summary": "Development of frameworks like 'Strengthening AI Governance through Techno-Legal Frameworks' embedding ethical principles from the start.",
                                            "code": 26
                                        },
                                        {
                                            "Category": "Embedding Ethics in AI Lifecycle",
                                            "Summary": "Ethical principles (human rights, fairness, transparency, privacy, accountability) must be integrated across the entire AI lifecycle, from data to deployment.",
                                            "code": 27
                                        }
                                    ],
                                    "Summary": "Examines the foundational principles and regulatory approaches guiding AI development and deployment in India, emphasizing a human-centric and ethical stance.",
                                    "code": 24
                                },
                                {
                                    "Subnet": "Trust, Transparency & Accountability",
                                    "Categories": [
                                        {
                                            "Category": "Public Trust as Key",
                                            "Summary": "Public trust is essential for AI technology adoption; without it, technology may not get deployed effectively.",
                                            "code": 29
                                        },
                                        {
                                            "Category": "Human Oversight & Consequences",
                                            "Summary": "Human oversight in AI is crucial to prevent catastrophes, recognizing that algorithmic choices carry real human consequences.",
                                            "code": 30
                                        },
                                        {
                                            "Category": "Risk & Legal Reviews",
                                            "Summary": "Recommendation for comprehensive evidence-based risk mapping and legal reviews to ensure informed AI governance across sectors.",
                                            "code": 31
                                        }
                                    ],
                                    "Summary": "Discusses the importance of building public confidence, ensuring clarity in AI operations, and establishing mechanisms for responsibility.",
                                    "code": 28
                                },
                                {
                                    "Subnet": "Human-Centric AI",
                                    "Categories": [
                                        {
                                            "Category": "'Do No Harm' Principle",
                                            "Summary": "AI must serve people, guided by a clear 'do no harm' principle to prevent negative societal impacts.",
                                            "code": 33
                                        },
                                        {
                                            "Category": "Fair, Trustworthy & Inclusive AI",
                                            "Summary": "Shaping AI to be fair, trustworthy, inclusive, and aligned with constitutional values, human dignity, and fundamental freedoms.",
                                            "code": 34
                                        },
                                        {
                                            "Category": "AI that Listens & Performs",
                                            "Summary": "Aim for AI that listens before it decides, performing well in the lived complexity of real communities, not just demos.",
                                            "code": 35
                                        }
                                    ],
                                    "Summary": "Focuses on the imperative of designing and deploying AI systems that prioritize human well-being, rights, and societal benefit.",
                                    "code": 32
                                }
                            ],
                            "Summary": "This topic delves into the core aspects of ethical AI governance, highlighting the principles and frameworks India is adopting to ensure AI development is human-centric, trustworthy, and aligns with societal values. It covers the need for ethical integration throughout the AI lifecycle, the importance of public trust, and the commitment to a 'do no harm' approach.",
                            "code": 23
                        },
                        {
                            "Net": "AI Development & Adoption Strategy",
                            "Subnets": [
                                {
                                    "Subnet": "Skilling & Capacity Building",
                                    "Categories": [
                                        {
                                            "Category": "AI Future-Ready Workforce",
                                            "Summary": "Strong call for making India's workforce AI future-ready through capacity building, ethics integration, and inclusive transitions.",
                                            "code": 38
                                        },
                                        {
                                            "Category": "Addressing Skilling Gaps (Urban/English)",
                                            "Summary": "AI skilling remains urban-centric and English-focused; need to extend to non-urban areas and informal workers through regional language offerings.",
                                            "code": 39
                                        },
                                        {
                                            "Category": "Public Officials' AI Literacy",
                                            "Summary": "Repeated calls for AI capacity building for public officials at all levels to effectively implement AI initiatives and policies.",
                                            "code": 40
                                        }
                                    ],
                                    "Summary": "Details the efforts and recommendations related to educating, training, and preparing the workforce and public officials for the AI era, with a focus on inclusivity.",
                                    "code": 37
                                },
                                {
                                    "Subnet": "Infrastructure & Ecosystem Development",
                                    "Categories": [
                                        {
                                            "Category": "Accessible & Affordable AI Infrastructure",
                                            "Summary": "Strong national focus on building accessible and affordable AI infrastructure, including GPU procurement initiatives.",
                                            "code": 42
                                        },
                                        {
                                            "Category": "Environmental Impact in Infrastructure",
                                            "Summary": "Environmental sustainability needs to be built more directly into AI infrastructure planning, considering issues like electricity reliability and water usage.",
                                            "code": 43
                                        },
                                        {
                                            "Category": "Vibrant Industry Ecosystem",
                                            "Summary": "India hosts a vibrant industry ecosystem with 2,000+ AI companies, particularly in tech hubs like Bangalore and Hyderabad.",
                                            "code": 44
                                        }
                                    ],
                                    "Summary": "Outlines the strategic development of the physical and digital foundations necessary for widespread AI adoption, including addressing sustainability concerns.",
                                    "code": 41
                                },
                                {
                                    "Subnet": "Funding & Incentives",
                                    "Categories": [
                                        {
                                            "Category": "Socially Beneficial AI Funding",
                                            "Summary": "Socially beneficial AI projects (e.g., sustainability, multilingual) were receiving insufficient funding, despite government's direction.",
                                            "code": 46
                                        },
                                        {
                                            "Category": "Targeted Incentives for Ethical AI",
                                            "Summary": "Stakeholders ask for targeted incentives to encourage firms to adopt more ethical approaches to inclusive AI development.",
                                            "code": 47
                                        },
                                        {
                                            "Category": "Subsidized Access to AI Compute",
                                            "Summary": "India is championing an alternate approach to AI compute, with the private sector building infrastructure and subsidizing access for those most in need.",
                                            "code": 48
                                        }
                                    ],
                                    "Summary": "Covers financial mechanisms, investment strategies, and incentives designed to promote ethical and inclusive AI development and adoption.",
                                    "code": 45
                                },
                                {
                                    "Subnet": "Application & Use Cases",
                                    "Categories": [
                                        {
                                            "Category": "Enterprise & Real-World Utility",
                                            "Summary": "Emphasis on converting use cases into actual solutions for enterprise acceptance and practical utility in the real world.",
                                            "code": 50
                                        },
                                        {
                                            "Category": "Focus on Priority Sectors",
                                            "Summary": "Experimentation with regulatory sandboxes in high-impact sectors like healthcare, agriculture, and pharmaceuticals.",
                                            "code": 51
                                        },
                                        {
                                            "Category": "Translational AI Investment",
                                            "Summary": "Investing meaningfully in building architecture that brings AI models to communities in the form of apps, services, and insights.",
                                            "code": 52
                                        }
                                    ],
                                    "Summary": "Focuses on the practical implementation and deployment of AI solutions, emphasizing real-world impact and utility across various sectors and communities.",
                                    "code": 49
                                }
                            ],
                            "Summary": "This topic outlines the practical strategies and approaches for fostering AI development and adoption in India, encompassing efforts in skilling, infrastructure build-out, funding mechanisms, and the crucial translation of AI models into tangible, real-world applications for societal benefit.",
                            "code": 36
                        },
                        {
                            "Net": "Collaboration & Global Leadership",
                            "Subnets": [
                                {
                                    "Subnet": "Stakeholder Collaboration",
                                    "Categories": [
                                        {
                                            "Category": "Whole-of-Society Approach",
                                            "Summary": "No single stakeholder can shape AI alone; government, industry, academia, civil society, and research communities all have vital roles.",
                                            "code": 55
                                        },
                                        {
                                            "Category": "India AI Mission & Partners",
                                            "Summary": "The India AI Mission emphasizes partnerships with diverse stakeholders to create an enabling environment for AI.",
                                            "code": 56
                                        },
                                        {
                                            "Category": "Center-State Coordination",
                                            "Summary": "Strengthening coordination between central and state governments for AI governance and implementation, with modular policy toolkits.",
                                            "code": 57
                                        }
                                    ],
                                    "Summary": "Highlights the importance of multi-stakeholder engagement and partnerships in steering India's AI journey.",
                                    "code": 54
                                },
                                {
                                    "Subnet": "India's Global Role & Learnings",
                                    "Categories": [
                                        {
                                            "Category": "Leadership vs. Adoption",
                                            "Summary": "Convergence of UNESCO's global ethical framework and India's national priorities provides a robust foundation for leadership, not just adoption, of AI.",
                                            "code": 59
                                        },
                                        {
                                            "Category": "Playbook for Global South",
                                            "Summary": "India's approach and the RAM report's findings can serve as a useful playbook for other countries, especially in the Global South.",
                                            "code": 60
                                        },
                                        {
                                            "Category": "Re-conceptualizing AI Governance Globally",
                                            "Summary": "India's model of AI governance, focusing on incentivizing positive vision and participatory mechanisms, offers a valuable lesson for the rest of the world.",
                                            "code": 61
                                        }
                                    ],
                                    "Summary": "Focuses on India's potential to influence global AI discourse and offer valuable models for ethical and inclusive AI development.",
                                    "code": 58
                                },
                                {
                                    "Subnet": "Future Outlook & Implementation",
                                    "Categories": [
                                        {
                                            "Category": "Assessment to Action",
                                            "Summary": "The transition from assessment findings to concrete actions and implementation is the crucial next phase.",
                                            "code": 63
                                        },
                                        {
                                            "Category": "Long-Term Vision",
                                            "Summary": "Looking forward to reviewing the outcomes of the assessment in the future, aiming for inclusive AI to be the norm.",
                                            "code": 64
                                        },
                                        {
                                            "Category": "Seizing the Moment",
                                            "Summary": "Encouragement to seize the current momentum to ensure AI makes a significant difference to society, especially at the last mile.",
                                            "code": 65
                                        }
                                    ],
                                    "Summary": "Addresses the forward-looking aspects, emphasizing the transition from planning to execution and the long-term vision for AI's societal impact.",
                                    "code": 62
                                }
                            ],
                            "Summary": "This topic emphasizes the collaborative spirit driving India's AI initiatives, recognizing the need for multi-stakeholder engagement. It highlights India's aspiration and potential to lead global ethical AI development, serving as a model, particularly for the Global South. The focus is also on the crucial next steps of implementing recommendations and realizing the long-term vision for AI's societal impact.",
                            "code": 53
                        }
                    ],
                    "Summary": "The provided text is a comprehensive discussion about the launch and findings of UNESCO's AI Readiness Assessment Methodology (RAM) report for India. It highlights India's progress and unique approach to AI adoption, focusing on ethical, inclusive, and sustainable development. Speakers from UNESCO, the Government of India, and the Patrick J. McGovern Foundation detail the report's diagnostic insights across legal, social, scientific, educational, economic, and technical dimensions. Key strengths include India's AI talent pool, multilingual AI efforts, and digital public services. Areas for improvement noted are ethical data use, environmental sustainability in AI infrastructure, mainstreaming AI ethics in education, and addressing urban-centric skilling gaps. The discussion emphasizes India's 'light-touch' governance approach, participatory mechanisms, and its potential to set a global example, especially for the Global South, through collaborative efforts and a focus on real-world, community-centric applications of AI."
                }
            },
            "summaryList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 9510,
                        "CandidatesTokenCount": 4205,
                        "TotalTokenCount": 15275,
                        "thoughtsTokenCount": 1560,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Summary": "<h2>Comprehensive Synthesis Report: India's AI Readiness Assessment</h2><p>This report presents a comprehensive synthesis of India's AI Readiness Assessment Methodology (RAM), a diagnostic tool launched to ensure the impact of AI is inclusive, ethical, and sustainable. Developed over 18 months with extensive stakeholder engagement, the RAM aims to translate high-level ambitions into a grounded, actionable roadmap for the nation. The assessment involved five regional consultations across Delhi, Bangalore, Hyderabad, and Guwahati, gathering insights from over 600 stakeholders, including government, start-ups, research, and civil society. This initiative, strongly supported by the India A.I. Mission and MEITY, underscores the foundational importance of inclusivity in a country as vast and diverse as India, with 1.4 billion people, thousands of languages, and deep regional and socioeconomic variations.</p><p>AI is already mediating access to critical services like welfare, employment, education, healthcare, and justice. Therefore, the design choices embedded in AI systems have profound human consequences, determining who benefits and who is marginalized. UNESCO's Recommendation on the Ethics of AI provides a clear baseline for trustworthy AI use, and the RAM helps map India's current standing against these global ethical principles, identifying areas of strength and opportunities for improvement.</p><h3>I. Key Findings from India's AI Readiness Assessment (RAM)</h3><h4>1. Overall AI Landscape and Strengths</h4><p>The RAM report indicates that India possesses a robust environment for responsible AI development and adoption. Key strengths include:</p><ul><li>India accounts for approximately 16% of the world's AI talent, demonstrating significant human capital in the field.</li><li>There is strong momentum in multilingual AI and digital public services, supported by efforts to expand Indian language datasets and models.</li><li>The country benefits from a vibrant innovation system underpinned by strong digital foundations.</li></ul><h4>2. Areas for Improvement and Challenges</h4><p>Despite these strengths, the assessment identified several areas where India could enhance its AI readiness:</p><ul><li><strong>Data Governance:</strong> Access to, ethical use, and sharing of data could be significantly strengthened.</li><li><strong>Environmental Sustainability:</strong> Environmental sustainability needs to be more directly integrated into AI infrastructure planning and development.</li><li><strong>Ethical Integration:</strong> AI ethics, while present, needs more consistent reflection in education and training programs, including offerings in regional languages.</li><li><strong>Funding for Socially Beneficial AI:</strong> Stakeholders noted insufficient funding for AI projects focused on social good, such as sustainability and multilingual applications.</li></ul><p>The report highlights that India's AI future will be shaped not merely by technological advancements but by *how* AI is built, *who* it serves, and the safeguards embedded from the outset. The goal is to foster AI that is responsive to real-world complexities and diverse community needs.</p><h3>II. Strategic Vision and Approach to AI Governance</h3><h4>1. India AI Mission's Flexible and Collaborative Approach</h4><p>Secretary Krishnan emphasized that the India-AI mission is designed with flexibility to accommodate mid-course corrections. Its vision is to create an enabling AI environment, prioritizing partnerships with various stakeholders—including the private sector—to build AI compute, develop skills, create models, and establish databases. This approach, where the private sector builds infrastructure with subsidized access for those in need, is championed as a potentially replicable model for other Global South countries to avoid massive upfront public investment.</p><h4>2. Ethical Imperatives and Core Questions for Thoughtful AI Shaping</h4><p>Dr. Ajay Kumar Sood, Principal Scientific Advisor, underscored that while AI offers immense opportunities, it also carries profound ethical, social, environmental, and human rights implications. He posed two fundamental questions guiding India's AI journey:</p><ol><li>How can AI truly serve people, guided by a clear "
                            }
                        ]
                    }
                }
            ]
        }
    ],
    "headerCols": {
        "col1": "Part Id",
        "col2": "FileName",
        "col3": "Content",
        "col4": "Status"
    },
    "fileName": "",
    "projectType": "VIDEO",
    "sourceLanguage": "English",
    "targetLanguage": "English",
    "__v": 0
},
{
    "_id": "69930d704c5e8f0974c8780f",
    "projectId": "585241",
    "projectName": "AI for Fraud Prevention and Financial Inclusion in BFSI AI India Summit 2026",
    "projectDescription": "AI for Fraud Prevention and Financial Inclusion in BFSI AI India Summit 2026",
     "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-16T12:28:31.181Z",
    "projectUpdatedDate": "2026-02-16T12:28:31.181Z",
    "projectStatus": "Active",
    "companyId": "69803c46f08f26d9c13921b1",
    "projectCreatedBy": "69803c46f08f26d9c13921b2",
    "projectUpdatedBy": "69803c46f08f26d9c13921b2",
    "projectLastAccessDate": "2026-02-16T12:28:31.181Z",
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "Content",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "AI for Financial Inclusion_ Fraud Prevention in BFSI(MP3_128K).mp3",
            "col3": " outcomes in underserved communities, optimizes agriculture for the small farmers, and delivers public services with transparency and efficiency, then AI can become a force multiplier for equity. And I would say the direction we choose will depend on the governance framework we build today. And in this context, I think India's digital journey offers useful lessons in building at scale.  while preserving democratic accountability. Open standards, interoperable platforms, and public-private collaborations have created ecosystems that foster innovation while promoting citizen rights. And as AI become embedded with these systems, our focus remains on building trust through safeguards, grievance, redressal mechanism, and responsible data governance practices.  Another important aspect which I want to highlight and has become very important is that geopolitical competition increasingly intersects with technological advancement and you all know that. AI governance is also emerging as an important dimension of strategic diplomacy nowadays and it is therefore important that we cooperate and collaborate. Cooperation in AI standards, ethical frameworks and research partnerships can build  bridges of trust among nations. On the other hand, fragmented regulatory regimes risk creating digital silos and widening technological divides. And in this context, I think dialogues such as this, bringing together policymakers from across the region are therefore vital in shaping a cooperative digital future. And despite all this, however,  I think the regulations alone would not ensure responsible AI. Capacity building is also equally critical. Policy makers need technological literacy. Regulator needs analytical tools. Judges need technical understanding. Citizen needs digital awareness. Democratizing AI resources, therefore, also means democratizing knowledge and institutional capacity. And without this, even the most well-designed legal frameworks would struggle.  implementation. So as we navigate this transition to my mind, three guiding ideas may help us stay on course. First, governance must remain human-centric, ensuring that technology serves people. Second, it must be inclusive, reflecting the diversity of the societies. Honorable Member of Parliament has alluded to a great respect on this. And third, I think it must be adaptive.  evolving alongside technological changes rather than lagging behind. And friends, the rule of law is not an obstacle to innovation, as often people think that rule of law is something which is going to stifle innovation and so on. But to my mind, it is the foundation. Trust enables adaptation. Predictability encourages investment. Accountability sustains legitimacy. And when citizens trust that AI systems are fair, transparent, and contestable,  they are far more willing to embrace technological transformation. Now let me conclude with a thought. The future of AI governance will not be determined only in laboratories or boardrooms. It will be shaped in legislatures, courtrooms, and democratic forums such as this one. And the choices which we make today will determine whether AI becomes an instrument of concentration or a catalyst for empowerment.  And if we align innovation with democratic values, we will not only regulate technology, we will shape a future where technological progress and human dignity advances together. Thank you. Thank you so much, sir. I invite Kazim to hand over the mementos to the speakers. Speakers, can you please take your places in front of the dais?  Thank you.  Thank you.",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771244904704.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:29",
                    "text": " outcomes in underserved communities, optimizes agriculture for the small farmers, and delivers public services with transparency and efficiency, then AI can become a force multiplier for equity. And I would say the direction we choose will depend on the governance framework we build today. And in this context, I think India's digital journey offers useful lessons in building at scale."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:29",
                    "end": "00:00:57",
                    "text": " while preserving democratic accountability. Open standards, interoperable platforms, and public-private collaborations have created ecosystems that foster innovation while promoting citizen rights. And as AI become embedded with these systems, our focus remains on building trust through safeguards, grievance, redressal mechanism, and responsible data governance practices."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:58",
                    "end": "00:01:28",
                    "text": " Another important aspect which I want to highlight and has become very important is that geopolitical competition increasingly intersects with technological advancement and you all know that. AI governance is also emerging as an important dimension of strategic diplomacy nowadays and it is therefore important that we cooperate and collaborate. Cooperation in AI standards, ethical frameworks and research partnerships can build"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:28",
                    "end": "00:01:55",
                    "text": " bridges of trust among nations. On the other hand, fragmented regulatory regimes risk creating digital silos and widening technological divides. And in this context, I think dialogues such as this, bringing together policymakers from across the region are therefore vital in shaping a cooperative digital future. And despite all this, however,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:55",
                    "end": "00:02:25",
                    "text": " I think the regulations alone would not ensure responsible AI. Capacity building is also equally critical. Policy makers need technological literacy. Regulator needs analytical tools. Judges need technical understanding. Citizen needs digital awareness. Democratizing AI resources, therefore, also means democratizing knowledge and institutional capacity. And without this, even the most well-designed legal frameworks would struggle."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:25",
                    "end": "00:02:56",
                    "text": " implementation. So as we navigate this transition to my mind, three guiding ideas may help us stay on course. First, governance must remain human-centric, ensuring that technology serves people. Second, it must be inclusive, reflecting the diversity of the societies. Honorable Member of Parliament has alluded to a great respect on this. And third, I think it must be adaptive."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:56",
                    "end": "00:03:26",
                    "text": " evolving alongside technological changes rather than lagging behind. And friends, the rule of law is not an obstacle to innovation, as often people think that rule of law is something which is going to stifle innovation and so on. But to my mind, it is the foundation. Trust enables adaptation. Predictability encourages investment. Accountability sustains legitimacy. And when citizens trust that AI systems are fair, transparent, and contestable,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:26",
                    "end": "00:03:55",
                    "text": " they are far more willing to embrace technological transformation. Now let me conclude with a thought. The future of AI governance will not be determined only in laboratories or boardrooms. It will be shaped in legislatures, courtrooms, and democratic forums such as this one. And the choices which we make today will determine whether AI becomes an instrument of concentration or a catalyst for empowerment."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:55",
                    "end": "00:04:23",
                    "text": " And if we align innovation with democratic values, we will not only regulate technology, we will shape a future where technological progress and human dignity advances together. Thank you. Thank you so much, sir. I invite Kazim to hand over the mementos to the speakers. Speakers, can you please take your places in front of the dais?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:05",
                    "end": "00:05:32",
                    "text": " Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:32",
                    "end": "00:06:00",
                    "text": " Thank you."
                }
            ]
        }
    ],
    "projectDeleted": false,
    "codeFrameColumn": [
        {
            "col1": "col1",
            "col2": "col3",
            "researchObjective": "",
            "filterColumn": [],
            "netSetting": true,
            "subNetSetting": true,
            "categorySetting": true,
            "sentimentSetting": true,
            "thinkingBudgetSetting": 2000,
            "questionType": "OPENEND",
            "otherBrand": false,
            "existingCodeFrame": [],
            "isList": false,
            "codeFrameWithDataList": [],
            "codeFrameList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 1479,
                        "CandidatesTokenCount": 2247,
                        "TotalTokenCount": 5169,
                        "thoughtsTokenCount": 1443,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Net": "AI Governance Principles & Vision",
                                "Subnets": [
                                    {
                                        "Subnet": "Core Guiding Ideas",
                                        "Categories": [
                                            {
                                                "Category": "Human-centric Governance",
                                                "Summary": "Ensuring that technology primarily serves people and human well-being."
                                            },
                                            {
                                                "Category": "Inclusive Governance",
                                                "Summary": "Reflecting the diversity of societies and ensuring broad representation in AI development and deployment."
                                            },
                                            {
                                                "Category": "Adaptive Governance",
                                                "Summary": "Evolving regulatory frameworks and policies alongside rapid technological changes, rather than lagging behind."
                                            }
                                        ],
                                        "Summary": "Fundamental principles that should direct the development and implementation of AI governance."
                                    },
                                    {
                                        "Subnet": "Role of Rule of Law & Trust",
                                        "Categories": [
                                            {
                                                "Category": "Foundation for Innovation",
                                                "Summary": "The rule of law is presented not as an obstacle but as a foundational element that enables technological advancement."
                                            },
                                            {
                                                "Category": "Trust Enables Adaptation",
                                                "Summary": "Citizens' willingness to embrace AI transformations is directly linked to their trust in fair, transparent, and contestable AI systems."
                                            },
                                            {
                                                "Category": "Predictability & Accountability",
                                                "Summary": "Predictability in AI systems encourages investment, while accountability sustains legitimacy and public acceptance."
                                            }
                                        ],
                                        "Summary": "The critical importance of legal frameworks and public trust in fostering responsible AI adoption and innovation."
                                    },
                                    {
                                        "Subnet": "Future Impact & Choice",
                                        "Categories": [
                                            {
                                                "Category": "Instrument of Concentration vs. Catalyst for Empowerment",
                                                "Summary": "The choice of governance will determine if AI centralizes power or broadly empowers individuals and communities."
                                            },
                                            {
                                                "Category": "Aligning Innovation with Democratic Values",
                                                "Summary": "The necessity of integrating technological progress with core democratic principles to shape a desirable future."
                                            },
                                            {
                                                "Category": "Technological Progress & Human Dignity",
                                                "Summary": "The ultimate goal of AI governance is to advance both technological capabilities and the dignity of humanity simultaneously."
                                            }
                                        ],
                                        "Summary": "The profound implications of current AI governance decisions on the future societal role and ethical alignment of AI."
                                    }
                                ],
                                "Summary": "This topic covers the foundational ideals and overarching objectives for how AI should be governed, focusing on ethical considerations, democratic values, and the ultimate societal impact."
                            },
                            {
                                "Net": "Frameworks for AI Governance",
                                "Subnets": [
                                    {
                                        "Subnet": "India's Digital Journey as a Model",
                                        "Categories": [
                                            {
                                                "Category": "Lessons in Building at Scale",
                                                "Summary": "Highlighting India's experience as a useful example for implementing large-scale digital initiatives."
                                            },
                                            {
                                                "Category": "Preserving Democratic Accountability",
                                                "Summary": "Emphasizing that large-scale digital development can be achieved while maintaining democratic oversight and citizen rights."
                                            },
                                            {
                                                "Category": "Open Standards & Interoperable Platforms",
                                                "Summary": "The use of open technologies and interconnected systems to foster innovation and accessibility."
                                            }
                                        ],
                                        "Summary": "Utilizing India's experience as a successful case study for developing and implementing digital infrastructure on a large scale."
                                    },
                                    {
                                        "Subnet": "Building Trust in AI Systems",
                                        "Categories": [
                                            {
                                                "Category": "Safeguards and Grievance Redressal",
                                                "Summary": "Implementing mechanisms to protect users and provide avenues for addressing concerns or issues with AI systems."
                                            },
                                            {
                                                "Category": "Responsible Data Governance Practices",
                                                "Summary": "Managing data ethically and securely to ensure privacy, fairness, and accountability in AI applications."
                                            },
                                            {
                                                "Category": "Embedding AI with Existing Systems",
                                                "Summary": "Integrating AI into established digital ecosystems with a focus on maintaining trust and accountability."
                                            }
                                        ],
                                        "Summary": "Strategies and mechanisms to ensure user confidence and ethical operation of AI systems within broader digital ecosystems."
                                    },
                                    {
                                        "Subnet": "Public-Private Collaborations",
                                        "Categories": [
                                            {
                                                "Category": "Ecosystems Fostering Innovation",
                                                "Summary": "The creation of environments where both public and private sectors work together to drive technological advancement."
                                            },
                                            {
                                                "Category": "Promoting Citizen Rights",
                                                "Summary": "Ensuring that collaborative efforts also prioritize and protect the rights and interests of citizens."
                                            },
                                            {
                                                "Category": "Multi-stakeholder Engagement",
                                                "Summary": "The involvement of various parties, including government, industry, and civil society, in developing AI frameworks."
                                            }
                                        ],
                                        "Summary": "The role of partnerships between government and private entities in fostering innovation while safeguarding public interest."
                                    }
                                ],
                                "Summary": "This topic explores the structural and operational components necessary for effective AI governance, drawing insights from existing models and emphasizing practical approaches to ensure trust and accountability."
                            },
                            {
                                "Net": "Challenges & Requirements for AI Governance",
                                "Subnets": [
                                    {
                                        "Subnet": "Beyond Regulation: Capacity Building",
                                        "Categories": [
                                            {
                                                "Category": "Technological Literacy for Policymakers",
                                                "Summary": "The need for governing bodies to understand the technical aspects of AI to create effective policies."
                                            },
                                            {
                                                "Category": "Analytical Tools for Regulators",
                                                "Summary": "Equipping regulatory bodies with the necessary instruments to analyze and oversee AI systems effectively."
                                            },
                                            {
                                                "Category": "Technical Understanding for Judges",
                                                "Summary": "Judicial systems requiring technical knowledge to adjudicate cases involving AI fairly and competently."
                                            },
                                            {
                                                "Category": "Digital Awareness for Citizens",
                                                "Summary": "Educating the public about AI to enable informed engagement and responsible use."
                                            }
                                        ],
                                        "Summary": "The necessity of developing knowledge and skills across various sectors of society to effectively manage AI, beyond just enacting laws."
                                    },
                                    {
                                        "Subnet": "Geopolitical Dimensions",
                                        "Categories": [
                                            {
                                                "Category": "Intersection with Technological Advancement",
                                                "Summary": "How global power dynamics and technological progress increasingly influence each other, especially with AI."
                                            },
                                            {
                                                "Category": "Strategic Diplomacy",
                                                "Summary": "AI governance is becoming a key element in international relations and negotiations."
                                            },
                                            {
                                                "Category": "Preventing Digital Silos",
                                                "Summary": "The importance of avoiding fragmented regulatory regimes that could isolate technological development and widen divides."
                                            }
                                        ],
                                        "Summary": "The intricate relationship between international politics, national security, and the development and governance of AI technologies."
                                    },
                                    {
                                        "Subnet": "Knowledge and Institutional Capacity",
                                        "Categories": [
                                            {
                                                "Category": "Democratizing AI Resources",
                                                "Summary": "Making AI tools and infrastructure accessible to a broader population, not just a select few."
                                            },
                                            {
                                                "Category": "Democratizing Knowledge",
                                                "Summary": "Spreading understanding and education about AI to a wider audience to ensure informed participation."
                                            },
                                            {
                                                "Category": "Strengthening Institutional Capacity",
                                                "Summary": "Building the capabilities of organizations and governments to effectively manage and implement AI initiatives."
                                            }
                                        ],
                                        "Summary": "The need to widely distribute AI knowledge and resources, and to strengthen organizational abilities to handle AI effectively."
                                    }
                                ],
                                "Summary": "This topic addresses the various obstacles and essential preconditions for establishing robust AI governance, focusing on human capital development and the geopolitical landscape."
                            },
                            {
                                "Net": "International Cooperation & Diplomacy",
                                "Subnets": [
                                    {
                                        "Subnet": "Importance of Collaboration",
                                        "Categories": [
                                            {
                                                "Category": "Building Bridges of Trust Among Nations",
                                                "Summary": "Cooperative efforts in AI foster mutual understanding and confidence between countries."
                                            },
                                            {
                                                "Category": "Cooperation in AI Standards",
                                                "Summary": "Harmonizing technical and operational guidelines for AI globally."
                                            },
                                            {
                                                "Category": "Ethical Frameworks",
                                                "Summary": "Developing shared principles for the responsible use and development of AI."
                                            },
                                            {
                                                "Category": "Research Partnerships",
                                                "Summary": "Joint scientific and technological endeavors in AI across borders."
                                            }
                                        ],
                                        "Summary": "The necessity of global teamwork to establish common ground and build confidence in AI governance."
                                    },
                                    {
                                        "Subnet": "Risks of Fragmentation",
                                        "Categories": [
                                            {
                                                "Category": "Creating Digital Silos",
                                                "Summary": "Fragmented regulatory approaches leading to isolated technological ecosystems and reduced interoperability."
                                            },
                                            {
                                                "Category": "Widening Technological Divides",
                                                "Summary": "Disparate regulations potentially exacerbating inequalities in access to and benefits from AI technologies."
                                            },
                                            {
                                                "Category": "Hindering Global Progress",
                                                "Summary": "Lack of unified standards and cooperation impeding the collective advancement of AI for common good."
                                            }
                                        ],
                                        "Summary": "The dangers associated with uncoordinated national approaches to AI governance, leading to isolation and inequality."
                                    },
                                    {
                                        "Subnet": "Dialogues & Trust Building",
                                        "Categories": [
                                            {
                                                "Category": "Shaping a Cooperative Digital Future",
                                                "Summary": "Engagements like international dialogues are crucial for forging a collaborative path in the digital realm."
                                            },
                                            {
                                                "Category": "Policymaker Engagement",
                                                "Summary": "Bringing together governmental leaders to discuss and agree upon common AI strategies."
                                            },
                                            {
                                                "Category": "Preventing a Race to the Bottom",
                                                "Summary": "Collaborative efforts can prevent countries from lowering ethical standards to gain competitive advantages in AI development."
                                            }
                                        ],
                                        "Summary": "The role of international discussions and shared understanding in fostering a united approach to AI governance and preventing negative outcomes."
                                    }
                                ],
                                "Summary": "This topic highlights the global dimension of AI governance, emphasizing the need for international collaboration to develop common standards, ethical frameworks, and avoid digital fragmentation."
                            },
                            {
                                "Net": "996 Non specific TOPIC",
                                "Subnets": [
                                    {
                                        "Subnet": "Irrelevant/Uncategorizable",
                                        "Categories": [
                                            {
                                                "Category": "Responses needing upcoding",
                                                "Summary": "Responses that are not directly relevant to the identified themes but may contain some discernible meaning."
                                            },
                                            {
                                                "Category": "Contextually Inapplicable",
                                                "Summary": "Verbatims that are comprehensible but do not fit any specific category within the codeframe."
                                            }
                                        ],
                                        "Summary": "Responses that are not clearly relevant to the main topics and require further review or specific handling."
                                    }
                                ],
                                "Summary": "Catch-all for responses that are irrelevant or cannot be easily coded into the defined hierarchical structure, possibly requiring further analysis."
                            },
                            {
                                "Net": "997 Don't Know Can't Say",
                                "Subnets": [
                                    {
                                        "Subnet": "Non-committal Responses",
                                        "Categories": [
                                            {
                                                "Category": "Vague/Uncertain",
                                                "Summary": "Responses indicating a lack of clear opinion, knowledge, or a preference not to state one."
                                            },
                                            {
                                                "Category": "Indecisive",
                                                "Summary": "Responses where the respondent expresses an inability to make a choice or provide a definite answer."
                                            }
                                        ],
                                        "Summary": "Responses that express uncertainty, indecisiveness, or a lack of opinion regarding the survey question."
                                    }
                                ],
                                "Summary": "This category captures responses where participants explicitly state they don't know or cannot provide a definitive answer."
                            },
                            {
                                "Net": "998 Not Applicable",
                                "Subnets": [
                                    {
                                        "Subnet": "Irrelevant to Question",
                                        "Categories": [
                                            {
                                                "Category": "Out of Scope",
                                                "Summary": "Responses that are clearly outside the purview of the survey question or project objectives."
                                            },
                                            {
                                                "Category": "Not Related to Topic",
                                                "Summary": "Statements that bear no relation to the subject matter being discussed."
                                            }
                                        ],
                                        "Summary": "Responses that are explicitly not relevant or pertinent to the specific question asked."
                                    }
                                ],
                                "Summary": "For responses that are understandable but are not relevant or applicable to the question asked."
                            },
                            {
                                "Net": "999 None-Nothing",
                                "Subnets": [
                                    {
                                        "Subnet": "Gibberish/Meaningless",
                                        "Categories": [
                                            {
                                                "Category": "Non-language characters",
                                                "Summary": "Responses composed of random symbols, numbers, or characters that do not form intelligible words."
                                            },
                                            {
                                                "Category": "Unintelligible Phrases",
                                                "Summary": "Words or phrases that do not convey any coherent meaning or are grammatically nonsensical."
                                            },
                                            {
                                                "Category": "Empty Responses",
                                                "Summary": "Blank submissions or responses indicating 'none' or 'nothing' without further explanation."
                                            }
                                        ],
                                        "Summary": "Responses that are unintelligible, nonsensical, or contain no meaningful content."
                                    }
                                ],
                                "Summary": "This category is for responses that are entirely meaningless, gibberish, or empty."
                            }
                        ],
                        "Summary": "The provided text is a speech highlighting the critical importance of effective AI governance to ensure AI acts as a force multiplier for equity and human dignity. It emphasizes human-centric, inclusive, and adaptive governance principles, drawing lessons from India's digital journey which prioritizes open standards, interoperable platforms, and public-private collaborations. The speaker identifies geopolitical competition and the need for international cooperation on AI standards and ethical frameworks as key challenges. Beyond regulation, the speech stresses the necessity of capacity building and technological literacy across all societal levels to ensure successful AI implementation. Ultimately, it argues that democratic values and the rule of law are foundational to building trust and shaping a future where technology empowers rather than concentrates power."
                    }
                }
            ],
            "onlyCodeFrame": true,
            "errorList": [],
            "codeFrameGenerateType": "CODE_FRAME_ONLY",
            "requestType": "CODEFRAME",
            "requestMode": "DEVELOPMENT",
            "sourceLanguage": "English",
            "targetLanguage": "English",
            "response": {
                "status": "OK",
                "data": {
                    "status": "OK",
                    "PromptTokenCount": 1479,
                    "CandidatesTokenCount": 2247,
                    "TotalTokenCount": 5169,
                    "thoughtsTokenCount": 1443,
                    "finishReason": "STOP",
                    "data": [
                        {
                            "Net": "AI Governance Principles & Vision",
                            "Subnets": [
                                {
                                    "Subnet": "Core Guiding Ideas",
                                    "Categories": [
                                        {
                                            "Category": "Human-centric Governance",
                                            "Summary": "Ensuring that technology primarily serves people and human well-being.",
                                            "code": 3
                                        },
                                        {
                                            "Category": "Inclusive Governance",
                                            "Summary": "Reflecting the diversity of societies and ensuring broad representation in AI development and deployment.",
                                            "code": 4
                                        },
                                        {
                                            "Category": "Adaptive Governance",
                                            "Summary": "Evolving regulatory frameworks and policies alongside rapid technological changes, rather than lagging behind.",
                                            "code": 5
                                        }
                                    ],
                                    "Summary": "Fundamental principles that should direct the development and implementation of AI governance.",
                                    "code": 2
                                },
                                {
                                    "Subnet": "Role of Rule of Law & Trust",
                                    "Categories": [
                                        {
                                            "Category": "Foundation for Innovation",
                                            "Summary": "The rule of law is presented not as an obstacle but as a foundational element that enables technological advancement.",
                                            "code": 7
                                        },
                                        {
                                            "Category": "Trust Enables Adaptation",
                                            "Summary": "Citizens' willingness to embrace AI transformations is directly linked to their trust in fair, transparent, and contestable AI systems.",
                                            "code": 8
                                        },
                                        {
                                            "Category": "Predictability & Accountability",
                                            "Summary": "Predictability in AI systems encourages investment, while accountability sustains legitimacy and public acceptance.",
                                            "code": 9
                                        }
                                    ],
                                    "Summary": "The critical importance of legal frameworks and public trust in fostering responsible AI adoption and innovation.",
                                    "code": 6
                                },
                                {
                                    "Subnet": "Future Impact & Choice",
                                    "Categories": [
                                        {
                                            "Category": "Instrument of Concentration vs. Catalyst for Empowerment",
                                            "Summary": "The choice of governance will determine if AI centralizes power or broadly empowers individuals and communities.",
                                            "code": 11
                                        },
                                        {
                                            "Category": "Aligning Innovation with Democratic Values",
                                            "Summary": "The necessity of integrating technological progress with core democratic principles to shape a desirable future.",
                                            "code": 12
                                        },
                                        {
                                            "Category": "Technological Progress & Human Dignity",
                                            "Summary": "The ultimate goal of AI governance is to advance both technological capabilities and the dignity of humanity simultaneously.",
                                            "code": 13
                                        }
                                    ],
                                    "Summary": "The profound implications of current AI governance decisions on the future societal role and ethical alignment of AI.",
                                    "code": 10
                                }
                            ],
                            "Summary": "This topic covers the foundational ideals and overarching objectives for how AI should be governed, focusing on ethical considerations, democratic values, and the ultimate societal impact.",
                            "code": 1
                        },
                        {
                            "Net": "Frameworks for AI Governance",
                            "Subnets": [
                                {
                                    "Subnet": "India's Digital Journey as a Model",
                                    "Categories": [
                                        {
                                            "Category": "Lessons in Building at Scale",
                                            "Summary": "Highlighting India's experience as a useful example for implementing large-scale digital initiatives.",
                                            "code": 16
                                        },
                                        {
                                            "Category": "Preserving Democratic Accountability",
                                            "Summary": "Emphasizing that large-scale digital development can be achieved while maintaining democratic oversight and citizen rights.",
                                            "code": 17
                                        },
                                        {
                                            "Category": "Open Standards & Interoperable Platforms",
                                            "Summary": "The use of open technologies and interconnected systems to foster innovation and accessibility.",
                                            "code": 18
                                        }
                                    ],
                                    "Summary": "Utilizing India's experience as a successful case study for developing and implementing digital infrastructure on a large scale.",
                                    "code": 15
                                },
                                {
                                    "Subnet": "Building Trust in AI Systems",
                                    "Categories": [
                                        {
                                            "Category": "Safeguards and Grievance Redressal",
                                            "Summary": "Implementing mechanisms to protect users and provide avenues for addressing concerns or issues with AI systems.",
                                            "code": 20
                                        },
                                        {
                                            "Category": "Responsible Data Governance Practices",
                                            "Summary": "Managing data ethically and securely to ensure privacy, fairness, and accountability in AI applications.",
                                            "code": 21
                                        },
                                        {
                                            "Category": "Embedding AI with Existing Systems",
                                            "Summary": "Integrating AI into established digital ecosystems with a focus on maintaining trust and accountability.",
                                            "code": 22
                                        }
                                    ],
                                    "Summary": "Strategies and mechanisms to ensure user confidence and ethical operation of AI systems within broader digital ecosystems.",
                                    "code": 19
                                },
                                {
                                    "Subnet": "Public-Private Collaborations",
                                    "Categories": [
                                        {
                                            "Category": "Ecosystems Fostering Innovation",
                                            "Summary": "The creation of environments where both public and private sectors work together to drive technological advancement.",
                                            "code": 24
                                        },
                                        {
                                            "Category": "Promoting Citizen Rights",
                                            "Summary": "Ensuring that collaborative efforts also prioritize and protect the rights and interests of citizens.",
                                            "code": 25
                                        },
                                        {
                                            "Category": "Multi-stakeholder Engagement",
                                            "Summary": "The involvement of various parties, including government, industry, and civil society, in developing AI frameworks.",
                                            "code": 26
                                        }
                                    ],
                                    "Summary": "The role of partnerships between government and private entities in fostering innovation while safeguarding public interest.",
                                    "code": 23
                                }
                            ],
                            "Summary": "This topic explores the structural and operational components necessary for effective AI governance, drawing insights from existing models and emphasizing practical approaches to ensure trust and accountability.",
                            "code": 14
                        },
                        {
                            "Net": "Challenges & Requirements for AI Governance",
                            "Subnets": [
                                {
                                    "Subnet": "Beyond Regulation: Capacity Building",
                                    "Categories": [
                                        {
                                            "Category": "Technological Literacy for Policymakers",
                                            "Summary": "The need for governing bodies to understand the technical aspects of AI to create effective policies.",
                                            "code": 29
                                        },
                                        {
                                            "Category": "Analytical Tools for Regulators",
                                            "Summary": "Equipping regulatory bodies with the necessary instruments to analyze and oversee AI systems effectively.",
                                            "code": 30
                                        },
                                        {
                                            "Category": "Technical Understanding for Judges",
                                            "Summary": "Judicial systems requiring technical knowledge to adjudicate cases involving AI fairly and competently.",
                                            "code": 31
                                        },
                                        {
                                            "Category": "Digital Awareness for Citizens",
                                            "Summary": "Educating the public about AI to enable informed engagement and responsible use.",
                                            "code": 32
                                        }
                                    ],
                                    "Summary": "The necessity of developing knowledge and skills across various sectors of society to effectively manage AI, beyond just enacting laws.",
                                    "code": 28
                                },
                                {
                                    "Subnet": "Geopolitical Dimensions",
                                    "Categories": [
                                        {
                                            "Category": "Intersection with Technological Advancement",
                                            "Summary": "How global power dynamics and technological progress increasingly influence each other, especially with AI.",
                                            "code": 34
                                        },
                                        {
                                            "Category": "Strategic Diplomacy",
                                            "Summary": "AI governance is becoming a key element in international relations and negotiations.",
                                            "code": 35
                                        },
                                        {
                                            "Category": "Preventing Digital Silos",
                                            "Summary": "The importance of avoiding fragmented regulatory regimes that could isolate technological development and widen divides.",
                                            "code": 36
                                        }
                                    ],
                                    "Summary": "The intricate relationship between international politics, national security, and the development and governance of AI technologies.",
                                    "code": 33
                                },
                                {
                                    "Subnet": "Knowledge and Institutional Capacity",
                                    "Categories": [
                                        {
                                            "Category": "Democratizing AI Resources",
                                            "Summary": "Making AI tools and infrastructure accessible to a broader population, not just a select few.",
                                            "code": 38
                                        },
                                        {
                                            "Category": "Democratizing Knowledge",
                                            "Summary": "Spreading understanding and education about AI to a wider audience to ensure informed participation.",
                                            "code": 39
                                        },
                                        {
                                            "Category": "Strengthening Institutional Capacity",
                                            "Summary": "Building the capabilities of organizations and governments to effectively manage and implement AI initiatives.",
                                            "code": 40
                                        }
                                    ],
                                    "Summary": "The need to widely distribute AI knowledge and resources, and to strengthen organizational abilities to handle AI effectively.",
                                    "code": 37
                                }
                            ],
                            "Summary": "This topic addresses the various obstacles and essential preconditions for establishing robust AI governance, focusing on human capital development and the geopolitical landscape.",
                            "code": 27
                        },
                        {
                            "Net": "International Cooperation & Diplomacy",
                            "Subnets": [
                                {
                                    "Subnet": "Importance of Collaboration",
                                    "Categories": [
                                        {
                                            "Category": "Building Bridges of Trust Among Nations",
                                            "Summary": "Cooperative efforts in AI foster mutual understanding and confidence between countries.",
                                            "code": 43
                                        },
                                        {
                                            "Category": "Cooperation in AI Standards",
                                            "Summary": "Harmonizing technical and operational guidelines for AI globally.",
                                            "code": 44
                                        },
                                        {
                                            "Category": "Ethical Frameworks",
                                            "Summary": "Developing shared principles for the responsible use and development of AI.",
                                            "code": 45
                                        },
                                        {
                                            "Category": "Research Partnerships",
                                            "Summary": "Joint scientific and technological endeavors in AI across borders.",
                                            "code": 46
                                        }
                                    ],
                                    "Summary": "The necessity of global teamwork to establish common ground and build confidence in AI governance.",
                                    "code": 42
                                },
                                {
                                    "Subnet": "Risks of Fragmentation",
                                    "Categories": [
                                        {
                                            "Category": "Creating Digital Silos",
                                            "Summary": "Fragmented regulatory approaches leading to isolated technological ecosystems and reduced interoperability.",
                                            "code": 48
                                        },
                                        {
                                            "Category": "Widening Technological Divides",
                                            "Summary": "Disparate regulations potentially exacerbating inequalities in access to and benefits from AI technologies.",
                                            "code": 49
                                        },
                                        {
                                            "Category": "Hindering Global Progress",
                                            "Summary": "Lack of unified standards and cooperation impeding the collective advancement of AI for common good.",
                                            "code": 50
                                        }
                                    ],
                                    "Summary": "The dangers associated with uncoordinated national approaches to AI governance, leading to isolation and inequality.",
                                    "code": 47
                                },
                                {
                                    "Subnet": "Dialogues & Trust Building",
                                    "Categories": [
                                        {
                                            "Category": "Shaping a Cooperative Digital Future",
                                            "Summary": "Engagements like international dialogues are crucial for forging a collaborative path in the digital realm.",
                                            "code": 52
                                        },
                                        {
                                            "Category": "Policymaker Engagement",
                                            "Summary": "Bringing together governmental leaders to discuss and agree upon common AI strategies.",
                                            "code": 53
                                        },
                                        {
                                            "Category": "Preventing a Race to the Bottom",
                                            "Summary": "Collaborative efforts can prevent countries from lowering ethical standards to gain competitive advantages in AI development.",
                                            "code": 54
                                        }
                                    ],
                                    "Summary": "The role of international discussions and shared understanding in fostering a united approach to AI governance and preventing negative outcomes.",
                                    "code": 51
                                }
                            ],
                            "Summary": "This topic highlights the global dimension of AI governance, emphasizing the need for international collaboration to develop common standards, ethical frameworks, and avoid digital fragmentation.",
                            "code": 41
                        }
                    ],
                    "Summary": "The provided text is a speech highlighting the critical importance of effective AI governance to ensure AI acts as a force multiplier for equity and human dignity. It emphasizes human-centric, inclusive, and adaptive governance principles, drawing lessons from India's digital journey which prioritizes open standards, interoperable platforms, and public-private collaborations. The speaker identifies geopolitical competition and the need for international cooperation on AI standards and ethical frameworks as key challenges. Beyond regulation, the speech stresses the necessity of capacity building and technological literacy across all societal levels to ensure successful AI implementation. Ultimately, it argues that democratic values and the rule of law are foundational to building trust and shaping a future where technology empowers rather than concentrates power."
                }
            },
            "summaryList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 1140,
                        "CandidatesTokenCount": 1421,
                        "TotalTokenCount": 3732,
                        "thoughtsTokenCount": 1171,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Summary": "<h2>Comprehensive Synthesis Report: Advancing Responsible AI Governance</h2><h3>Introduction: The Transformative Potential and Governance Imperative of AI</h3><p>The discourse commenced by highlighting the profound potential of Artificial Intelligence (AI) to act as a significant force multiplier for societal equity and efficiency. Specifically, AI can drive positive outcomes in underserved communities, optimize agricultural practices for small farmers, and deliver public services with enhanced transparency and efficiency. However, the realization of this equitable future is contingent upon the governance frameworks established today. The direction chosen for AI development and deployment &mdash; whether towards equity or concentration &mdash; will be determined by the architectural choices made in its governance.</p><h3>Section 1: Lessons from India's Digital Journey and Core Pillars of Trust</h3><h4>1.1. India's Digital Journey: A Blueprint for Scalable and Accountable AI Systems</h4><p>India's extensive digital transformation journey offers invaluable lessons for building AI systems at scale while rigorously preserving democratic accountability. This success has been underpinned by several critical components:</p><ul><li><b>Open Standards:</b> Promoting accessibility and interoperability across various platforms.</li><li><b>Interoperable Platforms:</b> Enabling seamless communication and data exchange between different systems.</li><li><b>Public-Private Collaborations:</b> Fostering ecosystems that stimulate innovation while upholding citizen rights.</li></ul><p>As AI becomes increasingly integrated into these systems, the central focus remains on cultivating and maintaining public trust.</p><h4>1.2. Building Trust in AI Systems: Essential Safeguards and Governance Practices</h4><p>Trust in AI is not inherent; it must be actively built and sustained through deliberate measures. The core strategies for building trust include:</p><ul><li><b>Robust Safeguards:</b> Implementing mechanisms to prevent misuse and ensure ethical operation.</li><li><b>Effective Grievance Redressal Mechanisms:</b> Providing accessible avenues for citizens to address concerns and seek resolutions.</li><li><b>Responsible Data Governance Practices:</b> Ensuring data privacy, security, and ethical handling throughout the AI lifecycle.</li></ul><h3>Section 2: The Geopolitical and Human Capacity Dimensions of AI Governance</h3><h4>2.1. AI Governance in the Geopolitical Landscape: Cooperation vs. Fragmentation</h4><p>The discussion underscored the growing intersection of geopolitical competition with technological advancement, positioning AI governance as a critical dimension of strategic diplomacy. The path forward necessitates international cooperation and collaboration to prevent digital fragmentation:</p><ul><li><b>Benefits of Cooperation:</b> Collaborating on AI standards, developing shared ethical frameworks, and fostering international research partnerships are vital for building bridges of trust among nations.</li><li><b>Risks of Fragmentation:</b> Conversely, fragmented regulatory regimes risk creating isolated digital silos and exacerbating existing technological divides, hindering global progress.</li></ul><p>Dialogues and forums such as the present one are therefore deemed essential for shaping a cooperative and inclusive digital future.</p><h4>2.2. Beyond Regulation: The Indispensable Role of Capacity Building</h4><p>A crucial insight highlighted was that regulations alone are insufficient to ensure responsible AI. Capacity building across various societal sectors is equally critical for effective implementation and oversight:</p><ul><li><b>Policymaker Literacy:</b> Policy makers require a deep understanding of technological nuances to craft effective legislation.</li><li><b>Regulator Tools:</b> Regulators need advanced analytical tools to monitor AI systems and enforce compliance.</li><li><b>Judicial Understanding:</b> Judges require technical comprehension to adjudicate cases involving AI with fairness and informed judgment.</li><li><b>Citizen Awareness:</b> Citizens need digital awareness to understand their rights, identify potential risks, and engage meaningfully with AI systems.</li></ul><p>The democratization of AI resources inherently means democratizing knowledge and institutional capacity. Without this foundational capacity, even the most meticulously designed legal frameworks will struggle in their implementation.</p><h3>Section 3: Guiding Principles and the Foundational Role of the Rule of Law</h3><h4>3.1. Three Guiding Principles for Navigating AI Transition</h4><p>As societies navigate the complex transition brought about by AI, three guiding ideas were proposed to ensure a constructive and beneficial trajectory:</p><ul><li><b>Human-Centric Governance:</b> Ensuring that technology consistently serves the needs and well-being of people, rather than the inverse.</li><li><b>Inclusive Governance:</b> Reflecting the diverse values, cultures, and needs of all segments of society, a point strongly echoed by the Honorable Member of Parliament.</li><li><b>Adaptive Governance:</b> The governance framework must evolve in tandem with technological changes, demonstrating agility and avoiding the pitfalls of lagging behind innovation.</li></ul><h4>3.2. The Rule of Law: A Foundation for Innovation and Trust</h4><p>Contrary to the common misconception that the rule of law impedes innovation, it was posited as its very foundation. A robust legal framework fosters an environment where innovation can thrive responsibly:</p><ul><li><b>Trust Enables Adaptation:</b> When citizens trust that AI systems are fair and accountable, they are more willing to embrace technological transformations.</li><li><b>Predictability Encourages Investment:</b> Clear legal guidelines provide predictability, which in turn encourages sustained investment in AI research and development.</li><li><b>Accountability Sustains Legitimacy:</b> Mechanisms for accountability ensure that AI systems remain legitimate in the eyes of the public.</li></ul><p>Citizens' willingness to embrace technological transformation is directly proportional to their trust in AI systems being fair, transparent, and contestable.</p><h3>Conclusion: Shaping the Future of AI Governance</h3><p>The future of AI governance will not be solely determined within laboratories or corporate boardrooms. Its trajectory will be profoundly shaped in legislatures, courtrooms, and democratic forums. The choices made today are pivotal in determining whether AI becomes an instrument of power concentration or a catalyst for widespread empowerment.</p><h4>Actionable Recommendations:</h4><ol><li><b>Establish Comprehensive Governance Frameworks:</b> Prioritize the immediate development of robust and adaptable governance frameworks that are human-centric, inclusive, and capable of evolving with technological advancements.</li><li><b>Foster International Collaboration:</b> Actively participate in and promote global cooperation on AI standards, ethical frameworks, and joint research initiatives to build trust and prevent digital fragmentation.</li><li><b>Invest in Multi-Stakeholder Capacity Building:</b> Implement programs to enhance technological literacy among policymakers, equip regulators with analytical tools, provide technical understanding for judges, and raise digital awareness among citizens.</li><li><b>Strengthen Rule of Law in AI Context:</b> Uphold and adapt existing legal frameworks to ensure transparency, accountability, and fairness in AI systems, thereby fostering public trust and encouraging responsible innovation.</li><li><b>Champion Democratic Dialogue:</b> Create and support forums for inclusive discussions on AI's societal impact, ensuring that its development aligns with democratic values and human dignity.</li></ol><p>By aligning innovation with democratic values, the global community can not only effectively regulate technology but also collectively shape a future where technological progress and human dignity advance in harmonious concert.</p>"
                            }
                        ]
                    }
                }
            ]
        }
    ],
    "headerCols": {
        "col1": "Part Id",
        "col2": "FileName",
        "col3": "Content",
        "col4": "Status"
    },
    "fileName": "",
    "projectType": "VIDEO",
    "sourceLanguage": "English",
    "targetLanguage": "English",
    "__v": 0
},
    {
    "_id": "6992ee344c5e8f0974c874a9",
    "projectId": "150421",
    "projectName": "Whose Language, Whose Model? Public-Interest Multilingual LLMs | India AI Impact Summit 2026",
    "projectDescription": "Whose Language, Whose Model? Public-Interest Multilingual LLMs | India AI Impact Summit 2026",
     "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-16T10:15:20.257Z",
    "projectUpdatedDate": "2026-02-16T10:15:20.257Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-16T10:15:20.257Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO_LINK",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames"
        },
        {
            "col1": "P1",
            "col2": "https://youtu.be/uPYN9MmBDg0",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEOLINK",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:10",
                    "text": "pivot your neck and participate. Um,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:05",
                    "end": "00:00:10",
                    "text": "okay. Please take your seat."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:22",
                    "end": "00:00:27",
                    "text": ">> Okay, great. Um, so thank you so much"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:24",
                    "end": "00:00:30",
                    "text": "for joining our session, Whose Language,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:27",
                    "end": "00:00:33",
                    "text": "Whose Model. Um, we're really upset that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:30",
                    "end": "00:00:37",
                    "text": "we are can't be joined by our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:33",
                    "end": "00:00:39",
                    "text": "facilitator, our main organizer, Marlina"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:37",
                    "end": "00:00:41",
                    "text": "Wiznjak from the European Center for Not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:39",
                    "end": "00:00:44",
                    "text": "for Profofit Law. But no worries, we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:41",
                    "end": "00:00:46",
                    "text": "have an incredible session planned. Um,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:44",
                    "end": "00:00:48",
                    "text": "and so to kick us off, my name is Aliyia"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:46",
                    "end": "00:00:49",
                    "text": "Bhhata. I am the senior policy analyst"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:48",
                    "end": "00:00:52",
                    "text": "at the Center for Democracy and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:49",
                    "end": "00:00:55",
                    "text": "Technology. We're a Washington-based"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:52",
                    "end": "00:00:56",
                    "text": "notfor-profit nonpartisan group. And"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:55",
                    "end": "00:00:59",
                    "text": "today we have an incredible set of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:56",
                    "end": "00:01:02",
                    "text": "speakers who are really going to contend"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:59",
                    "end": "00:01:05",
                    "text": "with this question of what does it mean"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:02",
                    "end": "00:01:06",
                    "text": "um to build to to to foster more"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:05",
                    "end": "00:01:09",
                    "text": "meaningful multistakeholder"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:06",
                    "end": "00:01:12",
                    "text": "participation in the development of AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:09",
                    "end": "00:01:14",
                    "text": "really across the life cycle of AI when"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:12",
                    "end": "00:01:17",
                    "text": "we're thinking about um AI systems from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:14",
                    "end": "00:01:20",
                    "text": "conception and design to development to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:17",
                    "end": "00:01:23",
                    "text": "deployment and post- deployment as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:20",
                    "end": "00:01:26",
                    "text": "Um, and we're going to start us off with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:23",
                    "end": "00:01:28",
                    "text": "a few sort of provocations and thoughts"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:26",
                    "end": "00:01:30",
                    "text": "from our esteemed panelists. Um, and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:28",
                    "end": "00:01:32",
                    "text": "then as I mentioned, sort of kick it off"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:30",
                    "end": "00:01:35",
                    "text": "to the group to think deeper about"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:32",
                    "end": "00:01:37",
                    "text": "either ways we can foster that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:35",
                    "end": "00:01:40",
                    "text": "participation. Um, but also maybe"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:37",
                    "end": "00:01:43",
                    "text": "initiatives um that in the room led by"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:40",
                    "end": "00:01:45",
                    "text": "organizations you represent um to sort"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:43",
                    "end": "00:01:49",
                    "text": "of learn and enable that sort of cross"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:45",
                    "end": "00:01:52",
                    "text": "sector collaboration. Um and um we'll"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:49",
                    "end": "00:01:54",
                    "text": "also be inviting folks to report out"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:52",
                    "end": "00:01:56",
                    "text": "after each of you sort of break out into"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:54",
                    "end": "00:01:59",
                    "text": "this discussion. So please amongst"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:56",
                    "end": "00:02:00",
                    "text": "yourself choose a few facilitators will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:59",
                    "end": "00:02:03",
                    "text": "be joining the groups as well to sort of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:00",
                    "end": "00:02:05",
                    "text": "see these questions. Um but before we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:03",
                    "end": "00:02:07",
                    "text": "start you know why is this important? As"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:05",
                    "end": "00:02:10",
                    "text": "we know, large language models and other"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:07",
                    "end": "00:02:13",
                    "text": "AI systems are already being procured"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:10",
                    "end": "00:02:15",
                    "text": "and used and sort of deployed in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:13",
                    "end": "00:02:18",
                    "text": "multiple settings um in public and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:15",
                    "end": "00:02:20",
                    "text": "private spaces. Um but more often than"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:18",
                    "end": "00:02:23",
                    "text": "not these conversations are being led"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:20",
                    "end": "00:02:26",
                    "text": "and sort of um centralized in the global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:23",
                    "end": "00:02:29",
                    "text": "north in industry or G2G government to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:26",
                    "end": "00:02:31",
                    "text": "government spaces and that loses out on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:29",
                    "end": "00:02:35",
                    "text": "a lot of very valuable expertise both"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:31",
                    "end": "00:02:39",
                    "text": "from civil society um and sort of policy"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:35",
                    "end": "00:02:41",
                    "text": "experts um but also a lot of experts who"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:39",
                    "end": "00:02:43",
                    "text": "are the most experts and people who are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:41",
                    "end": "00:02:45",
                    "text": "the most affected by the deployment of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:43",
                    "end": "00:02:47",
                    "text": "AI technologies. And so today we want to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:45",
                    "end": "00:02:51",
                    "text": "talk and spotlight a couple of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:47",
                    "end": "00:02:53",
                    "text": "initiatives that are bringing these um"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:51",
                    "end": "00:02:55",
                    "text": "these groups into the conversation. What"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:53",
                    "end": "00:02:57",
                    "text": "are some of the considerations we need"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:55",
                    "end": "00:02:59",
                    "text": "to have in mind when we try to foster"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:57",
                    "end": "00:03:01",
                    "text": "this meaningful participation towards"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:59",
                    "end": "00:03:04",
                    "text": "what end are we moving? You know what"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:01",
                    "end": "00:03:06",
                    "text": "what is participation for and how do we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:04",
                    "end": "00:03:08",
                    "text": "make ensure that it's not a box checking"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:06",
                    "end": "00:03:10",
                    "text": "exercise? And then what are policy and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:08",
                    "end": "00:03:12",
                    "text": "regulatory approaches to actually"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:10",
                    "end": "00:03:15",
                    "text": "incentivize or put teeth to these"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:12",
                    "end": "00:03:18",
                    "text": "efforts um by either giving uh more"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:15",
                    "end": "00:03:21",
                    "text": "standards for companies to adhere to and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:18",
                    "end": "00:03:23",
                    "text": "model developers to adhere to but also"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:21",
                    "end": "00:03:25",
                    "text": "um enforce some mechanisms when they um"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:23",
                    "end": "00:03:29",
                    "text": "to en ensure that when they don't we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:25",
                    "end": "00:03:32",
                    "text": "have actual means to um make them do so."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:29",
                    "end": "00:03:34",
                    "text": "Um so first I'll introduce our panelists"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:32",
                    "end": "00:03:37",
                    "text": "and invite them also to give a more"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:34",
                    "end": "00:03:40",
                    "text": "in-depth um introduction of themselves."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:37",
                    "end": "00:03:43",
                    "text": "Um to my right we have Jalak Kakar who"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:40",
                    "end": "00:03:45",
                    "text": "is the executive director of the center"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:43",
                    "end": "00:03:48",
                    "text": "for communication governance at the n"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:45",
                    "end": "00:03:51",
                    "text": "national law university in Delhi. Um,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:48",
                    "end": "00:03:53",
                    "text": "Jalok has written, co-written and also"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:51",
                    "end": "00:03:56",
                    "text": "steered the development of a lot of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:53",
                    "end": "00:03:58",
                    "text": "resources of uh, evaluating policy"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:56",
                    "end": "00:03:60",
                    "text": "approaches to the governance of AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:58",
                    "end": "00:04:03",
                    "text": "systems, but also thinking critically"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:60",
                    "end": "00:04:05",
                    "text": "about, you know, sort of who are stress"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:03",
                    "end": "00:04:06",
                    "text": "testing these claims of who these AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:05",
                    "end": "00:04:09",
                    "text": "systems sort of purport to help and in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:06",
                    "end": "00:04:12",
                    "text": "what contexts. Um and then we next to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:09",
                    "end": "00:04:15",
                    "text": "him next to her we have Vanraj Ther um"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:12",
                    "end": "00:04:18",
                    "text": "who is the sort of inaugural professor"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:15",
                    "end": "00:04:20",
                    "text": "professor um and director of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:18",
                    "end": "00:04:22",
                    "text": "emerging technology initiative at the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:20",
                    "end": "00:04:24",
                    "text": "George Washington University and I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:22",
                    "end": "00:04:28",
                    "text": "believe the George Washington University"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:24",
                    "end": "00:04:30",
                    "text": "Law School as well. Um so Jalik I'm"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:28",
                    "end": "00:04:34",
                    "text": "going to start off with you um to talk"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:30",
                    "end": "00:04:36",
                    "text": "us through a few of the sort of policy"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:34",
                    "end": "00:04:38",
                    "text": "or regulatory approaches where we are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:36",
                    "end": "00:04:41",
                    "text": "seeing meaningful participation sort of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:38",
                    "end": "00:04:44",
                    "text": "come up um and what these governance"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:41",
                    "end": "00:04:47",
                    "text": "mechanisms either through laws and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:44",
                    "end": "00:04:49",
                    "text": "policy but also soft laws uh voluntary"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:47",
                    "end": "00:04:52",
                    "text": "commitments etc. you know what do they"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:49",
                    "end": "00:04:56",
                    "text": "incentivize how have they been effective"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:52",
                    "end": "00:04:56",
                    "text": "and where are we left wanting more?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:60",
                    "end": "00:05:06",
                    "text": "Yeah. Um,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:03",
                    "end": "00:05:09",
                    "text": "you know, in in any new space, there's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:06",
                    "end": "00:05:12",
                    "text": "always a trajectory of figuring out what"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:09",
                    "end": "00:05:15",
                    "text": "is relative consensus"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:12",
                    "end": "00:05:18",
                    "text": "and where can we start to build norms"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:15",
                    "end": "00:05:21",
                    "text": "that then build into sort of voluntary"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:18",
                    "end": "00:05:24",
                    "text": "commitments, other forms of soft law"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:21",
                    "end": "00:05:26",
                    "text": "which start to then get codified in the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:24",
                    "end": "00:05:28",
                    "text": "shape of standards, in the shape of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:26",
                    "end": "00:05:30",
                    "text": "regulation,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:28",
                    "end": "00:05:33",
                    "text": "in the shape of other sort of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:30",
                    "end": "00:05:34",
                    "text": "international binding commitments and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:33",
                    "end": "00:05:38",
                    "text": "there's always like a process and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:34",
                    "end": "00:05:41",
                    "text": "trajectory to do that. I think um"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:38",
                    "end": "00:05:43",
                    "text": "there's value in having multistakeholder"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:41",
                    "end": "00:05:45",
                    "text": "conversations to start to build that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:43",
                    "end": "00:05:48",
                    "text": "consensus."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:45",
                    "end": "00:05:50",
                    "text": "But if we look at uh past experience"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:48",
                    "end": "00:05:53",
                    "text": "with say social media regulation, I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:50",
                    "end": "00:05:55",
                    "text": "think we stayed in the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:53",
                    "end": "00:05:59",
                    "text": "let innovation happen, we're building"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:55",
                    "end": "00:06:01",
                    "text": "consensus phase for too long. Um and I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:59",
                    "end": "00:06:03",
                    "text": "think there needs to be a cut off point"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:01",
                    "end": "00:06:07",
                    "text": "where we start moving beyond voluntary"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:03",
                    "end": "00:06:10",
                    "text": "commitments to more hard forms of law"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:07",
                    "end": "00:06:13",
                    "text": "whether that's standards uh whether"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:10",
                    "end": "00:06:15",
                    "text": "that's you know um sort of um laws and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:13",
                    "end": "00:06:18",
                    "text": "regulations"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:15",
                    "end": "00:06:22",
                    "text": "or even which which may not even mandate"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:18",
                    "end": "00:06:25",
                    "text": "particular sort of outcomes but may"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:22",
                    "end": "00:06:27",
                    "text": "mandate processes at this stage. Right?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:25",
                    "end": "00:06:30",
                    "text": "What are processes you need to integrate"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:27",
                    "end": "00:06:32",
                    "text": "into the design and deployment of these"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:30",
                    "end": "00:06:35",
                    "text": "systems so that we are able to at least"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:32",
                    "end": "00:06:38",
                    "text": "surface what are the kinds of harms and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:35",
                    "end": "00:06:41",
                    "text": "challenges uh that we are seeing because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:38",
                    "end": "00:06:43",
                    "text": "if we don't make that shift over the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:41",
                    "end": "00:06:46",
                    "text": "period of the next couple of years from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:43",
                    "end": "00:06:50",
                    "text": "voluntary commitments into harder uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:46",
                    "end": "00:06:53",
                    "text": "more uh structured forms of regulation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:50",
                    "end": "00:06:55",
                    "text": "we will sort of may see a repetition of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:53",
                    "end": "00:06:58",
                    "text": "what we have seen with the history of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:55",
                    "end": "00:07:01",
                    "text": "social media where um you know platforms"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:58",
                    "end": "00:07:04",
                    "text": "ran free. There was no mechanism to tr"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:01",
                    "end": "00:07:06",
                    "text": "hold them truly accountable"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:04",
                    "end": "00:07:09",
                    "text": "uh to these commitments that they were"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:06",
                    "end": "00:07:11",
                    "text": "repeatedly making in various forums and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:09",
                    "end": "00:07:14",
                    "text": "and we've seen the sort of mess that we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:11",
                    "end": "00:07:18",
                    "text": "have sort of landed up in because of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:14",
                    "end": "00:07:20",
                    "text": "that. So um yeah I think I'll pause here"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:18",
                    "end": "00:07:22",
                    "text": "and then you know we can come back for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:20",
                    "end": "00:07:25",
                    "text": "more. Yeah, I think that's a really good"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:22",
                    "end": "00:07:28",
                    "text": "place to pause because I think um it's a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:25",
                    "end": "00:07:30",
                    "text": "good sort of where we've come from and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:28",
                    "end": "00:07:34",
                    "text": "now when we're trying to build towards a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:30",
                    "end": "00:07:36",
                    "text": "vision of more people to have a role in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:34",
                    "end": "00:07:39",
                    "text": "the development or in the development of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:36",
                    "end": "00:07:40",
                    "text": "these new technologies but also truly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:39",
                    "end": "00:07:42",
                    "text": "saying when they when it doesn't work"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:40",
                    "end": "00:07:44",
                    "text": "for them you know what does it look like"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:42",
                    "end": "00:07:46",
                    "text": "to have the seat at the table I'd like"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:44",
                    "end": "00:07:49",
                    "text": "to bring Dhanj into that part you know"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:46",
                    "end": "00:07:52",
                    "text": "what are some of the effective"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:49",
                    "end": "00:07:53",
                    "text": "characteristics ics of meaningful"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:52",
                    "end": "00:07:55",
                    "text": "external and multistakeholder"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:53",
                    "end": "00:07:56",
                    "text": "participation. You know, what are we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:55",
                    "end": "00:07:58",
                    "text": "talking about when we're saying we need"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:56",
                    "end": "00:07:59",
                    "text": "to have a voice in the conversation and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:58",
                    "end": "00:08:00",
                    "text": "and who who is this we?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:59",
                    "end": "00:08:03",
                    "text": "[clears throat and cough]"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:00",
                    "end": "00:08:06",
                    "text": ">> Great. Thank you. Thank you, Alia. So,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:03",
                    "end": "00:08:07",
                    "text": "uh also thanks to the uh organizers to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:06",
                    "end": "00:08:10",
                    "text": "provide a space for this kind of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:07",
                    "end": "00:08:12",
                    "text": "discussion. Um"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:10",
                    "end": "00:08:15",
                    "text": "so when we talk about uh meaningful"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:12",
                    "end": "00:08:18",
                    "text": "participation and how we can uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:15",
                    "end": "00:08:21",
                    "text": "effectively engage communities in the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:18",
                    "end": "00:08:23",
                    "text": "design, deployment uh and evaluation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:21",
                    "end": "00:08:25",
                    "text": "around AI models, um I think it's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:23",
                    "end": "00:08:30",
                    "text": "important to recognize that this is a an"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:25",
                    "end": "00:08:31",
                    "text": "issue a problem that is not new. Um in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:30",
                    "end": "00:08:33",
                    "text": "the same way Jelik was referring to like"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:31",
                    "end": "00:08:35",
                    "text": "lessons from the social media regulation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:33",
                    "end": "00:08:36",
                    "text": "world, there are lessons that we can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:35",
                    "end": "00:08:39",
                    "text": "draw from when you look at other"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:36",
                    "end": "00:08:41",
                    "text": "technological interventions and how they"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:39",
                    "end": "00:08:45",
                    "text": "their practitioners and policymakers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:41",
                    "end": "00:08:47",
                    "text": "have sought to uh include uh uh wider"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:45",
                    "end": "00:08:50",
                    "text": "communities, targeted communities and so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:47",
                    "end": "00:08:51",
                    "text": "on in these questions. What ultimately"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:50",
                    "end": "00:08:54",
                    "text": "emerges though when you look at these"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:51",
                    "end": "00:08:55",
                    "text": "past lessons is a question about power."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:54",
                    "end": "00:08:57",
                    "text": "It's a question about the distribution"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:55",
                    "end": "00:08:59",
                    "text": "of power. When we talk about part"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:57",
                    "end": "00:09:00",
                    "text": "participation and collaboration and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:59",
                    "end": "00:09:04",
                    "text": "building partnerships with communities"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:00",
                    "end": "00:09:06",
                    "text": "with different uh constituencies, we're"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:04",
                    "end": "00:09:07",
                    "text": "in effect talking about how powers are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:06",
                    "end": "00:09:11",
                    "text": "distributed and what kinds of decisions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:07",
                    "end": "00:09:13",
                    "text": "are made made by who. Uh for a long time"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:11",
                    "end": "00:09:15",
                    "text": "when you look at the the field of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:13",
                    "end": "00:09:17",
                    "text": "international development and uh how"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:15",
                    "end": "00:09:19",
                    "text": "countries in the global south or the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:17",
                    "end": "00:09:22",
                    "text": "majority world would put in place"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:19",
                    "end": "00:09:24",
                    "text": "different kinds of development projects."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:22",
                    "end": "00:09:27",
                    "text": "What emerged from that those kinds of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:24",
                    "end": "00:09:29",
                    "text": "lessons uh was characterized by a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:27",
                    "end": "00:09:31",
                    "text": "scholar named Chambers back in the based"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:29",
                    "end": "00:09:33",
                    "text": "on his work in 1890s is in the supply"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:31",
                    "end": "00:09:35",
                    "text": "chain in the thought proc process"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:33",
                    "end": "00:09:38",
                    "text": "putting them first starting with their"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:35",
                    "end": "00:09:40",
                    "text": "needs and their concerns. That can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:38",
                    "end": "00:09:43",
                    "text": "ultimately mean for example communities,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:40",
                    "end": "00:09:45",
                    "text": "target groups defining what technology"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:43",
                    "end": "00:09:47",
                    "text": "use cases are best for them which is the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:45",
                    "end": "00:09:49",
                    "text": "reverse of what we see now where"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:47",
                    "end": "00:09:52",
                    "text": "frontier model companies uh deploy"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:49",
                    "end": "00:09:54",
                    "text": "models with the idea of um let people"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:52",
                    "end": "00:09:57",
                    "text": "figure out what the use case is best uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:54",
                    "end": "00:10:00",
                    "text": "for them um after the fact or"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:57",
                    "end": "00:10:02",
                    "text": "evaluations after the fact. Uh another"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:00",
                    "end": "00:10:04",
                    "text": "thing that uh emerged from a lot of that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:02",
                    "end": "00:10:05",
                    "text": "work around effective community"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:04",
                    "end": "00:10:07",
                    "text": "participation is recognizing that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:05",
                    "end": "00:10:10",
                    "text": "communities actually have a lot of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:07",
                    "end": "00:10:13",
                    "text": "expertise uh to to analyze their local"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:10",
                    "end": "00:10:15",
                    "text": "context and understand what's really"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:13",
                    "end": "00:10:17",
                    "text": "needed. Uh and so when we think about"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:15",
                    "end": "00:10:19",
                    "text": "the development of language models for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:17",
                    "end": "00:10:21",
                    "text": "example, how they're developed, what"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:19",
                    "end": "00:10:25",
                    "text": "kind of specific contexts, what kinds of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:21",
                    "end": "00:10:27",
                    "text": "uh linguistic uh uh concerns should be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:25",
                    "end": "00:10:28",
                    "text": "included in development of these models."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:27",
                    "end": "00:10:31",
                    "text": "um that is often where that kind of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:28",
                    "end": "00:10:34",
                    "text": "analysis can can lie. So uh that's kind"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:31",
                    "end": "00:10:36",
                    "text": "of a high level but I just wanted to uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:34",
                    "end": "00:10:39",
                    "text": "bring bring in this issue that there's a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:36",
                    "end": "00:10:41",
                    "text": "lot of uh debate and research around"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:39",
                    "end": "00:10:42",
                    "text": "this kind of uh what are the mechanisms"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:41",
                    "end": "00:10:45",
                    "text": "for effective participation that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:42",
                    "end": "00:10:47",
                    "text": "precedes AI but we can learn from that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:45",
                    "end": "00:10:49",
                    "text": "uh without necessarily needing to start"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:47",
                    "end": "00:10:52",
                    "text": "from scratch. No, I think that's really"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:49",
                    "end": "00:10:53",
                    "text": "helpful level setting to begin with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:52",
                    "end": "00:10:56",
                    "text": "because I think sometimes in the AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:53",
                    "end": "00:10:57",
                    "text": "policy conversation there is a lot of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:56",
                    "end": "00:10:59",
                    "text": "ahistoricizing"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:57",
                    "end": "00:11:01",
                    "text": "going on of like acting like the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:59",
                    "end": "00:11:03",
                    "text": "development of these technologies"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:01",
                    "end": "00:11:05",
                    "text": "necessitate is you know so complex and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:03",
                    "end": "00:11:07",
                    "text": "like uses this new jargon and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:05",
                    "end": "00:11:09",
                    "text": "necessitates only technical experts in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:07",
                    "end": "00:11:11",
                    "text": "the room. Although we've been here"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:09",
                    "end": "00:11:13",
                    "text": "before, we've built other sort of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:11",
                    "end": "00:11:15",
                    "text": "incredibly complex privacy enhancing"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:13",
                    "end": "00:11:18",
                    "text": "technologies, for example, with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:15",
                    "end": "00:11:18",
                    "text": "journalists and human"
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "69930ada4c5e8f0974c877af",
    "projectId": "578919",
    "projectName": "Genomics and AI for Global Health Empowering the Global South Through Secure Data ",
    "projectDescription": "Genomics and AI for Global Health Empowering the Global South Through Secure Data",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-16T12:17:28.890Z",
    "projectUpdatedDate": "2026-02-16T12:17:28.890Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-16T12:17:28.890Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "Genomics and AI for Global Health _ Empowering the Global South Through Secure Data Access(MP3_128K).mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771244057894.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:30",
                    "text": " Global South is everyone except those who live in Global North. Now what is Global North? People in the Western countries, right? Western Europe, United States of America, and some other countries. Only 15% of the people live in Global North. 85% of the people live in the Global South. Global South is usually less developed economically."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:30",
                    "end": "00:00:59",
                    "text": " okay, have less technology in their daily life, but those are the people who actually need the solution. We in India are one of them. Now, why is it important that the solution has to reach the Global South? How much time do we have? We have, thank you. So, if you see here, people in the Global South are under, are underrepresented in the large databases, right? All the large companies of the world, the Google and the Amazon, and the Facebook of the world,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:59",
                    "end": "00:01:26",
                    "text": " actually likes to get the data from the large number of people, but the models that they have built is not necessarily, I'm talking about the disease models, not necessarily using the data from the Global South. And the best example is the Human Genome Project 1, and I'll talk to you a little bit about it too, is the data which was actually using the Caucasian population. Now, in addition to that, we have a higher disease burden. What does it mean? All of you in this room,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:26",
                    "end": "00:01:50",
                    "text": " Indians either know someone or have someone in the family who is diabetic, right? Every single one of them. Wasting our money to the doctors, no, absolutely no benefit. It's only a lifestyle disease. And this is going to go on the same with coronary heart disease. So we have a higher disease burden. Now you can see, I'm not going to read all these things to you. These are the absolute..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:50",
                    "end": "00:02:20",
                    "text": " advantages of what we have today in Global South. But the databases that we use is not from Global South. And therefore, these are the two examples I've given. Everybody knows the example on the right. For the ones that are not from India, this is one of the very famous people's car. Unfortunately, it's not in production anymore. It's for Tata Nano. It was introduced with $1,500, like one lakh rupees, but became slowly two and a half lakhs. This is the most expensive car in the world."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:20",
                    "end": "00:02:47",
                    "text": " Rolls Royce, one of the model of the Rolls Royce, which is 30 million dollars. Now you can see here 200 crores. Now this is the solution of the global north that gives. Not necessarily a bad solution, but can we afford it? No. Do we need it? No. Do we want it? Yes. But this is what we want. We want a solution that is affordable, development for the people of the global south, easily accessible by the people in the global south."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:47",
                    "end": "00:03:17",
                    "text": " and is actually available openly to everyone. So it's very important that we do it. Now, that is the reason Global South must lead, not follow. Now, one of the example is Human Genome Project 2 and Webin is going to talk a lot about that during the panel discussion. So this is the second version of the Human Genome Project. Some of you are interested can actually just go to the website and find out and why it is going to be important for the next revolution in human genomics. And then finally, one slide about India. What is India?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:17",
                    "end": "00:03:44",
                    "text": " or what can we get you to lead this particular process and work with others in the Global South? Well, these are the different kinds of data in the healthcare. Okay? This is the information where it comes from. This is the AI tools, the APIs, the frameworks. And to use these AI tools and APIs and framework, we can actually come up with multiple different solutions in healthcare. Why we can do that? This side is..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:44",
                    "end": "00:04:13",
                    "text": " actually no trainer for you, those of you who are from India, that we have a large population, we have a large, you know, repertoire of digital public infrastructure, not just the Aadhaar, but all of those, and we have staff, so we can use all this information for healthcare data to build solution for Global South, not just for India, but also for Global South, but that will require a lot of collaboration and also a lot of talking with other people in the Global South."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:13",
                    "end": "00:05:13",
                    "text": " So therefore, I think the time of India is now. And if any of you is interested, it will be available after the session to talk to you or to discuss with you. Thank you. I think this mic is not working. OK, our next speaker is going to be Madhava Jay, who is a principal engineer at Open Mind Foundation. And he's going to tell us about BioVod, a privacy-first data visitation platform for equitable global collaboration in biomedicine. Thank you, everyone. Good afternoon. My name is Madhava. I'm a rare disease patient."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:13",
                    "end": "00:05:42",
                    "text": " I'm also a physical engineer of mines, where I've been working for the last five or six years on privacy technology. And this focus on biovolts, but I'm actually from Brisbane, Australia. But a very long time ago before I was born, my parents lived in India, very close to here. And they loved your beautiful culture and country so much that they gave me this name. And so I wear it with a badge of honor. So thank you for providing me."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:42",
                    "end": "00:06:11",
                    "text": " If there's one thing that I want you to take away from this talk, it's that you can collaborate globally to discover insights without ever losing control of your private data. So Binay was talking a little bit on the value of data in AI. This is an AI conference. So the question is, if AI has been trained on all the world's public data, what's the next destination? It's going to be the private data that they don't have."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:11",
                    "end": "00:06:39",
                    "text": " A lot of the conversations at this conference are going to be about what businesses could be integrated into a company, and what are some of the values involved. There's a little bit of concern that these state-of-the-art models could replace all of these different workflows. But I would urge you to read this blog post, which is a fantastic resource, which talks about one potential defense against this. And that defense is to have a cornered resource. And if you read it..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:39",
                    "end": "00:07:24",
                    "text": " It's not working. It's not working. It's not working. It's not working. It's not working. It's not working."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:28",
                    "end": "00:08:09",
                    "text": " Sorry guys. Everyone is starving here. Look at this. Everyone is starving. No one is telling us to talk to them. Pankaj bhai is here, right? What are you saying? He is not giving us any time. No, I called him. He said it's difficult to eat. Look, adjust. We will adjust."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:09",
                    "end": "00:08:43",
                    "text": " How do I adjust this?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:43",
                    "end": "00:09:09",
                    "text": " I think we're good. We're good. Thank you. Thank you very much, everyone. All right."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:09",
                    "end": "00:09:34",
                    "text": " And such a resource must be truly exclusive, not available for sale at any price in the type of operation. And so I think in the context of biomedical data, I would say that meets many of those requirements. It's very private, it's very valuable. Health data is probably among the most sensitive and valuable categories of data. So Vinay was asking if anyone here has people in their families who have diabetes."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:34",
                    "end": "00:10:02",
                    "text": " I think a lot of us here probably know some of these diseases unwell. The global rates of 20 people that are undiagnosed are quite high, up to 10% of the population. And I'm one of those people. So many years ago, I had chronic illness that I didn't know I had. And it turns out that even a few small letters in your genome can mean a difference between life and death. So my great-grandfather, my grandfather who had this disease, died at age 40. I'm 41."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:02",
                    "end": "00:10:30",
                    "text": " It's highly likely I would have died last year, probably, if I hadn't had early intervention. And so I should know about this. But these advances and benefits, they're not really being shared equally. To Vinay's point earlier, I live in a Western country, and I can get that diagnosis quite easily now. But we need to make that more globally equitable. So one of the challenges with data sharing in health care is that there are a lot of very important data protection laws to be around sovereignty. And these restrict cross-border"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:30",
                    "end": "00:10:53",
                    "text": " transfer of human data. But as you can imagine, you know, AI operates the best when you have more data, so sharing with biomedical data sets is necessary really to get to that next level of advancement. So the question is, how can we collaborate, and when I say we, it's a collective we, the north, the south, the whole world, while preserving our data sovereignty in each of our countries?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:53",
                    "end": "00:11:21",
                    "text": " Hello Hello Hello Hello Hello Hello Hello Hello Hello Hello Hello Hello Hello Hello Hello Hello Hello Hello Hello Hello Hello Hello Hello"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:21",
                    "end": "00:11:51",
                    "text": " At the moment, a lot of what's happening is by conducting cutting-edge research, we can get max utility, but we have to give up control of all that data. And so the alternative to say, well, I don't really want to lose my data because of all that value, I don't want to give it up. Importantly, data once copied cannot be uncopied. And so you should consider, you know, what does it mean to share and collaborate with data? But if I was to then keep it to myself, I get zero benefit. And so that's not very good either. And so really what we want to do is we want to try to occupy this space in the middle."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:51",
                    "end": "00:12:19",
                    "text": " and actually potentially see if we can get to that top right of the corner and just shoot right off the chart. So let's talk a little bit about some of these traditional data sharing models that people are currently using. So in this diagram we basically have three different institutions and you can see illustrated by their different colors and often what will happen is they will take their data and they will upload it and send it raw copies of the data to a data scientist who then brings it all together and does an analysis."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:19",
                    "end": "00:13:08",
                    "text": " One of the problems with this is that once you copy data, you can't uncopy it. And that means that while you might have a legal agreement with that data scientist, this can still lead to data leakage. But the most important thing actually here is not the trust. It's the fact that the original data owners may not benefit from that downstream result. You may not realize this, but often the output, the scientific output, the model training, the results are not always shared back to the people who own the data. And so they're being severed. We allow them to access that and kind of do their queries and analysis."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:08",
                    "end": "00:13:37",
                    "text": " still have to upload all that data, and now you've got to pay for it in a second location. If this is medical data, it's enormous. Some of these data sets are terabytes or petabytes, and you've got to pay for that as well. The other thing is that once you put it in the cloud, you have to protect it. So now you've got another thing where all your valuable data needs to be protected. And more importantly, like Vinay said, this is really only for well-resourced institutions who can afford this. But finally, it does not solve the problem we talked about, which is that data owners don't get visibility into how the analysis is happening."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:37",
                    "end": "00:14:07",
                    "text": " They don't get the granular control they need to be able to decide on a per-query basis or the outputs that are being produced, whether they want to collaborate or participate in that. And it also doesn't really scale to multipartiness, which we'll talk about in a minute. So there is an alternative way, a potential third way here, which has been coined as data visitation. And the idea in the name there is that we're visiting the data. The data is not visiting us. But it allows collaboration without movement of data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:07",
                    "end": "00:14:27",
                    "text": " And so on the left, we can see a data scientist or anybody working with data. And importantly, what they're doing is they're sending in code or a question or a prompt. And they have a question that they want answered. And on the right-hand side, we have a data owner. We call these data sites. And this is sort of like a website for data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:27",
                    "end": "00:14:56",
                    "text": " And they retain governance over that data at all times and never share a copy. All they return is the results of the question the data scientists wanted to ask. And if you think about it, that's all the green. You shouldn't have to have all the data, just have the answer, if that's possible. And it is possible. And so what we'll be working on in this project, and as I said earlier, I work at Open Mind. And so you're probably thinking, well, that's one name. So what is BioVault? BioVault is a implementation of these concepts for the biomedical space."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:56",
                    "end": "00:15:24",
                    "text": " And you can see on the right here we have a diagram of how you can sort of envisage what this is. It's a network, right, of open source privacy-first data visitation for equitable global collaboration. The way it works is, we've so built like an open source peer-to-peer network. The technology and the science and the math behind this is here and for a while. But what we're really focused on is bringing it, the engineering problem of bringing it to the people who need it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:24",
                    "end": "00:15:50",
                    "text": " So we spent a lot of time on developing things like a desktop utility that's really easy to use, that runs on Windows, for example, which a lot of people don't bother with. It has end-to-end encryption, and it supports arbitrary analysis through a variety of different extensions for technologies like Nextflow and Jupyter. But more importantly, it keeps all your data local. We don't ask you to upload data. There's no uploading data facility. You link the data, and you make the data available via sort of an interface, a facade."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:50",
                    "end": "00:16:13",
                    "text": " that allows people to then send in their computations that you can choose to execute against them. And more importantly, unlike a lot of the existing models, we also support federated encrypted computation. And that allows this key-to-key network to work in a point-to-point topology to allow more than two parties to basically collaborate together."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:13",
                    "end": "00:16:41",
                    "text": " We actually, I'll talk about this a little bit more at the end, but we have a preprint that we just launched on BioArchive a few days ago. There's a lot of work from some people in the panel here and some of the people in the audience. So thank you all. And you can see on the right-hand side, we check out the papers and figures, but some very specific use cases that we wanted to explore that we think are relevant and pertinent to real-world use. And so one of them is the analysis of single-cell RNA set. Now, this is, I'm not a biologist, so it's beyond me, but..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:41",
                    "end": "00:17:09",
                    "text": " very advanced technology related to analyzing cancer cells, looking at advanced genomics. We can also train machine learning models, which is pretty par for the course by now, but it's important we demonstrate that that's possible. And we've done remote inference on very large clinical data sets. And more importantly, there's also the option to do privacy-preserving analysis. And so actually we did some on my own genome. You can see in the bottom right-hand corner there, there's a little graph showing"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:09",
                    "end": "00:17:37",
                    "text": " or like a little count of some variants that I have that basically is to do with my disease. So that's cool. Finally, I want to talk a little bit about some real-world data and real-world use cases. So it's not enough to basically talk about a theoretical bit of technology and try to kind of tell people that they should use it. We need to actually demonstrate that it works in the real world. And we want to build a product and a sort of experience that is useful by working with real people."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:37",
                    "end": "00:17:58",
                    "text": " Here are two pilot projects that we just completed before this pre-print. We don't have time to go into both of them today, but you will actually hear from Dr. Rana Dajani later, in about a second. But I'd like to focus on the first one here, which is with Dr. Karika Weldon. She's the founder and CEO of Perigenetics, which is in Bermuda. And basically what happens is..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:58",
                    "end": "00:18:27",
                    "text": " She has a Oxford Nano4 machine she uses to do some sequencing. She also collects samples from a variety of participants on different islands in the Caribbean. And that data that gets gathered is stored by her. And in that earlier model that we discussed, there was the idea that traditionally to do outside research you have to share that data. And so if Karika wanted to do some of the more cutting-edge research, she would need to upload that data for research in America or something."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:27",
                    "end": "00:18:56",
                    "text": " There's a lot of, you know, computational requirements. There's expertise. There's a lot of complexity to this. It's not as simple as just having the data. But when we met Dr. Creaker, we understood that she had problems with, you know, working with these datasets. And so what we did was we built specific workflows that would work against her data so that we could do some very novel and exciting research. And so in this diagram, you can see what we actually did was we had the desktop application running in different islands."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:56",
                    "end": "00:19:26",
                    "text": " and we sent through queries, and we basically were able to analyze the allele frequencies for different populations within the Caribbean, as well as some disease classification. So there's a particularly pernicious disease called APL1, which creates a chronic kidney disease, and they have a very high prevalence in the Caribbean. And using our technology, we were able to sort of provide remote analysis and code so that Dr. Karika could"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:26",
                    "end": "00:19:53",
                    "text": " verify results that she already had and confirm new results without having to share that data with us. So we were actually able to take those allele frequencies and do something fairly cutting edge. So I believe this is a bit of a world first in a few ways. And what we did was some of this data has never been analyzed before because of the sensitivity of it. But we compared it to a thing called Nonad, which is a global"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:53",
                    "end": "00:20:15",
                    "text": " sort of ancestry and variant database that shows breakdowns of different ethnicity groups. And we're able to compare how it compares to those. And you can see those plot charts at the bottom. And yeah, we all did that all without uploading any data. So we think it's a bit of a well first. Anyway, so look, bioarchive has our preprint. There's a QR code there if you want to scan that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:15",
                    "end": "00:20:44",
                    "text": " I'll talk to us later if you want to hear more. We'd love to partner with some new people but also if you're interested in just getting involved in like open source or contributing or have questions about policy or anything, we have people here to talk to you about it. So thank you again for listening and enjoy the conference. And actually Dr. Rana was initially going to be here today but unfortunately she had a family emergency so she actually recorded a video for us that I couldn't play."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:47",
                    "end": "00:21:15",
                    "text": " Peace upon you all. My name is Rana Dajani. I'm a professor of molecular biology from the Hashemite University in Jordan. Unfortunately, I wasn't able to join you today in India at the AI Summit because my mother passed away recently. And this was a reminder of how our research and science cannot be isolated from our humanity and serving humanity. Ultimately, as scientists, our goal is to serve humanity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:15",
                    "end": "00:21:45",
                    "text": " I'm a molecular biologist and I work on genetics of ethnic populations in Jordan, the Circassians and the Chechens trying to uncover novel gene risk factors for complex diseases such as diabetes. I'm also an expert on the epigenetics of trauma in vulnerable communities such as refugees, Syrian refugees, Palestinian refugees as a result of the atrocities committed by oppressive regimes and powers around the world."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:46",
                    "end": "00:22:11",
                    "text": " Therefore, I generate many databases that enrich biomedical research globally. Now, biomedical databases and datasets usually represent or should represent diverse populations. And that diversity is essential for advancing precision medicine and for ensuring that AI systems in health are accurate, fair, and globally relevant. This is especially important."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:11",
                    "end": "00:22:39",
                    "text": " for the Global South, where populations remain underrepresented in biomedical research and AI development. Without equitable data sharing, health innovations risk reinforcing global disparities rather than reducing them. And that's the reason we're all here today, to try to reduce that. Yet much of this data remains siloed due to registry, regulatory, sovereignty and privacy constraints, existing data sharing models,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:39",
                    "end": "00:23:05",
                    "text": " all require complex infrastructure, specialized expertise, and significant financial investments, which as a country from Jordan, we do not have. These barriers, therefore, disproportionately impact institutions and resource-constrained settings, limiting our ability to participate fully in global collaborations. That is why I was so excited to join the BioVault team."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:06",
                    "end": "00:23:26",
                    "text": " BioVault provides a practical domain-ognostic framework for privacy-preserving biomedical collaboration, supporting AI development and genomic research, while advancing equity, sovereignty, and global participation in health innovation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:26",
                    "end": "00:23:56",
                    "text": " creators of Bible Vault and tested my data sets for the sarcasm in the Chessians and were able to reproduce authentically the results. So this is very promising and I hope you will learn more as this panel unfolds. I look forward to such approaches becoming the norm in the future. Enjoy the conference and the rest of your day and hopefully we will collaborate in the future to make this to do better science to serve humanity. Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:56",
                    "end": "00:24:31",
                    "text": " and goodbye. Next we actually have Andrew Sultan who is from University of Oxford and he's going to tell us about fair, scalable and private AI using routine collective hospital data. Thank you Dawn and thank you all for coming to our session today. One of my aims today is to talk through a scenario that we faced during the Covid-19 pandemic at Oxford University Hospitals which is one of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:31",
                    "end": "00:24:56",
                    "text": " constituent hospitals within the NHS and talk through a machine learning approach we took to try and solve that problem using data that was already routinely available to us. I'll then talk a little bit about an extension into how privacy preserving techniques, in our case federated learning, enabled us to achieve our objectives and talk about one of the strategies we use for deploying that technology, for making it available across the NHS and therefore..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:56",
                    "end": "00:25:23",
                    "text": " the relevance that that could have to other settings, including, for example, countries like India and elsewhere, and I'd be delighted to have conversations afterwards and take that forward. So to set the scene of one of the core challenges in clinical artificial intelligence is that data is fundamentally siloed. It exists within individual hospitals, and those hospitals are separate legal entities, and because of regulation within the European Union and the United Kingdom, that data can't be shared freely without..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:23",
                    "end": "00:25:52",
                    "text": " for example, the appropriate approvals, the appropriate ethical considerations, and a full understanding of the risk that is involved in that. And so the traditional process is to seek those approvals, to relocate that data within either a university or a company or the organization doing the development, to then clean that data, and you're already one step removed from the people who understand the nuance and the artifact in that data, to develop models, to validate them, and then to move towards deployment. And I'll start briefly with a case study."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:52",
                    "end": "00:26:08",
                    "text": " which is the work we did at the beginning of the COVID-19 pandemic to build a model that can be used to rule out COVID-19 in emergency departments. Now, the big challenge of course was access to testing and even within a relatively high resource setting like the United Kingdom."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:08",
                    "end": "00:26:37",
                    "text": " at the start of the pandemic it would take three days to get the result of the COVID PCR test now that came down to two days and then one day and then 12 hours as on-site testing but there were still significant challenges in being able to run a hospital operation i.e. how patients flow through the hospital with a 12-hour time lag to know if they have COVID-19 or not and unfortunately that was implicated in the spread of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:37",
                    "end": "00:27:02",
                    "text": " a set of routine data that is pretty much standard across middle-income countries and higher-income countries made up of your vital signs, your blood pressure, your heart rate, your temperatures, and things that are very easy to measure done very quickly, done very routinely. And also panels of blood tests that are routine and are standard of care across the developing as well as the developed world. So this is the real basics of blood testing."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:02",
                    "end": "00:27:27",
                    "text": " and so performed for almost every patient being admitted within one hour. Now patients afterwards are then transferred to the ward and that the selection of warders was guided by the results of their PCR tests and this is the challenge that we set out to face. And the clinical challenge is that a test that rules out COVID-19 within minutes is what is needed, that's what's demanded in the hospital to make sure that you have patients who have COVID away from patients who don't."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:27",
                    "end": "00:27:52",
                    "text": " And so we set out to build the Curial Screening Test for COVID-19, which is essentially working with that tabular data, data that can be represented in the table, and we went on to validate the model that we built across four NHS sites. So this would be Birmingham, Bedford, Portsmouth, and then for the second COVID-19 wave in Oxford across 72,000 patients. And we showed that that model could perform quite well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:52",
                    "end": "00:28:20",
                    "text": " We were fortunate enough to then go on and pilot our test. We deployed it at the Emergency Department of the John Necklick Hospital, and for a study of 520 patients showed that you can have a result within 45 minutes of the patient arriving in hospital. And we showed that we got those results faster than natural flow tests and faster than PCR tests by some hours, actually, considerably. And so I guess the question here is, what does this have to do with data security?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:20",
                    "end": "00:28:39",
                    "text": " it gets to the core of one of the challenges of this work. Now, a big challenge of working with data from multiple NHS sites is that although every site was aligned in its incentives, every site wanted to see a solution for COVID testing, we had significant challenges getting access to data, even within the NHS."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:39",
                    "end": "00:28:56",
                    "text": " even where people's incentives aligned and even in the pandemic situation. And so we asked the question, how can we deal with that data distribution problem? How can we deal with that core challenge that slows down medical research for understandable reasons and approach that with a nice technological solution? And of course,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:56",
                    "end": "00:29:13",
                    "text": " There have been several leaks in the past, several instances where data has not been handled properly, and so there is understanding that we can send around it. The technology of federated learning proposes to take out that data transfer step. So instead of you moving data from hospitals to your secure environment,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:13",
                    "end": "00:29:39",
                    "text": " and you perform training by bringing the model to the data and not the other way around and so in doing so you're sending the parameters or the weights within a model which are essentially mathematical numbers that you can't readily infer patient data from. You can do that in the way that's secure and encrypted and there are various technologies around it to improve that and so we wanted a platform that in the early days of federated learning where no one had heard of it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:39",
                    "end": "00:29:52",
                    "text": " We needed a platform that anybody could use. We needed it to be user-friendly. We needed it to be so that people who didn't have a background in federated learning could just get this platform up and running. And we used this device called the Raspberry Pi, which is a cheap"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:52",
                    "end": "00:30:17",
                    "text": " Small computer, it's the size of your credit card, just a little bit thicker, and it's sold for about £40 in Britain, so it's about 5,000 rupees, sorry, yes, 5,000 rupees or thereabouts. So it's relatively inexpensive in the world of healthcare technology. It's a relatively cheap device. And what's nice about these devices is they can run Linux. You can run a full operating system on it with a full user interface."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:17",
                    "end": "00:30:44",
                    "text": " And so we were able to put all of the software you need to take part in this study to develop a COVID-19 screening test onto these devices. And we were able to install Ubuntu onto them, give them a nice user interface that talked you through everything you had to do. So a hospital needed only plug it into power, plug it into a monitor, plug in a keyboard and a mouse, and off you go. You follow the instructions on the screen and take it from there. And of course, we can do all the governance approvals around it. What's particularly nice about this device is it uses a removable micro SD card."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:44",
                    "end": "00:31:04",
                    "text": " And these cards are very, very cheap. They're about four to five pounds of their cheapest in Britain, so that's around 600 rupees or so. And so the cards at the end of that work can be taken out and destroyed securely in the same way as heart rights are destroyed at the end of life. So you can be certain the data within the hospital has been safely controlled, fully controlled."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:04",
                    "end": "00:31:20",
                    "text": " And we were able to show that if we were to do that, so to send out devices to four hospitals and perform the training on these devices while they're physically within the hospitals, that we can produce a model that falls far back."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:20",
                    "end": "00:31:39",
                    "text": " that had more consistency in its performance as it had seen a greater range of data. We were very pleased with the results of that and particularly that we were able to show that we could use this technology to deploy a test on data from Bedford hospitals and it performed just as well as it did for data from hospitals who didn't participate."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:39",
                    "end": "00:32:00",
                    "text": " And we see this approach as a way that hospitals in areas that, for example, do not have access to the resources that we might in the UK to take part with a relatively low budget. They can join a federation like this, join a study like this, contribute to model training and model weights. And the value add of this is making it as easy as possible."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:00",
                    "end": "00:32:38",
                    "text": " to use it. And we were very fortunate to have significant coverage at the time. And so I suppose a thank you to my collaborators for this work, David Clifton over in Oxford, David Eyre in Oxford, and in the emergency department, Alex Novak. And that's it from me. Thank you, Andrew, for that wonderful talk. And our last speaker is going to be Francis Crowley, who is going to talk about piloting the global AI policy data visitation for quality in healthcare data. Thank you, Don."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:41",
                    "end": "00:33:10",
                    "text": " So essentially what we've been looking at with our distinguished panelists here today, under the leadership we've been, is being quiet. But what we're really looking at is a new technology and how we can do something better with data and interrogate that better for the purposes of health. And we've seen the slogans here for the conference here, welfare for all and happiness for all."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:10",
                    "end": "00:33:39",
                    "text": " That's wonderful. That's a hard road ahead. And certainly one of the things that we'll have to achieve on that road ahead is health for all as well. WHO still. So that's really what we're concerned with here. I think the entire AI summit here is really about the society we're moving into. So it's not for some small reason that..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:39",
                    "end": "00:34:09",
                    "text": " of the President. Prime Minister Modi is coming to open up your session here and that the whole of the Congress and the whole of the governments in the states as well are behind us here too. It's very important for the future of India but also for the future of the world. So it's an important thing that we have here too. And what I wanted to say is with these technologies it's not just the technology and that's what the..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:09",
                    "end": "00:34:39",
                    "text": " entire conference is reminding us. It's not just the technology. It has to do really with the whole of society. So when we talk about this data visitation, we're bringing in new technology to interrogate data. The data is the basis of our science, but it's also the basis of the information that we exchange for all the other things in our lives as well. So this data is important to us. It has personal value, it has societal value to us as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:39",
                    "end": "00:35:09",
                    "text": " So when we work on these technologies, we have to ensure now that we develop within the technologies the kinds of considerations that would lead to, let's say, welfare for all, happiness for all, or health for all. Things like this. So just a few thoughts for myself. Data visitation is a method. As we know in science, method is the key."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:09",
                    "end": "00:35:38",
                    "text": " It's not new. Method is what allows you to develop new knowledge. So when you develop a new method, and that's what we're presenting here, a really cutting-edge method as well, so congratulations for that. But with that method, we're able to interrogate data better and receive more information from it. And this is important here, too. But health-related research is bounded by and termed by far more than simply method."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:39",
                    "end": "00:36:08",
                    "text": " even far more than simply science. It's impregnated with society. So we have rules for it. We have rules about how you have to do it. And those rules are not accidental and they're not just there because it's like tax rules or something. They're really there because they want to protect people, protect the integrity of the people, protect what people think is important to them. And we know genetics is important to people because it tells a story about us that we believe is a really important story."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:08",
                    "end": "00:36:37",
                    "text": " So what we want to do with this technology is not only make it possible, let's say, to do the scientific part of it, but also to ensure that within that scientific part we have the considerations that we need to have for society. So we have to have things in that that have to do with not only the science, but also we're really concerned in a certain sense with reliability."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:37",
                    "end": "00:37:07",
                    "text": " with acceptability, and I would say most important, with veracity as well. In other words, our science has to be reliable. So the conclusions that these fantastic scientists are drawing from this data visitation models, we have to be able to know that those conclusions are ones that we can work with, that they actually give a picture of myself as a patient or ourselves in society. We have to know that that's correct."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:07",
                    "end": "00:37:36",
                    "text": " And we also have to know that the work that these folks are doing is acceptable, right? It has to be something that society will accept. India is very sensible about that. You have very strong ethics regulations in place and so forth. The ICMR and so forth, as you know, is a very strong organization, and they really protect the idea of what's important for the populations within, and that's the way it should be as well. So we need to think about that too."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:36",
                    "end": "00:38:03",
                    "text": " So thus we need to be able to, and going forward, we can't have ethics just in paper. We can't have ethics in PDFs. We can't have ethics just in committees. It can't be just an informed consent form. Going forward with AI, especially with AI, we have to put the ethics inside the tech name, inside the technology. It has to be built in. That's going to be hard to do, but we have to do it because we can't."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:03",
                    "end": "00:38:33",
                    "text": " We can't do it the way we've been doing AI just goes too fast. So we have to make sure that the machines that we're building and the Computational power that we're using within those machines also are done with not only within let's say scientific parameters, but also within parameters So I think that's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:33",
                    "end": "00:39:01",
                    "text": " But we did some publications on this within our European Open Science Club and Research Data Alliance working group to look at what kinds of considerations should an ethics committee take when they're looking at health-related research that involves AI. That's one thing we looked at. And then we also looked at what kinds of informed consent."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:01",
                    "end": "00:39:31",
                    "text": " Should we be thinking about going forward for patients when they're making their data available for the kind of research that they're doing here? So we just tried to think about that and we wrote this propaganda that we also Looked at then some also poverty issues as well. So that was that seemed to be important to us, too So and I think this is the wonderful thing about data visitation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:31",
                    "end": "00:39:58",
                    "text": " because I think what we can do, you've heard this already, so I don't have to repeat it, but I think what we can do with data visitation is not just analysis. We can do the analysis, but we can do much more than analysis with it. We can do quality control. And this is really important because we have to know what the value of that data is."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:58",
                    "end": "00:40:27",
                    "text": " In other words, we have to know the real world value, if you want, of that data. Because AI, however wonderful it is, the way it is today anyway, it has no access to the world. So if it doesn't have access to the world, it has no access to the truth. Right? It has no access to know that that data actually represents this person here. We don't know that that data actually represents you. We think it does, but we want to make sure it represents you or represents your..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:27",
                    "end": "00:40:57",
                    "text": " genome or whatever it is carefully. So we need to be able to not only use this methodology to analyze the data, to query the data with regard to, let's say, certain questions that we have, but we also can use it to do quality control. And we did some of that with another organization before. I won't talk about that too much. But I think that's what we can do."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:57",
                    "end": "00:41:26",
                    "text": " So our objective really, one of the objectives we have, not just to do the scientific analysis, but also to put this in place. So we've come up with, and we're just developing this, it's not yet propaganda, but we have these efficient principles that we're working out. But more importantly, and this gives us a kind of ethical background for what we're doing, but we're also working on these pilot principles."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:26",
                    "end": "00:41:53",
                    "text": " And what we want to do with the pilot principles is more or less like you have the fair principles for data, you have the care principles for data, and you have the trust principles for data repositories. Well, we want to have principles like this for artificial intelligence. That's what we're trying to do with both the invigis principles and the pilot principles. And that is to impregnate those with purpose, integrity, law, openness."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:53",
                    "end": "00:42:22",
                    "text": " whatever that means, and test it. That's what we thought there first here. So we want to move really from instruments, from pure instrumentation in a certain sense, to working purposely. But that means working purposely within society. Or, if you like, from moving from tautology to people and their health. That's what Waven is actually trying to lead us on in that path here. And with that, I'll just stop with what I think is one of the most important."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:22",
                    "end": "00:42:47",
                    "text": " remarks, and for me anyway, in life, is that morality is the basis of things, it's a quality in fact, and truth is the substance of all morality. One thing that ethics and science have in common, and it's really important, it is ferocity, it is truth. And we have to bring those things together within our AI technologies. Thank you very much."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:52",
                    "end": "00:43:21",
                    "text": " Yeah, thank you so much, Frances. And now we have a little bit of time for some panel discussion. I'm just going to put this up on the slide. Yeah, so I guess we don't really have a ton of time. So perhaps we'll just start off with one question, which is we have a lot of really amazing technologies that these panelists have already showed today. And what do you think are some of the challenges of translating some of those technologies to real implementations? And also, how can some of the people in this room who are sitting here help with that?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:21",
                    "end": "00:43:51",
                    "text": " Does anyone want to go ahead? I think at this stage, people probably realize that we're living in the post-AI world. Technology is amazing. And so I think we're in a sort of engineering bottleneck world. So bringing this technology to where it needs to go applies boots on the ground, you know, sort of FD, getting people there who have the expertise and capacity to integrate it with people who have the problems themselves. So technologists like me are not the main experts in problems, but we can help."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:51",
                    "end": "00:44:20",
                    "text": " Bring the technology to the ball. OK, yeah. I think one of the most important things that we should all think about, especially when it's related to human health, is we have to start from where it impacts everyone from in human terms. And that's not just about technology or about policy principles. All those are important. But ultimately, we have to start."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:20",
                    "end": "00:44:41",
                    "text": " from how does it affect each and every one of you? How does it affect your families and your health? And when you, technology is a critical enabler, but when you work backwards from there, you can also see all the additional support layers that must be in place to make it implemented responsibly and sensitively, including principles, including policy."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:41",
                    "end": "00:45:18",
                    "text": " So I think if we work backwards from the bigger goal of disruption, of making everyone's lives better, I think that would show us a path on how to implement it. What was the question again, Don? You know, I'll look at it slightly differently. I mean, we talked about data visitation. I'll just switch to Hindi."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:18",
                    "end": "00:45:49",
                    "text": " So the data's sovereignty will remain the same. So now I will ask you a question"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:49",
                    "end": "00:46:17",
                    "text": " So the important thing for you young audience particularly students here is there's a lot of things to do with data. Data is floating around then why are we not using data because either we don't know how to do it or we are bad at it or we just don't enjoy doing it. So I think if you enjoy working with data find a problem and also at the end of the day all this problem must lead to a solution and make some value. The value for person one could be money"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:17",
                    "end": "00:46:43",
                    "text": " Value for person 2 could be something else. So there must be a value attached to it. With that value, you will never work on that. So find out what you're good at, what you're passionate about. There's a lot of data. There are experts sitting there. Go and talk to them. Find what you can do with the data to get a solution. In fact, we are talking about health care. The other thing we were doing during the break talking about India and the Global South is a huge problem in agriculture. Water, clean drinking water."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:43",
                    "end": "00:47:12",
                    "text": " Agriculture. Agriculture is one huge area we're just waiting to be infused with data-driven solutions. What should farmers do? What type of seeds should the farmers sow? What is the right season to sow the seeds? What is the right season to actually harvest the crops? These are all data-driven solutions. These are local solutions. One solution which is in Eastern UP may not be the right thing for Kerala. One which is good for Gujarat might not be good for Nagaland. So what you need to do is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:12",
                    "end": "00:47:43",
                    "text": " Get the data, be passionate if you are, and the solutions will be there. So that's what I would say, just to turn things around. I really like Ganesh's answer. Now, I think if I reflect on examples of projects that have been successful, not just within the United Kingdom, but wider in Europe and around the world, what they all have in common is that there will always be a champion at the heart. There will be one individual or two individuals who had a vision and had the..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:43",
                    "end": "00:48:08",
                    "text": " Passion and the mouse to pursue that vision and that's why I think passion is so important in seeing it through now I think there is a second aspect which isn't around You know many of these technologies are well understood by people in this room well understood by Technologists well understood by people who are young and passionate about what they do Institutions that have been around a long time to them these technologies might be new"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:08",
                    "end": "00:48:38",
                    "text": " to them these technologies might conflict with existing organizational policy or with existing organizational direction of travel. And a huge role for people to play is in saying, I see an opportunity for this technology to really help with the problem and almost going out and selling that technology, not for money, but selling it in terms of the benefits it can bring, how it can align with what others want to do. And then there's the execution work. Now, I think being a professional plus a data scientist, I think is an excellent"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:38",
                    "end": "00:49:13",
                    "text": " place to be. I think being an engineer as well as a lawyer or a doctor or an accountant or pick your profession, pick your field, I think that's really important. It allows you to bring that deep domain expertise with the technology skills and that's recognising the two are inseparable. Yeah, so I think all great answers. Who here has used ChatGBT? Does it feel magical? Does it feel magical? I'm telling you, it feels magical."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:13",
                    "end": "00:49:42",
                    "text": " I think we need to realize we're in this post-AI world now. We should no longer be restricted by what we used to think was possible. We should now be thinking about what we want as a society, right? As a group of people, what do we dream of? Do we dream of having all of our family members be healthy? Do we want to cure all that disease? Let's do it. I mean, why are we waiting? Like, if we're restricted by the capacity and capability to implement things, that's no longer the restriction it used to be. So I just am very passionate about this, you know, as you know from what I said."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:42",
                    "end": "00:50:14",
                    "text": " And I think we should have action to get things done, right? We should take action, get those principles and those moral values that we can unite on, and then get together and just do it with the passion that we have to do it. I think, actually, we're running out of time for more questions. So please come up and talk to these very enthusiastic and passionate panelists afterwards. And actually, to end off the panel and our session, we're actually going to have Wade in to give us some concluding remarks."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:17",
                    "end": "00:50:47",
                    "text": " Thank you. So in 2003, even though most of you aren't in the biology and genomics field, but probably all of you have heard of the Human Genome Project, which brought humanity into the age of genomics, into the genomic era. And why was that important? Because the genome, as Binet had given everyone an explanation of, is..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:47",
                    "end": "00:51:19",
                    "text": " It's a blueprint for who we are. But all of us are very different. Even among Indians, there's a huge variation. So much variation that is unknown. So if the original intent of the Human Genome Project was that it's a public good, that it's something that everyone in the world can benefit from, then how can that be true if there's still so much gap in access, in representation?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:20",
                    "end": "00:51:51",
                    "text": " in useful knowledge, in implementation at the national level. What they did get right in the Human Genome Project was that they ensured it was a public good so that researchers could take this initial draft in an open way and openly work together to explore how to take that board. This was 23 years ago, 2003. But now in 2026, there's been so much"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:51",
                    "end": "00:52:20",
                    "text": " Progress. There's been incredible leaps in the technology development. There's been incredible breakthroughs in data generation and the science. There's been incredible exploration into precision medicine and how to use it in things like cancer treatment and in a lot of other things. But even though it's very impressive how the field of genomics and biology has developed in the last 20 years, but if you look at it in human terms,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:20",
                    "end": "00:52:50",
                    "text": " We have to go back to human terms. If it's a public good, it's not a public good just for researchers. It's not a public good for people to write papers or for companies to just make a bunch of money. For it to be truly a public good, it has to be able to touch each and every one of you and your families and your children. It has to be something that, because of it, one day we can all say we truly live longer and healthier because we fulfill the promise of the human genome that's a public good for everyone. So..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:52",
                    "end": "00:53:19",
                    "text": " The only way that this is going to happen is that we have to come together in a much bigger way and we have to do something much bolder. We have to do something where we remember the human impact as our measurement of success. And this is precisely the reason why a group of collaborators and scientists and leaders from 14 countries came together to write that editorial to ask for a human genome project too."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:20",
                    "end": "00:53:48",
                    "text": " Because the goal of the Human Genome Project 2, it's not just science, it's global precision health, precision public health. Its success is only measured by whether it really enabled whole populations around the world to live longer and healthier, to enable countries to become more sustainable and prosperous. And guess what? If it's something for humans, for humanity, who's most of humanity?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:48",
                    "end": "00:54:14",
                    "text": " What did Dene say in his talk? The Global South represents the vast majority of humanity, yet we have a big imbalance in terms of our participation, our access, and our ability to benefit in it. But this is a challenge that no one country can solve alone, which is why the scale of this project requires everyone to come together."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:14",
                    "end": "00:54:44",
                    "text": " and do it. So this is something for all of you. Think of it as, it doesn't matter if you're from genomics or not, this is something that one day wants to benefit you, your families, your children, your loved ones. And this is such a multi-stakeholder effort that it's not just about the scientists, it's not just about the companies trying to lower costs and make things available. We need to invent the future. This is why we need to build it. So,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:45",
                    "end": "00:55:13",
                    "text": " What you saw here today with the panelists here and what they share is an example of an attempt to solve some of those realistic problems. Because there are certain problems that are red lines that we can't cross. We have to protect sensitive data. We have to respect sovereignty and privacy. We cannot open that to risk. But there are innovative ways, as you have seen today, where you could actually use technology innovation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:13",
                    "end": "00:55:41",
                    "text": " to preserve privacy but at the same time still enable open science and benefit sharing. And so to conclude, to all the software engineers and to all the entrepreneurs and software developers, please be inspired by this and use these initial efforts to try to come together and build the future and solve the problems. We need your expertise."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:41",
                    "end": "00:56:10",
                    "text": " And also to all the clinicians and scientists, please find ways to support the software engineers to do cross domain collaboration. And finally to the policy makers, please find innovative ways to support this and future proof the implementation. Thank you very much."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:10",
                    "end": "00:56:41",
                    "text": " It's at an Indian International Center if you want to come over. Yeah, so thank you so much for listening and thank you so much for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:48",
                    "end": "00:57:45",
                    "text": " And next we have... And thirdly... And last but not least... Our moderator..."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6992eacf4c5e8f0974c87429",
    "projectId": "736926",
    "projectName": "BioAI: Integrating Artificial Intelligence & Biology for Breakthrough Innovation ",
    "projectDescription": "BioAI: Integrating Artificial Intelligence & Biology for Breakthrough Innovation ",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-16T10:00:46.446Z",
    "projectUpdatedDate": "2026-02-16T10:00:46.446Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-16T10:09:53.208Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "BioAI_ Integrating Artificial Intelligence _ Biology for Breakthrough Innovation _(M4A_128K).m4a.mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771235960485.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:28",
                    "text": " You know the structural data which was all very well found and presently available. You could actually I think some of the speakers will talk about it and how that really got converted into an AI model which is actually also led to the Nobel Prize activities. So therefore one needs to look at how the AI is functioning now. And if you really look at for the first time we are creating systems that learn, design and also interpret along with human beings."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:28",
                    "end": "00:00:46",
                    "text": " And I think in the biological context this is very very critical and important to understand because what we look at is life processes which are fairly dynamic and has large number of evolutionary pathways also embedded within it which actually survives in a models which are quite unique and unprecedented of its kind."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:46",
                    "end": "00:01:14",
                    "text": " The other thing one wants to understand from a biological function is that the scales are very different in biology. While you work at the molecular level, then you work at the protein level, you work at the cell level, you work at the organism level and you work at the ecosystem level as well. So these all have to be balanced out in different ways of looking at how to do this. From DBT BIRAC what we have been trying to do is to develop various programs to bring academia startup industry"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:14",
                    "end": "00:01:44",
                    "text": " such that we can accelerate the bio-manufacturing point of view of our country's ecosystem. And as you would have realized, the bio-manufacturing really got taken off during the COVID times. The number of vaccines that were created using bio-manufacturing capacity was unprecedented and never heard before. Like more than 2 million vaccines, 2 billion vaccines doses were available to be provided, to be empowering the entire country's ecosystem. And therefore, this is a much..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:44",
                    "end": "00:02:12",
                    "text": " time to really look at how to be able to ensure do this. So I think what we have is two sessions right now we have and one is the AI Powering Biomanufacturing for Next Generation Therapeutics and what department is trying to do is to set up large-scale genomics data in the Indian population which can provide the wonderful phenotypic phenotypes model as well as disease models that are available in different settings available as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:12",
                    "end": "00:02:41",
                    "text": " What has happened in the past is that 10,000 genomes have been identified and these are healthy genomes that are being provided whose genome sequence is available in biological data center. Now what we want to expand and vintage to a million genomes, which is a program that is going to be announced very soon from the government of India. At the same time, bio AI hubs actually allows you to put the pieces together. So there are various partners, as I showed you in this cartoon."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:41",
                    "end": "00:03:07",
                    "text": " One is a biologist, the second one is a computer, the third one is the, you know, the ecosystem IIT professionals that are working with and you need to cross talk across ecosystem to be able to ensure that that's sufficiently works together. The second panel has got to do with the AI driven bio factories for sustainable materials. And you can imagine that all the materials that we are coming in our country right now for the chemical industry that talks about."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:07",
                    "end": "00:03:36",
                    "text": " tremendous progress actually produces about 60,000 to 70,000 molecules. They all come from the petrochemical sector as a precursors come from the petrochemical sector and that as the petrochemical sector will undergo changes the dynamic changes that are associated with in terms of you know regenerative energy sources. You will have a challenge of becoming these petrochemical sectors to also resource activities to able to bring out into these kind of models."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:36",
                    "end": "00:08:30",
                    "text": " So for example, these are biodegradable polymers, biobased chemicals, sustainable materials, carbon capture technologies. All of this will undergo transitions in the next 20 to 25 years, which will be of unprecedented nature and cannot be imagined how they will actually transform. And that's where the Indian agro-biodegradable... Biology is no more an observational science alone, but now we can make, we can engineer biology of the future. People have talked about..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:32",
                    "end": "00:08:60",
                    "text": " People have talked about healthcare and other examples and when I was sharing this with a friend He said that if you looked at how we treated our people 100 years back We would laugh at the way they were treating our patients 100 years back, but if you look at the next 15-20 years I think people will laugh at how we treat our patients today in the context of what are the changes that are likely to happen So we know so little about biology and there is so much more for us to learn"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:60",
                    "end": "00:09:57",
                    "text": " And today with the kind of advent of technology and AI, we are now at a point where biology is not just materials. So there's a lot of activity along those areas as well. And in general, if we look at health or"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:57",
                    "end": "00:10:23",
                    "text": " plant health or any of those kinds of things. I think how we engineer next generation biology and how we make the interventions happen is going to define the future of how we are looking at these kinds of science. So I think we are at a very exciting age and I see a lot of students here. For those of you who are in this area, this is I think the best area for you to be if you look at the next 25-30 years and I'll share a personal story here. When I started"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:23",
                    "end": "00:10:51",
                    "text": " to be in the software industry in 1990. It just felt like very fuzzy and hazy. But now if you look at it today, it's a very large industry, and I think a similar opportunity exists for biology, especially because of the changes that we see here. The second thing that I'm going to mention here is that AI has become the microscope for modern biology, and it is allowing us to see patterns that we could have never seen before."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:51",
                    "end": "00:11:20",
                    "text": " So, AI is not just being used for various things, but because of the ability to look at micro-images, lots of other things. You can see structures beyond complex biology. There are signals that you can detect which you could not have detected through human eye or human observation. Now, we can look at much finer details and then connecting different concepts. This is again a very exciting thing with multimodality of AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:20",
                    "end": "00:11:50",
                    "text": " the ability to connect different concepts in biology and other sciences with each other is going to be possible for the very first time. So with AI as a companion for biology now, we have the ability to look at things that we had never possibly imagined. And maybe I should have added that it's not only the microscope, but it is also the telescope as well. So we can look at things that are far away, which we had never thought of. So this is where it becomes very exciting."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:50",
                    "end": "00:12:19",
                    "text": " Third point I want to make is the fact that AI has made it possible for us to build bio manufacturing which is both scalable, precise and more effective and we will have a bigger panel on this specific topic so I will not spend a whole lot of time trying to explain this because that's what we're going to do next. But essentially you can define in silicon, build in silicon or meaning build in a model first before you put it into manufacturing as we all know."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:19",
                    "end": "00:12:42",
                    "text": " One of the challenges in biological manufacturing is the search space, so to say, that you have for the number of proteins, molecules, how they can be combined is extremely large. And over evolution, biology has figured out the best ways for how these things communicate or connect. But this is very hard to do when you are starting to build new molecules or new proteins or new substances."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:42",
                    "end": "00:13:12",
                    "text": " and AI will allow us to have this opportunity to look at multiple different kinds of datasets and combine them in very very effective ways which is really what some of the work that has been done in alpha folds and all the other things that has happened through DeepMind you will see that but the ability to be more precise is what manufacturing is possible because of AI. The fourth thing that I'm going to mention is that this is actually an important time for us to invest in biology so we have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:12",
                    "end": "00:13:39",
                    "text": " three broad areas that I am going to share. One is of course research. There is so much more to do that has still not been figured out. So I think there is a time for us to invest in the science and the research. And some of the scientists that are in India and I am familiar with them, they are doing pretty amazing work in this area. So those students who are thinking about what is the future for them, there is a lot of opportunities for research in India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:39",
                    "end": "00:14:06",
                    "text": " and I would like to encourage you to take a look at the dbt labs because there's a lot of action there. Manufacturing we will spend some more time on just now and finally there is a need for setting up an ecosystem and the dbt has been working together with many educational institutions and other institutions to see how to build an ecosystem so that we get the right kinds of trained people who understand biology and computing and various other things. Now this is very important."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:06",
                    "end": "00:14:35",
                    "text": " Because many of us who are kind of my age, we actually had to make a choice between biology and mathematics in the ninth standard. So you know how things were at that time. Today, we cannot work unless we work collaboratively. So this is really a lot of activity going on here. And I want to mention this last thing, which is the number five item that I wanted to share, is that this is, you know, we are at the AI Summit here in India with lots of you here, and this is India's opportunity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:35",
                    "end": "00:14:60",
                    "text": " I think if you look at it we have large volumes of biodiversity within India whether it is genomics or other things and the secretary talked about the need for creating genomics related projects in India because we have unique diversity and there is an opportunity for us to find things that nobody else in the world can find. In addition to that we have a very healthy system that has come from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:60",
                    "end": "00:15:25",
                    "text": " ancient times including Ayurveda and others where we may not have had a chance to identify some of the things that we have always done if they're very what is the data backing on that and can we bring it to a scientific place where we can connect some of our ancient traditions where somehow we know that they work but we don't know why and how what are the pathways so this is an opportunity for us to differentiate as compared to the rest of the world"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:25",
                    "end": "00:15:52",
                    "text": " and we have in India an opportunity to build at scale. This is something that we have demonstrated through various of our projects and I think biology requires us to think about problems not just at small lab level but at population scale and this is where India can really be the driver for the rest of the world and finally I think this is our decade and this is our opportunity and I think I'm very excited to see so much interest here."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:52",
                    "end": "00:16:32",
                    "text": " And I'm also equally excited to learn more about it from the panelists. I'll stop here. And I want to thank you all for being here, and thank you for inviting me. Thank you, Dr. Despande, for your insightful keynote address. So now we now move to the next segment for our session, Engaging Panel Discussion on AI-Enabled Biomanufacturing for Next-Gen Medicine. So now I will invite our esteemed moderator for this panel, Dr. Anurag Agarwal, sir."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:32",
                    "end": "00:17:04",
                    "text": " and joining him are Dr. Depi Thakuralak ma'am, Dr. Arvind sir, Dr. Madhura ma'am. So, over to you Dr. Grawal. Thank you very much. I will not consume much time in introducing. We'll manage. We'll manage. First of all, I want to amplify two points made. All of you have chosen the right room to be in. In various good things about AI, Balaji is the most happy. So, Demis Hassab has basically spoke recently, just as math turned out to be the language of physics."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:04",
                    "end": "00:17:32",
                    "text": " AI will be the new language of biology. Why is that important? Dr. Gokri had mentioned about how so far we were looking at nature being inspired to build AI. Very soon with AI, we might be inspired to build new life. Good or bad, we will talk about that later. But basically we are already at a point where an entire synthetic chromosome can be made and bacteria can be made. If you could imagine what a new sequence could be, you could create primate life."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:32",
                    "end": "00:18:01",
                    "text": " And of course, in areas like bio-manufacturing, being able to create living factories is absolutely critical. But my panel is about bio-manufacturing for next-generation medicine. So I'm going to focus a bit on next-gen medicine, which is basically the type of medicine we have not seen before. And we have seen plenty of examples. We have seen AI-made antibiotics that are killing MDRs that we could find treatment for. We are coming up with genetic therapies by looking at genetic sequences."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:01",
                    "end": "00:18:27",
                    "text": " So I'm going to Arvind, I'm going to go to you first. Recently we saw that we had virtual labs basically creating new novel mutations never seen before. A team of AI agents building new nanobodies to solve school issues. Where do you see this entire direction with AI itself helps us discover new things in medicine? Thank you Anurag and thank you to DBT for giving this opportunity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:27",
                    "end": "00:18:56",
                    "text": " I'll start with a slightly humorous note in 2021 when the DeepMind team and David Baker lab put out Rosetta Fold and Alpha Fold. Many of my colleagues in other branches of biology, given that I'm an experimental structural biologist, started asking me, when are you losing your job? But five years down the line, I can reaffirm that AI has been deeply empowering to people, experimentalists like us."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:56",
                    "end": "00:19:19",
                    "text": " On a more serious note, we know that many of these processes are going to add immense value to the pharmaceutical sector, close to a few hundred billion dollars. And most of it is through biology-oriented discoveries. Now, these biology-oriented discoveries, as Anand had pointed out, he mentioned that biology can be engineered."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:19",
                    "end": "00:19:44",
                    "text": " But with the advent of AI, it is more and more accurate. So biology can be engineered more accurately. Now this can be applied to, as Anurag already pointed out, to building virtual labs where biologists who need not necessarily have a computational background can resource these AI-based integrated tools so that they can design molecules of their interest."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:44",
                    "end": "00:20:08",
                    "text": " One example is that of the antibody design that has come up where they integrate the diffusion models with the inverse folding models with simulations of first principle based studies in order to design higher and more improved therapeutics. Of course, it does not substitute for experiments, but then it brings you more accuracy towards what could be your starting molecules that you can screen."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:08",
                    "end": "00:20:35",
                    "text": " Now the advantage that is manifold because you are not necessarily speaking to animal derived antibodies which is the norm and now you can design antibodies in silico and test them in your labs and also develop an entire pipeline which does not require animal based molecules and this is not just empowering both from a biology standpoint but from a sustainability standpoint also."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:35",
                    "end": "00:21:04",
                    "text": " So I think this is where the field is going, and it's really, really going to get exponential as we go along. Thank you very much. Madhava, I'm going to come to you next. You have used Informatics to advance biomolecular design in pharma. You have seen the field grow from narrow AI to a generative AI, and you also recently received a grant in the Bio V3 AI mission. So tell us about, as a person from that side, industry, how do you see Gen. AI, the latest things, helping you?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:04",
                    "end": "00:21:30",
                    "text": " Okay, thank you Dr. Anurag and I also thank DBT for giving this opportunity to share the industrial perspective on this. Yes, it has been a transformational journey for me because when I started curating data manually to using automation talking about ML to eventually using AI to optimize a parameter to generate AI where we are talking about multi-parametric use case."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:30",
                    "end": "00:21:43",
                    "text": " For example, a current use case that you know we are working on is about enzyme engineering and enzyme not only that we are trying to the biology is proven here, biology is working but that does not mean that the."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:43",
                    "end": "00:22:12",
                    "text": " treatment is perfect because there are there has to be an improved efficacy then there is toxicity there is immunogenicity and there is a bio-manufacturability problem of this particular enzyme so what we have done is that we have developed a platform through which you know we have built a generative AI based platform so that multi-parametric assessment of all these not just stability or also stability it has to work in the human"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:12",
                    "end": "00:22:40",
                    "text": " environment right so protease should not attack it so all these parameters are worked upon simultaneously and the iterations which otherwise happens traditionally by mutagenesis experiment in the lab and takes several iterations including again failure in the test the testing of the molecule that is not now happening because we have reduced the timing to up to 50 percent it depends on use case but we have seen a 50 percent reduction."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:40",
                    "end": "00:23:05",
                    "text": " and that adds so much of value and we have built not just generative AI because there are hallucinations right hallucinations come with generative AI but this has to be grounded in experimental reality and that's what we have built a bio AI loop wherein we design we build we test and we share and the same information the same data goes back into the system and the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:05",
                    "end": "00:23:33",
                    "text": " that's where the beauty is that it keeps getting updated the models are updating themselves very rapidly and eventually in very few iterations we get we have seen a very great success and this is not for a single use case the platform has expandability to other enzymes therapeutics antibodies mimetics like many proteins and even cytokines so this can be used made use of as a hub"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:33",
                    "end": "00:23:58",
                    "text": " to which several spokes can join and utilize the facility. This has been a transformative journey for me to start from where I started and reach where I am. I would like to comment here before I end that Mark Andersen's comment 15 years back was that software will be eating the world. What it means is that it will eventually be everywhere. Now see where we are."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:58",
                    "end": "00:24:27",
                    "text": " And now we heard these two visionary people's comments. We know where AI in biology is heading. We are no more a biotech. We are going to be tech-bio. Thank you. Thank you very much. Predicting the future is famously said is a difficult business because it keeps on changing. So Lippy, I'm going to come to you. I remember the time when your molecular dynamics simulations, you took up 50% of IGIBs, used it on HPCs. And while AI has become quite good at predicting finding."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:27",
                    "end": "00:24:54",
                    "text": " You do full simulations even after the binding occurs. We have seen recently isomorphic labs come up with a brand new reasoning model type, vastly improving upon alpha forward. Where do you see this going where not everything is in the open source world? A lot of cutting edge reasoning models are coming. Developing this kind of AI takes compute at a very, very high level. And of course, we need to do a little bit more than just binding MSI. Over to you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:54",
                    "end": "00:25:23",
                    "text": " Thanks Anurag, and also thanks to DVD for this timely panel. I'm just going to introduce you to this concept of isomorphic and other AI labs that are based on the concept of drug delivery. Now drug delivery essentially relies on protein structure prediction. And for the folks who are not from the biology field, protein structure is nothing but the shape. So if you get the shape right, you essentially can get the drug. Under this tenet, I come from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:23",
                    "end": "00:25:51",
                    "text": " very first principle based methods where I did my PhD 20 years back and the title was the mechanism of protein coding. Meaning you need to understand how the proteins shape to find their meaning. But alpha code actually proved us all wrong. It essentially said that I'm gonna do a shortcut. We can predict just the shape and how is not essential. So now it has all the shape memory."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:51",
                    "end": "00:26:20",
                    "text": " isomorphic and other lab what they're saying is that if you have shape memory then we do shape reasoning and that's where it gets really interesting so now if you have shape you can get more complexes you can get more targets and the reasoning can tell you how the drug is binding to the protein so it's a very exciting phase and for example in our own lab we are trying to understand the biology in a realistic way computationally but also"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:20",
                    "end": "00:26:49",
                    "text": " things that are intractable experimentally. So I think that's where the space is going ahead. You do biology in a sense where experimentally things are seeing the forest from top, but inside forest you have a lot of things that are moving. So empty simulations is one of the next layer on isomorphic labs where you can also predict in a realistic environment how things are moving. So I think it's going to be very interesting in a way that we are capturing biology and how we are capturing it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:50",
                    "end": "00:27:21",
                    "text": " Mother will do my first follow up question with you because when you mentioned we are still thinking of the experimental part being human somebody is doing it on the bench and you get the feedback very similar that in medicine you know surgeons still think they are safe from AI because they are the one working with their hands but already in the West we are starting to see complete robotic loops where the design is starting to occur robots are synthesizing even the experiments are done robotically"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:21",
                    "end": "00:27:51",
                    "text": " Data is going back to the AI agents. Where do you see the human value additions roles in this particular group? They are, but I want you to bring it out. So everybody knows their job is very safe. I think one thing that is non-negotiable in this entire process, of course, I know what you are talking about and whether we will be reaching that stage. Yes, it's a matter of time for us. We may not be there right now in India, but we are."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:51",
                    "end": "00:28:19",
                    "text": " However, I think non-negotiable component here is trained resources for data because data is of utmost importance and I think proprietary data sets which are gold standard curated data sets are adding huge value to the platform that we are building. We all talk about large language models, yes. They are there and they're very useful. Almost everyone is finding it. It's daily usage."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:19",
                    "end": "00:28:48",
                    "text": " in their daily use. However, in biology, I feel, I'm sure others would agree, too, that the data need not be large to the extent where large language models we speak about. But it has to be specific. It has to be very accurate and highly curated with all the metadata associated with it. I think that's where humans will always be playing a big role. So there are two academics, Ukraine students."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:49",
                    "end": "00:29:08",
                    "text": " and you're not training them for this next generation of bio manufacturing for next generation of medicine. Where do you focus? So what we had painted in the previous discussion was the power of this methodology in doing a lot of the biological applications."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:08",
                    "end": "00:29:37",
                    "text": " But there's also the flip side, which is where many of these solutions that AIML-based algorithms put out are not necessarily fully customized to the experimental requirements, maybe as a therapeutic, maybe as a vaccine, or even as a diagnostic tool. So where the human element comes in, and especially to have students who have a future in this area, is particularly for making these models a lot more accurate at one level."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:37",
                    "end": "00:30:03",
                    "text": " and then subsequently developing high throughput platforms where many of these solutions, in silico solutions can be tested for proper experimental and therapeutic outcomes. So I think this is where the student human resource focus should be on. I think that is something very challenging at the same time exciting because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:03",
                    "end": "00:30:32",
                    "text": " Some of these departments in our Indian institution university setup have been always in silos, chemistry department, biology department and computer science department. And I think largely because of the advancements in the Alpha Fold, there's a lot of drifting which is happening naturally from the computer science folks to the biology. And as Dr. Deshpande was also saying, there is a realization that there are cutting edge algorithms that can really have a real world impact in biology."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:32",
                    "end": "00:30:57",
                    "text": " So I've had really a nice experience in interacting with poor computer science students lately who will take an initial learning curve to understand biology. But at the same time, they're really doing some phenomenal work in application development. So I think at the same time, biology students need to ramp up their understanding of basic algorithms, which is, as you all know, we are coding in English nowadays."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:57",
                    "end": "00:31:28",
                    "text": " So that is also patently happening. And I think it's really, really interesting how these two things are merging together. So I want to end by kind of quoting my friend of mine who wrote a book, Human Edge at the Age of AI. The point is there will never be a shortage for people who understand the problem. And I think those of you who understand the problem and leverage the power of AI will always have a very good life. Thank you. Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:28",
                    "end": "00:31:58",
                    "text": " Thank you Dr. Gawal and all our panelists for those three insights. We now move to our second panel discussion that is manufacturing materials with biology, the AI-driven biofactory. Now I will invite Prof. Santosh Chaudhary sir and joining him are Dr. Debasish Monthi sir and Dr. Ashish Senkarakwad. Over to you sir. Thank you. I think this is a very interesting panel and I am actually coordinating this panel coming completely from an engineering side and non-biological side."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:01",
                    "end": "00:32:32",
                    "text": " We are moving from the four biological questions he addressed to the health of the art to bring people to a manufacturing mindset. The engineering of the science that is discovered through a production process is another challenge. In fact, that challenge requires community. That gets there also. Today, biology comes in, and we need to have communication with that of the AI, because that defines the overall process."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:33",
                    "end": "00:33:02",
                    "text": " And if we have a process, now there are various new technologies emerging which have a key role in terms of managing this process. One is you have to think in a different way in the context of biology what you have thought in terms of manufacturing processes and when it comes to bio-manufacturing it is also another interesting part is that when you talk in terms of biology"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:03",
                    "end": "00:33:35",
                    "text": " The uncertainty of biology also impacts the production process. Now, the evolving biology, how it impacts the production process, how it is impacted, and that is something very important to look at. And the third is, how do you look qualitatively? Because in the industry, when you get a product, if there is a reliability, there are questions of quality and quality monitoring. So if you look at it from all these perspectives, I think we have very distinguished..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:35",
                    "end": "00:34:05",
                    "text": " who have actually been involved into this kind of research regarding bio manufacturing and actually having processes which are really involved in this space. So I'd first ask Ashish to, yeah, this side, okay. So Ashish, that with your experience of running an industry, you should really tell us then what are the challenges really these bio manufacturing processes have."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:05",
                    "end": "00:34:35",
                    "text": " Posing and you think that Absolutely First of all a very good afternoon everybody back all So we appreciate everybody's time and privilege to be a part of this August session I represent a company which is in the industrial biotech right so what we do is we take"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:35",
                    "end": "00:35:03",
                    "text": " the feed from the bio sources and we convert them into either fuels or into materials that are useful. That's what we do at Praj Industries, the company that I'm representing. And this has been a journey for the company for over four decades. And if we looked at the historical data and data is important as we just saw in the previous discussion."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:03",
                    "end": "00:35:32",
                    "text": " What it suggests is that in the past we had to really physically do a lot of stuff in terms of laboratory experiments, building pilot plants, taking those pilot plants to a slightly higher capacity plant before any of these industrial biotech technologies are brought for a real commercialization and a large production."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:32",
                    "end": "00:35:58",
                    "text": " And this meant hundreds of crores of rupees that needed to be spent and let alone the money that gets wasted if the experiment was not successful. And therefore it creates a deterrent because money is not always available cheap and you have multiple options to exercise when you have limited amount of money."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:58",
                    "end": "00:36:26",
                    "text": " So what is coming to the fore and why I gave that background to you is a concept of what is now being called as digital twins is coming up very very strongly. It's essentially nothing but a representation of the physical things in terms of science and mathematics that it can represent on a computer. So you need not build things physically."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:26",
                    "end": "00:36:54",
                    "text": " either in a lab or in a pilot plant or in a commercial plant. But way before you actually spend that kind of money, you can do multiple experiments just by creating creatively a model that represents those physical things. And those are very important when it comes to bio-manufacturing, the question that you are giving to. We've been working in technologies in the company where"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:54",
                    "end": "00:37:22",
                    "text": " Today, we can just use the models that we have created for, let's say, creating fuels out of the biological sources. For example, the agri-waste. You know, we can take rice paddy. We can take many other such agricultural produce that can convert the different chemicals that we can use into the industry."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:22",
                    "end": "00:37:51",
                    "text": " But all of these today can be first experimented with the models that we have. And then we can also tweak them. We can experiment with them what we call as the what-if scenarios. And before even building even one rupee or spending one rupee of capital expenditure, we are able to come up with multiple outcomes that we can then choose from."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:51",
                    "end": "00:38:20",
                    "text": " as to which ones will really yield us the best results. So in summary what I am saying is that there is a lot of computational technology and also there is a lot of knowledge that is available in the past. The trick would be for the students who are in the room today to ingeniously convert that into high fidelity models that will be used by the industries like us."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:20",
                    "end": "00:38:48",
                    "text": " so that we can carry on these experiments and with a fraction of the money that goes into either our R&D labs or in the capital expenditure for any kind of equipment, we will be able to produce much more efficiently. Thank you. Devarshi, let me just put this question in a slightly different context. In a sense that all these biological systems suffer from a challenge of virility."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:48",
                    "end": "00:39:14",
                    "text": " And when it comes to the industrial process, they have to be actually handling these variables in a very scientific way. So if you look at from a process point of view, how science can be used, what do you think of it? See the variability of course is in the natural system, it is part of it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:14",
                    "end": "00:39:40",
                    "text": " are but one is of course you try to decode and understand but under bio manufacturing what you try to do you are trying to engineer the system see let us say some of the example she gave let us say she will have carcinogen and you want to convert it to a sugar so now if you try to understand which enzymes are trying to make it which pathway will record you express that flow that pathway into another organism so now your variability would be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:40",
                    "end": "00:40:06",
                    "text": " much under your control. So you are not working on the actual bacteria or wherever on the plant where it comes from. You are cloning or removing the cell like trying to decode it and bringing it to another organism which is much under your control which is the E. coli or the yeast or something. And in that context how AI is going to help us like this has been done for the last 25 years. We started doing genome sequencing just for this purpose."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:06",
                    "end": "00:40:36",
                    "text": " genome-guided screening or something, we used our conventional bioinformatics and computational biology. But the limitation we had that where we could only deal with labeled data. So whichever pathways are characterized, whichever enzymes are experimentally characterized, we could do that. But we could not deal with unlabeled data. So what these foundation models, LLMs have come is our encodings were all biophysics-based or understanding-based."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:36",
                    "end": "00:40:47",
                    "text": " But with these LLMs and all these foundation models, you can use a lot of unlabeled data, like the sequence of not only the enzyme which you are interested in, all other enzymes are all other these things."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:47",
                    "end": "00:41:08",
                    "text": " build a foundation model and try to inherit the latent space what you call in this one and because of that a lot of power has come from unlabeled data and if we see the example of alpha 4-b inductor Gokhale said they not only use the sequence of proteins in PDB they use something called VFD big fantastic database which is a sequence of each and every protein which occurs anywhere in the world."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:08",
                    "end": "00:41:34",
                    "text": " So because of that two things and other thing which came for is this attention like with biophysis is new there are two residues that have to come closer but they were not able to decode it but attention which came from a computer transfer model which even a computer scientist may not have understood what is the role of attention for protein structure when these two things combined it became very very powerful so in that context if we do it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:34",
                    "end": "00:42:03",
                    "text": " from the enzyme sequence to pathway to this thing and plug it into each polycholine will have a good scope and that is where all the generative AI which are Amra gave or this we just discussed that is going to help in a big way like the designing of the mutations, designing of the pathway that's going to help a lot. So I think if you look at it in that sense that AI is also not leading to scientific discovery but also is a very useful tool in terms of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:03",
                    "end": "00:42:34",
                    "text": " Turing and making production process more efficient. Yeah, I'll just add to it. It is per se could lead to some scientific discovery because if you bring the explainable AI concepts in many cases, if you bring that in some scores and try to see in the context of biological understanding how we are discovering certain things which we would not have discovered. See, just one question to add to it is that when you are actually doing this whole process, you are making a product"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:34",
                    "end": "00:43:06",
                    "text": " Yeah, maybe I can attempt to answer this question. So I talked about digital prints."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:06",
                    "end": "00:43:34",
                    "text": " And I talked about how they, from the business perspective, help us. But also, when you get into the production side, one of the systems that we are developing is you can actually run the physical process of production and the digital twin, which is based on either the first principles or the first principles models that are aided by the AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:34",
                    "end": "00:44:03",
                    "text": " You know, because there are so many combinations possible, so many permutations possible, getting the right insights can be a problem if you do it in a conventional way. So we have a combination of both the first principles as well as the AI based engines as they are called. And we can run the physical processes, but we can run the digital processes even faster. So faster than the real time."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:03",
                    "end": "00:44:32",
                    "text": " because biological processes can be slower in terms of just making the production cycles and so by running them faster even before you can spoil a manufacturing batch by wrong operations or by wrong controls you can correct them because you will start to see some aberrations or some excursions in the parameters and therefore you can course correct them way before your entire batch can get spoiled."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:32",
                    "end": "00:45:03",
                    "text": " just as an example and so therefore combining these two you can really bring this whole technology that we are talking about to the actual industry applications that can start getting a lot more productivity through it. Just as an example I want to talk."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:03",
                    "end": "00:45:31",
                    "text": " which the chemical industry was making and we made it in a biosynthetic pathway. We took it from others but the actual product of course that could be predicted properties can be checked through AI model but once you start making a compound it has to go through the same sets of tests which the actual product goes through and then only it will be accepted but we can bring AI models which will reduce the actual screening part where a lot of things we can rule out using the AI model."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:31",
                    "end": "00:45:60",
                    "text": " So what you are telling is that this combination really increases efficiency of the production process. So I would just have a last question to address both of you. Because to run an industry, you require qualified manpower. So I think, do you suggest that this whole bioengineering that we are talking about, that requires repositioning itself to really meet the need of emerging bio-air industry?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:01",
                    "end": "00:46:31",
                    "text": " That's a great question. I'll give you a practical sort of challenge that we are going through. One of our processes is called as second-generation ethanol generation creation process where we take parali or the rice paddy and convert that into ethanol. Now there are not too many plants that are working like this worldwide and so getting trained manpower"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:31",
                    "end": "00:47:01",
                    "text": " to operate the plant in a manner that we really want it to is not easy and so again there the technology comes to our help we can create what are called as the operator training simulators which are again another type of digital twins but more oriented towards training the panel operators the field operators the people who sit on the control systems inside a control room and we can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:01",
                    "end": "00:47:30",
                    "text": " actually make use of these technologies which are like I said based on simulation and AI so that we can train the people just like how the new pilots are trained on a simulator before they take into a cockpit and fly a plane very similar concept for the operators in these new emerging bio industrial plants which are coming up and which will come up in the future and we can really make use of this for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:30",
                    "end": "00:47:51",
                    "text": " the startup sequence, the shutdown sequence, any malfunctions, how do you handle the plants, any kind of aberrations that the processes go through. So all of these can be very much utilized and therefore we can make use of the technology of practical applications in the industry."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:51",
                    "end": "00:48:09",
                    "text": " I think that's a very important point. I think we talked about it already. Ananth told even here. I picture we segregate like if you think of bio-AI, you need people with quantitative understanding, computer science understanding and biology. But we segregate too early into a physical science or a biological science."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:09",
                    "end": "00:48:32",
                    "text": " So we need to change that. That is one level of change. Even in 95 when I went to Israel for postdoc, most favorite subjects in their undergraduates was computer science, life science together. So but somehow this has not happened. Maybe some of the research varieties have done to some extent. So we need to bring that. That is one thing. But at this, but that will all take time."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:32",
                    "end": "00:49:01",
                    "text": " So at this stage, even if I am a professor in a biological institute, whatever I need, maybe I should also undergo in a training in a computer science person. And we should design such multi-tiered trainings depending on what is required at a different level. With that only, we can bridge this gap. And that's a real challenge. I think we can come to the end of this panel discussion. I think I brought this question specifically because I think in that total"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:01",
                    "end": "00:49:30",
                    "text": " academic ecosystem in the country, this convergence is not really practised. And it is absolutely important that this whole panel discussion is clearly shown from a science and production point of view."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:31",
                    "end": "00:50:07",
                    "text": " My question is open to the entire panel. First of all, amazing discussion that we had. Going back to the question, Maggie had mentioned that generative AI models, you use experimental data to grunt."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:07",
                    "end": "00:50:49",
                    "text": " those particular models. But aren't experimental data by its nature inaccurate? How do we make sure that it's not steering towards the wrong direction? Yes, it's a good question. So even if Jataki AI or AI weren't there, the data was still flawed. And still the discoveries were made, inventions were made, and we have reached the point."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:49",
                    "end": "00:51:18",
                    "text": " where we are now talking about this very high technology induced outcomes in terms of efficacy, in terms of time saving and in terms of cost saving because otherwise you know biology is expensive subject. So that's what I said a very highly trained data or gold standard data still stays the key to all of this right."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:18",
                    "end": "00:51:48",
                    "text": " And that's where human in the loop. So that's why the loop has to be developed. The human in the loop is imperative. We cannot at least right now get away from, you know, use of human. Also, one more thing that we need to establish here is that the data that comes from the experiment, okay, itself is either proven in the lab or not. So the laboratory experiments is also imperative. So unless..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:48",
                    "end": "00:52:11",
                    "text": " AI, generative AI, AI is grounded in these two things. There are chances that what you say would happen. But if we keep this loop continuously working, the chances that we get to reduce these errors would be very high. And still we are developing, right? AI is still a technology as maturing. So 15 years down the line, these problems may have some solutions."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:11",
                    "end": "00:52:39",
                    "text": " which are more easier. I'll just quickly add one thing. I think as a blanket statement, it's inaccurate to say that experiment data is not really correct. In fact, what I would say is ML architectures take this ground truth data, okay? Protein sequences, protein structures that are coming out of genomics and crystallography are as close as you can get to the ground truth data. In fact, to say that they are inaccurate is not true."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:39",
                    "end": "00:53:04",
                    "text": " Now you could say the other layers which are cell and tissue based and largely biology is conditional. So under a particular condition using high resolution data would give you really good data set. You could now say if I have different perturbations what is the result. So there is a lot of activity in that space."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:06",
                    "end": "00:53:35",
                    "text": " I have a question because I want to understand the industrial mindset. Sir, from Dr. Despande, while we recognize that AI is accelerating engineering biology as never before, while investment in big AI infrastructure is important, we are also focusing on small purpose-built model to empower distributed innovations. So from industry perspective."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:35",
                    "end": "00:54:11",
                    "text": " What do you think government as an enabler can do few things immediately to get off the block fast?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:11",
                    "end": "00:54:39",
                    "text": " you need very large volumes of data and large volumes of compute to be building these models. So I think bringing everyone together and investing in these kinds of projects with priority as I think the whole that we need to do and I know that dbt is kind of exploring these areas. Okay now as we are coming to the end and India again team is forcing us to close quickly so may I respectfully respectfully invite Dr. Jitin Kumar"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:39",
                    "end": "00:55:17",
                    "text": " Dr. Bhirek to kindly come forward for the presentation of memento. Sir, may I request you to kindly present the memento to Dr. Rajesh Gokhale sir for his inspiring opening remarks and continued leadership in advancing India's biodiversity system. Thank you, sir. May I now request our Dr. Rajesh Gokhale sir to kindly present the memento to Dr. Anand Deshpande sir and also to our moderator, Dr. Andhra Agarwal."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:19",
                    "end": "00:55:55",
                    "text": " Thank you Dr. Kalewani."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:55",
                    "end": "00:55:59",
                    "text": " Thank you Dr. Aidna and thank you Dr. Elka."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "699308f74c5e8f0974c87797",
    "projectId": "455675",
    "projectName": "Agentic Commerce Trust, Tokens and 'Know Your Agents' for the AI in the Economy",
    "projectDescription": "Agentic Commerce Trust, Tokens and 'Know Your Agents' for the AI in the Economy",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-16T12:09:25.546Z",
    "projectUpdatedDate": "2026-02-16T12:09:25.546Z",
    "projectStatus": "Active",
    "companyId": "699305cb4c5e8f0974c87767",
    "projectCreatedBy": "699305cb4c5e8f0974c87768",
    "projectUpdatedBy": "699305cb4c5e8f0974c87768",
    "projectLastAccessDate": "2026-02-16T12:09:25.546Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "Agentic Commerce_ Trust and Identity in the AI Economy(MP3_128K) (1).mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771243731266.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:25",
                    "text": " agent to execute your findings. So what you are doing when you look at 10 different things, in your mind you have an idea of what you like. And what you like is not so easy to transfer to data directly. The point I want to make it, there is something called tacit knowledge. Tacit knowledge is knowledge that is not necessarily captured in the data. It is this tacit knowledge that we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:25",
                    "end": "00:00:49",
                    "text": " will eventually need to transfer to our models. And during these conversations now, we will repeatedly come back to the data as the make or break. To your point, we don't know what you like unless you write it out or you somehow have it, and this is going to be an issue in the long term. And I will say we actually know a little bit of what you like through the loyalty programs, right, and through your transaction records. We kind of know."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:49",
                    "end": "00:01:14",
                    "text": " where you are at. And so that's a starting point. But agents can understand your preferences, your constraints, and build from there. So let me just summarize this particular point. It seems to me, I suppose most of you are in this, that for us, for those portions where we as humans spend a lot of time, and I don't think agents doing the final grunt work is a really big add value for me."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:14",
                    "end": "00:01:42",
                    "text": " At the time when I take spending for decision evaluating this versus this versus this, it's not just requires data. And till those things get captured in data, agentic commerce will have limitations in what it can do. She's saying that we know a little bit. She's saying that some other models will emerge. But therefore, while this is a good thing, I think we will still, I think."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:42",
                    "end": "00:02:11",
                    "text": " there are ways to go, if I may say so. Because till this portion gets... We shouldn't leave it at that, though. I think you're making a point. I think we shouldn't leave it that that agent in AI commerce can't, won't be able to do X, Y, Z. I think we know a lot about you. Not right now. In fact, no, even right now, we know a lot about a lot of people. By we, I mean there is so much data. Every time you touch your phone, you are generating data. Every time you browse and not click, you are generating data. And today,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:11",
                    "end": "00:02:27",
                    "text": " that data can make or break. So my point is we're not, the way you're ending this part is that we know little, it's going to take a long, and to your point Janet, we know a lot. We need to know more, but we know a lot. I can give an actual example."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:27",
                    "end": "00:02:55",
                    "text": " of how I personally used agentic technology. This is funny because, you know, I'm a computer science grad, spent decades in this field, and I was very nervous starting out with agents. But here comes December 29th, and by December 30th, I have to fund my 401K. Otherwise, I lack retirement for the year, because IRS taxes, you have to fund your retirement. So I realize on 29th of December, I haven't funded my own 401K, which is my retirement plan."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:55",
                    "end": "00:03:22",
                    "text": " And I'm panicked. I called my custodian, which is Fidelity, and Fidelity tells me I have to use an agent. And I'm nervous as hell. I've never used an agent, especially to transact money, move money from my bank to my IRS. But here's where the tipping point came. One, the agent was trusted. The custodian, it was the custodian telling me to use the agent. Two, the agent asked me three times."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:22",
                    "end": "00:03:49",
                    "text": " requested my consent three times for the exact amount of money to the dollar digits, like to the points, right, to the cents, what money needs to be moved from what account to what account. It saved me a trip, one hour to the bank, standing in line, no appointment. Last day, I was terribly in danger, risk of missing my 401k contributions. Agentic was able to do it fine, and the agent did not go any further with that. The agent was de-instantiated."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:49",
                    "end": "00:04:15",
                    "text": " after that transaction. It was a very trustworthy transaction and I would love the convenience of such agents to help out in the future in a trusted environment. And this is just one application, it could extend. Good point. Okay, let me just one question which maybe Prakhar can answer or someone else can. What kind of traction are you seeing from merchants?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:15",
                    "end": "00:04:31",
                    "text": " and users as of now about this. First, Prakhar, because it's a gesture, and then somebody else can. I think for, so merchants are directly..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:31",
                    "end": "00:04:49",
                    "text": " asking us that we need agents. In Janet's example of Fidelity, Fidelity wants to build this agent. This is probably the first time in technology that at least I have seen over a decade here that the merchant is saying, I want this. Generally, you build technology and then you try to sell it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:49",
                    "end": "00:05:18",
                    "text": " it's the other way around. If you were to do a TLDR of your first round, one can argue that agentic commerce is brittle, it's narrow, it's not fully scalable yet, the data is not there, the models are too generalized, the protocols are not. You can list down a million use cases as to why agentic commerce is not ready. But here, a merchant comes and says, look, I think I can at least try productivity. Or, you know what, I'm willing to put a merchant, an agent, on my website."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:18",
                    "end": "00:05:44",
                    "text": " And I want consumer to interact. You know why in my hypothesis is that why they want to do this is because it's probably the first time when you interact with an agent, you are creating something called unstructured data. And that has never been created before. Because all the data that has been created so far in technology was click data, was structured data, like here is what you're doing. Natural language processing was limited."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:44",
                    "end": "00:05:56",
                    "text": " But when the GPD came out, yes, OpenAI has a lot of unstructured data. So now, Fidelity wants to know this agent. A coffee seller needs to know, OK, do people actually like cappuccino?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:56",
                    "end": "00:06:26",
                    "text": " Which of the kurta will actually sell? Which of the necklaces actually have a demand? Because I might get it wrong on the first shot, but then on the second shot, I might fix my supply chain. And that is unprecedented. And so at PayPal, what we are trying to do is we're trying to say, OK, good, awesome. Merchants want it. We are going to support it. And to Janet's point, what are the three tenets for agent commerce to work? The first is trust, literally trust. If there is no trust, this game is over. So you need some support."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:26",
                    "end": "00:06:55",
                    "text": " some sort of Aadhaar equivalent for agentic commerce. That's first. Second is the money flow. How does money flow between two agents? So you need an UBI equivalent of agentic commerce. The third thing what you need is some sort of an ONDC equivalent. You need some interoperable platform of how this will work. And so at PayPal, we are trying to build all this infrastructure to make this happen."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:56",
                    "end": "00:07:10",
                    "text": " So we've sort of seen, you know, seven to nine times of a difference of doing commerce the agentic way versus without agentic. So what happens is a consumer comes, let's say it's a piece of apparel."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:10",
                    "end": "00:07:33",
                    "text": " Traditionally, just look at the apparel. Here we come back and say, hey, do you want us to try this on you? Yes. Do you want to now try this in a specific setting? Let's say I'm going to a cocktail party. Yes. Right? So that's conversation happening between the consumer and the agent. And maybe the consumer says, yes, I like. No, I don't like. Give me something else. Right? And then there comes a point where the consumer says, OK."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:33",
                    "end": "00:07:56",
                    "text": " I think this is good. I'm convinced this makes sense for me. And then you sort of hand it over to the merchant, right? The merchant e-commerce platform. And the difference we are seeing is you just send a user the old fashioned way and this way, right? That conversion is seeing a seven to nine times difference, right? So the merchants are getting users that are a lot more interested than they got in the past."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:56",
                    "end": "00:08:26",
                    "text": " That's happening because the indecision you spoke of, right? That's something I'm iterating with the agent. Works doesn't work, modify. And so that's how we got here. And I can add to this conversation. I was in Dubai last week, and we were in a roundtable with the merchants. And one of the things the merchants are finding out that AI is disrupting everybody at a very, very large scale. So before, the merchants were places where we went to transact."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:26",
                    "end": "00:08:52",
                    "text": " We did the transaction fundamentally. And now transaction is happening worldwide. So people are buying things from the internet, Amazon, everywhere that they can. So we no longer go to the merchants to transact. So what is the role of merchants when the transaction can happen anywhere in the world and through many different versions rather than through a storefront? So now merchants are starting to think about what is the role of a storefront, really?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:52",
                    "end": "00:09:08",
                    "text": " The role of a storefront then becomes about intent, community. Why do people come there? To make network with other people similar, similarities, things they like. My community, if I'm a golf player, I go to the golf."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:08",
                    "end": "00:09:35",
                    "text": " storefront to find other golf players. What does that community look like? How do I have an intent around that? So those experience, what is the experience of going to a storefront? These things now matter far more than really going to a merchant for transaction. What the merchant sells, you can find out through so many different ways. All right, great. So let's move on since, so it seems not only agents will help us."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:35",
                    "end": "00:10:03",
                    "text": " All the merchants want it. Agents are helping. And what was clarified, even in the area where humans need to make decisions, you can have an agent helping you make decisions. And a lot of data is already there. Good. Now let's move on to the next point, which she referred to, and also trust and safety. So that, of course, seems to be the key point. So there's a question for Parag about"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:03",
                    "end": "00:10:27",
                    "text": " What evidence of identity authorization should be made mandatory so that trust is built in? Hello, you can hear me? I think everyone's right to point out. Look, we are not going to be transacting with any money unless we have trust in these systems. It's as simple as that. And today, generative AI models very much need to..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:27",
                    "end": "00:10:48",
                    "text": " act on our behalf with that kind of trust. So there's three things that we point out, and I have some notes here. Today's chat tools are becoming our digital proxy. That's a big ask. We are moving from just them chatting to us, to them being a proxy of us in the digital world. So that's a different ball game. From passwords,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:48",
                    "end": "00:11:04",
                    "text": " We are moving on to machine-grade trust frameworks, like you're mentioning. You no longer can have a password to log in, and you have many agents running on your behalf. The agents need to have their own identity. And like you point out, there are three things. There's identity itself."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:04",
                    "end": "00:11:23",
                    "text": " and we need to ensure that each agent has a cryptographically proven identity that stays with that agent for the period of time that it transacts in. That agent's identity should in no way be able to be compromised by other agents or in fact by other users. So the key piece is building up identity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:23",
                    "end": "00:11:47",
                    "text": " for agents, and there are already frameworks like that. The SPIFFE SPIF, for example, is an interesting framework. You couldn't use X.509 search in this space to, again, go down the path of identifying agents. That's one thing. In addition to identity, authorization is key. And where we're heading with authorization is granular authorization. Gone are the days where you gave, you know..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:47",
                    "end": "00:12:11",
                    "text": " access to your database to do whatever you can. As an example, if you give your agent read access to your emails and now you need it to write or send an email, this should be done dynamically within agents for the use case or the task at hand, and that is the kind of frameworks that are being developed now. So think granular access that is dynamic in nature. This is no easy task."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:11",
                    "end": "00:12:40",
                    "text": " This has to be done quickly. And finally, the third part is provenance. You need to be able to have a chain of custody, as you know. Blockchain is a really good example of that. But more generally, agents will need a chain of custody. They will need a way to prove that the agents are who they are, and they continue to be who they are. If you can do these three things, you then have an accountability layer. You have a layered responsibility, which is interesting. You also need redress mechanisms."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:40",
                    "end": "00:13:02",
                    "text": " This is why the world of transactions, the online world works, because you have strong redress mechanisms. In fact, MasterCard is an excellent example of how that works. We can use MasterCard's cards anywhere, for example. So we need agents to have, if an agent does something on your behalf, we need those redress mechanisms."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:02",
                    "end": "00:13:24",
                    "text": " pointing not just to you. So in the future, it is not me, Prag Sharma, interacting. It's Prag Sharma and his agent, Y, interacting. And my personality online is going to become me and my agents doing, I'm hopefully, some nice things on my behalf and not wasting my money or spending my money inappropriately."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:24",
                    "end": "00:13:51",
                    "text": " If I may add one thing, which is very short, yeah. I think if you believe the modality of the conversation of agent commerce, will it be voice or will it be chat? Because if it is voice, then you need to somehow figure out how to identify it as genuine and not fake. And then even if somebody is inadvertently passing sensitive information, like my Aadhaar card number or social security number, how do I not allow it to hit LLM?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:51",
                    "end": "00:14:21",
                    "text": " So apart from identity, there is a security layer that is highly dependent on the mode of how you're operating. I think that needs to be solved. Okay, and just one question to Arvind in this section of trust and safety. Where do you think governance needs to be kicked in? So one is the technology solutions of identity, security, trail, et cetera. Where do you think governance kicks in?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:21",
                    "end": "00:14:49",
                    "text": " Look, as people building consumer products and technologies, it's very easy to keep reducing frictions. Oh, I know how to do an API integration, B2B partnerships. Everything happens without the human ever being in the foreground. That's very tempting. Now, when you do that, someday you're going to have an unhappy user saying, I did not want this to happen, right? And so it's counterintuitive, but it's actually valuable to, by design, have elements where we want to have friction."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:49",
                    "end": "00:15:11",
                    "text": " have friction or require the end user to participate to be in the loop like how Janet had when she was trying to do her tax returns to come back and say, this decision seems like a high stakes decision. You may be very unhappy if you are not in the loop. Not that I can't automate it. It's just that I want you to actively participate. So."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:11",
                    "end": "00:15:40",
                    "text": " That really is a very big element, which is don't get carried away by, I can write code to remove friction. Stop and ask, will the consumer in 1% or 0.1% of the situations be very unhappy saying, no, I did not want that. If so, intentionally have friction. That's the very layman version. I'm sure there is a framework version of doing governance, but this is how I understand it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:40",
                    "end": "00:16:01",
                    "text": " OK, good point. I think this is a nice way to say. There is technology solutions happening for ensuring that the agents can genuinely be your proxy and security solutions. And his point is really, really very good, that human in the loop or user in the loop is essential to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:01",
                    "end": "00:16:29",
                    "text": " avoid the user who's unhappy, but in general to make sure there's a certain amount of visibility that is always needed, and that's best done by using the loop. Okay, great. Let's move on to the next section, which is about platform and architectures that can enable it. Some of you must be technical people as well, so we'll try to have..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:29",
                    "end": "00:16:49",
                    "text": " The questions are technical. Let's see if we can keep it that it is accessible to most of us. So for Janet, if your question is minimal architecture for secure and explainable agents, what are two or three non-negotiables that must be there in the architecture?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:49",
                    "end": "00:17:10",
                    "text": " Okay, so to explain what architecture foundations we need, it is very important to understand the technology we are using behind these foundations. So when we go into agentic architectures, we're talking about high potency in AI. And what do I mean by high potency? These can learn."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:10",
                    "end": "00:17:38",
                    "text": " just like humans can learn. So it's very non-deterministic type of learning, which means you can't control what the agent has learned. And so when you can't control what the agent has learned, how do you use agents very effectively? This is where the architectural foundations come into play. So you have to build foundations with non-negotiables. For example, you authenticate the identity of the agent. You understand the agent."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:38",
                    "end": "00:17:59",
                    "text": " on what data it trained. So the transparency in its training models. You understand its toxicity level. You understand whether the data is biased from which the agent has learned. You understand what jailbreaks you need to put around the agent. Again, part of the architecture that you build. You understand..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:59",
                    "end": "00:18:27",
                    "text": " toxicity, you understand jail breaks, you understand whether there's data leakage. Is the agent actually leaking any data? So you're monitoring the agent from birth to almost depth, if you will. There's a lot of observability built into it. Now, I will say where the complexity exists, right? Because agents work with memory, just like human beings. So we have working memory, long-term, short-term, persistent memory, and now Amazon introduced"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:27",
                    "end": "00:18:57",
                    "text": " something called episodic memory. Now these are long-term memories that these agents can have. And so it's very important to monitor the foundations so you know how to flush the memory of the agent. How do you de-instantiate the agent, just as much as you have created these agents. So these architectural foundations are non-negotiable when you actually build out agents. So question for Prague. How do you see token-like constructs?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:57",
                    "end": "00:19:24",
                    "text": " to facilitate agentic commerce, without going into too much details of this thing, but please. Yeah, so the idea for token is that it's a digital representation of something, right? And in fact, we have been digitizing our lives and our world the whole time. The paperless office was about digitizing. What token is bringing is digitization with traceability or identity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:24",
                    "end": "00:19:54",
                    "text": " The idea is that the token now represents something specific in some cases or something more generic like currency in terms of stable coins for example. But the whole idea of a token is to make something specific that represents something in the real world. Now if you make something digital in the real world, digital, there are so many benefits from it. Specifically if you digitize money, for example. So stable coins are a way to digitize money and the idea is that if you can digitize it,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:54",
                    "end": "00:20:19",
                    "text": " and you can safely and securely use it, then you have opened up a whole world of commerce that can be actioned through agents. So now you're combining the world of money, you're combining the world of payments, you're combining the world of technology into action. And that is amazing and very important. So just to give you an example, you can have programmable agents."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:19",
                    "end": "00:20:48",
                    "text": " that say, suppose you're buying something from your local grocer, and you say, look, if you don't deliver this before 6 p.m., you will not get paid X. You will get paid a much lesser amount. And that can be programmed. And the way it's possible that you can program is maybe through smart contracts or through other pieces of code. But the way you would pay that person is through tokens. The tokens could represent payments in the real world. It doesn't have to be on the blockchain. It could represent the real world payments. And you could still use traditional rails."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:48",
                    "end": "00:21:16",
                    "text": " that are repeatedly used to complete these transactions. So that's, I think, what it is. And one of the use cases around this is exactly that. Use tokens for anything to manage your financial health, all the way to make little purchases. But with tokenization, you can now look at the agents as unique entities. You can look at your payment rails as unique entities. You can even have tokens of yourself, if you like, in the future. The possibilities are endless."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:17",
                    "end": "00:21:33",
                    "text": " If you can quickly explain to us all, we already have digital things, whether it's my card number and so on and so forth, where I don't really transact with real money. How is token going to add to that aspect? Just quickly."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:33",
                    "end": "00:21:52",
                    "text": " Yeah, it's a good question. I mean, already in many countries, if you're traveling, you may not need to take any money out. I'm trying to do that with a rupee here. I haven't taken any rupees out yet, so I'm hoping I can pass the couple of weeks without doing that. I think tokenization brings another layer of independence beyond"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:52",
                    "end": "00:22:21",
                    "text": " digitization because when you're looking at your grant, you still have, you know, all the traditional rails behind that bank account that you're looking at works exactly like that. With tokenization, there is a little more freedom to maneuver and maybe go beyond your traditional assets and look in different ways. And with the tokenization, the agents never touch your credentials. They touch the tokens. So your credentials are safe. They are embedded within the token."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:21",
                    "end": "00:22:51",
                    "text": " So the question for Prakhar, follow up on this, is there are already existing trusted payment systems, artifacts, authentication, all that is there. Would they be useful in the agentic world? And how do you think tokens are going to eat up on that? Yeah, I think tokens are foundational to agentic commerce, as Janet said, because for agents to, if agent is doing a transaction,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:51",
                    "end": "00:23:18",
                    "text": " If you take a step back and say, what is an agent? The building lowest atomic unit of an agent is a large language model, which is a probabilistic instrument. It's not a deterministic instrument. So you actually want to make it as deterministic as possible. That's the first rule. If you're designing, you ask the architecture question. That's what you do at the first. You do that through protocols. Protocols are like highways. Go at the speed of 65 kilometers per hour. Don't go beyond that. So that protocols give determinism."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:18",
                    "end": "00:23:44",
                    "text": " Tokens add security. So if I am basically trying to buy a purchase ticket and that transaction is basically taking two hours now, so that token has been issued, agents already have been authenticated for those two hours to do that. So it could be asynchronous, could be synchronous, right? So now coming to your question, you asked, will the current payment infrastructure, at the lowest atomic level, yes, but a lot of work needs to happen on top of it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:44",
                    "end": "00:24:13",
                    "text": " Just because your large language models are probabilistic, so you need protocols. You need somehow protocol to interact with tokens. You need to issue short-term tokens, long-term tokens, and you need to just keep going there. So that's how I see it. Okay. So this section we end. So basically, it seems like the existing infrastructure will help move forward with this. But in future, if you want to do more, it's probably better to have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:16",
                    "end": "00:24:45",
                    "text": " Plus, you just add the time aspect of it, which my credit card information doesn't, and so on. Okay, great. So thanks. So the last question is more general, and after that, we still will have time for the... So inclusion and growth. I think at the end of the day, while we are all very excited about agent e-commerce, we know finally it should benefit by economic growth."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:45",
                    "end": "00:25:15",
                    "text": " as well as inclusion that not just growth for some, but exclusion of others. So this topic is about inclusion and growth, and my question for Arvind first is to ensure small merchants, rural users, and first-time digital participants, will they benefit from agentic commerce, or how will they benefit, whichever way you want to address it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:15",
                    "end": "00:25:41",
                    "text": " You're right. Eventually, it has to benefit everyone at population scale. I think the decision-making complexity exists for different transactions at different strata of the society. For example, some of the lower strata, maybe buying detergent is a hard decision because a 100-gram packet, if you think about it, unit cost-wise, per gram is cheaper or more expensive. It's more expensive than buying 1 kg."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:41",
                    "end": "00:26:08",
                    "text": " per gram. But they may not have the absolute cash liquidity to go buy the 1 kg packet. The money they have today only allows them to buy 100 grams. So even though they're trying to optimize for cost, it is not unit cost. It is the actual absolute affordability they have. Someone else might be space constrained. They may not have storage space to buy everything in large quantity and have them."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:08",
                    "end": "00:26:33",
                    "text": " The assistance needed for decision-making in transaction, I think, is universal and present everywhere. The challenge is the underlying technology cost for answering this. Is that cheap enough to also support a low-value transaction? That's what is going to determine, will it reach them or not? And I think that's where we need to invest time to make sure that the agentic AI is also cheap enough."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:33",
                    "end": "00:27:02",
                    "text": " to be able to support all of this. On the merchant side, I think something analogous to ONDC will help them participate. So the consumer story, it's going to be a consumer led story, ultimately bottlenecked on how cheap can you run agentic AI. If we crack that, there is an opportunity even for those small transactions for every strata of society. I'll add to this. If you see, agentic commerce makes literally any merchant global."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:03",
                    "end": "00:27:31",
                    "text": " And so you are now selling to Earth. And that is dramatic because LLMs can understand any language. So language is not a barrier. Agents can understand any language. You can create a website overnight. You can move money if you get tokens right and protocols right. You can move money universally across the globe. So yeah, I mean, a merchant in Jaipur can now sell to a globe."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:31",
                    "end": "00:28:00",
                    "text": " And that was unprecedented. That was not possible. And which is why agentic commerce is not an incremental shift to e-commerce. It's a different paradigm. A lot of plumbing needs to happen. But that's the promise, right? And so that's my hope. And last question to Janet in the same inclusion and growth. So going forward, which standard of public good would accelerate safe global agentic commerce?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:00",
                    "end": "00:28:22",
                    "text": " So we need global governance, right? The way I think about it, if you think about AI, AI knows no color. AI knows no country. AI knows no borders. So in a way, it brings a lot of equality. Equality to the large merchant, equality to the small merchant, equality to all. And if we have..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:22",
                    "end": "00:28:49",
                    "text": " global standards, if we have global governance, then we can make transaction and other aspects of agentic technology very, very safe. So we need the governance. And I would say inclusion is super important, but responsibility is equally important. So we have to figure out how to create responsible agentic AI. Great. So this ends our side of it. You heard it. We still have 15 minutes."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:49",
                    "end": "00:29:19",
                    "text": " though we started five minutes late. If, please make sure that whoever is circulating the mic goes there. If you have a question, please be very sharp. And if you want to make a statement, less than one minute, please. Just so people can know. Whoever is moving the microphone around, I don't know who has that. Are there volunteers who are moving the microphone?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:19",
                    "end": "00:29:42",
                    "text": " Okay, please go ahead. Question or comment? Question. Okay, Shah. So my end goal is that when you bring these agents, right, they learn behavior patterns of how people are buying things and what their intent is essentially to buy also. Let's say if I am buying a particular type of furniture and that becomes a general..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:42",
                    "end": "00:30:11",
                    "text": " understanding of the agents that most of the people are buying this furniture. So the merchant also knows that only this type of furniture is selling, so I'll make this type of furniture. Now, there might be some other person who is in a very small niche who needs a different type of furniture. Now, due to having these kind of streamlined intent recognition systems in place, we'll end up reinforcing those biases as well, right? Which people of majority using those agents will incorporate and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:11",
                    "end": "00:30:40",
                    "text": " inculcate into the demand and supply pipeline itself. I think we got it. There are one answer so we can, I know many of you would have answers, but let's just take one, whoever you want to go ahead. Look, we already live in an information bubble to some degree, okay? Many algorithms feed what you want to watch, and if you click on it, you watch again. Having said that, you know, what will evolve here is what makes sense both for the merchant to sell and for the consumer to buy. And the idea is that your niche products and services"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:40",
                    "end": "00:31:08",
                    "text": " algorithms have to be attuned to that, they have to follow it, you have to generate that data, and eventually those algorithms may cater to you depending on the kind of data you feed them. Okay, next question or comment, whoever gets the microphone, I'm sorry I'm not controlling the microphone, so whoever gets it. Question or comment? Question. I'm the founder of Karada.ai. The question I have..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:08",
                    "end": "00:31:36",
                    "text": " One thing that we are solving for is discoverability of MCPs and all that. We do auto-generation, creation, and deployment of MCP, ACP, to agent. And in the discovery phase, what we want is microtransactions. For example, let's say iLovePDF creates an MCP using our platform. And some user types into their AI that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:36",
                    "end": "00:32:02",
                    "text": " convert this PDF to JPG. I want iLovePDF to make a small transaction that I could take from the user and give to iLovePDF. There are solutions in the crypto space like x402 that is there, but we have all major transaction companies here. So are there any traditional..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:02",
                    "end": "00:32:28",
                    "text": " solutions that are coming. Okay, one answer. This is a classical case, like agent to agent payments. There's a lot of work going on on agent to agent payments. Literally, you need protocols, tokens. My bet is it will get solved in six months. Yeah, okay. Can we also have one microphone on this side? Otherwise, all are going that side. Is there a microphone this side? Hello?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:28",
                    "end": "00:32:49",
                    "text": " Yes, I have a question. After this, can you please take it this side and then we still have another 10 minutes. Yeah. My name is Piyush. I have a very short question. How do you see the role of biometrics in the world of agentic payments? Role of biometrics? Who wants to take it up? I can try. It ties back to the identity."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:49",
                    "end": "00:33:10",
                    "text": " And also authorization, right? If I can exactly know who you are and authenticate you, life becomes much, much easier. So it goes back to other. Agents don't have biometrics, so it's not a thing. We have biometrics, but they need an equivalent, and I agree with you on that."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:10",
                    "end": "00:33:34",
                    "text": " OK, who's next? Hi, this is Neeraj. Quick question. So in today's world, a customer can call up for a dispute. In tomorrow's world, how would a customer having a Citibank card running on MasterCard and a merchandise on PayPal, how would the agents dispute in the agent equal? They'll work it out like we did. But I think get used to calling agents and get used to agents answering you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:34",
                    "end": "00:34:04",
                    "text": " Or in the near future, get used to your agent calling other agents and other agents calling your agents. It's all coming to a cinema near you. So wait for it. All right. OK. We also have heard this thing and get some lady questions. After this, can we have some ladies or some elder people maybe ask? After this, yeah. Go ahead, please. Question? Yeah, I have a follow-up question on the dispute management process."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:04",
                    "end": "00:34:33",
                    "text": " Let's say an agent does a transaction with the customer, right? So what kind of record-keeping will you do which will aid in the dispute management process? What are your thoughts on dispute management evaluation? Good point. That goes to the earlier discussion of keeping track and traces and blockchain, but please do answer. I mean, the dispute, so the merchant of, who is the merchant of record is very important. So if a human did not authorize it, then somebody has to underwrite that risk."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:33",
                    "end": "00:35:03",
                    "text": " Right like and so and so back in e-commerce days literally just why PayPal was formed because we could underwrite that risk Right like and so in agentic commerce well somebody has to underwrite that risk because bad actors will happen right like it So so that's that's the answer actually more generally for everyone liability and where it sits becomes a key component so that question probably could be Differently said that you know where is the liability in this transaction? And this is kind of an emerging This is an emerging field to some degree if you buy a car"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:03",
                    "end": "00:35:32",
                    "text": " and you drive out with the car and then you crash the car, that's your problem. But if there's a mechanical failure in a car, that's the problem of the owner and you can still go back to them. With agentic AI or with generative AI, this is becoming a bit more complicated to separate out. If you provide a foundation model and you at the end user are using it or giving it to your clients, then you are responsible for some aspect of it. But if you fine tune the model,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:32",
                    "end": "00:35:58",
                    "text": " then you have now suddenly become more responsible for the outcome of the model. And those of you who have been following this, there have been some really, really interesting cases. There was an airlines company in Canada, I won't name the company, but the point was they had a chat pod on their website that gave wrong advice to somebody to buy a plane ticket now and we'll reimburse you later. Actually, that was not the policy."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:58",
                    "end": "00:36:23",
                    "text": " The policy was you have to reach out to them first for this emergency ticket, and then you can buy it. So they took him to court. The company said, this is nothing to do with us. It's a chatbot from a third party. You know, not our chatbot. We didn't train it with the model. And the judge said, it's your chatbot, sir. It's on your website. So liability is interesting. So as of now, it seems to me you may have agents working for you, but you're still liable."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:23",
                    "end": "00:36:42",
                    "text": " Just like in commerce, you had liability. You will have a third party in between. Quick question on cyber security and risk. I think we can discuss that a little bit more. I can't think of a major company who's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:42",
                    "end": "00:37:11",
                    "text": " data was not breached and did not have cybersecurity issue in the recent past. So we talk about AI agents, but there are also rogue AI agents which have better incentives to game the system. So how do you deal with that? We're seeing that even biometrics are being compromised to a certain extent. Somebody asked that wonderful question. So how do you make those authentication layers and trust layers more and more secure? And how do you deal with cybersecurity? Excellent question. If you can answer, I'm told we have to leave."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:11",
                    "end": "00:37:24",
                    "text": " Agents we are thinking they're all doing good stuff for us But she's saying hey look they could be rogue agents and they could be yeah, so the top security cyber security Consortium has found the top ten security breaches"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:24",
                    "end": "00:37:52",
                    "text": " Of the top 10 security breaches, number one was a misuse of credentials. Number two was publishing on behalf of you without your consent on a public website, like a social website. And then number three was going out there and crawling websites and getting infected and bringing the infection back home. So that's the trifactor that they talked about in the security councils. The fourth most potent one is this concept of persistent memory."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:52",
                    "end": "00:38:22",
                    "text": " So in addition to all of these things that you would have in the software world, you bring additional risks into the AI world. This is why it is super important to understand the birth and death of the agent. How was the agent born? How is the agent transacting? What data are we keeping? At what point of time did it turn toxic? And because of the earlier gentleman's question, the disputes with everything, the world we lived in before agents was fragmented data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:22",
                    "end": "00:38:50",
                    "text": " Now we have data, and it's living on a foundation. So from the minute that data is born, like you can promote and demote the data, so you know all the details, and so it's more controllable when you know the data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:50",
                    "end": "00:39:16",
                    "text": " You want to show some enthusiasm and give FOMO to the people outside. Clap louder, they won't know what's happening here. And the organizers are awesome. They took that photo in the beginning and now they've created a memento out of that photograph. So another clap for all the people behind the scenes, from the security guards to the people out here and so on."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6992fe894c5e8f0974c876dd",
    "projectId": "917933",
    "projectName": "Redesigning the AI Economy_ Ethical Data Pipelines at National Scale",
    "projectDescription": "Redesigning the AI Economy_ Ethical Data Pipelines at National Scale",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-16T11:24:55.881Z",
    "projectUpdatedDate": "2026-02-16T11:24:55.881Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-16T11:24:55.881Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "Redesigning the AI Economy_ Ethical Data Pipelines at National Scale(MP3_128K).mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771240935932.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:11",
                    "text": " और में को प्रवलम नहीं नहीं चाल बड़ा लाज नहीं ही है।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:11",
                    "end": "00:00:42",
                    "text": " अम्में चालू कर रहा था उन्हें। स्टार्टिंग उन्हें।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:42",
                    "end": "00:01:11",
                    "text": " आगिन तोल दिन स्टाटिं तुम्मिद्स अब अब लग तुम्मिद्स के लिए आप आप आप आप आप तुम्मिद्स करें।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:21",
                    "end": "00:02:17",
                    "text": " पिजास्टिक। मेरे तो भिख लिए है।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:17",
                    "end": "00:02:32",
                    "text": " सूपर्इक्साइटिट्टु वार्ख रहे हैं। वेरी वेरी हाँपी हैं। सूब हाँपी देखे हैं। सूब हाँपी देखे हैं। सूब हाँपी देखे हैं। सूब हाँपी देखे हैं। सूब हाँपी देखे हैं। सूब हाँपी देखे है।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:32",
                    "end": "00:02:57",
                    "text": " साउट कार एक साउट के बभी आगे हैं। किस चेज़ के लिए हैं? किस चेज़ के लिए हैं? किस चेज़ के लिए हैं? किस चेज़ के लिए हैं?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:57",
                    "end": "00:03:27",
                    "text": " आपको सामसार्ड और एक सलार्ड के लिए आगर था। यह एक फाई है, इच्मी त्रांस्लेशण हो अगर स्ट्रेइन में दिखे रहा है। तो यह एक सामसार्ड और एक सलार्ड के लिए आगर था। आप ही नहीं दिखाना हैं, जब दिखाना हूं और में भुता है। मैं इसक"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:27",
                    "end": "00:03:52",
                    "text": " And today I am honored to join with two other people on the panel, who have not only thought about it, but have actually led India to make better decisions in the AI space. I have Sunaina ji with me from Microsoft Research. Sunaina and I have known each other for 9 years, which is incredible. I also think Kalika, Dr. Kalika is there. No, no, E-X is not there."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:52",
                    "end": "00:04:21",
                    "text": " दॉक्रिबाल्यो और आपकुर्स आपकुर्स आपकुर्स आपकुर्स आपकुर्स आपकुर्स आपकुर्स आपकुर्स आपकुर्स आपकुर्स"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:21",
                    "end": "00:04:46",
                    "text": " इंडिया के लिए पूर्ग्रसिक पाब्लिक पोलिसी के लिए हैं। माल्टालिंग्मुल एआइन के लिए हैं। अपने आपके लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:46",
                    "end": "00:05:34",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:34",
                    "end": "00:06:02",
                    "text": " आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आपके आ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:02",
                    "end": "00:06:31",
                    "text": " So here we are really looking at how communities can be part of the evaluation process itself. And if you have been following some of the AI news, then you know that evaluations have become very big over the last year or so. Everyone is talking about building benchmarks, even in the previous session by Gates Foundation where they were looking at agriculture, they mentioned how important it is to have benchmarks and evaluate for impact and so on."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:31",
                    "end": "00:07:01",
                    "text": " और आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:01",
                    "end": "00:07:30",
                    "text": " तास्क के लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:30",
                    "end": "00:07:60",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:60",
                    "end": "00:08:28",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:28",
                    "end": "00:08:56",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:56",
                    "end": "00:09:26",
                    "text": " And also most importantly, evaluations create accountability. At least in the short term, there is signal to show us whether something is working as intended or not. And in the long term, we need to follow that up with longer term impact evaluations and so on. So here in this talk, I am going to be focusing mainly on model level evaluation. But I also recognize that system evaluation, user evaluation and impact evaluation is extremely critical to know whether these technologies will really be adopted and whether they will have the outcomes that we want them to have."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:26",
                    "end": "00:09:52",
                    "text": " So, what makes AI evaluation, particularly you know, GenAI or LLM evaluation so challenging to them, right? As we all know, the outputs of these systems are extremely hard to find, right? And so we don't really know what they will say given a response or given a question or a small tweak to the question, you can have a completely different response, right? And so this makes it really hard to build an evaluation data set with which you can be sure that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:52",
                    "end": "00:10:11",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:11",
                    "end": "00:10:39",
                    "text": " और अपने आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:39",
                    "end": "00:11:07",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:07",
                    "end": "00:11:37",
                    "text": " अपने लिए लिए लिए लिए लिए लिए लिए लिए"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:37",
                    "end": "00:12:09",
                    "text": " प्रोटोपाल्ज के लिए आपके लिए करें। और आपके लिए आपके लिए प्रोटोपाल्ज के लिए आपके लिए करें।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:09",
                    "end": "00:12:27",
                    "text": " अपने इवालुइषण मेथार्थ के लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:27",
                    "end": "00:12:52",
                    "text": " And found that they were flawed in some way. Either the data was flawed or the method that they were using, the metrics were flawed or the interpretation was flawed. What they interpreted the outcome to be from the study itself was flawed. Similarly, AI agents have been found to cheat on evaluation. So this was a NIST report that showed that some of these agents actually look up the answer on the web while the evaluation is going on."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:52",
                    "end": "00:13:21",
                    "text": " And so they try to cut off the web access when they are doing evaluation and so on. It's a really interesting read. I highly recommend you check it out. Similarly, there have been critiques of benchmarks themselves where, as I said, the interpretation of a benchmark score is not valid because the benchmark is very constrained in some way. There's a very, very popular paradigm called the LLM as judge paradigm where an LLM or a JNAI model itself does the evaluation for you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:21",
                    "end": "00:13:51",
                    "text": " यह आट्ट्स लिए आट्ट्स लिए आट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट्ट।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:52",
                    "end": "00:14:21",
                    "text": " नहीं जब जब ज़ेनरली अबार्ट इवालुइशण रहा हैं और यह चालेंजी हैं। और ज़ेनरली अबार्ट इवालुइशण रहा हैं और यह चालेंजी हैं। और यह चालेंजी हैं। और यह चालेंजी हैं। और यह चालेंजी हैं। और यह चालेंजी हैं। और यह चाल"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:21",
                    "end": "00:14:50",
                    "text": " और हमें भी लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:51",
                    "end": "00:15:16",
                    "text": " आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:16",
                    "end": "00:15:44",
                    "text": " How that system is actually going to be used in India, how that model is going to be used in India, how well it can work in the Indian scenario of culture, context and so on. And this is actually a problem with most benchmarks today that look at multilingual evaluation. And this is something that we've really been trying to change over the last couple of years that we need to build benchmarks separately for each language, culture and context. If we really want to be able to tell how well these models will do on those languages, culture and context."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:44",
                    "end": "00:16:08",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:08",
                    "end": "00:16:36",
                    "text": " आपके लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:36",
                    "end": "00:17:05",
                    "text": " और आप आप नहीं लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए लिए।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:05",
                    "end": "00:17:30",
                    "text": " And trying to ensure that there is trust and rigor in the benchmarking process over the last few months where we have been building Samiksha. And before Samiksha, I just want to show you another project that we did with Karya, which was last year, last year to last year now, which is called Pariksha. And so that's again, you know, a word that means exam. And this was actually the very first time that we engaged Karya workers in doing the evaluation process."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:30",
                    "end": "00:18:00",
                    "text": " आप आप आपकार्या के लिए आपकार्या करते हैं, आपकार्या के लिए आपकार्या करते हैं। यह आपकार्या के लिए आपकार्या करते हैं। यह आपकार्या के लिए आपकार्या करते हैं। यह आपकार्या के लिए आपकार्या करते हैं। यह आपकार्या के लिए आपकार्"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:00",
                    "end": "00:18:21",
                    "text": " So we are also working on automated methods in order to try and make them closer to human evaluation as well. Because there is also a need for some automation over here. So with that, I want to introduce Samik Shah, which is what we are all here for. And this is the new benchmark that we've created for Indian languages, contexts and cultures."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:21",
                    "end": "00:18:51",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:51",
                    "end": "00:19:21",
                    "text": " आप आप नहीं नहीं नहीं नहीं नहीं नहीं नहीं नहीं नहीं नहीं नहीं"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:21",
                    "end": "00:19:51",
                    "text": " हमें कार्या के 150,000 कार्या के लिए कार्या के लिए कार्या के लिए कार्या के लिए कार्या के लिए कार्या के लिए कार्या के लिए कार्या के लिए कार्या के लिए कार्या के लिए कार्या के लिए कार्या के लिए कार्या के लिए कार्या के लिए कार्या के लिए कार्या के लिए क"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:51",
                    "end": "00:20:15",
                    "text": " We did 11 languages. We want to go to 22 languages and beyond in the next iteration. But we've also done a lot of depth. So we've really analyzed all of the examples. We've done lots of statistical tests. You can find all of that in the report. We've also done a lot of qualitative analysis to look at each data point and see why it was rated a certain way, what kind of score it got, what the problem was with the answer and so on."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:15",
                    "end": "00:20:40",
                    "text": " So we hope that that will really be useful to model developers to try and figure out why these models are doing well or not so well in specific languages and settings. So this is how we built the benchmark. So there were three different phases of our benchmark and the idea here was really to involve the community in multiple ways in order to construct the benchmark. The very first way that we reached out to communities is through civil society organizations."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:41",
                    "end": "00:21:02",
                    "text": " So in order to determine what goes into the benchmark, we actually leaned on the knowledge and experience of CSOs that work in specific domains. Some of them already have AI based solutions in these languages and markets. Some of them are about to deploy and so they have a really good idea of what kinds of questions people are asking AI models today."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:02",
                    "end": "00:21:30",
                    "text": " And our benchmark has four domains in it. And we are going to expand all the domains in the future as well. Right now we are focusing on healthcare, education, legal and finance. And these are not questions that say a lawyer would be asking an AI system. But this is the questions from the community. To somebody who has say a legal dispute with somebody like their neighbor or something is asking a question to an AI model and advice seeking question. Those are the kinds of questions that we are capturing in Samiksha."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:30",
                    "end": "00:21:56",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:56",
                    "end": "00:22:23",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:23",
                    "end": "00:22:49",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:49",
                    "end": "00:23:18",
                    "text": " आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:18",
                    "end": "00:23:46",
                    "text": " और अच्छाने के लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:51",
                    "end": "00:24:21",
                    "text": " Yes, so in finance the questions and I'll show you some examples of questions. So finance the questions are like should I do this kind, should I invest in this or should I take this loan and things like that. So these are everyday questions that people would ask, advice seeking questions that they would ask. And the kind of model errors that we find are, we still find that there are some models that use awkward languages that almost look translated. They translate frozen expressions from English. These are the things that raters typically rate."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:21",
                    "end": "00:24:50",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:50",
                    "end": "00:25:18",
                    "text": " और आपका कार्या इवालुइशन्स के लिए हैं. और अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:18",
                    "end": "00:25:39",
                    "text": " आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:39",
                    "end": "00:26:06",
                    "text": " And the interesting thing is that the answer that did well over here, sorry, the answer that did poorly over here was geometric assessments, job shadowing, industry visits, internships and so on. And so this did not, the person who evaluated, who was also a member of the community, did not connect with this answer because it didn't seem relevant to them. So these are the kinds of differences we see between models that do well and not. And similarly on the right, if you look at the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:06",
                    "end": "00:26:33",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:33",
                    "end": "00:26:57",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:57",
                    "end": "00:27:22",
                    "text": " आपका बांगलाडेश का आपका बांगलाडेश का बांगलाडेश का बांगलाडेश का बांगलाडेश का बांगलाडेश का बांगलाडेश का बांगलाडेश का बांगलाडेश का बांगलाडेश का बांगलाडेश का बांगलाडेश का बांगलाडेश का बांगलाडेश का ब"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:22",
                    "end": "00:27:36",
                    "text": " In some cases, the model assumed that it was a Bangladeshi context. But in other cases, it actually mixes the two contexts. And so it is referencing this objection from both India and Bangladesh in the same answer, which causes a lot of confusion. So this is a low-scoring answer for us."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:36",
                    "end": "00:28:06",
                    "text": " और आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:06",
                    "end": "00:28:34",
                    "text": " और इसकार्या के लिए इसकार्या के लिए इसकार्या के लिए इसकार्या के इसकार्या के इसकार्या के इसकार्या के इसकार्या के इसकार्या के इसकार्या के इसकार्या के इसकार्या के इसकार्या के इसकार्या के इसकार्या के इसकार्या के इसकार्या के इसकार्या के इसकार्य"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:34",
                    "end": "00:29:36",
                    "text": " So I'll end here with the link to the page that has the report as well as the leaderboards for you to look at. Thank you so much and I'm happy to take questions now or later."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:48",
                    "end": "00:30:21",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:21",
                    "end": "00:30:50",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:50",
                    "end": "00:31:19",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:19",
                    "end": "00:32:22",
                    "text": " और क्यों आप आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:22",
                    "end": "00:32:49",
                    "text": " 6 to 7 languages occupying the most of the digital data which was available, while as rest of the languages did not have any digital data, and even the 6 to 7 languages which were there, which had digital data, they were approximately, you know, about 2% or 3%, you know, anybody can guess that amount of digital data which was available."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:49",
                    "end": "00:33:18",
                    "text": " अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:18",
                    "end": "00:33:47",
                    "text": " आपलिकेजियों के लिए आपलिकेजियों के लिए आपलिकेजियों के लिए आपलिकेजियों के लिए आपलिकेजियों के लिए आपलिकेजियों के लिए आपलिकेजियों के लिए आपलिकेजियों के लिए आपलिकेजियों के लिए आपलिकेजियों के लिए आपलिकेजियों"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:48",
                    "end": "00:34:14",
                    "text": " And they speak, they transcribe, they translate and they move all of that which is there to build the digital data. With that we built the corpus which was barely sufficient to build the models and obviously we could have waited for very long to actually see that these models have enough data to become 100% accurate and then we moved on to the fields."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:14",
                    "end": "00:34:43",
                    "text": " और हम आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आ�"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:43",
                    "end": "00:35:11",
                    "text": " पंचायत्स के लिए आपने पंचायत्स के लिए आपने पंचायत्स के लिए आपने पंचायत्स के लिए आपने पंचायत्स के लिए आपने पंचायत्स के लिए आपने पंचायत्स के लिए आपने पंचायत्स के लिए आपने पंचायत्स के लिए आपने पंचायत्स के लिए"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:11",
                    "end": "00:35:41",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:41",
                    "end": "00:36:09",
                    "text": " अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:09",
                    "end": "00:36:35",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:35",
                    "end": "00:37:04",
                    "text": " अपने आपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:04",
                    "end": "00:37:31",
                    "text": " अपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आपने आ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:32",
                    "end": "00:37:59",
                    "text": " So we had to go into a specific tribe of involving the agriculture students to actually collect those addresses to pay them on so that things start moving fine in that direction. So that's another intervention which we worked upon. The third intervention was of course where we touch every citizen across the country is UID and where the place names are translated and other names are translated."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:59",
                    "end": "00:38:27",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:27",
                    "end": "00:38:54",
                    "text": " 16 lakhs to 18 lakhs and currently those are not in the digital center and so we have to bring that back and hence we are working with survey of India and then again those places have to be on the local dialect because those places are known on the local dialect to where the people eat there and the idea of creating that particular database is to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:54",
                    "end": "00:39:23",
                    "text": " आपका आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:23",
                    "end": "00:39:54",
                    "text": " वाल्यू चेन अपने लिए चाहिंच पर लिए चाहिंच पर लिए चाहिंच पर लिए चाहिंच पर लिए चाहिंच पर लिए चाहिंच पर लिए चाहिंच पर लिए चाहिंच"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:54",
                    "end": "00:40:22",
                    "text": " अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:22",
                    "end": "00:40:50",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:50",
                    "end": "00:41:21",
                    "text": " आप स्टार्टे लेरिंग तुट्राती हैं। तो आप नहीं आपने कुछ करता हैं। तो आपने कुछ करता हैं। तो आपने कुछ करता हैं। तो आपने कुछ करता हैं। तो आपने कुछ करता हैं। तो आपने कुछ करता हैं। तो आपने कुछ करता हैं। तो आपने कुछ कर"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:21",
                    "end": "00:41:50",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:50",
                    "end": "00:42:18",
                    "text": " आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको आपको।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:18",
                    "end": "00:42:49",
                    "text": " आद्रवाइज आद्रवाइज आद्रवाइज आद्रवाइज आद्रवाइज आद्रवाइज आद्रवाइज आद्रवाइज आद्रवाइज आद्रवाइज"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:49",
                    "end": "00:43:18",
                    "text": " आपने आपने आपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:18",
                    "end": "00:43:46",
                    "text": " आपका देटावार्कर के लिए वाल्यू चेन के लिए वाल्यू चेन के लिए वाल्यू चेन के लिए वाल्यू चेन के लिए वाल्यू चेन के लिए वाल्यू चेन के लिए वाल्यू चेन के लिए वाल्यू चेन के लिए वाल्यू चेन के लिए वाल्यू चेन के लिए वाल्यू"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:43:46",
                    "end": "00:44:16",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:16",
                    "end": "00:44:45",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:44:45",
                    "end": "00:45:16",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:16",
                    "end": "00:45:43",
                    "text": " और वेरी वेरी इक्साइटिन बाद समख दराई, सर्थ बाट प्रॉजेक वेरोद माहाराश्ट्र। और वेरी वेराश्ट्र। और वेरी वेराश्ट्र। और वेरी वेराश्ट्र। और वेरी वेरी वेरी वेरी वेरी वेरी वेरी वेरी वेरी वेरी वेरी वेरी वेरी वेरी व"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:45:43",
                    "end": "00:46:12",
                    "text": " आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:12",
                    "end": "00:46:52",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:46:52",
                    "end": "00:47:20",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:20",
                    "end": "00:47:45",
                    "text": " और में समिक्षा के लिए कुछ है। और हम आप आपके लिए कुछ के लिए कुछ है। और हम आपके लिए कुछ के लिए कुछ के लिए कुछ है। और हम आपके लिए कुछ के लिए कुछ हैं। और हम आपके लिए कुछ के लिए कुछ हैं। और हम आपके लिए कुछ के लिए"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:47:45",
                    "end": "00:48:13",
                    "text": " और अच्छानी प्राइर्यार के लिए प्राइर्यार के लिए प्राइर्यार के लिए प्राइर्यार के लिए प्राइर्यार के लिए प्राइर्यार के लिए प्राइर्यार के लिए प्राइर्यार के लिए प्राइर्यार के लिए प्राइर्यार के लिए प्राइर्यार के लिए प"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:13",
                    "end": "00:48:42",
                    "text": " आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:48:42",
                    "end": "00:49:08",
                    "text": " अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:08",
                    "end": "00:49:37",
                    "text": " And from her perspective, we'll actually get to learn about what does community-centered AI mean. For her, I also want to call out Shannon Farley, who is in the room. I think many of us know and love Shannon, but Shannon runs Fast Forward, one of the most prolific AI for good investors in the world. Shannon, I want to quickly bring you in. Could you talk to us about what were the first two of you when the first checks into Tarya, into CIP, into Nura Health, into Rocket Learning, so many organizations that have been highlighted at the summit today."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:49:37",
                    "end": "00:50:01",
                    "text": " जब आप आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:01",
                    "end": "00:50:16",
                    "text": " इंदियन अर्गिनिसाइशन्स के लिए अपने लिए अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने अपने"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:16",
                    "end": "00:50:45",
                    "text": " इंफ्रास्ट्रॉक्षर के लिए आप लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए आपके लिए"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:50:45",
                    "end": "00:51:14",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:14",
                    "end": "00:51:45",
                    "text": " अंट्रप्रणूर्शिक और फिलान्थर्पी के विश्ट करता हैं। आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपक"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:51:45",
                    "end": "00:52:07",
                    "text": " आप आप आप आप आप आप आप आप आप आप आप आप आप आप"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:07",
                    "end": "00:52:28",
                    "text": " सब्रदाय अर्फाँस्यू और हाँ आप रिलाइपर्टेगा। और हाँ दिज्टिल वो कपच्रेच्यों दिर्ट्रेट्यों दिर्ट्रेट्यों दिर्ट्रेट्यों दिर्ट्रेट्यों दिर्ट्रेट्यों दिर्ट्रेट्यों दिर्ट्रेट्यों दिर्ट्रेट्यों दिर्ट्र"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:28",
                    "end": "00:52:59",
                    "text": " आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप आप"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:52:59",
                    "end": "00:53:06",
                    "text": " अच्छानी प्राज़ाइंगा प्राज़ाइंगा"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:06",
                    "end": "00:53:36",
                    "text": " वो तो लेके आओ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:36",
                    "end": "00:53:54",
                    "text": " पालूप्ते लिए आपके लिए के रहा है।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:53:54",
                    "end": "00:54:23",
                    "text": " तेखे तेखे आरी आरी"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:23",
                    "end": "00:54:33",
                    "text": " अच्छा पेखाँ पेखाँ"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:33",
                    "end": "00:54:55",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:54:55",
                    "end": "00:55:21",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:21",
                    "end": "00:55:42",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:55:42",
                    "end": "00:56:10",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:10",
                    "end": "00:56:18",
                    "text": " across Kenya, South Africa, Ethiopia, the US, Indonesia. So please check out the mastercard book."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:18",
                    "end": "00:56:44",
                    "text": " भाश्मी के लिए भाश्मी के लिए भाश्मी के लिए भाश्मी के लिए भाश्मी के लिए भाश्मी के लिए भाश्मी के लिए भाश्मी के लिए भाश्मी के लिए भाश्मी के लिए"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:56:52",
                    "end": "00:57:21",
                    "text": " कारिया टीम से कोई सीने मांगा तो आपका से नहीं है। कौन सी टीम से? कारिया के देगी से, ये आउर्गनाइसिंग टीम. तीक, आउर्गनाइसिंग टीम से कोई मांगे रहा हो, देना है या मना कर देना है? देना है या मना कर देना है? नहीं, देना है नहीं. यहीं वाली फा"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:57:22",
                    "end": "00:57:45",
                    "text": " आप आप आप आप आप आप आप आप"
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6992f1aa4c5e8f0974c87513",
    "projectId": "141003",
    "projectName": "Sovereign AI and National Security: India’s Digital Path - Hindi",
    "projectDescription": "Sovereign AI and National Security: India’s Digital Path - Hindi",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-16T10:30:06.327Z",
    "projectUpdatedDate": "2026-02-16T10:30:06.327Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-16T10:30:06.327Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO_LINK",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames"
        },
        {
            "col1": "P1",
            "col2": "https://youtu.be/58V4tbgvnwc",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEOLINK",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:04",
                    "text": "You saw Russia Ukraine war was going on."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:02",
                    "end": "00:00:07",
                    "text": "Swift was stopped for Russia. So you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:04",
                    "end": "00:00:09",
                    "text": "will have to with the way we own digital"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:07",
                    "end": "00:00:11",
                    "text": "public infrastructure in India. We'll"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:09",
                    "end": "00:00:13",
                    "text": "have to create a cognitive public"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:11",
                    "end": "00:00:15",
                    "text": "infrastructure. How soever difficult it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:13",
                    "end": "00:00:17",
                    "text": "is and that's why pair it's visceral"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:15",
                    "end": "00:00:19",
                    "text": "it's existential and we will have to do"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:17",
                    "end": "00:00:22",
                    "text": "it and probably it's time that we should"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:19",
                    "end": "00:00:23",
                    "text": "do it. I yesterday wrote an article in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:22",
                    "end": "00:00:25",
                    "text": "uh the Sunday gajin which where where I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:23",
                    "end": "00:00:28",
                    "text": "called it the the architecture of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:25",
                    "end": "00:00:30",
                    "text": "independence right it is the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:28",
                    "end": "00:00:32",
                    "text": "architecture of independence. There's no"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:30",
                    "end": "00:00:34",
                    "text": "der of you know experts in India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:32",
                    "end": "00:00:36",
                    "text": "There's no der of talent in India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:34",
                    "end": "00:00:38",
                    "text": "There's no der of only thing is compute"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:36",
                    "end": "00:00:41",
                    "text": "which I think geopolitically we should"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:38",
                    "end": "00:00:44",
                    "text": "solve. That again is sovereignty right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:41",
                    "end": "00:00:46",
                    "text": "because people are using GPUs to control"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:44",
                    "end": "00:00:49",
                    "text": "geopolitics"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:46",
                    "end": "00:00:51",
                    "text": "and that is why we will have to be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:49",
                    "end": "00:00:52",
                    "text": "sovereign at all you know all the layers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:51",
                    "end": "00:00:52",
                    "text": "of the cake"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:52",
                    "end": "00:00:55",
                    "text": ">> right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:52",
                    "end": "00:00:57",
                    "text": ">> and I believe that uh when you apply"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:55",
                    "end": "00:01:01",
                    "text": "these systems to policing you will have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:57",
                    "end": "00:01:03",
                    "text": "to go for trust and transparency"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:01",
                    "end": "00:01:04",
                    "text": "uh and it has created problems on the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:03",
                    "end": "00:01:07",
                    "text": "other side it's become a force"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:04",
                    "end": "00:01:09",
                    "text": "multiplier it has hugely enabled"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:07",
                    "end": "00:01:12",
                    "text": "criminal ecosystem we polymorphic"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:09",
                    "end": "00:01:14",
                    "text": "malware. You have, you know, there was a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:12",
                    "end": "00:01:17",
                    "text": "a company called Arup. I think it's a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:14",
                    "end": "00:01:20",
                    "text": "advertisement company or something where"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:17",
                    "end": "00:01:22",
                    "text": "the CEO came on a live zoom call and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:20",
                    "end": "00:01:26",
                    "text": "they lost uh $30 million. Right? This is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:22",
                    "end": "00:01:28",
                    "text": "a state at which AI is evolving. Do we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:26",
                    "end": "00:01:30",
                    "text": "have realtime policing? Sir, you know,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:28",
                    "end": "00:01:32",
                    "text": "explained that policing is not so real"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:30",
                    "end": "00:01:36",
                    "text": "time. So, you will have to increase both"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:32",
                    "end": "00:01:38",
                    "text": "capability and capacity and this needs a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:36",
                    "end": "00:01:40",
                    "text": "serious thought right now. So pay we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:38",
                    "end": "00:01:43",
                    "text": "need sovereign AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:40",
                    "end": "00:01:46",
                    "text": ">> Thank you. Thank you sir. And again as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:43",
                    "end": "00:01:49",
                    "text": "expected very very relevant points sir"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:46",
                    "end": "00:01:51",
                    "text": "that reminds me of of interview by"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:49",
                    "end": "00:01:53",
                    "text": "Jensen Huang a few years back. He said I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:51",
                    "end": "00:01:56",
                    "text": "met your prime minister all before this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:53",
                    "end": "00:01:58",
                    "text": "AI hype uh when it's about in very"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:56",
                    "end": "00:01:59",
                    "text": "initial stages and he said your prime"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:58",
                    "end": "00:02:02",
                    "text": "minister told me that I don't want to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:59",
                    "end": "00:02:04",
                    "text": "export data to import intelligence. He"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:02",
                    "end": "00:02:07",
                    "text": "said that was really remarkable for a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:04",
                    "end": "00:02:08",
                    "text": "world leader to think in those times. So"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:07",
                    "end": "00:02:11",
                    "text": "I think very relevant points you also"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:08",
                    "end": "00:02:13",
                    "text": "echoed but good that sir you brought"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:11",
                    "end": "00:02:14",
                    "text": "compute in in picture because compute is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:13",
                    "end": "00:02:16",
                    "text": "something where we really have to make"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:14",
                    "end": "00:02:19",
                    "text": "pro in fact the rest of the world and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:16",
                    "end": "00:02:22",
                    "text": "then I would love to bring pair maybe at"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:19",
                    "end": "00:02:24",
                    "text": "later stage. So uh with this then I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:22",
                    "end": "00:02:26",
                    "text": "would like to move to Martin. Martin"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:24",
                    "end": "00:02:28",
                    "text": "thanks a lot uh first of all traveling"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:26",
                    "end": "00:02:31",
                    "text": "all the way to India to attend the AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:28",
                    "end": "00:02:35",
                    "text": "summit and being with us. M in your role"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:31",
                    "end": "00:02:38",
                    "text": "as uh in terod data as a leader in in in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:35",
                    "end": "00:02:41",
                    "text": "data analytics big data implementation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:38",
                    "end": "00:02:43",
                    "text": "uh you have been quoted twice in data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:41",
                    "end": "00:02:45",
                    "text": "IQ's 100 most influential people in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:43",
                    "end": "00:02:47",
                    "text": "datadriven business and what's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:45",
                    "end": "00:02:49",
                    "text": "remarkable I was looking at your website"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:47",
                    "end": "00:02:53",
                    "text": "where you are now saying that uh stop"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:49",
                    "end": "00:02:55",
                    "text": "managing data activate intelligence so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:53",
                    "end": "00:02:57",
                    "text": "and I also read your article about the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:55",
                    "end": "00:02:58",
                    "text": "large scale compute requirement which is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:57",
                    "end": "00:03:01",
                    "text": "going to come for example you give a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:58",
                    "end": "00:03:03",
                    "text": "healthcare example where every patient"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:01",
                    "end": "00:03:05",
                    "text": "will be followed by an agent and he'll"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:03",
                    "end": "00:03:07",
                    "text": "be firing like thousands hundreds and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:05",
                    "end": "00:03:09",
                    "text": "thousands of queries to make sure that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:07",
                    "end": "00:03:12",
                    "text": "you have that hypers uh personalization"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:09",
                    "end": "00:03:15",
                    "text": "experience. So with that kind of compute"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:12",
                    "end": "00:03:18",
                    "text": "requirement Martin do you think that uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:15",
                    "end": "00:03:21",
                    "text": "we are ready for so sovereignity from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:18",
                    "end": "00:03:24",
                    "text": "day one or is going or as everybody said"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:21",
                    "end": "00:03:25",
                    "text": "you have to tackle it layer wise so what"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:24",
                    "end": "00:03:28",
                    "text": "do you think especially when you come to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:25",
                    "end": "00:03:30",
                    "text": "the layer of data you know how the data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:28",
                    "end": "00:03:32",
                    "text": "structures uh a country like India can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:30",
                    "end": "00:03:35",
                    "text": "make sure that they are sovereign in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:32",
                    "end": "00:03:36",
                    "text": "those aspects yeah so over to you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:35",
                    "end": "00:03:38",
                    "text": ">> thank you and uh and good morning"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:36",
                    "end": "00:03:40",
                    "text": "everybody lovely to be here um there's a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:38",
                    "end": "00:03:41",
                    "text": "lot to unpack in your question and we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:40",
                    "end": "00:03:43",
                    "text": "should come back and we should talk"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:41",
                    "end": "00:03:46",
                    "text": "about the implications of potentially"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:43",
                    "end": "00:03:47",
                    "text": "millions of agents running thousands of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:46",
                    "end": "00:03:49",
                    "text": "queries because I think that is a very"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:47",
                    "end": "00:03:50",
                    "text": "important consequence uh of where we're"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:49",
                    "end": "00:03:52",
                    "text": "going that we should think about. But"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:50",
                    "end": "00:03:54",
                    "text": "you're absolutely right at Terodata we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:52",
                    "end": "00:03:56",
                    "text": "think that the first and the most"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:54",
                    "end": "00:03:58",
                    "text": "fundamental layer is the data layer."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:56",
                    "end": "00:03:60",
                    "text": "There is no good AI without good data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:58",
                    "end": "00:04:02",
                    "text": "So it's incredibly important we get that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:60",
                    "end": "00:04:03",
                    "text": "right. And there are some principles"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:02",
                    "end": "00:04:06",
                    "text": "that have proven to be more or less"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:03",
                    "end": "00:04:08",
                    "text": "timeless. And if I speak about timeless"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:06",
                    "end": "00:04:09",
                    "text": "in in terms of it, I'm speaking they've"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:08",
                    "end": "00:04:11",
                    "text": "proven to be good over three or four"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:09",
                    "end": "00:04:13",
                    "text": "decades. That's the kind of event"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:11",
                    "end": "00:04:15",
                    "text": "horizon, isn't it, for technology. Um,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:13",
                    "end": "00:04:17",
                    "text": "so we need to focus on capturing the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:15",
                    "end": "00:04:18",
                    "text": "most detailed data that we can. We need"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:17",
                    "end": "00:04:21",
                    "text": "to make sure that we are capable of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:18",
                    "end": "00:04:23",
                    "text": "integrating that data so that we can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:21",
                    "end": "00:04:26",
                    "text": "solve complex problems across domains. I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:23",
                    "end": "00:04:28",
                    "text": "guess in in defense, air, space,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:26",
                    "end": "00:04:31",
                    "text": "maritime, land, cyber increasingly that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:28",
                    "end": "00:04:33",
                    "text": "requires that we can connect the dots."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:31",
                    "end": "00:04:36",
                    "text": "Um and then we need to expose a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:33",
                    "end": "00:04:39",
                    "text": "simplified representation of that data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:36",
                    "end": "00:04:42",
                    "text": "um so that a hard-pressed individual uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:39",
                    "end": "00:04:44",
                    "text": "operating under pressure uh can make"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:42",
                    "end": "00:04:46",
                    "text": "sense of the data and also so that we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:44",
                    "end": "00:04:49",
                    "text": "can as we increasingly open that data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:46",
                    "end": "00:04:51",
                    "text": "platform to agents so we reduce the risk"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:49",
                    "end": "00:04:53",
                    "text": "of hallucination for the agents as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:51",
                    "end": "00:04:55",
                    "text": "So those are three principles that I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:53",
                    "end": "00:04:57",
                    "text": "think have stood the test of time. When"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:55",
                    "end": "00:04:58",
                    "text": "we think about a modern data foundation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:57",
                    "end": "00:05:00",
                    "text": "and we tend to use this word data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:58",
                    "end": "00:05:02",
                    "text": "modernization a lot right now in the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:00",
                    "end": "00:05:04",
                    "text": "industry"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:02",
                    "end": "00:05:06",
                    "text": "a a difficult term I think I uh I I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:04",
                    "end": "00:05:08",
                    "text": "would say that uh data modernization can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:06",
                    "end": "00:05:09",
                    "text": "mean whatever the vendor sat across you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:08",
                    "end": "00:05:11",
                    "text": "wants to sell you that day. So we should"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:09",
                    "end": "00:05:13",
                    "text": "be careful with the term but I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:11",
                    "end": "00:05:14",
                    "text": "there are two things that are really"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:13",
                    "end": "00:05:17",
                    "text": "important in in this idea of data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:14",
                    "end": "00:05:19",
                    "text": "modernization. The first is this idea of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:17",
                    "end": "00:05:21",
                    "text": "open table formats and and the the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:19",
                    "end": "00:05:24",
                    "text": "promise of open table formats is that we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:21",
                    "end": "00:05:26",
                    "text": "can store more data more reliably and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:24",
                    "end": "00:05:28",
                    "text": "crucially more cost-ffectively and cost"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:26",
                    "end": "00:05:29",
                    "text": "effective is important. I was speaking"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:28",
                    "end": "00:05:32",
                    "text": "to the ministry of defense at one of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:29",
                    "end": "00:05:34",
                    "text": "smaller European nations last week their"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:32",
                    "end": "00:05:37",
                    "text": "next generation AI data platform that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:34",
                    "end": "00:05:39",
                    "text": "they envisaged to integrate those five"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:37",
                    "end": "00:05:42",
                    "text": "domains they think will be a minimum of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:39",
                    "end": "00:05:45",
                    "text": "2.5 pabytes of data. for a country of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:42",
                    "end": "00:05:47",
                    "text": "the size complexity of India um I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:45",
                    "end": "00:05:49",
                    "text": "you can multiply that by at least 10. So"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:47",
                    "end": "00:05:51",
                    "text": "being able to store large volumes of of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:49",
                    "end": "00:05:53",
                    "text": "data and to to fuse that data that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:51",
                    "end": "00:05:55",
                    "text": "sensor data cost effectively is very"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:53",
                    "end": "00:05:57",
                    "text": "important. But the more important"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:55",
                    "end": "00:05:60",
                    "text": "consequence of open table formats is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:57",
                    "end": "00:06:01",
                    "text": "this idea of interoperability."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:60",
                    "end": "00:06:04",
                    "text": "This idea that we cannot possibly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:01",
                    "end": "00:06:06",
                    "text": "predict sat here in this room today"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:04",
                    "end": "00:06:08",
                    "text": "which models and which technologies will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:06",
                    "end": "00:06:10",
                    "text": "be the most important four or five years"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:08",
                    "end": "00:06:12",
                    "text": "from now. And this idea that different"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:10",
                    "end": "00:06:13",
                    "text": "technologies, different processing"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:12",
                    "end": "00:06:15",
                    "text": "engines have different strengths and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:13",
                    "end": "00:06:17",
                    "text": "weaknesses for different applications."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:15",
                    "end": "00:06:19",
                    "text": "And so we need to be able to support"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:17",
                    "end": "00:06:20",
                    "text": "many of them. And so the idea of having"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:19",
                    "end": "00:06:22",
                    "text": "a data foundation that we can bring"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:20",
                    "end": "00:06:24",
                    "text": "multiple tools to is incredibly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:22",
                    "end": "00:06:26",
                    "text": "important in that conversation. The"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:24",
                    "end": "00:06:28",
                    "text": "other modernization conversation that's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:26",
                    "end": "00:06:29",
                    "text": "very important is data products, but we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:28",
                    "end": "00:06:32",
                    "text": "probably don't have time for that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:29",
                    "end": "00:06:33",
                    "text": "discussion today. And this now brings us"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:32",
                    "end": "00:06:35",
                    "text": "to why are we building this data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:33",
                    "end": "00:06:37",
                    "text": "foundation? Well, we're building this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:35",
                    "end": "00:06:38",
                    "text": "data foundation so that we can bring AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:37",
                    "end": "00:06:40",
                    "text": "to that data foundation and increasingly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:38",
                    "end": "00:06:42",
                    "text": "so we can bring agentic AI to that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:40",
                    "end": "00:06:43",
                    "text": "foundation. And I think there are a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:42",
                    "end": "00:06:46",
                    "text": "couple of things that are important to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:43",
                    "end": "00:06:48",
                    "text": "to think through there. The first is I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:46",
                    "end": "00:06:50",
                    "text": "think we can get"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:48",
                    "end": "00:06:52",
                    "text": "very wrapped up around the idea of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:50",
                    "end": "00:06:55",
                    "text": "training models around training the best"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:52",
                    "end": "00:06:56",
                    "text": "model around training our own model. I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:55",
                    "end": "00:06:58",
                    "text": "think that's a little bit of a trap."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:56",
                    "end": "00:07:00",
                    "text": "Honestly, when I look at um the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:58",
                    "end": "00:07:03",
                    "text": "performance of the open- source models,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:00",
                    "end": "00:07:06",
                    "text": "they're generally only six months behind"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:03",
                    "end": "00:07:07",
                    "text": "the proprietary models. So, right now,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:06",
                    "end": "00:07:09",
                    "text": "the open source models are catching up"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:07",
                    "end": "00:07:11",
                    "text": "with the proprietary models inside six"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:09",
                    "end": "00:07:12",
                    "text": "months. And those open source models, we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:11",
                    "end": "00:07:15",
                    "text": "can inspect the architecture, we can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:12",
                    "end": "00:07:18",
                    "text": "inspect the weights. So, we can we can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:15",
                    "end": "00:07:20",
                    "text": "be pretty confident in those models. And"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:18",
                    "end": "00:07:22",
                    "text": "so, I think we should focus less on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:20",
                    "end": "00:07:25",
                    "text": "training our own models and think more"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:22",
                    "end": "00:07:27",
                    "text": "about inference. A good model that I can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:25",
                    "end": "00:07:31",
                    "text": "score quickly and cost-effectively"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:27",
                    "end": "00:07:33",
                    "text": "is better nine times out of 10 than the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:31",
                    "end": "00:07:35",
                    "text": "best model that I can't get out of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:33",
                    "end": "00:07:37",
                    "text": "lab and into production. So we should we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:35",
                    "end": "00:07:38",
                    "text": "should focus on inference first. We"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:37",
                    "end": "00:07:40",
                    "text": "should also think about the diversity of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:38",
                    "end": "00:07:43",
                    "text": "models. There was a really interesting"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:40",
                    "end": "00:07:45",
                    "text": "paper by Nvidia last year called small"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:43",
                    "end": "00:07:47",
                    "text": "models of the future of agentic AI and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:45",
                    "end": "00:07:50",
                    "text": "in that paper they pointed out that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:47",
                    "end": "00:07:54",
                    "text": "smaller models are often cheaper to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:50",
                    "end": "00:07:55",
                    "text": "score faster to score and can be less"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:54",
                    "end": "00:07:58",
                    "text": "prone to hallucination when they're"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:55",
                    "end": "00:07:59",
                    "text": "focused on a task specific problem. So"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:58",
                    "end": "00:08:00",
                    "text": "better, cheaper, faster and more"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:59",
                    "end": "00:08:02",
                    "text": "reliable. That's a pretty good trade,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:00",
                    "end": "00:08:04",
                    "text": "right? So we should think about not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:02",
                    "end": "00:08:06",
                    "text": "having one single model to rule them"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:04",
                    "end": "00:08:08",
                    "text": "all. We need to start thinking about the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:06",
                    "end": "00:08:10",
                    "text": "models that are available to us as a box"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:08",
                    "end": "00:08:12",
                    "text": "of tools. And when we have a box of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:10",
                    "end": "00:08:14",
                    "text": "tools, we shouldn't always reach for the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:12",
                    "end": "00:08:16",
                    "text": "hammer just because it's the biggest"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:14",
                    "end": "00:08:17",
                    "text": "shiniest tool in the box. And then the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:16",
                    "end": "00:08:19",
                    "text": "last thing that I think is particularly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:17",
                    "end": "00:08:22",
                    "text": "relevant in a defense and a security"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:19",
                    "end": "00:08:24",
                    "text": "context is for I've been in I've been in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:22",
                    "end": "00:08:27",
                    "text": "this industry for a little over 30 years"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:24",
                    "end": "00:08:29",
                    "text": "now and mostly we've been concerned with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:27",
                    "end": "00:08:31",
                    "text": "structured data with rows and columns of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:29",
                    "end": "00:08:34",
                    "text": "numeric data. And the great promise of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:31",
                    "end": "00:08:36",
                    "text": "AI that we're seeing play out already is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:34",
                    "end": "00:08:38",
                    "text": "that we can now apply these models to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:36",
                    "end": "00:08:40",
                    "text": "unstructured data to images to audio to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:38",
                    "end": "00:08:42",
                    "text": "text and we can have multimodal"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:40",
                    "end": "00:08:44",
                    "text": "intelligence and this is incredibly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:42",
                    "end": "00:08:46",
                    "text": "important. So at Terodata for example"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:44",
                    "end": "00:08:47",
                    "text": "we've always worked with defense"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:46",
                    "end": "00:08:49",
                    "text": "agencies around supply chain"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:47",
                    "end": "00:08:51",
                    "text": "optimization to your point around"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:49",
                    "end": "00:08:52",
                    "text": "predictive maintenance. Today in terod"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:51",
                    "end": "00:08:54",
                    "text": "data we can support very complex"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:52",
                    "end": "00:08:58",
                    "text": "maritime surveillance applications where"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:54",
                    "end": "00:09:02",
                    "text": "we take large volumes of detailed data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:58",
                    "end": "00:09:03",
                    "text": "across multiple sensor types a and fuse"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:02",
                    "end": "00:09:05",
                    "text": "that intelligence for maritime"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:03",
                    "end": "00:09:06",
                    "text": "surveillance. We can assess camouflage"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:05",
                    "end": "00:09:08",
                    "text": "effectiveness. I think you made the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:06",
                    "end": "00:09:09",
                    "text": "point earlier that that highv value"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:08",
                    "end": "00:09:11",
                    "text": "assets are increasingly vulnerable to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:09",
                    "end": "00:09:14",
                    "text": "loitering munitions relatively"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:11",
                    "end": "00:09:17",
                    "text": "inexpensive loitering munitions. So we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:14",
                    "end": "00:09:19",
                    "text": "can capture images of camouflaged assets"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:17",
                    "end": "00:09:21",
                    "text": "like tanks and assess camouflage"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:19",
                    "end": "00:09:22",
                    "text": "effectiveness to protect those assets"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:21",
                    "end": "00:09:24",
                    "text": "and these are applications that would"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:22",
                    "end": "00:09:26",
                    "text": "not have been possible even 3 four five"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:24",
                    "end": "00:09:30",
                    "text": "years ago. So the the pace of change and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:26",
                    "end": "00:09:31",
                    "text": "innovation is spectacular and exciting."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:30",
                    "end": "00:09:34",
                    "text": ">> Thank you. Thank you Martin. I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:31",
                    "end": "00:09:37",
                    "text": "what really stood out to me is that in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:34",
                    "end": "00:09:39",
                    "text": "our quest for sovereignity I think using"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:37",
                    "end": "00:09:41",
                    "text": "open source data data model is really"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:39",
                    "end": "00:09:44",
                    "text": "important because you get to inspect the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:41",
                    "end": "00:09:46",
                    "text": "ar architecture no matter who's created"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:44",
                    "end": "00:09:49",
                    "text": "it at least that aspect can we bring"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:46",
                    "end": "00:09:50",
                    "text": "into soy very early on and also it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:49",
                    "end": "00:09:52",
                    "text": "resonates with general shiba earlier"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:50",
                    "end": "00:09:55",
                    "text": "said you know building superior"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:52",
                    "end": "00:09:60",
                    "text": "algorithms on top of it yeah great so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:55",
                    "end": "00:10:02",
                    "text": "now I come come to uh pri saxa pri is a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:60",
                    "end": "00:10:05",
                    "text": "director of global data analytics in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:02",
                    "end": "00:10:08",
                    "text": "concentrics uh recognized as one of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:05",
                    "end": "00:10:11",
                    "text": "10 most influential leader awarded as uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:08",
                    "end": "00:10:14",
                    "text": "one of the women leaders in AI and looks"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:11",
                    "end": "00:10:16",
                    "text": "after the entire genai initiative across"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:14",
                    "end": "00:10:19",
                    "text": "the world from concentrics. So pri great"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:16",
                    "end": "00:10:22",
                    "text": "to have you here and thanks for coming."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:19",
                    "end": "00:10:25",
                    "text": "So pri uh what we would like your inputs"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:22",
                    "end": "00:10:27",
                    "text": "is is on the public and private"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:25",
                    "end": "00:10:29",
                    "text": "collaboration and you have been at the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:27",
                    "end": "00:10:31",
                    "text": "forefront of it. So what do you think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:29",
                    "end": "00:10:34",
                    "text": "are the right models where both the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:31",
                    "end": "00:10:35",
                    "text": "teams can work together with a larger"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:34",
                    "end": "00:10:37",
                    "text": "focus on bringing sovereignty across"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:35",
                    "end": "00:10:40",
                    "text": "layers of AI? Over to you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:37",
                    "end": "00:10:42",
                    "text": ">> Sure. Thank you. Thanks Abishek. Am I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:40",
                    "end": "00:10:45",
                    "text": "audible? Okay. Great. So good good"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:42",
                    "end": "00:10:48",
                    "text": "morning everyone. So I would just like"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:45",
                    "end": "00:10:51",
                    "text": "to answer it by having very small blocks"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:48",
                    "end": "00:10:54",
                    "text": "to the entire ecosystem of data AI and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:51",
                    "end": "00:10:56",
                    "text": "analytics. So starting from data which"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:54",
                    "end": "00:10:59",
                    "text": "is the main thing the bread and butter"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:56",
                    "end": "00:11:02",
                    "text": "on which all the layers are added on top"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:59",
                    "end": "00:11:04",
                    "text": "of it as as you mentioned rightly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:02",
                    "end": "00:11:06",
                    "text": "capturing the right data having the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:04",
                    "end": "00:11:10",
                    "text": "right form of governance is absolutely"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:06",
                    "end": "00:11:12",
                    "text": "important. I think I think starting from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:10",
                    "end": "00:11:14",
                    "text": "residency the data residency aspect in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:12",
                    "end": "00:11:17",
                    "text": "which India has already initiated the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:14",
                    "end": "00:11:19",
                    "text": "DPDP act which is very very absolutely"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:17",
                    "end": "00:11:22",
                    "text": "relevant and that would be the first"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:19",
                    "end": "00:11:24",
                    "text": "foundational layers to it. So once we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:22",
                    "end": "00:11:27",
                    "text": "have ensured that all our healthcare"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:24",
                    "end": "00:11:29",
                    "text": "data PI information all the transactions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:27",
                    "end": "00:11:32",
                    "text": "we are made are residing in our own"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:29",
                    "end": "00:11:34",
                    "text": "country that's a foundational layer in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:32",
                    "end": "00:11:36",
                    "text": "terms of residency then we need to have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:34",
                    "end": "00:11:39",
                    "text": "the right layer of governance on top of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:36",
                    "end": "00:11:40",
                    "text": "it and when I say governance like do we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:39",
                    "end": "00:11:42",
                    "text": "have any biases in the data are we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:40",
                    "end": "00:11:45",
                    "text": "capturing in the right formats right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:42",
                    "end": "00:11:47",
                    "text": "shape so that is absolutely important"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:45",
                    "end": "00:11:50",
                    "text": "because data governance is going to be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:47",
                    "end": "00:11:52",
                    "text": "the next key thing for your success for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:50",
                    "end": "00:11:56",
                    "text": "any program any analytics or AI program"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:52",
                    "end": "00:11:58",
                    "text": "that you run across. Now here when we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:56",
                    "end": "00:12:01",
                    "text": "have the data residency solved,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:58",
                    "end": "00:12:04",
                    "text": "governance solved and then on top of it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:01",
                    "end": "00:12:06",
                    "text": "we try to apply any AI models on top of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:04",
                    "end": "00:12:08",
                    "text": "it and I completely agree to your point"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:06",
                    "end": "00:12:11",
                    "text": "which you said like open source are much"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:08",
                    "end": "00:12:13",
                    "text": "speedy to market we should leverage that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:11",
                    "end": "00:12:16",
                    "text": "but with all the relevant guard rails."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:13",
                    "end": "00:12:18",
                    "text": "So here I believe public and private"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:16",
                    "end": "00:12:21",
                    "text": "partnership can have a great role like"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:18",
                    "end": "00:12:24",
                    "text": "can we have a challenge based programs."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:21",
                    "end": "00:12:26",
                    "text": "Can we have some hackathons in which"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:24",
                    "end": "00:12:28",
                    "text": "public sector also and private sector"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:26",
                    "end": "00:12:30",
                    "text": "also come across with various prototypes"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:28",
                    "end": "00:12:33",
                    "text": "and those prototypes can later on be put"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:30",
                    "end": "00:12:35",
                    "text": "into mass productions but that can be a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:33",
                    "end": "00:12:38",
                    "text": "great collaborative formats and because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:35",
                    "end": "00:12:40",
                    "text": "open source these days are like very"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:38",
                    "end": "00:12:42",
                    "text": "very speedy to the market. they are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:40",
                    "end": "00:12:45",
                    "text": "actually solving the real problems and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:42",
                    "end": "00:12:46",
                    "text": "they are coming up with very innovative"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:45",
                    "end": "00:12:48",
                    "text": "concepts that we haven't thought of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:46",
                    "end": "00:12:50",
                    "text": "enough. So here I feel like public"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:48",
                    "end": "00:12:53",
                    "text": "private collaboration can really be of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:50",
                    "end": "00:12:56",
                    "text": "help and there are various uh you can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:53",
                    "end": "00:12:59",
                    "text": "say uh unique models to curl that out be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:56",
                    "end": "00:13:02",
                    "text": "challenge based be hackathons or in any"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:59",
                    "end": "00:13:04",
                    "text": "other formats. So once we have the AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:02",
                    "end": "00:13:06",
                    "text": "layer sorted out and have the relevant"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:04",
                    "end": "00:13:10",
                    "text": "models built then comes the consumption"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:06",
                    "end": "00:13:12",
                    "text": "layer. Now as as uh my fellow panelist"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:10",
                    "end": "00:13:14",
                    "text": "was saying like there is no industry"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:12",
                    "end": "00:13:17",
                    "text": "today right now which is not affected by"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:14",
                    "end": "00:13:20",
                    "text": "AI. You you name any industry we have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:17",
                    "end": "00:13:21",
                    "text": "the analytics use cases. Why? Because we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:20",
                    "end": "00:13:24",
                    "text": "have the right set of techniques"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:21",
                    "end": "00:13:27",
                    "text": "especially in defense area any anomaly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:24",
                    "end": "00:13:29",
                    "text": "detection any pattern recognition we can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:27",
                    "end": "00:13:31",
                    "text": "put the right alarms right setups. Okay"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:29",
                    "end": "00:13:33",
                    "text": "this this is something which is going"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:31",
                    "end": "00:13:36",
                    "text": "off the track and should highlight we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:33",
                    "end": "00:13:38",
                    "text": "should raise the relevant alerts. So in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:36",
                    "end": "00:13:41",
                    "text": "those mechanisms the user especially in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:38",
                    "end": "00:13:43",
                    "text": "the defense sector is absolutely"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:41",
                    "end": "00:13:47",
                    "text": "important and can be done on real time."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:43",
                    "end": "00:13:50",
                    "text": "Now all of this is en you can say"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:47",
                    "end": "00:13:52",
                    "text": "encapsulated with the ecosystem layer"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:50",
                    "end": "00:13:55",
                    "text": "and what we call it as like you can call"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:52",
                    "end": "00:13:58",
                    "text": "it as an ESG layer because as we are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:55",
                    "end": "00:14:00",
                    "text": "building the Indian soverent track to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:58",
                    "end": "00:14:02",
                    "text": "build our own AIS we need to be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:00",
                    "end": "00:14:05",
                    "text": "absolutely cognizant of the fact what"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:02",
                    "end": "00:14:06",
                    "text": "kind of carbon footprints uh while we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:05",
                    "end": "00:14:08",
                    "text": "are creating our own data centers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:06",
                    "end": "00:14:10",
                    "text": "building our own clouds what kind of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:08",
                    "end": "00:14:13",
                    "text": "footprints we are having. So somewhere"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:10",
                    "end": "00:14:16",
                    "text": "in back of this journey we should be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:13",
                    "end": "00:14:18",
                    "text": "cognizant of this facts as well how ESG"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:16",
                    "end": "00:14:21",
                    "text": "can actually have an impact in this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:18",
                    "end": "00:14:22",
                    "text": "entire ball game. So that's that's all"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:21",
                    "end": "00:14:24",
                    "text": "my thoughts abishek."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:22",
                    "end": "00:14:27",
                    "text": ">> Thank you. Thank you so much P. I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:24",
                    "end": "00:14:29",
                    "text": "uh yeah"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:27",
                    "end": "00:14:31",
                    "text": "my my takeaway from you is that open"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:29",
                    "end": "00:14:33",
                    "text": "source but with guardrails"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:31",
                    "end": "00:14:35",
                    "text": "um you know quick to go to market"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:33",
                    "end": "00:14:37",
                    "text": "innovative. Yeah and you also talked"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:35",
                    "end": "00:14:39",
                    "text": "about the consumption layer but most"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:37",
                    "end": "00:14:41",
                    "text": "importantly you came towards the end on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:39",
                    "end": "00:14:43",
                    "text": "the ESG side of it. I'm very sure we are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:41",
                    "end": "00:14:45",
                    "text": "going to have one more session sometime"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:43",
                    "end": "00:14:46",
                    "text": "in future on ESG and we'd love to invite"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:45",
                    "end": "00:14:48",
                    "text": "you and express your thoughts because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:46",
                    "end": "00:14:51",
                    "text": "after all data centers wherever we read"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:48",
                    "end": "00:14:53",
                    "text": "US or anywhere local communities are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:51",
                    "end": "00:14:55",
                    "text": "they they are really affected by it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:53",
                    "end": "00:14:59",
                    "text": "Thank you so much Pri. Okay. So uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:55",
                    "end": "00:15:01",
                    "text": "towards the end I come to Mandar. Uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:59",
                    "end": "00:15:03",
                    "text": "last but not the least, Mandar is is the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:01",
                    "end": "00:15:06",
                    "text": "national security officer of Microsoft"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:03",
                    "end": "00:15:08",
                    "text": "India and Asia Pacific and is at the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:06",
                    "end": "00:15:12",
                    "text": "forefront of Microsoft's uh security"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:08",
                    "end": "00:15:15",
                    "text": "initiatives including cyber. So Mandard"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:12",
                    "end": "00:15:17",
                    "text": "your thoughts about adoption of AI in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:15",
                    "end": "00:15:20",
                    "text": "general because your platform as we say"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:17",
                    "end": "00:15:21",
                    "text": "you know various layers while that layer"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:20",
                    "end": "00:15:24",
                    "text": "can be addressed later but what is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:21",
                    "end": "00:15:26",
                    "text": "whatever is built on top of it is is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:24",
                    "end": "00:15:29",
                    "text": "actually what we debated can be you know"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:26",
                    "end": "00:15:30",
                    "text": "the early adopters of sovereign AI so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:29",
                    "end": "00:15:33",
                    "text": "over to you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:30",
                    "end": "00:15:34",
                    "text": ">> right thanks I think interesting and uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:33",
                    "end": "00:15:38",
                    "text": "I have the advantage of speaking after"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:34",
                    "end": "00:15:40",
                    "text": "everybody so so I will uh I'll just take"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:38",
                    "end": "00:15:42",
                    "text": "one more minute right I think a lot of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:40",
                    "end": "00:15:45",
                    "text": "speakers before me talked about"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:42",
                    "end": "00:15:47",
                    "text": "sovereignity and uh different words used"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:45",
                    "end": "00:15:50",
                    "text": "visceral and emotional and all that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:47",
                    "end": "00:15:52",
                    "text": "right I think in my opinion um"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:50",
                    "end": "00:15:54",
                    "text": "sovereignity has been obscured or"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:52",
                    "end": "00:15:56",
                    "text": "confused by a lot of interest right and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:54",
                    "end": "00:15:60",
                    "text": "interest from everybody from government"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:56",
                    "end": "00:16:02",
                    "text": "from tech companies and um of course uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:60",
                    "end": "00:16:04",
                    "text": "general audience right but sovereignty"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:02",
                    "end": "00:16:06",
                    "text": "and uh in my role I work with lot of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:04",
                    "end": "00:16:09",
                    "text": "governments including Indian government"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:06",
                    "end": "00:16:11",
                    "text": "I have I had privilege of working with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:09",
                    "end": "00:16:13",
                    "text": "some of my uh some of the Indian defense"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:11",
                    "end": "00:16:15",
                    "text": "folks as well. Solverity is fairly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:13",
                    "end": "00:16:17",
                    "text": "simple, right? Essentially, it's four"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:15",
                    "end": "00:16:20",
                    "text": "layers. The layer number one is data,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:17",
                    "end": "00:16:23",
                    "text": "right? Data sovereignity. We whole"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:20",
                    "end": "00:16:25",
                    "text": "started with data residency. But then"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:23",
                    "end": "00:16:26",
                    "text": "given the geopolitical situation and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:25",
                    "end": "00:16:28",
                    "text": "some of the incidents, we actually now"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:26",
                    "end": "00:16:30",
                    "text": "talk about data sovereignty. How is my"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:28",
                    "end": "00:16:33",
                    "text": "data protected? Nobody else has access."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:30",
                    "end": "00:16:36",
                    "text": "Nobody else has jurisdictional access to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:33",
                    "end": "00:16:38",
                    "text": "my data. Right? That's data sovereignty."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:36",
                    "end": "00:16:40",
                    "text": "The second layer is operational"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:38",
                    "end": "00:16:43",
                    "text": "sovereignty. uh couple of speakers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:40",
                    "end": "00:16:45",
                    "text": "before me spoke spoke about how can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:43",
                    "end": "00:16:47",
                    "text": "nobody else switch me off right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:45",
                    "end": "00:16:51",
                    "text": "operational so also is fairly simple so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:47",
                    "end": "00:16:54",
                    "text": "that my business can continue uh I can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:51",
                    "end": "00:16:55",
                    "text": "continue to trust that cloud or AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:54",
                    "end": "00:16:58",
                    "text": "without somebody else having ability to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:55",
                    "end": "00:16:59",
                    "text": "switch off tech this is where things"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:58",
                    "end": "00:17:02",
                    "text": "become little complicated and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:59",
                    "end": "00:17:03",
                    "text": "interesting right tech soy means I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:02",
                    "end": "00:17:06",
                    "text": "should be able to build on top of that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:03",
                    "end": "00:17:08",
                    "text": "right I'll not get into bit of clos"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:06",
                    "end": "00:17:10",
                    "text": "source open source commercial right but"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:08",
                    "end": "00:17:13",
                    "text": "that's where the issue starts happening"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:10",
                    "end": "00:17:15",
                    "text": "because tech also means the tech has to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:13",
                    "end": "00:17:17",
                    "text": "be not only in my control but it has to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:15",
                    "end": "00:17:19",
                    "text": "be good right the classic example"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:17",
                    "end": "00:17:20",
                    "text": "classic debate that we get again we have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:19",
                    "end": "00:17:23",
                    "text": "eminent defense folks so I'll refer to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:20",
                    "end": "00:17:25",
                    "text": "that right is India willing to wait 10"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:23",
                    "end": "00:17:27",
                    "text": "years for AMA to come fifth generation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:25",
                    "end": "00:17:30",
                    "text": "fighter or is India willing to buy"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:27",
                    "end": "00:17:32",
                    "text": "Rafael to fill that gap right uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:30",
                    "end": "00:17:34",
                    "text": "sovereign aircraft maybe we wait 10"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:32",
                    "end": "00:17:36",
                    "text": "years but we have needs today that's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:34",
                    "end": "00:17:38",
                    "text": "where the tech sovereignty becomes a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:36",
                    "end": "00:17:40",
                    "text": "very very interesting But sorry to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:38",
                    "end": "00:17:43",
                    "text": "pitching but just interesting input. The"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:40",
                    "end": "00:17:44",
                    "text": "AMA is coming with 50% indigenization."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:43",
                    "end": "00:17:47",
                    "text": ">> Sure. Sir would agree with me."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:44",
                    "end": "00:17:47",
                    "text": ">> No no absolutely I'm saying AMA drawing"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:47",
                    "end": "00:17:50",
                    "text": "a parallel."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:47",
                    "end": "00:17:51",
                    "text": ">> No no I'm saying AMA is great but AMA"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:50",
                    "end": "00:17:53",
                    "text": "will come maybe 10 years."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:51",
                    "end": "00:17:54",
                    "text": ">> Sorry not AMA. Rafal is coming with 50%"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:53",
                    "end": "00:17:56",
                    "text": "Indianization."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:54",
                    "end": "00:17:58",
                    "text": ">> Yeah of course of course. Yeah. So uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:56",
                    "end": "00:17:60",
                    "text": "but the point I'm making is AMA is 10"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:58",
                    "end": "00:18:03",
                    "text": "years away. Would we want or would we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:60",
                    "end": "00:18:05",
                    "text": "want 50% Indianization and use uh Rafal"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:03",
                    "end": "00:18:08",
                    "text": "today? Now that's the balance on tech"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:05",
                    "end": "00:18:10",
                    "text": "right? If we insist on every layer has"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:08",
                    "end": "00:18:12",
                    "text": "to be India then we probably will have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:10",
                    "end": "00:18:14",
                    "text": "to wait right that's the balance and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:12",
                    "end": "00:18:16",
                    "text": "finally the AI sovereignity right the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:14",
                    "end": "00:18:18",
                    "text": "fourth layer which was not on top of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:16",
                    "end": "00:18:20",
                    "text": "everybody's mind right the three layers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:18",
                    "end": "00:18:22",
                    "text": "have been talked for pretty much for a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:20",
                    "end": "00:18:24",
                    "text": "long time the data the operation and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:22",
                    "end": "00:18:26",
                    "text": "tech so even in the cloud world but AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:24",
                    "end": "00:18:29",
                    "text": "already came now AI so is a little bit"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:26",
                    "end": "00:18:32",
                    "text": "more complicated than cloud sovereignity"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:29",
                    "end": "00:18:34",
                    "text": "or or or any other tech soy right why"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:32",
                    "end": "00:18:37",
                    "text": "because I think speakers before me spoke"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:34",
                    "end": "00:18:39",
                    "text": "so I'll not go detail there but AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:37",
                    "end": "00:18:41",
                    "text": "brings a perspective right that means"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:39",
                    "end": "00:18:43",
                    "text": "the data that it is trained on has to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:41",
                    "end": "00:18:46",
                    "text": "have the Indian perspective right it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:43",
                    "end": "00:18:48",
                    "text": "needs to have all the nuances which in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:46",
                    "end": "00:18:50",
                    "text": "the first three layers were not so much"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:48",
                    "end": "00:18:52",
                    "text": "so data solidity is not only whether it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:50",
                    "end": "00:18:54",
                    "text": "is running in India whether it is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:52",
                    "end": "00:18:56",
                    "text": "developed by Indians it's also is it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:54",
                    "end": "00:18:58",
                    "text": "trained on Indian ethics is it trained"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:56",
                    "end": "00:19:01",
                    "text": "on Indian languages and all that so I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:58",
                    "end": "00:19:04",
                    "text": "think AI so is a step beyond that right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:01",
                    "end": "00:19:06",
                    "text": "and because most of the models are uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:04",
                    "end": "00:19:08",
                    "text": "like black blocks explanative"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:06",
                    "end": "00:19:09",
                    "text": "explanability is a challenge. I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:08",
                    "end": "00:19:11",
                    "text": "that's where AI already becomes"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:09",
                    "end": "00:19:13",
                    "text": "extremely important, right?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:11",
                    "end": "00:19:15",
                    "text": ">> Um and that's where I again agree with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:13",
                    "end": "00:19:17",
                    "text": "what I think Peter you mentioned the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:15",
                    "end": "00:19:19",
                    "text": "small language models Martin sorry you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:17",
                    "end": "00:19:21",
                    "text": "mentioned small language model"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:19",
                    "end": "00:19:23",
                    "text": "purpose-built models are probably making"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:21",
                    "end": "00:19:26",
                    "text": "more sense. I think that's the way we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:23",
                    "end": "00:19:28",
                    "text": "think generally of sovereignity right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:26",
                    "end": "00:19:30",
                    "text": "the question that we need to all look at"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:28",
                    "end": "00:19:32",
                    "text": "is solidity is not binary it's not zero"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:30",
                    "end": "00:19:35",
                    "text": "or one right as it is typically made out"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:32",
                    "end": "00:19:37",
                    "text": "to be it's a spectrum different"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:35",
                    "end": "00:19:40",
                    "text": "departments different organizations"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:37",
                    "end": "00:19:42",
                    "text": "different applications need different"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:40",
                    "end": "00:19:44",
                    "text": "level of sovereignty somewhere you need"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:42",
                    "end": "00:19:45",
                    "text": "to have operations control versus"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:44",
                    "end": "00:19:47",
                    "text": "somebody somewhere you will have to end"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:45",
                    "end": "00:19:50",
                    "text": "to end the stack like in defense needs"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:47",
                    "end": "00:19:53",
                    "text": "to be developed ourself right uh the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:50",
                    "end": "00:19:56",
                    "text": "final point that I'll make here is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:53",
                    "end": "00:19:59",
                    "text": "see life has come full circle. Uh those"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:56",
                    "end": "00:20:01",
                    "text": "who are in IT for maybe 25 odd years we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:59",
                    "end": "00:20:03",
                    "text": "saw mainframe where computing was"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:01",
                    "end": "00:20:05",
                    "text": "centralized to distributed computing"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:03",
                    "end": "00:20:07",
                    "text": "where computing was distributed to cloud"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:05",
                    "end": "00:20:09",
                    "text": "where again that computing become little"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:07",
                    "end": "00:20:12",
                    "text": "centralized. Now we are talking when"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:09",
                    "end": "00:20:13",
                    "text": "cloud started 10 15 years back right all"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:12",
                    "end": "00:20:15",
                    "text": "the big techs believed that there are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:13",
                    "end": "00:20:16",
                    "text": "going to be five regions in the world"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:15",
                    "end": "00:20:19",
                    "text": "right somewhere in Europe somewhere in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:16",
                    "end": "00:20:21",
                    "text": "Asia somewhere in US and everybody will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:19",
                    "end": "00:20:24",
                    "text": "use cloud from those five regions today"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:21",
                    "end": "00:20:26",
                    "text": "reality is Microsoft has 75 regions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:24",
                    "end": "00:20:29",
                    "text": "right so we moved from five regions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:26",
                    "end": "00:20:31",
                    "text": "literally 10 years back to 75 regions I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:29",
                    "end": "00:20:33",
                    "text": "think it's time to take the next step"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:31",
                    "end": "00:20:35",
                    "text": "where we are saying the region is not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:33",
                    "end": "00:20:37",
                    "text": "only in my control but region is also in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:35",
                    "end": "00:20:40",
                    "text": "customers data center so I think what we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:37",
                    "end": "00:20:42",
                    "text": "will see happening in tech is that the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:40",
                    "end": "00:20:45",
                    "text": "sovereign clouds will become a reality"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:42",
                    "end": "00:20:47",
                    "text": "right for example in Microsoft we have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:45",
                    "end": "00:20:49",
                    "text": "now created something something called a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:47",
                    "end": "00:20:51",
                    "text": "sovereign private cloud which can run in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:49",
                    "end": "00:20:55",
                    "text": "customers own data center under complete"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:51",
                    "end": "00:20:56",
                    "text": "customer control disconnected air gap we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:55",
                    "end": "00:20:58",
                    "text": "did not do that maybe two years back"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:56",
                    "end": "00:21:01",
                    "text": "that's the reality right we we talk kept"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:58",
                    "end": "00:21:03",
                    "text": "on talking about softwaredefined"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:01",
                    "end": "00:21:05",
                    "text": "sovereignty until about a year back"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:03",
                    "end": "00:21:07",
                    "text": ">> today we are saying okay we have a full"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:05",
                    "end": "00:21:09",
                    "text": "stack sovereign private cloud and I'm"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:07",
                    "end": "00:21:11",
                    "text": "taking Microsoft as an example. Our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:09",
                    "end": "00:21:11",
                    "text": "competition and other companies also"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:11",
                    "end": "00:21:13",
                    "text": "have similar."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:11",
                    "end": "00:21:15",
                    "text": ">> No, that's a very valid point. In fact,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:13",
                    "end": "00:21:17",
                    "text": "competitions like SAP already deployed"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:15",
                    "end": "00:21:17",
                    "text": "sovereign cloud and that's the way to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:17",
                    "end": "00:21:18",
                    "text": "go."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:17",
                    "end": "00:21:21",
                    "text": ">> That's sovereign public. What I'm saying"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:18",
                    "end": "00:21:22",
                    "text": "is sovereign private which can run in an"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:21",
                    "end": "00:21:23",
                    "text": "aircapped environment in defense of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:22",
                    "end": "00:21:25",
                    "text": "course of course"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:23",
                    "end": "00:21:28",
                    "text": ">> and that's a reality because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:25",
                    "end": "00:21:30",
                    "text": ">> last point we will have to take us out"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:28",
                    "end": "00:21:32",
                    "text": "of the equation the zero trust model"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:30",
                    "end": "00:21:34",
                    "text": "right you don't even need to trust the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:32",
                    "end": "00:21:35",
                    "text": "company that gives you tech. That's the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:34",
                    "end": "00:21:36",
                    "text": "only way we can build content."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:35",
                    "end": "00:21:38",
                    "text": ">> Thank you. Thank you. Right. Very valid"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:36",
                    "end": "00:21:40",
                    "text": "point Mandar and it was important to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:38",
                    "end": "00:21:42",
                    "text": "bring the industry perspective here to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:40",
                    "end": "00:21:44",
                    "text": "complete the full picture. Uh ladies and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:42",
                    "end": "00:21:46",
                    "text": "gentlemen we have almost run out of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:44",
                    "end": "00:21:48",
                    "text": "time. I would request Pier to give his"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:46",
                    "end": "00:21:50",
                    "text": "closing remarks and then we'll uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:48",
                    "end": "00:21:53",
                    "text": "conclude the session. P what you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:50",
                    "end": "00:21:56",
                    "text": ">> okay so try to keep it brief. So what we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:53",
                    "end": "00:21:58",
                    "text": "have been discussing here it's we would"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:56",
                    "end": "00:22:01",
                    "text": "say it's when we talk about sovereignity"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:58",
                    "end": "00:22:04",
                    "text": "and AI it's openness versus security the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:01",
                    "end": "00:22:05",
                    "text": "trade-offs speed of innovation versus"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:04",
                    "end": "00:22:09",
                    "text": "governance"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:05",
                    "end": "00:22:11",
                    "text": "global integration versus autonomy"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:09",
                    "end": "00:22:12",
                    "text": "public infrastructure versus private"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:11",
                    "end": "00:22:16",
                    "text": "innovation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:12",
                    "end": "00:22:18",
                    "text": "So these are the extremes and um what"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:16",
                    "end": "00:22:21",
                    "text": "what what I wish was would be for India"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:18",
                    "end": "00:22:23",
                    "text": "would be if they if you could have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:21",
                    "end": "00:22:26",
                    "text": "something in the middle of these"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:23",
                    "end": "00:22:31",
                    "text": "extremes um so the best of let's say the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:26",
                    "end": "00:22:35",
                    "text": "EU and the US EU very much regula"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:31",
                    "end": "00:22:37",
                    "text": "regulatory driven uh regulatory first"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:35",
                    "end": "00:22:40",
                    "text": "innovation second etc and the US"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:37",
                    "end": "00:22:43",
                    "text": "probably innovation first and then well"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:40",
                    "end": "00:22:46",
                    "text": "some control from the permanent um"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:43",
                    "end": "00:22:48",
                    "text": "anyway that comes. So if if that would"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:46",
                    "end": "00:22:51",
                    "text": "be something uh for India, I think this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:48",
                    "end": "00:22:55",
                    "text": "is a is a good rate good route to take."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:51",
                    "end": "00:22:59",
                    "text": "You said it uh collaborative models um"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:55",
                    "end": "00:23:03",
                    "text": "in uh in to invest in in infrastructure"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:59",
                    "end": "00:23:05",
                    "text": "something like um centers of excellence"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:03",
                    "end": "00:23:08",
                    "text": "um where public and private come"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:05",
                    "end": "00:23:11",
                    "text": "together to solve the real world"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:08",
                    "end": "00:23:13",
                    "text": "problems in architecture in healthcare"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:11",
                    "end": "00:23:15",
                    "text": "and all these kind of things. So if that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:13",
                    "end": "00:23:17",
                    "text": "would be a route for India I would be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:15",
                    "end": "00:23:21",
                    "text": "more than glad and it brings the best"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:17",
                    "end": "00:23:23",
                    "text": "the best together from both worlds."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:21",
                    "end": "00:23:25",
                    "text": "Thank you. Thank you Pier. So towards"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:23",
                    "end": "00:23:26",
                    "text": "the end of the session I'll request all"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:25",
                    "end": "00:23:28",
                    "text": "speakers to please come forward. We are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:26",
                    "end": "00:23:30",
                    "text": "releasing a background paper and I would"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:28",
                    "end": "00:23:34",
                    "text": "request Nachal Mishra and Akilshuta from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:30",
                    "end": "00:23:34",
                    "text": "KPMG to please come up as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:34",
                    "end": "00:23:38",
                    "text": "Show it please."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:47",
                    "end": "00:23:50",
                    "text": "Let's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:02",
                    "end": "00:24:05",
                    "text": "get out of this."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Hindi"
},
{
    "_id": "6992f2d04c5e8f0974c87539",
    "projectId": "777018",
    "projectName": "Sovereign AI and National Security: India’s Digital Path - Spanish",
    "projectDescription": "Sovereign AI and National Security: India’s Digital Path - Spanish",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-16T10:35:00.870Z",
    "projectUpdatedDate": "2026-02-16T10:35:00.870Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-16T11:30:01.492Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO_LINK",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames"
        },
        {
            "col1": "P1",
            "col2": "https://youtu.be/58V4tbgvnwc",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEOLINK",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:04",
                    "text": "You saw Russia Ukraine war was going on."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:02",
                    "end": "00:00:07",
                    "text": "Swift was stopped for Russia. So you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:04",
                    "end": "00:00:09",
                    "text": "will have to with the way we own digital"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:07",
                    "end": "00:00:11",
                    "text": "public infrastructure in India. We'll"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:09",
                    "end": "00:00:13",
                    "text": "have to create a cognitive public"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:11",
                    "end": "00:00:15",
                    "text": "infrastructure. How soever difficult it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:13",
                    "end": "00:00:17",
                    "text": "is and that's why pair it's visceral"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:15",
                    "end": "00:00:19",
                    "text": "it's existential and we will have to do"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:17",
                    "end": "00:00:22",
                    "text": "it and probably it's time that we should"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:19",
                    "end": "00:00:23",
                    "text": "do it. I yesterday wrote an article in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:22",
                    "end": "00:00:25",
                    "text": "uh the Sunday gajin which where where I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:23",
                    "end": "00:00:28",
                    "text": "called it the the architecture of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:25",
                    "end": "00:00:30",
                    "text": "independence right it is the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:28",
                    "end": "00:00:32",
                    "text": "architecture of independence. There's no"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:30",
                    "end": "00:00:34",
                    "text": "der of you know experts in India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:32",
                    "end": "00:00:36",
                    "text": "There's no der of talent in India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:34",
                    "end": "00:00:38",
                    "text": "There's no der of only thing is compute"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:36",
                    "end": "00:00:41",
                    "text": "which I think geopolitically we should"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:38",
                    "end": "00:00:44",
                    "text": "solve. That again is sovereignty right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:41",
                    "end": "00:00:46",
                    "text": "because people are using GPUs to control"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:44",
                    "end": "00:00:49",
                    "text": "geopolitics"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:46",
                    "end": "00:00:51",
                    "text": "and that is why we will have to be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:49",
                    "end": "00:00:52",
                    "text": "sovereign at all you know all the layers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:51",
                    "end": "00:00:52",
                    "text": "of the cake"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:52",
                    "end": "00:00:55",
                    "text": ">> right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:52",
                    "end": "00:00:57",
                    "text": ">> and I believe that uh when you apply"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:55",
                    "end": "00:01:01",
                    "text": "these systems to policing you will have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:57",
                    "end": "00:01:03",
                    "text": "to go for trust and transparency"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:01",
                    "end": "00:01:04",
                    "text": "uh and it has created problems on the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:03",
                    "end": "00:01:07",
                    "text": "other side it's become a force"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:04",
                    "end": "00:01:09",
                    "text": "multiplier it has hugely enabled"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:07",
                    "end": "00:01:12",
                    "text": "criminal ecosystem we polymorphic"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:09",
                    "end": "00:01:14",
                    "text": "malware. You have, you know, there was a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:12",
                    "end": "00:01:17",
                    "text": "a company called Arup. I think it's a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:14",
                    "end": "00:01:20",
                    "text": "advertisement company or something where"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:17",
                    "end": "00:01:22",
                    "text": "the CEO came on a live zoom call and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:20",
                    "end": "00:01:26",
                    "text": "they lost uh $30 million. Right? This is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:22",
                    "end": "00:01:28",
                    "text": "a state at which AI is evolving. Do we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:26",
                    "end": "00:01:30",
                    "text": "have realtime policing? Sir, you know,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:28",
                    "end": "00:01:32",
                    "text": "explained that policing is not so real"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:30",
                    "end": "00:01:36",
                    "text": "time. So, you will have to increase both"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:32",
                    "end": "00:01:38",
                    "text": "capability and capacity and this needs a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:36",
                    "end": "00:01:40",
                    "text": "serious thought right now. So pay we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:38",
                    "end": "00:01:43",
                    "text": "need sovereign AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:40",
                    "end": "00:01:46",
                    "text": ">> Thank you. Thank you sir. And again as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:43",
                    "end": "00:01:49",
                    "text": "expected very very relevant points sir"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:46",
                    "end": "00:01:51",
                    "text": "that reminds me of of interview by"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:49",
                    "end": "00:01:53",
                    "text": "Jensen Huang a few years back. He said I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:51",
                    "end": "00:01:56",
                    "text": "met your prime minister all before this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:53",
                    "end": "00:01:58",
                    "text": "AI hype uh when it's about in very"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:56",
                    "end": "00:01:59",
                    "text": "initial stages and he said your prime"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:58",
                    "end": "00:02:02",
                    "text": "minister told me that I don't want to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:59",
                    "end": "00:02:04",
                    "text": "export data to import intelligence. He"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:02",
                    "end": "00:02:07",
                    "text": "said that was really remarkable for a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:04",
                    "end": "00:02:08",
                    "text": "world leader to think in those times. So"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:07",
                    "end": "00:02:11",
                    "text": "I think very relevant points you also"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:08",
                    "end": "00:02:13",
                    "text": "echoed but good that sir you brought"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:11",
                    "end": "00:02:14",
                    "text": "compute in in picture because compute is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:13",
                    "end": "00:02:16",
                    "text": "something where we really have to make"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:14",
                    "end": "00:02:19",
                    "text": "pro in fact the rest of the world and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:16",
                    "end": "00:02:22",
                    "text": "then I would love to bring pair maybe at"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:19",
                    "end": "00:02:24",
                    "text": "later stage. So uh with this then I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:22",
                    "end": "00:02:26",
                    "text": "would like to move to Martin. Martin"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:24",
                    "end": "00:02:28",
                    "text": "thanks a lot uh first of all traveling"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:26",
                    "end": "00:02:31",
                    "text": "all the way to India to attend the AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:28",
                    "end": "00:02:35",
                    "text": "summit and being with us. M in your role"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:31",
                    "end": "00:02:38",
                    "text": "as uh in terod data as a leader in in in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:35",
                    "end": "00:02:41",
                    "text": "data analytics big data implementation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:38",
                    "end": "00:02:43",
                    "text": "uh you have been quoted twice in data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:41",
                    "end": "00:02:45",
                    "text": "IQ's 100 most influential people in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:43",
                    "end": "00:02:47",
                    "text": "datadriven business and what's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:45",
                    "end": "00:02:49",
                    "text": "remarkable I was looking at your website"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:47",
                    "end": "00:02:53",
                    "text": "where you are now saying that uh stop"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:49",
                    "end": "00:02:55",
                    "text": "managing data activate intelligence so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:53",
                    "end": "00:02:57",
                    "text": "and I also read your article about the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:55",
                    "end": "00:02:58",
                    "text": "large scale compute requirement which is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:57",
                    "end": "00:03:01",
                    "text": "going to come for example you give a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:58",
                    "end": "00:03:03",
                    "text": "healthcare example where every patient"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:01",
                    "end": "00:03:05",
                    "text": "will be followed by an agent and he'll"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:03",
                    "end": "00:03:07",
                    "text": "be firing like thousands hundreds and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:05",
                    "end": "00:03:09",
                    "text": "thousands of queries to make sure that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:07",
                    "end": "00:03:12",
                    "text": "you have that hypers uh personalization"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:09",
                    "end": "00:03:15",
                    "text": "experience. So with that kind of compute"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:12",
                    "end": "00:03:18",
                    "text": "requirement Martin do you think that uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:15",
                    "end": "00:03:21",
                    "text": "we are ready for so sovereignity from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:18",
                    "end": "00:03:24",
                    "text": "day one or is going or as everybody said"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:21",
                    "end": "00:03:25",
                    "text": "you have to tackle it layer wise so what"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:24",
                    "end": "00:03:28",
                    "text": "do you think especially when you come to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:25",
                    "end": "00:03:30",
                    "text": "the layer of data you know how the data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:28",
                    "end": "00:03:32",
                    "text": "structures uh a country like India can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:30",
                    "end": "00:03:35",
                    "text": "make sure that they are sovereign in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:32",
                    "end": "00:03:36",
                    "text": "those aspects yeah so over to you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:35",
                    "end": "00:03:38",
                    "text": ">> thank you and uh and good morning"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:36",
                    "end": "00:03:40",
                    "text": "everybody lovely to be here um there's a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:38",
                    "end": "00:03:41",
                    "text": "lot to unpack in your question and we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:40",
                    "end": "00:03:43",
                    "text": "should come back and we should talk"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:41",
                    "end": "00:03:46",
                    "text": "about the implications of potentially"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:43",
                    "end": "00:03:47",
                    "text": "millions of agents running thousands of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:46",
                    "end": "00:03:49",
                    "text": "queries because I think that is a very"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:47",
                    "end": "00:03:50",
                    "text": "important consequence uh of where we're"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:49",
                    "end": "00:03:52",
                    "text": "going that we should think about. But"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:50",
                    "end": "00:03:54",
                    "text": "you're absolutely right at Terodata we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:52",
                    "end": "00:03:56",
                    "text": "think that the first and the most"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:54",
                    "end": "00:03:58",
                    "text": "fundamental layer is the data layer."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:56",
                    "end": "00:03:60",
                    "text": "There is no good AI without good data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:58",
                    "end": "00:04:02",
                    "text": "So it's incredibly important we get that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:60",
                    "end": "00:04:03",
                    "text": "right. And there are some principles"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:02",
                    "end": "00:04:06",
                    "text": "that have proven to be more or less"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:03",
                    "end": "00:04:08",
                    "text": "timeless. And if I speak about timeless"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:06",
                    "end": "00:04:09",
                    "text": "in in terms of it, I'm speaking they've"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:08",
                    "end": "00:04:11",
                    "text": "proven to be good over three or four"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:09",
                    "end": "00:04:13",
                    "text": "decades. That's the kind of event"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:11",
                    "end": "00:04:15",
                    "text": "horizon, isn't it, for technology. Um,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:13",
                    "end": "00:04:17",
                    "text": "so we need to focus on capturing the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:15",
                    "end": "00:04:18",
                    "text": "most detailed data that we can. We need"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:17",
                    "end": "00:04:21",
                    "text": "to make sure that we are capable of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:18",
                    "end": "00:04:23",
                    "text": "integrating that data so that we can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:21",
                    "end": "00:04:26",
                    "text": "solve complex problems across domains. I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:23",
                    "end": "00:04:28",
                    "text": "guess in in defense, air, space,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:26",
                    "end": "00:04:31",
                    "text": "maritime, land, cyber increasingly that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:28",
                    "end": "00:04:33",
                    "text": "requires that we can connect the dots."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:31",
                    "end": "00:04:36",
                    "text": "Um and then we need to expose a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:33",
                    "end": "00:04:39",
                    "text": "simplified representation of that data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:36",
                    "end": "00:04:42",
                    "text": "um so that a hard-pressed individual uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:39",
                    "end": "00:04:44",
                    "text": "operating under pressure uh can make"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:42",
                    "end": "00:04:46",
                    "text": "sense of the data and also so that we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:44",
                    "end": "00:04:49",
                    "text": "can as we increasingly open that data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:46",
                    "end": "00:04:51",
                    "text": "platform to agents so we reduce the risk"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:49",
                    "end": "00:04:53",
                    "text": "of hallucination for the agents as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:51",
                    "end": "00:04:55",
                    "text": "So those are three principles that I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:53",
                    "end": "00:04:57",
                    "text": "think have stood the test of time. When"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:55",
                    "end": "00:04:58",
                    "text": "we think about a modern data foundation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:57",
                    "end": "00:05:00",
                    "text": "and we tend to use this word data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:58",
                    "end": "00:05:02",
                    "text": "modernization a lot right now in the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:00",
                    "end": "00:05:04",
                    "text": "industry"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:02",
                    "end": "00:05:06",
                    "text": "a a difficult term I think I uh I I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:04",
                    "end": "00:05:08",
                    "text": "would say that uh data modernization can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:06",
                    "end": "00:05:09",
                    "text": "mean whatever the vendor sat across you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:08",
                    "end": "00:05:11",
                    "text": "wants to sell you that day. So we should"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:09",
                    "end": "00:05:13",
                    "text": "be careful with the term but I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:11",
                    "end": "00:05:14",
                    "text": "there are two things that are really"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:13",
                    "end": "00:05:17",
                    "text": "important in in this idea of data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:14",
                    "end": "00:05:19",
                    "text": "modernization. The first is this idea of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:17",
                    "end": "00:05:21",
                    "text": "open table formats and and the the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:19",
                    "end": "00:05:24",
                    "text": "promise of open table formats is that we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:21",
                    "end": "00:05:26",
                    "text": "can store more data more reliably and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:24",
                    "end": "00:05:28",
                    "text": "crucially more cost-ffectively and cost"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:26",
                    "end": "00:05:29",
                    "text": "effective is important. I was speaking"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:28",
                    "end": "00:05:32",
                    "text": "to the ministry of defense at one of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:29",
                    "end": "00:05:34",
                    "text": "smaller European nations last week their"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:32",
                    "end": "00:05:37",
                    "text": "next generation AI data platform that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:34",
                    "end": "00:05:39",
                    "text": "they envisaged to integrate those five"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:37",
                    "end": "00:05:42",
                    "text": "domains they think will be a minimum of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:39",
                    "end": "00:05:45",
                    "text": "2.5 pabytes of data. for a country of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:42",
                    "end": "00:05:47",
                    "text": "the size complexity of India um I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:45",
                    "end": "00:05:49",
                    "text": "you can multiply that by at least 10. So"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:47",
                    "end": "00:05:51",
                    "text": "being able to store large volumes of of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:49",
                    "end": "00:05:53",
                    "text": "data and to to fuse that data that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:51",
                    "end": "00:05:55",
                    "text": "sensor data cost effectively is very"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:53",
                    "end": "00:05:57",
                    "text": "important. But the more important"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:55",
                    "end": "00:05:60",
                    "text": "consequence of open table formats is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:57",
                    "end": "00:06:01",
                    "text": "this idea of interoperability."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:60",
                    "end": "00:06:04",
                    "text": "This idea that we cannot possibly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:01",
                    "end": "00:06:06",
                    "text": "predict sat here in this room today"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:04",
                    "end": "00:06:08",
                    "text": "which models and which technologies will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:06",
                    "end": "00:06:10",
                    "text": "be the most important four or five years"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:08",
                    "end": "00:06:12",
                    "text": "from now. And this idea that different"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:10",
                    "end": "00:06:13",
                    "text": "technologies, different processing"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:12",
                    "end": "00:06:15",
                    "text": "engines have different strengths and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:13",
                    "end": "00:06:17",
                    "text": "weaknesses for different applications."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:15",
                    "end": "00:06:19",
                    "text": "And so we need to be able to support"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:17",
                    "end": "00:06:20",
                    "text": "many of them. And so the idea of having"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:19",
                    "end": "00:06:22",
                    "text": "a data foundation that we can bring"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:20",
                    "end": "00:06:24",
                    "text": "multiple tools to is incredibly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:22",
                    "end": "00:06:26",
                    "text": "important in that conversation. The"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:24",
                    "end": "00:06:28",
                    "text": "other modernization conversation that's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:26",
                    "end": "00:06:29",
                    "text": "very important is data products, but we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:28",
                    "end": "00:06:32",
                    "text": "probably don't have time for that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:29",
                    "end": "00:06:33",
                    "text": "discussion today. And this now brings us"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:32",
                    "end": "00:06:35",
                    "text": "to why are we building this data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:33",
                    "end": "00:06:37",
                    "text": "foundation? Well, we're building this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:35",
                    "end": "00:06:38",
                    "text": "data foundation so that we can bring AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:37",
                    "end": "00:06:40",
                    "text": "to that data foundation and increasingly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:38",
                    "end": "00:06:42",
                    "text": "so we can bring agentic AI to that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:40",
                    "end": "00:06:43",
                    "text": "foundation. And I think there are a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:42",
                    "end": "00:06:46",
                    "text": "couple of things that are important to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:43",
                    "end": "00:06:48",
                    "text": "to think through there. The first is I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:46",
                    "end": "00:06:50",
                    "text": "think we can get"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:48",
                    "end": "00:06:52",
                    "text": "very wrapped up around the idea of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:50",
                    "end": "00:06:55",
                    "text": "training models around training the best"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:52",
                    "end": "00:06:56",
                    "text": "model around training our own model. I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:55",
                    "end": "00:06:58",
                    "text": "think that's a little bit of a trap."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:56",
                    "end": "00:07:00",
                    "text": "Honestly, when I look at um the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:58",
                    "end": "00:07:03",
                    "text": "performance of the open- source models,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:00",
                    "end": "00:07:06",
                    "text": "they're generally only six months behind"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:03",
                    "end": "00:07:07",
                    "text": "the proprietary models. So, right now,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:06",
                    "end": "00:07:09",
                    "text": "the open source models are catching up"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:07",
                    "end": "00:07:11",
                    "text": "with the proprietary models inside six"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:09",
                    "end": "00:07:12",
                    "text": "months. And those open source models, we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:11",
                    "end": "00:07:15",
                    "text": "can inspect the architecture, we can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:12",
                    "end": "00:07:18",
                    "text": "inspect the weights. So, we can we can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:15",
                    "end": "00:07:20",
                    "text": "be pretty confident in those models. And"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:18",
                    "end": "00:07:22",
                    "text": "so, I think we should focus less on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:20",
                    "end": "00:07:25",
                    "text": "training our own models and think more"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:22",
                    "end": "00:07:27",
                    "text": "about inference. A good model that I can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:25",
                    "end": "00:07:31",
                    "text": "score quickly and cost-effectively"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:27",
                    "end": "00:07:33",
                    "text": "is better nine times out of 10 than the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:31",
                    "end": "00:07:35",
                    "text": "best model that I can't get out of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:33",
                    "end": "00:07:37",
                    "text": "lab and into production. So we should we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:35",
                    "end": "00:07:38",
                    "text": "should focus on inference first. We"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:37",
                    "end": "00:07:40",
                    "text": "should also think about the diversity of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:38",
                    "end": "00:07:43",
                    "text": "models. There was a really interesting"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:40",
                    "end": "00:07:45",
                    "text": "paper by Nvidia last year called small"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:43",
                    "end": "00:07:47",
                    "text": "models of the future of agentic AI and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:45",
                    "end": "00:07:50",
                    "text": "in that paper they pointed out that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:47",
                    "end": "00:07:54",
                    "text": "smaller models are often cheaper to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:50",
                    "end": "00:07:55",
                    "text": "score faster to score and can be less"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:54",
                    "end": "00:07:58",
                    "text": "prone to hallucination when they're"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:55",
                    "end": "00:07:59",
                    "text": "focused on a task specific problem. So"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:58",
                    "end": "00:08:00",
                    "text": "better, cheaper, faster and more"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:59",
                    "end": "00:08:02",
                    "text": "reliable. That's a pretty good trade,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:00",
                    "end": "00:08:04",
                    "text": "right? So we should think about not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:02",
                    "end": "00:08:06",
                    "text": "having one single model to rule them"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:04",
                    "end": "00:08:08",
                    "text": "all. We need to start thinking about the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:06",
                    "end": "00:08:10",
                    "text": "models that are available to us as a box"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:08",
                    "end": "00:08:12",
                    "text": "of tools. And when we have a box of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:10",
                    "end": "00:08:14",
                    "text": "tools, we shouldn't always reach for the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:12",
                    "end": "00:08:16",
                    "text": "hammer just because it's the biggest"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:14",
                    "end": "00:08:17",
                    "text": "shiniest tool in the box. And then the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:16",
                    "end": "00:08:19",
                    "text": "last thing that I think is particularly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:17",
                    "end": "00:08:22",
                    "text": "relevant in a defense and a security"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:19",
                    "end": "00:08:24",
                    "text": "context is for I've been in I've been in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:22",
                    "end": "00:08:27",
                    "text": "this industry for a little over 30 years"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:24",
                    "end": "00:08:29",
                    "text": "now and mostly we've been concerned with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:27",
                    "end": "00:08:31",
                    "text": "structured data with rows and columns of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:29",
                    "end": "00:08:34",
                    "text": "numeric data. And the great promise of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:31",
                    "end": "00:08:36",
                    "text": "AI that we're seeing play out already is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:34",
                    "end": "00:08:38",
                    "text": "that we can now apply these models to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:36",
                    "end": "00:08:40",
                    "text": "unstructured data to images to audio to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:38",
                    "end": "00:08:42",
                    "text": "text and we can have multimodal"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:40",
                    "end": "00:08:44",
                    "text": "intelligence and this is incredibly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:42",
                    "end": "00:08:46",
                    "text": "important. So at Terodata for example"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:44",
                    "end": "00:08:47",
                    "text": "we've always worked with defense"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:46",
                    "end": "00:08:49",
                    "text": "agencies around supply chain"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:47",
                    "end": "00:08:51",
                    "text": "optimization to your point around"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:49",
                    "end": "00:08:52",
                    "text": "predictive maintenance. Today in terod"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:51",
                    "end": "00:08:54",
                    "text": "data we can support very complex"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:52",
                    "end": "00:08:58",
                    "text": "maritime surveillance applications where"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:54",
                    "end": "00:09:02",
                    "text": "we take large volumes of detailed data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:58",
                    "end": "00:09:03",
                    "text": "across multiple sensor types a and fuse"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:02",
                    "end": "00:09:05",
                    "text": "that intelligence for maritime"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:03",
                    "end": "00:09:06",
                    "text": "surveillance. We can assess camouflage"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:05",
                    "end": "00:09:08",
                    "text": "effectiveness. I think you made the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:06",
                    "end": "00:09:09",
                    "text": "point earlier that that highv value"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:08",
                    "end": "00:09:11",
                    "text": "assets are increasingly vulnerable to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:09",
                    "end": "00:09:14",
                    "text": "loitering munitions relatively"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:11",
                    "end": "00:09:17",
                    "text": "inexpensive loitering munitions. So we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:14",
                    "end": "00:09:19",
                    "text": "can capture images of camouflaged assets"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:17",
                    "end": "00:09:21",
                    "text": "like tanks and assess camouflage"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:19",
                    "end": "00:09:22",
                    "text": "effectiveness to protect those assets"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:21",
                    "end": "00:09:24",
                    "text": "and these are applications that would"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:22",
                    "end": "00:09:26",
                    "text": "not have been possible even 3 four five"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:24",
                    "end": "00:09:30",
                    "text": "years ago. So the the pace of change and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:26",
                    "end": "00:09:31",
                    "text": "innovation is spectacular and exciting."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:30",
                    "end": "00:09:34",
                    "text": ">> Thank you. Thank you Martin. I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:31",
                    "end": "00:09:37",
                    "text": "what really stood out to me is that in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:34",
                    "end": "00:09:39",
                    "text": "our quest for sovereignity I think using"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:37",
                    "end": "00:09:41",
                    "text": "open source data data model is really"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:39",
                    "end": "00:09:44",
                    "text": "important because you get to inspect the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:41",
                    "end": "00:09:46",
                    "text": "ar architecture no matter who's created"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:44",
                    "end": "00:09:49",
                    "text": "it at least that aspect can we bring"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:46",
                    "end": "00:09:50",
                    "text": "into soy very early on and also it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:49",
                    "end": "00:09:52",
                    "text": "resonates with general shiba earlier"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:50",
                    "end": "00:09:55",
                    "text": "said you know building superior"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:52",
                    "end": "00:09:60",
                    "text": "algorithms on top of it yeah great so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:55",
                    "end": "00:10:02",
                    "text": "now I come come to uh pri saxa pri is a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:60",
                    "end": "00:10:05",
                    "text": "director of global data analytics in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:02",
                    "end": "00:10:08",
                    "text": "concentrics uh recognized as one of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:05",
                    "end": "00:10:11",
                    "text": "10 most influential leader awarded as uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:08",
                    "end": "00:10:14",
                    "text": "one of the women leaders in AI and looks"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:11",
                    "end": "00:10:16",
                    "text": "after the entire genai initiative across"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:14",
                    "end": "00:10:19",
                    "text": "the world from concentrics. So pri great"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:16",
                    "end": "00:10:22",
                    "text": "to have you here and thanks for coming."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:19",
                    "end": "00:10:25",
                    "text": "So pri uh what we would like your inputs"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:22",
                    "end": "00:10:27",
                    "text": "is is on the public and private"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:25",
                    "end": "00:10:29",
                    "text": "collaboration and you have been at the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:27",
                    "end": "00:10:31",
                    "text": "forefront of it. So what do you think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:29",
                    "end": "00:10:34",
                    "text": "are the right models where both the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:31",
                    "end": "00:10:35",
                    "text": "teams can work together with a larger"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:34",
                    "end": "00:10:37",
                    "text": "focus on bringing sovereignty across"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:35",
                    "end": "00:10:40",
                    "text": "layers of AI? Over to you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:37",
                    "end": "00:10:42",
                    "text": ">> Sure. Thank you. Thanks Abishek. Am I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:40",
                    "end": "00:10:45",
                    "text": "audible? Okay. Great. So good good"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:42",
                    "end": "00:10:48",
                    "text": "morning everyone. So I would just like"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:45",
                    "end": "00:10:51",
                    "text": "to answer it by having very small blocks"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:48",
                    "end": "00:10:54",
                    "text": "to the entire ecosystem of data AI and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:51",
                    "end": "00:10:56",
                    "text": "analytics. So starting from data which"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:54",
                    "end": "00:10:59",
                    "text": "is the main thing the bread and butter"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:56",
                    "end": "00:11:02",
                    "text": "on which all the layers are added on top"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:59",
                    "end": "00:11:04",
                    "text": "of it as as you mentioned rightly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:02",
                    "end": "00:11:06",
                    "text": "capturing the right data having the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:04",
                    "end": "00:11:10",
                    "text": "right form of governance is absolutely"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:06",
                    "end": "00:11:12",
                    "text": "important. I think I think starting from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:10",
                    "end": "00:11:14",
                    "text": "residency the data residency aspect in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:12",
                    "end": "00:11:17",
                    "text": "which India has already initiated the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:14",
                    "end": "00:11:19",
                    "text": "DPDP act which is very very absolutely"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:17",
                    "end": "00:11:22",
                    "text": "relevant and that would be the first"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:19",
                    "end": "00:11:24",
                    "text": "foundational layers to it. So once we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:22",
                    "end": "00:11:27",
                    "text": "have ensured that all our healthcare"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:24",
                    "end": "00:11:29",
                    "text": "data PI information all the transactions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:27",
                    "end": "00:11:32",
                    "text": "we are made are residing in our own"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:29",
                    "end": "00:11:34",
                    "text": "country that's a foundational layer in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:32",
                    "end": "00:11:36",
                    "text": "terms of residency then we need to have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:34",
                    "end": "00:11:39",
                    "text": "the right layer of governance on top of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:36",
                    "end": "00:11:40",
                    "text": "it and when I say governance like do we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:39",
                    "end": "00:11:42",
                    "text": "have any biases in the data are we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:40",
                    "end": "00:11:45",
                    "text": "capturing in the right formats right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:42",
                    "end": "00:11:47",
                    "text": "shape so that is absolutely important"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:45",
                    "end": "00:11:50",
                    "text": "because data governance is going to be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:47",
                    "end": "00:11:52",
                    "text": "the next key thing for your success for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:50",
                    "end": "00:11:56",
                    "text": "any program any analytics or AI program"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:52",
                    "end": "00:11:58",
                    "text": "that you run across. Now here when we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:56",
                    "end": "00:12:01",
                    "text": "have the data residency solved,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:58",
                    "end": "00:12:04",
                    "text": "governance solved and then on top of it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:01",
                    "end": "00:12:06",
                    "text": "we try to apply any AI models on top of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:04",
                    "end": "00:12:08",
                    "text": "it and I completely agree to your point"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:06",
                    "end": "00:12:11",
                    "text": "which you said like open source are much"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:08",
                    "end": "00:12:13",
                    "text": "speedy to market we should leverage that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:11",
                    "end": "00:12:16",
                    "text": "but with all the relevant guard rails."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:13",
                    "end": "00:12:18",
                    "text": "So here I believe public and private"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:16",
                    "end": "00:12:21",
                    "text": "partnership can have a great role like"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:18",
                    "end": "00:12:24",
                    "text": "can we have a challenge based programs."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:21",
                    "end": "00:12:26",
                    "text": "Can we have some hackathons in which"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:24",
                    "end": "00:12:28",
                    "text": "public sector also and private sector"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:26",
                    "end": "00:12:30",
                    "text": "also come across with various prototypes"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:28",
                    "end": "00:12:33",
                    "text": "and those prototypes can later on be put"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:30",
                    "end": "00:12:35",
                    "text": "into mass productions but that can be a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:33",
                    "end": "00:12:38",
                    "text": "great collaborative formats and because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:35",
                    "end": "00:12:40",
                    "text": "open source these days are like very"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:38",
                    "end": "00:12:42",
                    "text": "very speedy to the market. they are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:40",
                    "end": "00:12:45",
                    "text": "actually solving the real problems and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:42",
                    "end": "00:12:46",
                    "text": "they are coming up with very innovative"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:45",
                    "end": "00:12:48",
                    "text": "concepts that we haven't thought of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:46",
                    "end": "00:12:50",
                    "text": "enough. So here I feel like public"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:48",
                    "end": "00:12:53",
                    "text": "private collaboration can really be of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:50",
                    "end": "00:12:56",
                    "text": "help and there are various uh you can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:53",
                    "end": "00:12:59",
                    "text": "say uh unique models to curl that out be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:56",
                    "end": "00:13:02",
                    "text": "challenge based be hackathons or in any"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:59",
                    "end": "00:13:04",
                    "text": "other formats. So once we have the AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:02",
                    "end": "00:13:06",
                    "text": "layer sorted out and have the relevant"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:04",
                    "end": "00:13:10",
                    "text": "models built then comes the consumption"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:06",
                    "end": "00:13:12",
                    "text": "layer. Now as as uh my fellow panelist"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:10",
                    "end": "00:13:14",
                    "text": "was saying like there is no industry"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:12",
                    "end": "00:13:17",
                    "text": "today right now which is not affected by"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:14",
                    "end": "00:13:20",
                    "text": "AI. You you name any industry we have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:17",
                    "end": "00:13:21",
                    "text": "the analytics use cases. Why? Because we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:20",
                    "end": "00:13:24",
                    "text": "have the right set of techniques"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:21",
                    "end": "00:13:27",
                    "text": "especially in defense area any anomaly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:24",
                    "end": "00:13:29",
                    "text": "detection any pattern recognition we can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:27",
                    "end": "00:13:31",
                    "text": "put the right alarms right setups. Okay"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:29",
                    "end": "00:13:33",
                    "text": "this this is something which is going"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:31",
                    "end": "00:13:36",
                    "text": "off the track and should highlight we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:33",
                    "end": "00:13:38",
                    "text": "should raise the relevant alerts. So in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:36",
                    "end": "00:13:41",
                    "text": "those mechanisms the user especially in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:38",
                    "end": "00:13:43",
                    "text": "the defense sector is absolutely"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:41",
                    "end": "00:13:47",
                    "text": "important and can be done on real time."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:43",
                    "end": "00:13:50",
                    "text": "Now all of this is en you can say"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:47",
                    "end": "00:13:52",
                    "text": "encapsulated with the ecosystem layer"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:50",
                    "end": "00:13:55",
                    "text": "and what we call it as like you can call"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:52",
                    "end": "00:13:58",
                    "text": "it as an ESG layer because as we are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:55",
                    "end": "00:14:00",
                    "text": "building the Indian soverent track to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:58",
                    "end": "00:14:02",
                    "text": "build our own AIS we need to be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:00",
                    "end": "00:14:05",
                    "text": "absolutely cognizant of the fact what"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:02",
                    "end": "00:14:06",
                    "text": "kind of carbon footprints uh while we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:05",
                    "end": "00:14:08",
                    "text": "are creating our own data centers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:06",
                    "end": "00:14:10",
                    "text": "building our own clouds what kind of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:08",
                    "end": "00:14:13",
                    "text": "footprints we are having. So somewhere"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:10",
                    "end": "00:14:16",
                    "text": "in back of this journey we should be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:13",
                    "end": "00:14:18",
                    "text": "cognizant of this facts as well how ESG"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:16",
                    "end": "00:14:21",
                    "text": "can actually have an impact in this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:18",
                    "end": "00:14:22",
                    "text": "entire ball game. So that's that's all"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:21",
                    "end": "00:14:24",
                    "text": "my thoughts abishek."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:22",
                    "end": "00:14:27",
                    "text": ">> Thank you. Thank you so much P. I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:24",
                    "end": "00:14:29",
                    "text": "uh yeah"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:27",
                    "end": "00:14:31",
                    "text": "my my takeaway from you is that open"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:29",
                    "end": "00:14:33",
                    "text": "source but with guardrails"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:31",
                    "end": "00:14:35",
                    "text": "um you know quick to go to market"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:33",
                    "end": "00:14:37",
                    "text": "innovative. Yeah and you also talked"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:35",
                    "end": "00:14:39",
                    "text": "about the consumption layer but most"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:37",
                    "end": "00:14:41",
                    "text": "importantly you came towards the end on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:39",
                    "end": "00:14:43",
                    "text": "the ESG side of it. I'm very sure we are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:41",
                    "end": "00:14:45",
                    "text": "going to have one more session sometime"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:43",
                    "end": "00:14:46",
                    "text": "in future on ESG and we'd love to invite"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:45",
                    "end": "00:14:48",
                    "text": "you and express your thoughts because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:46",
                    "end": "00:14:51",
                    "text": "after all data centers wherever we read"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:48",
                    "end": "00:14:53",
                    "text": "US or anywhere local communities are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:51",
                    "end": "00:14:55",
                    "text": "they they are really affected by it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:53",
                    "end": "00:14:59",
                    "text": "Thank you so much Pri. Okay. So uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:55",
                    "end": "00:15:01",
                    "text": "towards the end I come to Mandar. Uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:59",
                    "end": "00:15:03",
                    "text": "last but not the least, Mandar is is the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:01",
                    "end": "00:15:06",
                    "text": "national security officer of Microsoft"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:03",
                    "end": "00:15:08",
                    "text": "India and Asia Pacific and is at the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:06",
                    "end": "00:15:12",
                    "text": "forefront of Microsoft's uh security"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:08",
                    "end": "00:15:15",
                    "text": "initiatives including cyber. So Mandard"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:12",
                    "end": "00:15:17",
                    "text": "your thoughts about adoption of AI in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:15",
                    "end": "00:15:20",
                    "text": "general because your platform as we say"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:17",
                    "end": "00:15:21",
                    "text": "you know various layers while that layer"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:20",
                    "end": "00:15:24",
                    "text": "can be addressed later but what is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:21",
                    "end": "00:15:26",
                    "text": "whatever is built on top of it is is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:24",
                    "end": "00:15:29",
                    "text": "actually what we debated can be you know"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:26",
                    "end": "00:15:30",
                    "text": "the early adopters of sovereign AI so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:29",
                    "end": "00:15:33",
                    "text": "over to you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:30",
                    "end": "00:15:34",
                    "text": ">> right thanks I think interesting and uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:33",
                    "end": "00:15:38",
                    "text": "I have the advantage of speaking after"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:34",
                    "end": "00:15:40",
                    "text": "everybody so so I will uh I'll just take"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:38",
                    "end": "00:15:42",
                    "text": "one more minute right I think a lot of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:40",
                    "end": "00:15:45",
                    "text": "speakers before me talked about"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:42",
                    "end": "00:15:47",
                    "text": "sovereignity and uh different words used"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:45",
                    "end": "00:15:50",
                    "text": "visceral and emotional and all that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:47",
                    "end": "00:15:52",
                    "text": "right I think in my opinion um"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:50",
                    "end": "00:15:54",
                    "text": "sovereignity has been obscured or"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:52",
                    "end": "00:15:56",
                    "text": "confused by a lot of interest right and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:54",
                    "end": "00:15:60",
                    "text": "interest from everybody from government"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:56",
                    "end": "00:16:02",
                    "text": "from tech companies and um of course uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:60",
                    "end": "00:16:04",
                    "text": "general audience right but sovereignty"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:02",
                    "end": "00:16:06",
                    "text": "and uh in my role I work with lot of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:04",
                    "end": "00:16:09",
                    "text": "governments including Indian government"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:06",
                    "end": "00:16:11",
                    "text": "I have I had privilege of working with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:09",
                    "end": "00:16:13",
                    "text": "some of my uh some of the Indian defense"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:11",
                    "end": "00:16:15",
                    "text": "folks as well. Solverity is fairly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:13",
                    "end": "00:16:17",
                    "text": "simple, right? Essentially, it's four"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:15",
                    "end": "00:16:20",
                    "text": "layers. The layer number one is data,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:17",
                    "end": "00:16:23",
                    "text": "right? Data sovereignity. We whole"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:20",
                    "end": "00:16:25",
                    "text": "started with data residency. But then"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:23",
                    "end": "00:16:26",
                    "text": "given the geopolitical situation and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:25",
                    "end": "00:16:28",
                    "text": "some of the incidents, we actually now"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:26",
                    "end": "00:16:30",
                    "text": "talk about data sovereignty. How is my"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:28",
                    "end": "00:16:33",
                    "text": "data protected? Nobody else has access."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:30",
                    "end": "00:16:36",
                    "text": "Nobody else has jurisdictional access to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:33",
                    "end": "00:16:38",
                    "text": "my data. Right? That's data sovereignty."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:36",
                    "end": "00:16:40",
                    "text": "The second layer is operational"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:38",
                    "end": "00:16:43",
                    "text": "sovereignty. uh couple of speakers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:40",
                    "end": "00:16:45",
                    "text": "before me spoke spoke about how can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:43",
                    "end": "00:16:47",
                    "text": "nobody else switch me off right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:45",
                    "end": "00:16:51",
                    "text": "operational so also is fairly simple so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:47",
                    "end": "00:16:54",
                    "text": "that my business can continue uh I can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:51",
                    "end": "00:16:55",
                    "text": "continue to trust that cloud or AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:54",
                    "end": "00:16:58",
                    "text": "without somebody else having ability to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:55",
                    "end": "00:16:59",
                    "text": "switch off tech this is where things"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:58",
                    "end": "00:17:02",
                    "text": "become little complicated and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:59",
                    "end": "00:17:03",
                    "text": "interesting right tech soy means I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:02",
                    "end": "00:17:06",
                    "text": "should be able to build on top of that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:03",
                    "end": "00:17:08",
                    "text": "right I'll not get into bit of clos"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:06",
                    "end": "00:17:10",
                    "text": "source open source commercial right but"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:08",
                    "end": "00:17:13",
                    "text": "that's where the issue starts happening"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:10",
                    "end": "00:17:15",
                    "text": "because tech also means the tech has to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:13",
                    "end": "00:17:17",
                    "text": "be not only in my control but it has to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:15",
                    "end": "00:17:19",
                    "text": "be good right the classic example"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:17",
                    "end": "00:17:20",
                    "text": "classic debate that we get again we have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:19",
                    "end": "00:17:23",
                    "text": "eminent defense folks so I'll refer to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:20",
                    "end": "00:17:25",
                    "text": "that right is India willing to wait 10"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:23",
                    "end": "00:17:27",
                    "text": "years for AMA to come fifth generation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:25",
                    "end": "00:17:30",
                    "text": "fighter or is India willing to buy"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:27",
                    "end": "00:17:32",
                    "text": "Rafael to fill that gap right uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:30",
                    "end": "00:17:34",
                    "text": "sovereign aircraft maybe we wait 10"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:32",
                    "end": "00:17:36",
                    "text": "years but we have needs today that's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:34",
                    "end": "00:17:38",
                    "text": "where the tech sovereignty becomes a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:36",
                    "end": "00:17:40",
                    "text": "very very interesting But sorry to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:38",
                    "end": "00:17:43",
                    "text": "pitching but just interesting input. The"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:40",
                    "end": "00:17:44",
                    "text": "AMA is coming with 50% indigenization."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:43",
                    "end": "00:17:47",
                    "text": ">> Sure. Sir would agree with me."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:44",
                    "end": "00:17:47",
                    "text": ">> No no absolutely I'm saying AMA drawing"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:47",
                    "end": "00:17:50",
                    "text": "a parallel."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:47",
                    "end": "00:17:51",
                    "text": ">> No no I'm saying AMA is great but AMA"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:50",
                    "end": "00:17:53",
                    "text": "will come maybe 10 years."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:51",
                    "end": "00:17:54",
                    "text": ">> Sorry not AMA. Rafal is coming with 50%"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:53",
                    "end": "00:17:56",
                    "text": "Indianization."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:54",
                    "end": "00:17:58",
                    "text": ">> Yeah of course of course. Yeah. So uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:56",
                    "end": "00:17:60",
                    "text": "but the point I'm making is AMA is 10"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:58",
                    "end": "00:18:03",
                    "text": "years away. Would we want or would we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:60",
                    "end": "00:18:05",
                    "text": "want 50% Indianization and use uh Rafal"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:03",
                    "end": "00:18:08",
                    "text": "today? Now that's the balance on tech"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:05",
                    "end": "00:18:10",
                    "text": "right? If we insist on every layer has"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:08",
                    "end": "00:18:12",
                    "text": "to be India then we probably will have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:10",
                    "end": "00:18:14",
                    "text": "to wait right that's the balance and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:12",
                    "end": "00:18:16",
                    "text": "finally the AI sovereignity right the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:14",
                    "end": "00:18:18",
                    "text": "fourth layer which was not on top of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:16",
                    "end": "00:18:20",
                    "text": "everybody's mind right the three layers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:18",
                    "end": "00:18:22",
                    "text": "have been talked for pretty much for a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:20",
                    "end": "00:18:24",
                    "text": "long time the data the operation and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:22",
                    "end": "00:18:26",
                    "text": "tech so even in the cloud world but AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:24",
                    "end": "00:18:29",
                    "text": "already came now AI so is a little bit"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:26",
                    "end": "00:18:32",
                    "text": "more complicated than cloud sovereignity"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:29",
                    "end": "00:18:34",
                    "text": "or or or any other tech soy right why"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:32",
                    "end": "00:18:37",
                    "text": "because I think speakers before me spoke"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:34",
                    "end": "00:18:39",
                    "text": "so I'll not go detail there but AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:37",
                    "end": "00:18:41",
                    "text": "brings a perspective right that means"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:39",
                    "end": "00:18:43",
                    "text": "the data that it is trained on has to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:41",
                    "end": "00:18:46",
                    "text": "have the Indian perspective right it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:43",
                    "end": "00:18:48",
                    "text": "needs to have all the nuances which in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:46",
                    "end": "00:18:50",
                    "text": "the first three layers were not so much"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:48",
                    "end": "00:18:52",
                    "text": "so data solidity is not only whether it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:50",
                    "end": "00:18:54",
                    "text": "is running in India whether it is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:52",
                    "end": "00:18:56",
                    "text": "developed by Indians it's also is it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:54",
                    "end": "00:18:58",
                    "text": "trained on Indian ethics is it trained"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:56",
                    "end": "00:19:01",
                    "text": "on Indian languages and all that so I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:58",
                    "end": "00:19:04",
                    "text": "think AI so is a step beyond that right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:01",
                    "end": "00:19:06",
                    "text": "and because most of the models are uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:04",
                    "end": "00:19:08",
                    "text": "like black blocks explanative"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:06",
                    "end": "00:19:09",
                    "text": "explanability is a challenge. I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:08",
                    "end": "00:19:11",
                    "text": "that's where AI already becomes"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:09",
                    "end": "00:19:13",
                    "text": "extremely important, right?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:11",
                    "end": "00:19:15",
                    "text": ">> Um and that's where I again agree with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:13",
                    "end": "00:19:17",
                    "text": "what I think Peter you mentioned the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:15",
                    "end": "00:19:19",
                    "text": "small language models Martin sorry you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:17",
                    "end": "00:19:21",
                    "text": "mentioned small language model"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:19",
                    "end": "00:19:23",
                    "text": "purpose-built models are probably making"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:21",
                    "end": "00:19:26",
                    "text": "more sense. I think that's the way we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:23",
                    "end": "00:19:28",
                    "text": "think generally of sovereignity right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:26",
                    "end": "00:19:30",
                    "text": "the question that we need to all look at"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:28",
                    "end": "00:19:32",
                    "text": "is solidity is not binary it's not zero"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:30",
                    "end": "00:19:35",
                    "text": "or one right as it is typically made out"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:32",
                    "end": "00:19:37",
                    "text": "to be it's a spectrum different"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:35",
                    "end": "00:19:40",
                    "text": "departments different organizations"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:37",
                    "end": "00:19:42",
                    "text": "different applications need different"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:40",
                    "end": "00:19:44",
                    "text": "level of sovereignty somewhere you need"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:42",
                    "end": "00:19:45",
                    "text": "to have operations control versus"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:44",
                    "end": "00:19:47",
                    "text": "somebody somewhere you will have to end"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:45",
                    "end": "00:19:50",
                    "text": "to end the stack like in defense needs"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:47",
                    "end": "00:19:53",
                    "text": "to be developed ourself right uh the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:50",
                    "end": "00:19:56",
                    "text": "final point that I'll make here is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:53",
                    "end": "00:19:59",
                    "text": "see life has come full circle. Uh those"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:56",
                    "end": "00:20:01",
                    "text": "who are in IT for maybe 25 odd years we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:59",
                    "end": "00:20:03",
                    "text": "saw mainframe where computing was"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:01",
                    "end": "00:20:05",
                    "text": "centralized to distributed computing"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:03",
                    "end": "00:20:07",
                    "text": "where computing was distributed to cloud"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:05",
                    "end": "00:20:09",
                    "text": "where again that computing become little"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:07",
                    "end": "00:20:12",
                    "text": "centralized. Now we are talking when"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:09",
                    "end": "00:20:13",
                    "text": "cloud started 10 15 years back right all"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:12",
                    "end": "00:20:15",
                    "text": "the big techs believed that there are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:13",
                    "end": "00:20:16",
                    "text": "going to be five regions in the world"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:15",
                    "end": "00:20:19",
                    "text": "right somewhere in Europe somewhere in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:16",
                    "end": "00:20:21",
                    "text": "Asia somewhere in US and everybody will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:19",
                    "end": "00:20:24",
                    "text": "use cloud from those five regions today"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:21",
                    "end": "00:20:26",
                    "text": "reality is Microsoft has 75 regions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:24",
                    "end": "00:20:29",
                    "text": "right so we moved from five regions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:26",
                    "end": "00:20:31",
                    "text": "literally 10 years back to 75 regions I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:29",
                    "end": "00:20:33",
                    "text": "think it's time to take the next step"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:31",
                    "end": "00:20:35",
                    "text": "where we are saying the region is not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:33",
                    "end": "00:20:37",
                    "text": "only in my control but region is also in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:35",
                    "end": "00:20:40",
                    "text": "customers data center so I think what we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:37",
                    "end": "00:20:42",
                    "text": "will see happening in tech is that the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:40",
                    "end": "00:20:45",
                    "text": "sovereign clouds will become a reality"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:42",
                    "end": "00:20:47",
                    "text": "right for example in Microsoft we have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:45",
                    "end": "00:20:49",
                    "text": "now created something something called a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:47",
                    "end": "00:20:51",
                    "text": "sovereign private cloud which can run in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:49",
                    "end": "00:20:55",
                    "text": "customers own data center under complete"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:51",
                    "end": "00:20:56",
                    "text": "customer control disconnected air gap we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:55",
                    "end": "00:20:58",
                    "text": "did not do that maybe two years back"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:56",
                    "end": "00:21:01",
                    "text": "that's the reality right we we talk kept"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:58",
                    "end": "00:21:03",
                    "text": "on talking about softwaredefined"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:01",
                    "end": "00:21:05",
                    "text": "sovereignty until about a year back"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:03",
                    "end": "00:21:07",
                    "text": ">> today we are saying okay we have a full"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:05",
                    "end": "00:21:09",
                    "text": "stack sovereign private cloud and I'm"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:07",
                    "end": "00:21:11",
                    "text": "taking Microsoft as an example. Our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:09",
                    "end": "00:21:11",
                    "text": "competition and other companies also"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:11",
                    "end": "00:21:13",
                    "text": "have similar."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:11",
                    "end": "00:21:15",
                    "text": ">> No, that's a very valid point. In fact,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:13",
                    "end": "00:21:17",
                    "text": "competitions like SAP already deployed"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:15",
                    "end": "00:21:17",
                    "text": "sovereign cloud and that's the way to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:17",
                    "end": "00:21:18",
                    "text": "go."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:17",
                    "end": "00:21:21",
                    "text": ">> That's sovereign public. What I'm saying"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:18",
                    "end": "00:21:22",
                    "text": "is sovereign private which can run in an"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:21",
                    "end": "00:21:23",
                    "text": "aircapped environment in defense of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:22",
                    "end": "00:21:25",
                    "text": "course of course"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:23",
                    "end": "00:21:28",
                    "text": ">> and that's a reality because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:25",
                    "end": "00:21:30",
                    "text": ">> last point we will have to take us out"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:28",
                    "end": "00:21:32",
                    "text": "of the equation the zero trust model"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:30",
                    "end": "00:21:34",
                    "text": "right you don't even need to trust the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:32",
                    "end": "00:21:35",
                    "text": "company that gives you tech. That's the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:34",
                    "end": "00:21:36",
                    "text": "only way we can build content."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:35",
                    "end": "00:21:38",
                    "text": ">> Thank you. Thank you. Right. Very valid"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:36",
                    "end": "00:21:40",
                    "text": "point Mandar and it was important to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:38",
                    "end": "00:21:42",
                    "text": "bring the industry perspective here to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:40",
                    "end": "00:21:44",
                    "text": "complete the full picture. Uh ladies and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:42",
                    "end": "00:21:46",
                    "text": "gentlemen we have almost run out of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:44",
                    "end": "00:21:48",
                    "text": "time. I would request Pier to give his"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:46",
                    "end": "00:21:50",
                    "text": "closing remarks and then we'll uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:48",
                    "end": "00:21:53",
                    "text": "conclude the session. P what you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:50",
                    "end": "00:21:56",
                    "text": ">> okay so try to keep it brief. So what we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:53",
                    "end": "00:21:58",
                    "text": "have been discussing here it's we would"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:56",
                    "end": "00:22:01",
                    "text": "say it's when we talk about sovereignity"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:58",
                    "end": "00:22:04",
                    "text": "and AI it's openness versus security the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:01",
                    "end": "00:22:05",
                    "text": "trade-offs speed of innovation versus"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:04",
                    "end": "00:22:09",
                    "text": "governance"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:05",
                    "end": "00:22:11",
                    "text": "global integration versus autonomy"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:09",
                    "end": "00:22:12",
                    "text": "public infrastructure versus private"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:11",
                    "end": "00:22:16",
                    "text": "innovation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:12",
                    "end": "00:22:18",
                    "text": "So these are the extremes and um what"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:16",
                    "end": "00:22:21",
                    "text": "what what I wish was would be for India"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:18",
                    "end": "00:22:23",
                    "text": "would be if they if you could have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:21",
                    "end": "00:22:26",
                    "text": "something in the middle of these"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:23",
                    "end": "00:22:31",
                    "text": "extremes um so the best of let's say the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:26",
                    "end": "00:22:35",
                    "text": "EU and the US EU very much regula"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:31",
                    "end": "00:22:37",
                    "text": "regulatory driven uh regulatory first"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:35",
                    "end": "00:22:40",
                    "text": "innovation second etc and the US"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:37",
                    "end": "00:22:43",
                    "text": "probably innovation first and then well"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:40",
                    "end": "00:22:46",
                    "text": "some control from the permanent um"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:43",
                    "end": "00:22:48",
                    "text": "anyway that comes. So if if that would"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:46",
                    "end": "00:22:51",
                    "text": "be something uh for India, I think this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:48",
                    "end": "00:22:55",
                    "text": "is a is a good rate good route to take."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:51",
                    "end": "00:22:59",
                    "text": "You said it uh collaborative models um"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:55",
                    "end": "00:23:03",
                    "text": "in uh in to invest in in infrastructure"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:59",
                    "end": "00:23:05",
                    "text": "something like um centers of excellence"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:03",
                    "end": "00:23:08",
                    "text": "um where public and private come"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:05",
                    "end": "00:23:11",
                    "text": "together to solve the real world"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:08",
                    "end": "00:23:13",
                    "text": "problems in architecture in healthcare"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:11",
                    "end": "00:23:15",
                    "text": "and all these kind of things. So if that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:13",
                    "end": "00:23:17",
                    "text": "would be a route for India I would be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:15",
                    "end": "00:23:21",
                    "text": "more than glad and it brings the best"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:17",
                    "end": "00:23:23",
                    "text": "the best together from both worlds."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:21",
                    "end": "00:23:25",
                    "text": "Thank you. Thank you Pier. So towards"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:23",
                    "end": "00:23:26",
                    "text": "the end of the session I'll request all"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:25",
                    "end": "00:23:28",
                    "text": "speakers to please come forward. We are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:26",
                    "end": "00:23:30",
                    "text": "releasing a background paper and I would"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:28",
                    "end": "00:23:34",
                    "text": "request Nachal Mishra and Akilshuta from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:30",
                    "end": "00:23:34",
                    "text": "KPMG to please come up as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:34",
                    "end": "00:23:38",
                    "text": "Show it please."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:47",
                    "end": "00:23:50",
                    "text": "Let's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:02",
                    "end": "00:24:05",
                    "text": "get out of this."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "Spanish"
},
    {
    "_id": "6992c1804c5e8f0974c86fbc",
    "projectId": "892208",
    "projectName": "Sovereign AI and National Security: India’s Digital Path",
    "projectDescription": "Sovereign AI and National Security: India’s Digital Path",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-16T07:04:36.912Z",
    "projectUpdatedDate": "2026-02-16T07:04:36.912Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-16T07:04:36.912Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO_LINK",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames"
        },
        {
            "col1": "P1",
            "col2": "https://youtu.be/58V4tbgvnwc",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEOLINK",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:04",
                    "text": "You saw Russia Ukraine war was going on."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:02",
                    "end": "00:00:07",
                    "text": "Swift was stopped for Russia. So you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:04",
                    "end": "00:00:09",
                    "text": "will have to with the way we own digital"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:07",
                    "end": "00:00:11",
                    "text": "public infrastructure in India. We'll"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:09",
                    "end": "00:00:13",
                    "text": "have to create a cognitive public"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:11",
                    "end": "00:00:15",
                    "text": "infrastructure. How soever difficult it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:13",
                    "end": "00:00:17",
                    "text": "is and that's why pair it's visceral"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:15",
                    "end": "00:00:19",
                    "text": "it's existential and we will have to do"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:17",
                    "end": "00:00:22",
                    "text": "it and probably it's time that we should"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:19",
                    "end": "00:00:23",
                    "text": "do it. I yesterday wrote an article in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:22",
                    "end": "00:00:25",
                    "text": "uh the Sunday gajin which where where I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:23",
                    "end": "00:00:28",
                    "text": "called it the the architecture of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:25",
                    "end": "00:00:30",
                    "text": "independence right it is the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:28",
                    "end": "00:00:32",
                    "text": "architecture of independence. There's no"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:30",
                    "end": "00:00:34",
                    "text": "der of you know experts in India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:32",
                    "end": "00:00:36",
                    "text": "There's no der of talent in India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:34",
                    "end": "00:00:38",
                    "text": "There's no der of only thing is compute"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:36",
                    "end": "00:00:41",
                    "text": "which I think geopolitically we should"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:38",
                    "end": "00:00:44",
                    "text": "solve. That again is sovereignty right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:41",
                    "end": "00:00:46",
                    "text": "because people are using GPUs to control"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:44",
                    "end": "00:00:49",
                    "text": "geopolitics"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:46",
                    "end": "00:00:51",
                    "text": "and that is why we will have to be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:49",
                    "end": "00:00:52",
                    "text": "sovereign at all you know all the layers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:51",
                    "end": "00:00:52",
                    "text": "of the cake"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:52",
                    "end": "00:00:55",
                    "text": ">> right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:52",
                    "end": "00:00:57",
                    "text": ">> and I believe that uh when you apply"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:55",
                    "end": "00:01:01",
                    "text": "these systems to policing you will have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:57",
                    "end": "00:01:03",
                    "text": "to go for trust and transparency"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:01",
                    "end": "00:01:04",
                    "text": "uh and it has created problems on the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:03",
                    "end": "00:01:07",
                    "text": "other side it's become a force"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:04",
                    "end": "00:01:09",
                    "text": "multiplier it has hugely enabled"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:07",
                    "end": "00:01:12",
                    "text": "criminal ecosystem we polymorphic"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:09",
                    "end": "00:01:14",
                    "text": "malware. You have, you know, there was a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:12",
                    "end": "00:01:17",
                    "text": "a company called Arup. I think it's a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:14",
                    "end": "00:01:20",
                    "text": "advertisement company or something where"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:17",
                    "end": "00:01:22",
                    "text": "the CEO came on a live zoom call and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:20",
                    "end": "00:01:26",
                    "text": "they lost uh $30 million. Right? This is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:22",
                    "end": "00:01:28",
                    "text": "a state at which AI is evolving. Do we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:26",
                    "end": "00:01:30",
                    "text": "have realtime policing? Sir, you know,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:28",
                    "end": "00:01:32",
                    "text": "explained that policing is not so real"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:30",
                    "end": "00:01:36",
                    "text": "time. So, you will have to increase both"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:32",
                    "end": "00:01:38",
                    "text": "capability and capacity and this needs a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:36",
                    "end": "00:01:40",
                    "text": "serious thought right now. So pay we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:38",
                    "end": "00:01:43",
                    "text": "need sovereign AI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:40",
                    "end": "00:01:46",
                    "text": ">> Thank you. Thank you sir. And again as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:43",
                    "end": "00:01:49",
                    "text": "expected very very relevant points sir"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:46",
                    "end": "00:01:51",
                    "text": "that reminds me of of interview by"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:49",
                    "end": "00:01:53",
                    "text": "Jensen Huang a few years back. He said I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:51",
                    "end": "00:01:56",
                    "text": "met your prime minister all before this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:53",
                    "end": "00:01:58",
                    "text": "AI hype uh when it's about in very"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:56",
                    "end": "00:01:59",
                    "text": "initial stages and he said your prime"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:58",
                    "end": "00:02:02",
                    "text": "minister told me that I don't want to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:59",
                    "end": "00:02:04",
                    "text": "export data to import intelligence. He"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:02",
                    "end": "00:02:07",
                    "text": "said that was really remarkable for a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:04",
                    "end": "00:02:08",
                    "text": "world leader to think in those times. So"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:07",
                    "end": "00:02:11",
                    "text": "I think very relevant points you also"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:08",
                    "end": "00:02:13",
                    "text": "echoed but good that sir you brought"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:11",
                    "end": "00:02:14",
                    "text": "compute in in picture because compute is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:13",
                    "end": "00:02:16",
                    "text": "something where we really have to make"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:14",
                    "end": "00:02:19",
                    "text": "pro in fact the rest of the world and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:16",
                    "end": "00:02:22",
                    "text": "then I would love to bring pair maybe at"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:19",
                    "end": "00:02:24",
                    "text": "later stage. So uh with this then I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:22",
                    "end": "00:02:26",
                    "text": "would like to move to Martin. Martin"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:24",
                    "end": "00:02:28",
                    "text": "thanks a lot uh first of all traveling"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:26",
                    "end": "00:02:31",
                    "text": "all the way to India to attend the AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:28",
                    "end": "00:02:35",
                    "text": "summit and being with us. M in your role"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:31",
                    "end": "00:02:38",
                    "text": "as uh in terod data as a leader in in in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:35",
                    "end": "00:02:41",
                    "text": "data analytics big data implementation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:38",
                    "end": "00:02:43",
                    "text": "uh you have been quoted twice in data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:41",
                    "end": "00:02:45",
                    "text": "IQ's 100 most influential people in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:43",
                    "end": "00:02:47",
                    "text": "datadriven business and what's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:45",
                    "end": "00:02:49",
                    "text": "remarkable I was looking at your website"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:47",
                    "end": "00:02:53",
                    "text": "where you are now saying that uh stop"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:49",
                    "end": "00:02:55",
                    "text": "managing data activate intelligence so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:53",
                    "end": "00:02:57",
                    "text": "and I also read your article about the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:55",
                    "end": "00:02:58",
                    "text": "large scale compute requirement which is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:57",
                    "end": "00:03:01",
                    "text": "going to come for example you give a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:58",
                    "end": "00:03:03",
                    "text": "healthcare example where every patient"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:01",
                    "end": "00:03:05",
                    "text": "will be followed by an agent and he'll"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:03",
                    "end": "00:03:07",
                    "text": "be firing like thousands hundreds and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:05",
                    "end": "00:03:09",
                    "text": "thousands of queries to make sure that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:07",
                    "end": "00:03:12",
                    "text": "you have that hypers uh personalization"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:09",
                    "end": "00:03:15",
                    "text": "experience. So with that kind of compute"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:12",
                    "end": "00:03:18",
                    "text": "requirement Martin do you think that uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:15",
                    "end": "00:03:21",
                    "text": "we are ready for so sovereignity from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:18",
                    "end": "00:03:24",
                    "text": "day one or is going or as everybody said"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:21",
                    "end": "00:03:25",
                    "text": "you have to tackle it layer wise so what"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:24",
                    "end": "00:03:28",
                    "text": "do you think especially when you come to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:25",
                    "end": "00:03:30",
                    "text": "the layer of data you know how the data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:28",
                    "end": "00:03:32",
                    "text": "structures uh a country like India can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:30",
                    "end": "00:03:35",
                    "text": "make sure that they are sovereign in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:32",
                    "end": "00:03:36",
                    "text": "those aspects yeah so over to you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:35",
                    "end": "00:03:38",
                    "text": ">> thank you and uh and good morning"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:36",
                    "end": "00:03:40",
                    "text": "everybody lovely to be here um there's a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:38",
                    "end": "00:03:41",
                    "text": "lot to unpack in your question and we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:40",
                    "end": "00:03:43",
                    "text": "should come back and we should talk"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:41",
                    "end": "00:03:46",
                    "text": "about the implications of potentially"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:43",
                    "end": "00:03:47",
                    "text": "millions of agents running thousands of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:46",
                    "end": "00:03:49",
                    "text": "queries because I think that is a very"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:47",
                    "end": "00:03:50",
                    "text": "important consequence uh of where we're"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:49",
                    "end": "00:03:52",
                    "text": "going that we should think about. But"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:50",
                    "end": "00:03:54",
                    "text": "you're absolutely right at Terodata we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:52",
                    "end": "00:03:56",
                    "text": "think that the first and the most"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:54",
                    "end": "00:03:58",
                    "text": "fundamental layer is the data layer."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:56",
                    "end": "00:03:60",
                    "text": "There is no good AI without good data."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:58",
                    "end": "00:04:02",
                    "text": "So it's incredibly important we get that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:60",
                    "end": "00:04:03",
                    "text": "right. And there are some principles"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:02",
                    "end": "00:04:06",
                    "text": "that have proven to be more or less"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:03",
                    "end": "00:04:08",
                    "text": "timeless. And if I speak about timeless"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:06",
                    "end": "00:04:09",
                    "text": "in in terms of it, I'm speaking they've"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:08",
                    "end": "00:04:11",
                    "text": "proven to be good over three or four"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:09",
                    "end": "00:04:13",
                    "text": "decades. That's the kind of event"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:11",
                    "end": "00:04:15",
                    "text": "horizon, isn't it, for technology. Um,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:13",
                    "end": "00:04:17",
                    "text": "so we need to focus on capturing the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:15",
                    "end": "00:04:18",
                    "text": "most detailed data that we can. We need"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:17",
                    "end": "00:04:21",
                    "text": "to make sure that we are capable of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:18",
                    "end": "00:04:23",
                    "text": "integrating that data so that we can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:21",
                    "end": "00:04:26",
                    "text": "solve complex problems across domains. I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:23",
                    "end": "00:04:28",
                    "text": "guess in in defense, air, space,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:26",
                    "end": "00:04:31",
                    "text": "maritime, land, cyber increasingly that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:28",
                    "end": "00:04:33",
                    "text": "requires that we can connect the dots."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:31",
                    "end": "00:04:36",
                    "text": "Um and then we need to expose a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:33",
                    "end": "00:04:39",
                    "text": "simplified representation of that data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:36",
                    "end": "00:04:42",
                    "text": "um so that a hard-pressed individual uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:39",
                    "end": "00:04:44",
                    "text": "operating under pressure uh can make"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:42",
                    "end": "00:04:46",
                    "text": "sense of the data and also so that we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:44",
                    "end": "00:04:49",
                    "text": "can as we increasingly open that data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:46",
                    "end": "00:04:51",
                    "text": "platform to agents so we reduce the risk"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:49",
                    "end": "00:04:53",
                    "text": "of hallucination for the agents as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:51",
                    "end": "00:04:55",
                    "text": "So those are three principles that I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:53",
                    "end": "00:04:57",
                    "text": "think have stood the test of time. When"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:55",
                    "end": "00:04:58",
                    "text": "we think about a modern data foundation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:57",
                    "end": "00:05:00",
                    "text": "and we tend to use this word data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:58",
                    "end": "00:05:02",
                    "text": "modernization a lot right now in the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:00",
                    "end": "00:05:04",
                    "text": "industry"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:02",
                    "end": "00:05:06",
                    "text": "a a difficult term I think I uh I I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:04",
                    "end": "00:05:08",
                    "text": "would say that uh data modernization can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:06",
                    "end": "00:05:09",
                    "text": "mean whatever the vendor sat across you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:08",
                    "end": "00:05:11",
                    "text": "wants to sell you that day. So we should"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:09",
                    "end": "00:05:13",
                    "text": "be careful with the term but I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:11",
                    "end": "00:05:14",
                    "text": "there are two things that are really"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:13",
                    "end": "00:05:17",
                    "text": "important in in this idea of data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:14",
                    "end": "00:05:19",
                    "text": "modernization. The first is this idea of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:17",
                    "end": "00:05:21",
                    "text": "open table formats and and the the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:19",
                    "end": "00:05:24",
                    "text": "promise of open table formats is that we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:21",
                    "end": "00:05:26",
                    "text": "can store more data more reliably and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:24",
                    "end": "00:05:28",
                    "text": "crucially more cost-ffectively and cost"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:26",
                    "end": "00:05:29",
                    "text": "effective is important. I was speaking"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:28",
                    "end": "00:05:32",
                    "text": "to the ministry of defense at one of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:29",
                    "end": "00:05:34",
                    "text": "smaller European nations last week their"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:32",
                    "end": "00:05:37",
                    "text": "next generation AI data platform that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:34",
                    "end": "00:05:39",
                    "text": "they envisaged to integrate those five"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:37",
                    "end": "00:05:42",
                    "text": "domains they think will be a minimum of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:39",
                    "end": "00:05:45",
                    "text": "2.5 pabytes of data. for a country of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:42",
                    "end": "00:05:47",
                    "text": "the size complexity of India um I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:45",
                    "end": "00:05:49",
                    "text": "you can multiply that by at least 10. So"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:47",
                    "end": "00:05:51",
                    "text": "being able to store large volumes of of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:49",
                    "end": "00:05:53",
                    "text": "data and to to fuse that data that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:51",
                    "end": "00:05:55",
                    "text": "sensor data cost effectively is very"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:53",
                    "end": "00:05:57",
                    "text": "important. But the more important"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:55",
                    "end": "00:05:60",
                    "text": "consequence of open table formats is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:57",
                    "end": "00:06:01",
                    "text": "this idea of interoperability."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:60",
                    "end": "00:06:04",
                    "text": "This idea that we cannot possibly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:01",
                    "end": "00:06:06",
                    "text": "predict sat here in this room today"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:04",
                    "end": "00:06:08",
                    "text": "which models and which technologies will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:06",
                    "end": "00:06:10",
                    "text": "be the most important four or five years"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:08",
                    "end": "00:06:12",
                    "text": "from now. And this idea that different"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:10",
                    "end": "00:06:13",
                    "text": "technologies, different processing"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:12",
                    "end": "00:06:15",
                    "text": "engines have different strengths and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:13",
                    "end": "00:06:17",
                    "text": "weaknesses for different applications."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:15",
                    "end": "00:06:19",
                    "text": "And so we need to be able to support"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:17",
                    "end": "00:06:20",
                    "text": "many of them. And so the idea of having"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:19",
                    "end": "00:06:22",
                    "text": "a data foundation that we can bring"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:20",
                    "end": "00:06:24",
                    "text": "multiple tools to is incredibly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:22",
                    "end": "00:06:26",
                    "text": "important in that conversation. The"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:24",
                    "end": "00:06:28",
                    "text": "other modernization conversation that's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:26",
                    "end": "00:06:29",
                    "text": "very important is data products, but we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:28",
                    "end": "00:06:32",
                    "text": "probably don't have time for that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:29",
                    "end": "00:06:33",
                    "text": "discussion today. And this now brings us"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:32",
                    "end": "00:06:35",
                    "text": "to why are we building this data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:33",
                    "end": "00:06:37",
                    "text": "foundation? Well, we're building this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:35",
                    "end": "00:06:38",
                    "text": "data foundation so that we can bring AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:37",
                    "end": "00:06:40",
                    "text": "to that data foundation and increasingly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:38",
                    "end": "00:06:42",
                    "text": "so we can bring agentic AI to that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:40",
                    "end": "00:06:43",
                    "text": "foundation. And I think there are a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:42",
                    "end": "00:06:46",
                    "text": "couple of things that are important to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:43",
                    "end": "00:06:48",
                    "text": "to think through there. The first is I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:46",
                    "end": "00:06:50",
                    "text": "think we can get"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:48",
                    "end": "00:06:52",
                    "text": "very wrapped up around the idea of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:50",
                    "end": "00:06:55",
                    "text": "training models around training the best"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:52",
                    "end": "00:06:56",
                    "text": "model around training our own model. I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:55",
                    "end": "00:06:58",
                    "text": "think that's a little bit of a trap."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:56",
                    "end": "00:07:00",
                    "text": "Honestly, when I look at um the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:58",
                    "end": "00:07:03",
                    "text": "performance of the open- source models,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:00",
                    "end": "00:07:06",
                    "text": "they're generally only six months behind"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:03",
                    "end": "00:07:07",
                    "text": "the proprietary models. So, right now,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:06",
                    "end": "00:07:09",
                    "text": "the open source models are catching up"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:07",
                    "end": "00:07:11",
                    "text": "with the proprietary models inside six"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:09",
                    "end": "00:07:12",
                    "text": "months. And those open source models, we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:11",
                    "end": "00:07:15",
                    "text": "can inspect the architecture, we can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:12",
                    "end": "00:07:18",
                    "text": "inspect the weights. So, we can we can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:15",
                    "end": "00:07:20",
                    "text": "be pretty confident in those models. And"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:18",
                    "end": "00:07:22",
                    "text": "so, I think we should focus less on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:20",
                    "end": "00:07:25",
                    "text": "training our own models and think more"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:22",
                    "end": "00:07:27",
                    "text": "about inference. A good model that I can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:25",
                    "end": "00:07:31",
                    "text": "score quickly and cost-effectively"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:27",
                    "end": "00:07:33",
                    "text": "is better nine times out of 10 than the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:31",
                    "end": "00:07:35",
                    "text": "best model that I can't get out of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:33",
                    "end": "00:07:37",
                    "text": "lab and into production. So we should we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:35",
                    "end": "00:07:38",
                    "text": "should focus on inference first. We"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:37",
                    "end": "00:07:40",
                    "text": "should also think about the diversity of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:38",
                    "end": "00:07:43",
                    "text": "models. There was a really interesting"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:40",
                    "end": "00:07:45",
                    "text": "paper by Nvidia last year called small"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:43",
                    "end": "00:07:47",
                    "text": "models of the future of agentic AI and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:45",
                    "end": "00:07:50",
                    "text": "in that paper they pointed out that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:47",
                    "end": "00:07:54",
                    "text": "smaller models are often cheaper to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:50",
                    "end": "00:07:55",
                    "text": "score faster to score and can be less"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:54",
                    "end": "00:07:58",
                    "text": "prone to hallucination when they're"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:55",
                    "end": "00:07:59",
                    "text": "focused on a task specific problem. So"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:58",
                    "end": "00:08:00",
                    "text": "better, cheaper, faster and more"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:59",
                    "end": "00:08:02",
                    "text": "reliable. That's a pretty good trade,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:00",
                    "end": "00:08:04",
                    "text": "right? So we should think about not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:02",
                    "end": "00:08:06",
                    "text": "having one single model to rule them"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:04",
                    "end": "00:08:08",
                    "text": "all. We need to start thinking about the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:06",
                    "end": "00:08:10",
                    "text": "models that are available to us as a box"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:08",
                    "end": "00:08:12",
                    "text": "of tools. And when we have a box of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:10",
                    "end": "00:08:14",
                    "text": "tools, we shouldn't always reach for the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:12",
                    "end": "00:08:16",
                    "text": "hammer just because it's the biggest"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:14",
                    "end": "00:08:17",
                    "text": "shiniest tool in the box. And then the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:16",
                    "end": "00:08:19",
                    "text": "last thing that I think is particularly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:17",
                    "end": "00:08:22",
                    "text": "relevant in a defense and a security"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:19",
                    "end": "00:08:24",
                    "text": "context is for I've been in I've been in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:22",
                    "end": "00:08:27",
                    "text": "this industry for a little over 30 years"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:24",
                    "end": "00:08:29",
                    "text": "now and mostly we've been concerned with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:27",
                    "end": "00:08:31",
                    "text": "structured data with rows and columns of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:29",
                    "end": "00:08:34",
                    "text": "numeric data. And the great promise of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:31",
                    "end": "00:08:36",
                    "text": "AI that we're seeing play out already is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:34",
                    "end": "00:08:38",
                    "text": "that we can now apply these models to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:36",
                    "end": "00:08:40",
                    "text": "unstructured data to images to audio to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:38",
                    "end": "00:08:42",
                    "text": "text and we can have multimodal"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:40",
                    "end": "00:08:44",
                    "text": "intelligence and this is incredibly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:42",
                    "end": "00:08:46",
                    "text": "important. So at Terodata for example"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:44",
                    "end": "00:08:47",
                    "text": "we've always worked with defense"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:46",
                    "end": "00:08:49",
                    "text": "agencies around supply chain"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:47",
                    "end": "00:08:51",
                    "text": "optimization to your point around"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:49",
                    "end": "00:08:52",
                    "text": "predictive maintenance. Today in terod"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:51",
                    "end": "00:08:54",
                    "text": "data we can support very complex"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:52",
                    "end": "00:08:58",
                    "text": "maritime surveillance applications where"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:54",
                    "end": "00:09:02",
                    "text": "we take large volumes of detailed data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:58",
                    "end": "00:09:03",
                    "text": "across multiple sensor types a and fuse"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:02",
                    "end": "00:09:05",
                    "text": "that intelligence for maritime"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:03",
                    "end": "00:09:06",
                    "text": "surveillance. We can assess camouflage"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:05",
                    "end": "00:09:08",
                    "text": "effectiveness. I think you made the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:06",
                    "end": "00:09:09",
                    "text": "point earlier that that highv value"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:08",
                    "end": "00:09:11",
                    "text": "assets are increasingly vulnerable to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:09",
                    "end": "00:09:14",
                    "text": "loitering munitions relatively"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:11",
                    "end": "00:09:17",
                    "text": "inexpensive loitering munitions. So we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:14",
                    "end": "00:09:19",
                    "text": "can capture images of camouflaged assets"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:17",
                    "end": "00:09:21",
                    "text": "like tanks and assess camouflage"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:19",
                    "end": "00:09:22",
                    "text": "effectiveness to protect those assets"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:21",
                    "end": "00:09:24",
                    "text": "and these are applications that would"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:22",
                    "end": "00:09:26",
                    "text": "not have been possible even 3 four five"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:24",
                    "end": "00:09:30",
                    "text": "years ago. So the the pace of change and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:26",
                    "end": "00:09:31",
                    "text": "innovation is spectacular and exciting."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:30",
                    "end": "00:09:34",
                    "text": ">> Thank you. Thank you Martin. I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:31",
                    "end": "00:09:37",
                    "text": "what really stood out to me is that in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:34",
                    "end": "00:09:39",
                    "text": "our quest for sovereignity I think using"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:37",
                    "end": "00:09:41",
                    "text": "open source data data model is really"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:39",
                    "end": "00:09:44",
                    "text": "important because you get to inspect the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:41",
                    "end": "00:09:46",
                    "text": "ar architecture no matter who's created"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:44",
                    "end": "00:09:49",
                    "text": "it at least that aspect can we bring"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:46",
                    "end": "00:09:50",
                    "text": "into soy very early on and also it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:49",
                    "end": "00:09:52",
                    "text": "resonates with general shiba earlier"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:50",
                    "end": "00:09:55",
                    "text": "said you know building superior"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:52",
                    "end": "00:09:60",
                    "text": "algorithms on top of it yeah great so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:55",
                    "end": "00:10:02",
                    "text": "now I come come to uh pri saxa pri is a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:60",
                    "end": "00:10:05",
                    "text": "director of global data analytics in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:02",
                    "end": "00:10:08",
                    "text": "concentrics uh recognized as one of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:05",
                    "end": "00:10:11",
                    "text": "10 most influential leader awarded as uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:08",
                    "end": "00:10:14",
                    "text": "one of the women leaders in AI and looks"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:11",
                    "end": "00:10:16",
                    "text": "after the entire genai initiative across"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:14",
                    "end": "00:10:19",
                    "text": "the world from concentrics. So pri great"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:16",
                    "end": "00:10:22",
                    "text": "to have you here and thanks for coming."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:19",
                    "end": "00:10:25",
                    "text": "So pri uh what we would like your inputs"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:22",
                    "end": "00:10:27",
                    "text": "is is on the public and private"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:25",
                    "end": "00:10:29",
                    "text": "collaboration and you have been at the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:27",
                    "end": "00:10:31",
                    "text": "forefront of it. So what do you think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:29",
                    "end": "00:10:34",
                    "text": "are the right models where both the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:31",
                    "end": "00:10:35",
                    "text": "teams can work together with a larger"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:34",
                    "end": "00:10:37",
                    "text": "focus on bringing sovereignty across"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:35",
                    "end": "00:10:40",
                    "text": "layers of AI? Over to you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:37",
                    "end": "00:10:42",
                    "text": ">> Sure. Thank you. Thanks Abishek. Am I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:40",
                    "end": "00:10:45",
                    "text": "audible? Okay. Great. So good good"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:42",
                    "end": "00:10:48",
                    "text": "morning everyone. So I would just like"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:45",
                    "end": "00:10:51",
                    "text": "to answer it by having very small blocks"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:48",
                    "end": "00:10:54",
                    "text": "to the entire ecosystem of data AI and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:51",
                    "end": "00:10:56",
                    "text": "analytics. So starting from data which"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:54",
                    "end": "00:10:59",
                    "text": "is the main thing the bread and butter"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:56",
                    "end": "00:11:02",
                    "text": "on which all the layers are added on top"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:59",
                    "end": "00:11:04",
                    "text": "of it as as you mentioned rightly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:02",
                    "end": "00:11:06",
                    "text": "capturing the right data having the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:04",
                    "end": "00:11:10",
                    "text": "right form of governance is absolutely"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:06",
                    "end": "00:11:12",
                    "text": "important. I think I think starting from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:10",
                    "end": "00:11:14",
                    "text": "residency the data residency aspect in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:12",
                    "end": "00:11:17",
                    "text": "which India has already initiated the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:14",
                    "end": "00:11:19",
                    "text": "DPDP act which is very very absolutely"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:17",
                    "end": "00:11:22",
                    "text": "relevant and that would be the first"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:19",
                    "end": "00:11:24",
                    "text": "foundational layers to it. So once we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:22",
                    "end": "00:11:27",
                    "text": "have ensured that all our healthcare"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:24",
                    "end": "00:11:29",
                    "text": "data PI information all the transactions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:27",
                    "end": "00:11:32",
                    "text": "we are made are residing in our own"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:29",
                    "end": "00:11:34",
                    "text": "country that's a foundational layer in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:32",
                    "end": "00:11:36",
                    "text": "terms of residency then we need to have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:34",
                    "end": "00:11:39",
                    "text": "the right layer of governance on top of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:36",
                    "end": "00:11:40",
                    "text": "it and when I say governance like do we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:39",
                    "end": "00:11:42",
                    "text": "have any biases in the data are we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:40",
                    "end": "00:11:45",
                    "text": "capturing in the right formats right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:42",
                    "end": "00:11:47",
                    "text": "shape so that is absolutely important"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:45",
                    "end": "00:11:50",
                    "text": "because data governance is going to be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:47",
                    "end": "00:11:52",
                    "text": "the next key thing for your success for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:50",
                    "end": "00:11:56",
                    "text": "any program any analytics or AI program"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:52",
                    "end": "00:11:58",
                    "text": "that you run across. Now here when we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:56",
                    "end": "00:12:01",
                    "text": "have the data residency solved,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:58",
                    "end": "00:12:04",
                    "text": "governance solved and then on top of it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:01",
                    "end": "00:12:06",
                    "text": "we try to apply any AI models on top of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:04",
                    "end": "00:12:08",
                    "text": "it and I completely agree to your point"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:06",
                    "end": "00:12:11",
                    "text": "which you said like open source are much"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:08",
                    "end": "00:12:13",
                    "text": "speedy to market we should leverage that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:11",
                    "end": "00:12:16",
                    "text": "but with all the relevant guard rails."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:13",
                    "end": "00:12:18",
                    "text": "So here I believe public and private"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:16",
                    "end": "00:12:21",
                    "text": "partnership can have a great role like"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:18",
                    "end": "00:12:24",
                    "text": "can we have a challenge based programs."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:21",
                    "end": "00:12:26",
                    "text": "Can we have some hackathons in which"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:24",
                    "end": "00:12:28",
                    "text": "public sector also and private sector"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:26",
                    "end": "00:12:30",
                    "text": "also come across with various prototypes"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:28",
                    "end": "00:12:33",
                    "text": "and those prototypes can later on be put"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:30",
                    "end": "00:12:35",
                    "text": "into mass productions but that can be a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:33",
                    "end": "00:12:38",
                    "text": "great collaborative formats and because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:35",
                    "end": "00:12:40",
                    "text": "open source these days are like very"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:38",
                    "end": "00:12:42",
                    "text": "very speedy to the market. they are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:40",
                    "end": "00:12:45",
                    "text": "actually solving the real problems and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:42",
                    "end": "00:12:46",
                    "text": "they are coming up with very innovative"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:45",
                    "end": "00:12:48",
                    "text": "concepts that we haven't thought of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:46",
                    "end": "00:12:50",
                    "text": "enough. So here I feel like public"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:48",
                    "end": "00:12:53",
                    "text": "private collaboration can really be of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:50",
                    "end": "00:12:56",
                    "text": "help and there are various uh you can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:53",
                    "end": "00:12:59",
                    "text": "say uh unique models to curl that out be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:56",
                    "end": "00:13:02",
                    "text": "challenge based be hackathons or in any"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:59",
                    "end": "00:13:04",
                    "text": "other formats. So once we have the AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:02",
                    "end": "00:13:06",
                    "text": "layer sorted out and have the relevant"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:04",
                    "end": "00:13:10",
                    "text": "models built then comes the consumption"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:06",
                    "end": "00:13:12",
                    "text": "layer. Now as as uh my fellow panelist"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:10",
                    "end": "00:13:14",
                    "text": "was saying like there is no industry"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:12",
                    "end": "00:13:17",
                    "text": "today right now which is not affected by"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:14",
                    "end": "00:13:20",
                    "text": "AI. You you name any industry we have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:17",
                    "end": "00:13:21",
                    "text": "the analytics use cases. Why? Because we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:20",
                    "end": "00:13:24",
                    "text": "have the right set of techniques"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:21",
                    "end": "00:13:27",
                    "text": "especially in defense area any anomaly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:24",
                    "end": "00:13:29",
                    "text": "detection any pattern recognition we can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:27",
                    "end": "00:13:31",
                    "text": "put the right alarms right setups. Okay"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:29",
                    "end": "00:13:33",
                    "text": "this this is something which is going"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:31",
                    "end": "00:13:36",
                    "text": "off the track and should highlight we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:33",
                    "end": "00:13:38",
                    "text": "should raise the relevant alerts. So in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:36",
                    "end": "00:13:41",
                    "text": "those mechanisms the user especially in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:38",
                    "end": "00:13:43",
                    "text": "the defense sector is absolutely"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:41",
                    "end": "00:13:47",
                    "text": "important and can be done on real time."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:43",
                    "end": "00:13:50",
                    "text": "Now all of this is en you can say"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:47",
                    "end": "00:13:52",
                    "text": "encapsulated with the ecosystem layer"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:50",
                    "end": "00:13:55",
                    "text": "and what we call it as like you can call"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:52",
                    "end": "00:13:58",
                    "text": "it as an ESG layer because as we are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:55",
                    "end": "00:14:00",
                    "text": "building the Indian soverent track to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:58",
                    "end": "00:14:02",
                    "text": "build our own AIS we need to be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:00",
                    "end": "00:14:05",
                    "text": "absolutely cognizant of the fact what"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:02",
                    "end": "00:14:06",
                    "text": "kind of carbon footprints uh while we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:05",
                    "end": "00:14:08",
                    "text": "are creating our own data centers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:06",
                    "end": "00:14:10",
                    "text": "building our own clouds what kind of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:08",
                    "end": "00:14:13",
                    "text": "footprints we are having. So somewhere"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:10",
                    "end": "00:14:16",
                    "text": "in back of this journey we should be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:13",
                    "end": "00:14:18",
                    "text": "cognizant of this facts as well how ESG"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:16",
                    "end": "00:14:21",
                    "text": "can actually have an impact in this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:18",
                    "end": "00:14:22",
                    "text": "entire ball game. So that's that's all"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:21",
                    "end": "00:14:24",
                    "text": "my thoughts abishek."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:22",
                    "end": "00:14:27",
                    "text": ">> Thank you. Thank you so much P. I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:24",
                    "end": "00:14:29",
                    "text": "uh yeah"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:27",
                    "end": "00:14:31",
                    "text": "my my takeaway from you is that open"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:29",
                    "end": "00:14:33",
                    "text": "source but with guardrails"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:31",
                    "end": "00:14:35",
                    "text": "um you know quick to go to market"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:33",
                    "end": "00:14:37",
                    "text": "innovative. Yeah and you also talked"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:35",
                    "end": "00:14:39",
                    "text": "about the consumption layer but most"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:37",
                    "end": "00:14:41",
                    "text": "importantly you came towards the end on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:39",
                    "end": "00:14:43",
                    "text": "the ESG side of it. I'm very sure we are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:41",
                    "end": "00:14:45",
                    "text": "going to have one more session sometime"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:43",
                    "end": "00:14:46",
                    "text": "in future on ESG and we'd love to invite"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:45",
                    "end": "00:14:48",
                    "text": "you and express your thoughts because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:46",
                    "end": "00:14:51",
                    "text": "after all data centers wherever we read"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:48",
                    "end": "00:14:53",
                    "text": "US or anywhere local communities are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:51",
                    "end": "00:14:55",
                    "text": "they they are really affected by it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:53",
                    "end": "00:14:59",
                    "text": "Thank you so much Pri. Okay. So uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:55",
                    "end": "00:15:01",
                    "text": "towards the end I come to Mandar. Uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:59",
                    "end": "00:15:03",
                    "text": "last but not the least, Mandar is is the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:01",
                    "end": "00:15:06",
                    "text": "national security officer of Microsoft"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:03",
                    "end": "00:15:08",
                    "text": "India and Asia Pacific and is at the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:06",
                    "end": "00:15:12",
                    "text": "forefront of Microsoft's uh security"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:08",
                    "end": "00:15:15",
                    "text": "initiatives including cyber. So Mandard"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:12",
                    "end": "00:15:17",
                    "text": "your thoughts about adoption of AI in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:15",
                    "end": "00:15:20",
                    "text": "general because your platform as we say"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:17",
                    "end": "00:15:21",
                    "text": "you know various layers while that layer"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:20",
                    "end": "00:15:24",
                    "text": "can be addressed later but what is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:21",
                    "end": "00:15:26",
                    "text": "whatever is built on top of it is is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:24",
                    "end": "00:15:29",
                    "text": "actually what we debated can be you know"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:26",
                    "end": "00:15:30",
                    "text": "the early adopters of sovereign AI so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:29",
                    "end": "00:15:33",
                    "text": "over to you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:30",
                    "end": "00:15:34",
                    "text": ">> right thanks I think interesting and uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:33",
                    "end": "00:15:38",
                    "text": "I have the advantage of speaking after"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:34",
                    "end": "00:15:40",
                    "text": "everybody so so I will uh I'll just take"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:38",
                    "end": "00:15:42",
                    "text": "one more minute right I think a lot of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:40",
                    "end": "00:15:45",
                    "text": "speakers before me talked about"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:42",
                    "end": "00:15:47",
                    "text": "sovereignity and uh different words used"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:45",
                    "end": "00:15:50",
                    "text": "visceral and emotional and all that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:47",
                    "end": "00:15:52",
                    "text": "right I think in my opinion um"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:50",
                    "end": "00:15:54",
                    "text": "sovereignity has been obscured or"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:52",
                    "end": "00:15:56",
                    "text": "confused by a lot of interest right and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:54",
                    "end": "00:15:60",
                    "text": "interest from everybody from government"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:56",
                    "end": "00:16:02",
                    "text": "from tech companies and um of course uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:60",
                    "end": "00:16:04",
                    "text": "general audience right but sovereignty"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:02",
                    "end": "00:16:06",
                    "text": "and uh in my role I work with lot of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:04",
                    "end": "00:16:09",
                    "text": "governments including Indian government"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:06",
                    "end": "00:16:11",
                    "text": "I have I had privilege of working with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:09",
                    "end": "00:16:13",
                    "text": "some of my uh some of the Indian defense"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:11",
                    "end": "00:16:15",
                    "text": "folks as well. Solverity is fairly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:13",
                    "end": "00:16:17",
                    "text": "simple, right? Essentially, it's four"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:15",
                    "end": "00:16:20",
                    "text": "layers. The layer number one is data,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:17",
                    "end": "00:16:23",
                    "text": "right? Data sovereignity. We whole"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:20",
                    "end": "00:16:25",
                    "text": "started with data residency. But then"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:23",
                    "end": "00:16:26",
                    "text": "given the geopolitical situation and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:25",
                    "end": "00:16:28",
                    "text": "some of the incidents, we actually now"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:26",
                    "end": "00:16:30",
                    "text": "talk about data sovereignty. How is my"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:28",
                    "end": "00:16:33",
                    "text": "data protected? Nobody else has access."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:30",
                    "end": "00:16:36",
                    "text": "Nobody else has jurisdictional access to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:33",
                    "end": "00:16:38",
                    "text": "my data. Right? That's data sovereignty."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:36",
                    "end": "00:16:40",
                    "text": "The second layer is operational"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:38",
                    "end": "00:16:43",
                    "text": "sovereignty. uh couple of speakers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:40",
                    "end": "00:16:45",
                    "text": "before me spoke spoke about how can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:43",
                    "end": "00:16:47",
                    "text": "nobody else switch me off right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:45",
                    "end": "00:16:51",
                    "text": "operational so also is fairly simple so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:47",
                    "end": "00:16:54",
                    "text": "that my business can continue uh I can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:51",
                    "end": "00:16:55",
                    "text": "continue to trust that cloud or AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:54",
                    "end": "00:16:58",
                    "text": "without somebody else having ability to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:55",
                    "end": "00:16:59",
                    "text": "switch off tech this is where things"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:58",
                    "end": "00:17:02",
                    "text": "become little complicated and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:59",
                    "end": "00:17:03",
                    "text": "interesting right tech soy means I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:02",
                    "end": "00:17:06",
                    "text": "should be able to build on top of that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:03",
                    "end": "00:17:08",
                    "text": "right I'll not get into bit of clos"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:06",
                    "end": "00:17:10",
                    "text": "source open source commercial right but"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:08",
                    "end": "00:17:13",
                    "text": "that's where the issue starts happening"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:10",
                    "end": "00:17:15",
                    "text": "because tech also means the tech has to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:13",
                    "end": "00:17:17",
                    "text": "be not only in my control but it has to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:15",
                    "end": "00:17:19",
                    "text": "be good right the classic example"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:17",
                    "end": "00:17:20",
                    "text": "classic debate that we get again we have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:19",
                    "end": "00:17:23",
                    "text": "eminent defense folks so I'll refer to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:20",
                    "end": "00:17:25",
                    "text": "that right is India willing to wait 10"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:23",
                    "end": "00:17:27",
                    "text": "years for AMA to come fifth generation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:25",
                    "end": "00:17:30",
                    "text": "fighter or is India willing to buy"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:27",
                    "end": "00:17:32",
                    "text": "Rafael to fill that gap right uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:30",
                    "end": "00:17:34",
                    "text": "sovereign aircraft maybe we wait 10"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:32",
                    "end": "00:17:36",
                    "text": "years but we have needs today that's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:34",
                    "end": "00:17:38",
                    "text": "where the tech sovereignty becomes a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:36",
                    "end": "00:17:40",
                    "text": "very very interesting But sorry to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:38",
                    "end": "00:17:43",
                    "text": "pitching but just interesting input. The"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:40",
                    "end": "00:17:44",
                    "text": "AMA is coming with 50% indigenization."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:43",
                    "end": "00:17:47",
                    "text": ">> Sure. Sir would agree with me."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:44",
                    "end": "00:17:47",
                    "text": ">> No no absolutely I'm saying AMA drawing"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:47",
                    "end": "00:17:50",
                    "text": "a parallel."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:47",
                    "end": "00:17:51",
                    "text": ">> No no I'm saying AMA is great but AMA"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:50",
                    "end": "00:17:53",
                    "text": "will come maybe 10 years."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:51",
                    "end": "00:17:54",
                    "text": ">> Sorry not AMA. Rafal is coming with 50%"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:53",
                    "end": "00:17:56",
                    "text": "Indianization."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:54",
                    "end": "00:17:58",
                    "text": ">> Yeah of course of course. Yeah. So uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:56",
                    "end": "00:17:60",
                    "text": "but the point I'm making is AMA is 10"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:58",
                    "end": "00:18:03",
                    "text": "years away. Would we want or would we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:60",
                    "end": "00:18:05",
                    "text": "want 50% Indianization and use uh Rafal"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:03",
                    "end": "00:18:08",
                    "text": "today? Now that's the balance on tech"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:05",
                    "end": "00:18:10",
                    "text": "right? If we insist on every layer has"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:08",
                    "end": "00:18:12",
                    "text": "to be India then we probably will have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:10",
                    "end": "00:18:14",
                    "text": "to wait right that's the balance and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:12",
                    "end": "00:18:16",
                    "text": "finally the AI sovereignity right the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:14",
                    "end": "00:18:18",
                    "text": "fourth layer which was not on top of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:16",
                    "end": "00:18:20",
                    "text": "everybody's mind right the three layers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:18",
                    "end": "00:18:22",
                    "text": "have been talked for pretty much for a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:20",
                    "end": "00:18:24",
                    "text": "long time the data the operation and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:22",
                    "end": "00:18:26",
                    "text": "tech so even in the cloud world but AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:24",
                    "end": "00:18:29",
                    "text": "already came now AI so is a little bit"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:26",
                    "end": "00:18:32",
                    "text": "more complicated than cloud sovereignity"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:29",
                    "end": "00:18:34",
                    "text": "or or or any other tech soy right why"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:32",
                    "end": "00:18:37",
                    "text": "because I think speakers before me spoke"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:34",
                    "end": "00:18:39",
                    "text": "so I'll not go detail there but AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:37",
                    "end": "00:18:41",
                    "text": "brings a perspective right that means"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:39",
                    "end": "00:18:43",
                    "text": "the data that it is trained on has to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:41",
                    "end": "00:18:46",
                    "text": "have the Indian perspective right it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:43",
                    "end": "00:18:48",
                    "text": "needs to have all the nuances which in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:46",
                    "end": "00:18:50",
                    "text": "the first three layers were not so much"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:48",
                    "end": "00:18:52",
                    "text": "so data solidity is not only whether it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:50",
                    "end": "00:18:54",
                    "text": "is running in India whether it is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:52",
                    "end": "00:18:56",
                    "text": "developed by Indians it's also is it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:54",
                    "end": "00:18:58",
                    "text": "trained on Indian ethics is it trained"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:56",
                    "end": "00:19:01",
                    "text": "on Indian languages and all that so I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:58",
                    "end": "00:19:04",
                    "text": "think AI so is a step beyond that right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:01",
                    "end": "00:19:06",
                    "text": "and because most of the models are uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:04",
                    "end": "00:19:08",
                    "text": "like black blocks explanative"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:06",
                    "end": "00:19:09",
                    "text": "explanability is a challenge. I think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:08",
                    "end": "00:19:11",
                    "text": "that's where AI already becomes"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:09",
                    "end": "00:19:13",
                    "text": "extremely important, right?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:11",
                    "end": "00:19:15",
                    "text": ">> Um and that's where I again agree with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:13",
                    "end": "00:19:17",
                    "text": "what I think Peter you mentioned the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:15",
                    "end": "00:19:19",
                    "text": "small language models Martin sorry you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:17",
                    "end": "00:19:21",
                    "text": "mentioned small language model"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:19",
                    "end": "00:19:23",
                    "text": "purpose-built models are probably making"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:21",
                    "end": "00:19:26",
                    "text": "more sense. I think that's the way we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:23",
                    "end": "00:19:28",
                    "text": "think generally of sovereignity right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:26",
                    "end": "00:19:30",
                    "text": "the question that we need to all look at"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:28",
                    "end": "00:19:32",
                    "text": "is solidity is not binary it's not zero"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:30",
                    "end": "00:19:35",
                    "text": "or one right as it is typically made out"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:32",
                    "end": "00:19:37",
                    "text": "to be it's a spectrum different"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:35",
                    "end": "00:19:40",
                    "text": "departments different organizations"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:37",
                    "end": "00:19:42",
                    "text": "different applications need different"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:40",
                    "end": "00:19:44",
                    "text": "level of sovereignty somewhere you need"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:42",
                    "end": "00:19:45",
                    "text": "to have operations control versus"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:44",
                    "end": "00:19:47",
                    "text": "somebody somewhere you will have to end"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:45",
                    "end": "00:19:50",
                    "text": "to end the stack like in defense needs"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:47",
                    "end": "00:19:53",
                    "text": "to be developed ourself right uh the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:50",
                    "end": "00:19:56",
                    "text": "final point that I'll make here is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:53",
                    "end": "00:19:59",
                    "text": "see life has come full circle. Uh those"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:56",
                    "end": "00:20:01",
                    "text": "who are in IT for maybe 25 odd years we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:59",
                    "end": "00:20:03",
                    "text": "saw mainframe where computing was"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:01",
                    "end": "00:20:05",
                    "text": "centralized to distributed computing"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:03",
                    "end": "00:20:07",
                    "text": "where computing was distributed to cloud"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:05",
                    "end": "00:20:09",
                    "text": "where again that computing become little"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:07",
                    "end": "00:20:12",
                    "text": "centralized. Now we are talking when"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:09",
                    "end": "00:20:13",
                    "text": "cloud started 10 15 years back right all"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:12",
                    "end": "00:20:15",
                    "text": "the big techs believed that there are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:13",
                    "end": "00:20:16",
                    "text": "going to be five regions in the world"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:15",
                    "end": "00:20:19",
                    "text": "right somewhere in Europe somewhere in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:16",
                    "end": "00:20:21",
                    "text": "Asia somewhere in US and everybody will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:19",
                    "end": "00:20:24",
                    "text": "use cloud from those five regions today"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:21",
                    "end": "00:20:26",
                    "text": "reality is Microsoft has 75 regions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:24",
                    "end": "00:20:29",
                    "text": "right so we moved from five regions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:26",
                    "end": "00:20:31",
                    "text": "literally 10 years back to 75 regions I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:29",
                    "end": "00:20:33",
                    "text": "think it's time to take the next step"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:31",
                    "end": "00:20:35",
                    "text": "where we are saying the region is not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:33",
                    "end": "00:20:37",
                    "text": "only in my control but region is also in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:35",
                    "end": "00:20:40",
                    "text": "customers data center so I think what we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:37",
                    "end": "00:20:42",
                    "text": "will see happening in tech is that the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:40",
                    "end": "00:20:45",
                    "text": "sovereign clouds will become a reality"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:42",
                    "end": "00:20:47",
                    "text": "right for example in Microsoft we have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:45",
                    "end": "00:20:49",
                    "text": "now created something something called a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:47",
                    "end": "00:20:51",
                    "text": "sovereign private cloud which can run in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:49",
                    "end": "00:20:55",
                    "text": "customers own data center under complete"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:51",
                    "end": "00:20:56",
                    "text": "customer control disconnected air gap we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:55",
                    "end": "00:20:58",
                    "text": "did not do that maybe two years back"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:56",
                    "end": "00:21:01",
                    "text": "that's the reality right we we talk kept"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:58",
                    "end": "00:21:03",
                    "text": "on talking about softwaredefined"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:01",
                    "end": "00:21:05",
                    "text": "sovereignty until about a year back"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:03",
                    "end": "00:21:07",
                    "text": ">> today we are saying okay we have a full"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:05",
                    "end": "00:21:09",
                    "text": "stack sovereign private cloud and I'm"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:07",
                    "end": "00:21:11",
                    "text": "taking Microsoft as an example. Our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:09",
                    "end": "00:21:11",
                    "text": "competition and other companies also"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:11",
                    "end": "00:21:13",
                    "text": "have similar."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:11",
                    "end": "00:21:15",
                    "text": ">> No, that's a very valid point. In fact,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:13",
                    "end": "00:21:17",
                    "text": "competitions like SAP already deployed"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:15",
                    "end": "00:21:17",
                    "text": "sovereign cloud and that's the way to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:17",
                    "end": "00:21:18",
                    "text": "go."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:17",
                    "end": "00:21:21",
                    "text": ">> That's sovereign public. What I'm saying"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:18",
                    "end": "00:21:22",
                    "text": "is sovereign private which can run in an"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:21",
                    "end": "00:21:23",
                    "text": "aircapped environment in defense of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:22",
                    "end": "00:21:25",
                    "text": "course of course"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:23",
                    "end": "00:21:28",
                    "text": ">> and that's a reality because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:25",
                    "end": "00:21:30",
                    "text": ">> last point we will have to take us out"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:28",
                    "end": "00:21:32",
                    "text": "of the equation the zero trust model"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:30",
                    "end": "00:21:34",
                    "text": "right you don't even need to trust the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:32",
                    "end": "00:21:35",
                    "text": "company that gives you tech. That's the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:34",
                    "end": "00:21:36",
                    "text": "only way we can build content."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:35",
                    "end": "00:21:38",
                    "text": ">> Thank you. Thank you. Right. Very valid"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:36",
                    "end": "00:21:40",
                    "text": "point Mandar and it was important to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:38",
                    "end": "00:21:42",
                    "text": "bring the industry perspective here to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:40",
                    "end": "00:21:44",
                    "text": "complete the full picture. Uh ladies and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:42",
                    "end": "00:21:46",
                    "text": "gentlemen we have almost run out of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:44",
                    "end": "00:21:48",
                    "text": "time. I would request Pier to give his"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:46",
                    "end": "00:21:50",
                    "text": "closing remarks and then we'll uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:48",
                    "end": "00:21:53",
                    "text": "conclude the session. P what you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:50",
                    "end": "00:21:56",
                    "text": ">> okay so try to keep it brief. So what we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:53",
                    "end": "00:21:58",
                    "text": "have been discussing here it's we would"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:56",
                    "end": "00:22:01",
                    "text": "say it's when we talk about sovereignity"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:58",
                    "end": "00:22:04",
                    "text": "and AI it's openness versus security the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:01",
                    "end": "00:22:05",
                    "text": "trade-offs speed of innovation versus"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:04",
                    "end": "00:22:09",
                    "text": "governance"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:05",
                    "end": "00:22:11",
                    "text": "global integration versus autonomy"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:09",
                    "end": "00:22:12",
                    "text": "public infrastructure versus private"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:11",
                    "end": "00:22:16",
                    "text": "innovation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:12",
                    "end": "00:22:18",
                    "text": "So these are the extremes and um what"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:16",
                    "end": "00:22:21",
                    "text": "what what I wish was would be for India"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:18",
                    "end": "00:22:23",
                    "text": "would be if they if you could have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:21",
                    "end": "00:22:26",
                    "text": "something in the middle of these"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:23",
                    "end": "00:22:31",
                    "text": "extremes um so the best of let's say the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:26",
                    "end": "00:22:35",
                    "text": "EU and the US EU very much regula"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:31",
                    "end": "00:22:37",
                    "text": "regulatory driven uh regulatory first"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:35",
                    "end": "00:22:40",
                    "text": "innovation second etc and the US"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:37",
                    "end": "00:22:43",
                    "text": "probably innovation first and then well"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:40",
                    "end": "00:22:46",
                    "text": "some control from the permanent um"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:43",
                    "end": "00:22:48",
                    "text": "anyway that comes. So if if that would"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:46",
                    "end": "00:22:51",
                    "text": "be something uh for India, I think this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:48",
                    "end": "00:22:55",
                    "text": "is a is a good rate good route to take."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:51",
                    "end": "00:22:59",
                    "text": "You said it uh collaborative models um"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:55",
                    "end": "00:23:03",
                    "text": "in uh in to invest in in infrastructure"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:59",
                    "end": "00:23:05",
                    "text": "something like um centers of excellence"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:03",
                    "end": "00:23:08",
                    "text": "um where public and private come"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:05",
                    "end": "00:23:11",
                    "text": "together to solve the real world"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:08",
                    "end": "00:23:13",
                    "text": "problems in architecture in healthcare"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:11",
                    "end": "00:23:15",
                    "text": "and all these kind of things. So if that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:13",
                    "end": "00:23:17",
                    "text": "would be a route for India I would be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:15",
                    "end": "00:23:21",
                    "text": "more than glad and it brings the best"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:17",
                    "end": "00:23:23",
                    "text": "the best together from both worlds."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:21",
                    "end": "00:23:25",
                    "text": "Thank you. Thank you Pier. So towards"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:23",
                    "end": "00:23:26",
                    "text": "the end of the session I'll request all"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:25",
                    "end": "00:23:28",
                    "text": "speakers to please come forward. We are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:26",
                    "end": "00:23:30",
                    "text": "releasing a background paper and I would"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:28",
                    "end": "00:23:34",
                    "text": "request Nachal Mishra and Akilshuta from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:30",
                    "end": "00:23:34",
                    "text": "KPMG to please come up as well."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:34",
                    "end": "00:23:38",
                    "text": "Show it please."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:47",
                    "end": "00:23:50",
                    "text": "Let's"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:02",
                    "end": "00:24:05",
                    "text": "get out of this."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6992d40c4c5e8f0974c871a5",
    "projectId": "112150",
    "projectName": "AI for Bharat",
    "projectDescription": "AI for Bharat | Impact Summit 2026",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-16T08:23:38.608Z",
    "projectUpdatedDate": "2026-02-16T08:23:38.608Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-16T08:23:38.608Z",
    "projectDeleted": false,
    "projectDataLength": 3,
    "projectType": "DOCUMENT",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "AI for Bharat program.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        },
        {
            "col1": "P1_Part_2",
            "col2": "AI for Bharat program.docx",
            "col3": "",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6992b7154c5e8f0974c86ebf",
    "projectId": "106150",
    "projectName": "The Future Of Employability In The Age Of AI | AI Impact Summit 2026, New Delhi",
    "projectDescription": "The Future Of Employability In The Age Of AI | AI Impact Summit 2026, New Delhi",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-16T06:20:04.215Z",
    "projectUpdatedDate": "2026-02-16T06:20:04.215Z",
    "projectStatus": "Active",
    "companyId": "69453bf2e603a0a401b6f35b",
    "projectCreatedBy": "69453bf2e603a0a401b6f35c",
    "projectUpdatedBy": "69453bf2e603a0a401b6f35c",
    "projectLastAccessDate": "2026-02-16T06:20:04.215Z",
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "Content",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "The Future Of Employability In The Age Of AI.docx",
            "col3": "Transcript 0:00 You know that arm jent and learn a computer was a new concept. 0:05 Now it happened. This is the first story. So institutions respond slowly. Don't expect it to be very different. 0:11 Now number two rather institution that are not accountable to markets respond 0:17 slowly. Number two uh second story 1985 0:23 I am now 21 22 years old. 22 years old. uh Rajiv Gandhi Sarati to it's announced 0:31 that you know we are going to introduce computers in banks 0:37 right now the all India bank employees association was a very powerful trade union then 0:43 they got very worried and there were all sorts of uh blowback saying we'll lose 0:48 jobs if you lose computers but the government said we are going ahead regardless this is modern India this is 0:54 new India there must be computers and bikes So computers were bought and put into banks. For a while they were not 1:00 used but they began to be used and guess what? Nobody lost jobs. People became 1:06 more productive. They serve customers better. Right? Nobody lost jobs. Uh so 1:12 second lesson is when new technology comes in there is almost always great fears of job loss. 1:19 But very often productivity goes up and people are doing more things. The same number of people are doing more 1:24 things. They're doing different things. They're doing more useful things. Uh, can it be different this time? It's 1:31 possible. Anything's possible. But will it necessarily be different? Answer is not 1:37 necessarily. People might learn to do different things. People might learn to do stuff that is not getting done might 1:42 get done. So I'll give you an example from my own company. Uh, we've got about 1:47 at no, we've got about one and a half lakh clients, 1.2, 1.3 lakh clients. 1:54 We have a sales team. We have a sales effort. So the top 10 20% of clients 2:00 they get a field sales force calling on them because they are more valuable. We can afford a field sales force call and 2:05 it gives us an opportunity to upsell. So it makes a lot of sense to have field sales force. The next 30% of clients 2:11 they get te sales right you can sit in an office and you can make phone calls. 2:17 Now the bottom 50% of clients who really don't pay that much. It doesn't make 2:23 financial sense to have a a person calling on them. It doesn't make financial sense for a uh a person even 2:30 phoning up. So, we put a chatbot, a voice, automated calls. You can't make out. 2:36 It's not a human being. That's how it's advanced, right? And suddenly we are calling them up. 2:44 Now, what is happening here is stuff that is not getting done. We have served an underserved segment, underserved 2:49 market by using AI. Thus far at least nobody's lost a job in our company because of AI. I don't know 2:55 what will happen going forward but right now it's being used to increase productivity. It is being used to do stuff better. It's being used to do 3:01 stuff in a matter. Okay. So uh simple example 3:08 one more uh so you know there's a 3:15 surface of content on video on YouTube podcast 3:20 right you don't have time to follow all the podcasts but podcast man I'm going I 3:26 have to listen and watch for 50 minutes it won't work I don't have time so every time an interesting podcast comes or 3:32 interesting interview comes I just asked a colleague of mine claude get me a summary. 3:40 So Vishal Sikka spoke about AI. It was a 30 40m minute interview and I said yeah 3:47 uh I can't go 30 40 minutes. I gave it to Rishaba a colleague of mur within 10 3:52 minutes gave it back to me. He had three minute summary. Now now this is not something this is 3:58 not something I would have done or even watched 40 minutes. See, I lost the click. This is just what I was saying. 4:05 This is how I'm getting beaten. Okay. No, you're saying useful stuff which I will not watch because I don't have 40 4:11 minutes. But at least I get to absorb your content. But I may not give you advertising, right? Okay. So, so you 4:17 know u so this is stuff being done that has not been done earlier. It's being used done by AI. Nobody's lost a job 4:23 yet. Uh the final story, the final story I will talk about is uh 1989. 4:31 So I've finished college. I've worked for 3 years in advertising. I've done business school and I'm now in my first 4:38 job after business school. I'm in marketing. I'm managing the brand holics for a company called HMM. It's now 4:44 called Black Smithline after various mergers and acquisitions. Now that marketing team was 10 15 people. All 4:50 were MBAs from the IM or FMS or Baj or the best business schools. Okay. Nine 10 4:56 of us. Now what is the difference? I was the newest. I was 89 batch. The others 5:01 were 8 5:42 There are four pieces. There were four pieces in the marketing department. One was with the secretary of the head 5:48 of marketing. So nobody had access to it. Other was secretary of a marketing manager. Nobody had access to it. Two 5:56 was for the general pool. Problem was nobody knew how to use them because they not studied it as part of the course. I 6:03 was like like a fish to water, right? A power presentation. 6:10 So in those days there was no PowerPoint. There was something called Harvard Graphics software called Harvard Graphics to all pirated. 6:19 Okay. Uh they were using HCL PC. Uh I don't know. I don't recall. But it had it had to be branded P. It had to be 6:25 a branded PC. It must be an NCL. Okay. Uh okay. So uh I was the only guy who 6:31 knew how to use it. So before that people would take these overhead transparencies and write slides by hand, 6:39 right? And I came in with this Howard graphics and you used to take a computer print out on the on the transparency. Suddenly there's a new tech and it was 6:45 sexy. So I became the go-to guy. 6:50 Okay. So I was the most useful guy at the ground level because I had a skill with a technology which is actually easy 6:56 to use. Actually easy to use but you had never learned it. I learned it because it was there in the institute. Now AI is 7:03 like that. You don't have to build LLMs, right? And I tell you what to all the 7:09 young people here, you don't worry about systemic problems. You don't worry about policy issues. You just worry about your 7:14 job. Yeah. And your career, your individual job and career. What should you do? What should you do to make sure AI doesn't make you lose your job or in 7:22 fact AI enables you to get your job? Just learn 5 10 15 useful AI tools 7:28 because let me assure you the older people in any company will not know them because they are not quick learners, 7:35 right? But if you learn them right, you will get a job. Now, now so the one message I want to leave is look AI is 7:40 happening. It is relentless. So if you don't do AI, AI will be done 7:46 to you. So you better do AI. Yeah. Thank you. I think Satish, let me 7:52 bring you in because Mr. Big Chan spoke about claude. There's been so much of 7:57 buzz. Stock markets tanked, market caps went down. What is clot doing to your 8:03 business at Infosys and IT services? and and also on the question of enterprise automation which you are which is your 8:09 product area. Thank you. Hi everyone and glad to be here. Thanks for the invite. Uh I'll 8:16 share a few thoughts specifically from an enterprise IT perspective. Um 8:22 to start with I just want to mention that AI to all technologist on all every 8:28 worker in the world it should be seen as a capability multiplier. 8:33 It is not necessarily a a tool to replace jobs, but it is something that 8:40 you can use as he rightly mentioned to deliver faster, better and incredible outcome in 8:47 the same jobs that you're currently doing. So it's a capability multiplier. When we look at this as a capability 8:53 multiplier, it impacts you know both sides of the enterprises, right? One is the people who are building the software 8:59 which are engineers. It should also impact the people who are using the software. All right? And so both of 9:05 their lives will get transformed. And I have given this example many places today. You know all of you have phones 9:12 today. You open your phones and take a photograph. The incredible photographs have been made so simple simply because 9:18 of technology. That's a you know it's a capability multiplier in the hands of every human being. see AI as one such 9:24 more another tool a great tool to make decisions that we were not able to do 9:30 earlier maybe even faster with far more clarity I think that's how we should absorb this technology one 9:37 if you look at enterprises there's always a concern you know Shashi mentioned stock market and all that let 9:43 me not comment about that but I want to let you know that it is not like technology is going away 9:49 right if any with AI technology techology will get far more entrenched 9:54 in enterprises of the future. It is not going to be less technology, it's going to be more technology. 10:01 Now a template that we can look at are digital native companies that are around us. If you look at these digital digital 10:07 natives which were born out of software, right, which were not necessarily brick and mortar companies. If you look at how 10:13 they run their firms, there are you know employees who understand the business 10:18 strategy who have different roles today. But they design the logic. They design the flows. They design the thinking 10:26 behind the business and code that into the software and let the software run this business. So software is the heart 10:33 of the business. It runs the business and there are a lot of people who are there to take care of the software. 10:38 Right? That's how digital native companies are. So if you fast forward let's say next 5 years and 10 years I 10:44 think where will AI take all these enterprises to become more and more digitally native. 10:50 Now that means more technology jobs. I don't think it's going to be less technology but different kinds of jobs 10:57 and I don't think anybody should be paranoid about it. Right? Like any other technology that has come by that I think 11:04 we've all being Indians or otherwise worldwide embraced it and made it good 11:09 for ourselves and our lives and our families and the society. I think AI will be one more. So enterprises in the 11:15 long run will continue to use more technology. I just want to say that the complexity if any of these enterprises 11:21 is huge right they are they are like civilizations by their own right it's 11:26 not going to change overnight so I don't think anybody should fear almost like AI will take away jobs tomorrow day after 11:32 no it's not going to happen it has it's it has been evolving over you know for a very long period of time it'll continue 11:38 to evolve the speed has accelerated obviously so you can see this in three 11:44 phases right there will be automation that was mentioned multiple time next two to three there'll be incredible automation. 11:50 I think if you were to look another 3 to 5 years, there'll be a lot of workforce restructuring that will happen due to I 11:57 would say automation and the last leg of it is the business model transformation and the business model transformation at 12:02 least I believe is going to start mirroring and mirroring closer to the digital native operating models. I think 12:08 that's where AI really will take everybody which means more value creation, faster value creation, etc. 12:16 Lastly, I think you know because Shashi said you know Sish you need to speak about the technologies and the coders 12:21 and the engineers I can tell you from our own organization which is a software product company 12:28 AI has generated incredible productivity 70 80% more productivity in the last 2 12:35 to 3 years is significant but there's nothing to fear about it it's just that you can do more innovation faster you 12:42 can deliver higher quality of products to your customers So, so that is one and if and and if I 12:48 don't know how many of you are engineers just so you guys can understand if you wait to look at agile development as a 12:54 methodology that is used and you develop agile teams scrum teams right many of 13:00 you are nodding so I'm assuming you're technologist here and you look at there are seven or eight members in a 13:06 simplistic way and that's how work gets done tomorrow there'll be three human beings and five agents you see what I 13:12 mean Humans will be not only doing the defining the problem, we'll also be managing, monitoring what agents do. And 13:19 finally, let's not forget it's the human beings who take accountability for all these technologies. 13:25 So human beings are not going anywhere. Jobs are not going anywhere. It's just going to change. The nature of job is 13:30 just going to change. Let me pause here. So Anorak, so Satish said, you know, 13:36 three humans, seven agents. So the next time I go to a doctor, is it going to be a human or an agent? 13:42 Great question. You know how many of uh the folks in the in the audience are students? Can I see a show of show of 13:48 hands? You know, I'm a professor, so I I need to know my audience. Uh you know, 13:53 there are three points I want to very quickly make and and excellent points made by everybody here here on the panel. Um first first of all, I mean I 14:02 don't know how many of you read this book of mine. This is reimagining health and what we really talking about reimagining health employment which is 14:09 what I'm going to talk about. uh there are 8 billion people in the world out of which 5 billion people do 14:15 not have meaningful access to healthare and part of the reason they do not have meaningful access to healthare is because we have become more and more 14:21 sickentric model where tertiary care institutions like as and Stanford 14:26 hospitals are the ones providing care to most of our people whereas where are most of the people they're in the homes 14:32 and communities and and villages and and smaller towns and and cities 14:37 so that's that's point number one that there is a there's a need for imag health care or health system to be 14:43 reimagined right bringing it closer to people closer to humanity uh as it were. 14:49 Second point I would make is that the cost of care is going to continue to 14:55 grow go up if we stay with the sickare model. Right? So if you want India to be a vixit bharat by 2047 or we want rest 15:03 of the LMIC settings the 5 billion people I talked about for them to emerge 15:08 out of poverty emerge out of their their uh resource constraints they need to make sure that they are able to keep their populations healthy and productive 15:15 at a cost in a cost effective way. What that means is you need innovation, technology innovation is part of it, 15:21 process innovation to enable that shift. And then the last thing I would say is that we often think of AI uh or any 15:29 technological tool for that matter as taking away jobs. In fact, we should we should flip that especially in 15:34 healthcare. Healthare is a is a is a very strong use case for AI for technology in general. In the US, 15% of 15:41 the population is uh involved with healthcare one way or another. If you translate that number to India, you're 15:47 talking about 150 million plus jobs. Okay? So if you think of AI as a job 15:52 creator, a net job creator, because you have 1.4 billion people whose health 15:57 needs to be cared for, you need a lot of people who who will provide care for, you know, as our population ages and as 16:04 our healthcare moves to closer and closer to the to the smaller uh towns and cities and and and villages, we need 16:11 more and more people. But how are you going to do that? The one uh way you can do that is by enabling them through 16:16 technology. Now it could be devices, diagnostics or digital health and the latest tool in the toolkit is AI enabled 16:23 uh systems. You know I was I was just looking at the new kind of jobs that would be created because of AI. Um how 16:30 many of you have heard um care navigator for high-risk cohorts diabetes, TB, 16:35 pregnancy, heart failure? Care navigator for high-risisk patients. I mean those 16:40 jobs don't exist today. So you in fact are going to create new kind of jobs and and a good number of you folks in the in 16:47 the audience who are students will likely be doing those jobs. So so the way I think about AI in healthcare and 16:53 other use cases too I I suspect this might be the case in other uh industries other applications that AI will actually 16:59 net net create jobs. So the last point I'll quickly make is that India shouldn't really be um catching up with 17:07 AI AI in healthcare in particular or AI in other use cases. India should be leaprogging it. That's what India should 17:13 be should be aiming for. Very interesting perspective and you also are on the board of the IT Bombay 17:20 Alumni Association in the United States. What are you hearing from fresh graduates who are now coming out of 17:27 universities within the IT family and of course the broader student ecosystem as well. What are they seeing in the job 17:33 market? It's challenging for them because I think I think you made this point uh Sanjie that you know um folks 17:40 are not ready when a technology wave hits they're not ready for it and then they quickly try to pivot and our 17:46 academic institutions mine included even though Stanford is at the at the hub of technology innovation and we have a lot 17:53 of professors a lot of students are involved in it honestly we are not doing as good a job as as we should in 17:58 preparing this next generation for the kinds of jobs that are coming. So they are they are concerned but they're also 18:04 excited. Uh I think the the youngest population that's graduating today uh 18:09 feels a lot more excited about AI technology and and feels they could do something with it. But we have to show 18:15 them we have to show them I think our generation has to show them the path. We have to show them how this these tools 18:21 could potentially be useful for creating more productivity, more impact, more outcome at a lower cost. Um, you know, 18:27 that's the other thing. Most most folks when they see the numbers around AI, LLMs in particular, they're they're 18:32 looking at numbers, trillions of dollars being invested. I think the ROI for that, I don't know how that's going to be, but I think from use case 18:38 perspective, which is where most of the ROI is going to come from, I see tremendous impact. You know, a trillion 18:43 dollar healthcare uh health technology ecosystem requires 10 million uh trained 18:48 professionals. 10 million, which is very much what India's opportunity is. So I think we are in the last uh 7 18:56 minutes or so. I'll just quickly go around the rest of the panel very short responses. So Satish with you starting 19:02 you know Infosys when in nowadays would recruit in the thousands how is it changing now? What are you seeing? Is it 19:08 very targeted recruitment? You know what what's changed? I think from a core 19:13 capability I'd say two things that everybody should 19:18 have. It's whether it's technologies or otherwise. One is lifelong learnability. I think it's an becomes an essential 19:24 ingredient that students should have right if any with AI I think that's far 19:30 more important than ever before right because one thing is for sure things around you will change will continuously 19:36 change and the speed with which it's going to change is also accelerating so which means learnability has to be the 19:43 core capability for everyone to harness on and I think if I were to hire someone 19:48 si I'll be looking for that number one sir you're Closing thoughts. What do you see as it's a big chain? 19:57 Well, I mean uh for each person here individually, I think set yourself a personal target of learning three how to 20:04 use three AI platforms within the next 3 months and applying them in your work. The more you do that, uh the more your 20:11 job is safe. Congratulations. Um even though uh you are the moderator 20:20 Shashi being you know it's second nature to me to do the so uh 20:28 so um couple of points which I want to make is one uh Satish when you're talking about uh you know uh capability 20:35 multiplier and that AI may not take over many jobs I kind of uh at least in my 20:41 profession I find that uh jobs are going because Um long ago there was a TEL 20:47 operator he lost his job to the fax machine. Uh there was a guy who was a gopher who would take the tape from the 20:53 cameraman to the newsroom that went away when chip technology came in. Um once 20:58 chip technology came in we thought that that's what going to be then FTP started off. So there are many many jobs which 21:04 have gone. those who upskilled and learned they are the ones who are able 21:09 to do and which you know we we heard in the healthcare sector when there are new if you do not upskill you don't I use 21:15 for healthcare sector I'll tell you I've gone through a plethora of uh dieticians 21:20 it's it's an uphill battle to you know as you age for controlling weight diabetes all these kind of things so 21:27 what do I do I went into an AI model and it's helped me for 6 months there's a 21:32 person who's not judging me because I ate that extra samosa and she's not looking at me with that disapproving 21:38 look. It's just telling me what is the solution. So I'm using AI for health. Um have I taken away the um the healthcare 21:46 uh uh person's job? Maybe I have. She's lost a client and not just she 15 others 21:51 who have tried before that. Um on LLM models which you were talking about whether the government is going to give 21:57 incentives when the India will come up with it. I'm sorry but the IT sector has lost out on that. the LLM models have 22:04 are not hallucinating as much as they used to six months back or a year ago or even two months back. Um we have lost 22:12 the race on LLMs. We have to make sure that the LLM doesn't strip us of our uh 22:18 intellectual property rights as Mr. Bchandani which you said that you're watching my podcast in 20inut clips. Let 22:23 me tell you what I did when I started it for the first year. I was resisting it. I said everybody has to watch that 1 22:30 hour interview that I'm doing with Shashi and he better give me a 1 hour content on that and everybody watched 22:35 that 1 hour. Then I realized there are people like Sanjie who's listening to it for 20 minutes and I'm not getting the 22:41 clicks or the revenue. So what did I do the second time? I had Shashi on the podcast. I made ress out of it. So 22:48 people like Sanjie watch the reels and I got the revenue. So I beat him. So that 22:53 I think what you please also give plot summaries in text of two pages three pages so you 23:00 know that can supplement the res. Yeah. Okay. But I'll charge you for that mina 23:12 or dark spots or dull smart and handsome daily brightening cream 23:22 or perfection. Smart and handsome brightening cream for men. We'll have 50% more jobs than we have 23:29 today. And if you combine these two, the issue is not the number of jobs that 23:34 will exist. The issue is the skills in which we are in are going to get 23:40 obsolete and hence jobs are going to get obsolete like the TX like the operator sheet talked about. Therefore, this is a 23:46 mega opportunity for us to not re not upgrade our skills. I don't like that 23:52 word but to reskill ourselves and to ask ourselves what skill is going to be relevant for tomorrow. And as a 23:58 60-year-old man, I can do it. At a 20-year-old person, you can do it. And that is a critical debate we need to 24:04 have in the country. Not about technology, not about LLMs because SLMs 24:09 can do exactly the same thing. It is about what skills are going to be relevant for tomorrow which is going to 24:15 use this use the technology to be able to do that because there will be 50% more jobs. Now whether those jobs will 24:22 be in India or in US or in Vietnam, nobody knows. Remember what Indian IT 24:28 did? Indian IT migrated most IT jobs across the world to India because they 24:34 had a proposition which the world did not have. So India needs to develop that proposition so that we are competitive 24:40 on a global scale. So this fight is not about India. This fight is going to be fought in the international space of who 24:47 is going to use technology to create jobs and get a major share of the 50%. 24:52 Thank you sir. Uh just before I conclude uh you know when we uh designed our AI 24:58 for India calendar we thought we'll do do it using AI and uh the whole output 25:03 the design output was so dry so robotic so mechanical that you know we were not very happy and then we reached out to a 25:10 young designer who used AI but with human creativity and created the colorful presentation that was running 25:15 at the beginning of the session. So I'd like to call Ankita you know the designer who did that 25:23 and I'll request smith gi to felicitate her 25:40 with that we conclude the session. A big round of applause for all the speakers please. 25:51 Thank you. 26:06 Thank you. Thank you. immediate 26:16 action at all levels of the society. Thank you very much and wish you all the 26:21 best. So thanks to Dr. Nageshwan who couldn't be here but uh we have our last speaker 26:29 who's made it from all the way from the US from Stanford University Dr. Anra Mal. You missed the photo up but we'll 26:35 do one more at the end. So, so to kick off the conversation, let me 26:40 start with uh Smith Prakash. Uh you know, you have uh two very interesting 26:48 perspectives. You are also seeing the news that's happening all over in India. 26:53 You recently had a news report on how AI is impacting jobs. And then you're also 26:58 a big media employer as well. So you're also looking at how AI is impacting your own business. So if you could you know 27:05 kind of set the context and then we'll take the conversation around the panel. Thank you. Thank you Mr. Shaker and 27:10 thank you for the organizers to having me here. Uh quite right. I did a report recently on u on the use of AI and the 27:18 impact of AI on employability and I just found the need to do that because I um 27:24 you know um as uh Mr. Nageshwaran said that you know it's a stress test. He 27:29 said uh on employability of of uh journalists, of reporters, of camera 27:36 persons. Um there is uh you know like you said uh that the the students these 27:42 days are doing uh or learning uh artificial intelligence and the tools surreptitiously because there's a push 27:49 back by educational institutions who don't have that vision that this is 27:54 something that it's a skill that they need to have right from middle school onwards. So there are number of people 28:01 who are applying for jobs using cheap models and they're all sounding the 28:07 same. So when I get a resume, it sounds like 15 others which have come in and 28:12 when when that first firewall is cleared and we go through the first level of sifting and when you actually meet with 28:19 them they do not have the skill which they have said in their LinkedIn profile 28:24 or when they have applied for jobs and there's a kind of a lazy attitude of just using a chart GPT to make their 28:32 profile and just send it across to 20 different media organizations. It's then that I realized that um the skill 28:40 training which should be there is not there. Similarly, journalists who are working with us, you know, um 25 years 28:47 ago, I know that I was one of those who said Google use matkarna. Don't use Google to do your story. Go to the 28:54 ground and report yourself. Find out the truth from the ground. Don't just Google it from different I probably a dinosaur 29:02 very early in life in adapting to Google. But it took me probably 29:07 the movement causing all sorts of trouble there. Cricket season workout in no time 29:14 with anytime fitness a fitness all the time. What Google was doing was different from 29:20 what AI is doing today. Um there are many reporters who just you know many 29:25 news organizations today who don't have the budget to send out reporters on the ground. they're just using um various AI 29:34 tools to do it. And this is an existential threat, something that both the speakers before me spoke about. Uh 29:40 this is a threat which has hit the media. And if you've been watching some 29:45 of the conversations which are happening at the Munich uh security conference where they are saying that the old order 29:52 has broken down and uh you know with with regard to national security the old 29:58 order in the media too has broken down in education systems it has broken down 30:03 and if we don't adapt soon you're going to have the situation in India like you're having in America where nobody's 30:10 reading the the newspaper the physical newspaper Many in this room if I even do a show of 30:16 hands have not read the newspaper before coming to work as in a physical newspaper. I think our generation grew 30:23 up having readers digest in our homes or we read the time and newsweek to understand we read the times of India 30:29 and Hindustan times to get to know what what is happening around the world before we got to work. But that's not 30:36 happening with with I'm sure most of the audience out here. So the media is going 30:41 to change a a lot. Existential threat to many anchors who today are ruling the 30:47 roost because uh you never know like you know all of you are watching Arnab Gowami on television and uh your 30:55 cortisol levels are rising uh by the minute. You never know that uh it's possible that Republic has a milder 31:01 version of uh uses AI tool and gets a milder version of Arnab Gowami uh for the 8:00 p.m. and the uh the more 31:09 excitable version at 9:00 p.m. and you can choose which version of uh Arnab you want to watch. It's going to be a 31:14 reality soon. Uh these AI tools are going to be used in the news organization. So you can choose which 31:21 version of the anchor you like uh just like you you know and it's going that selection is going to happen for you 31:28 before you realize that you have made the selection just like the news feed is coming in just before I conclude just 31:35 one more point I want to make uh 31:42 is about um intellectual property rights. Unfortunately all the big 31:50 giants the big fives um they are scraping content from Indian media organizations and not paying us for IPR 31:58 whereas they're paying millions abroad and that's where I feel that we are losing out we are losing the battle even 32:06 before it's begun in India we need something to be done to protect our IPR 32:12 for our news organizations otherwise media is going because it's all clicking right. Uh why would anybody want the 32:19 long- form article in Times of India when a short form of that similar kind 32:24 of article is available from Times of India, Hindustan Times, the Hindu uh the Sun newspaper translated real time and 32:31 given to you and the gist out there. So the clicks are going to go, the advertising is going to go and media in 32:38 its long form is going to disappear and India is going to face an existential threat as far as reporting and media 32:45 organizations are concerned. Thank you. Thank you. Let me uh now go to Mr. Vinit 32:51 Na. Uh sir, you've been a veteran of the Indian IT industry. Uh you do you're working extensively with schools and 32:58 school children through your Sukak Foundation. if you could set a you know macro perspective on employment 33:04 employability how this whole debate is shaping up in the context of AI. Thank you so much and thank you for 33:10 inviting me. Uh let me try and go back and understand the AI phenomena from a 33:16 different lens. Uh when industrial age came in, one of 33:22 the things which the reason industrial age created the kind of value it created 33:27 was because of a management idea of the fact that you can break a long process 33:33 into subprocesses and you can scale a person on a subprocess. So if the cars were 33:40 manufactured where one car one one guy would create a car in industrial age we broke that process into a guy who is 33:46 very good with engines very good with tire very good with paints. So we dummyified the skills to a large extent 33:54 and because of which the number of people who came into the employment and the people who were employable were very 34:00 large. our education institution postindustrial age started catering to 34:06 these knowledge workers or what I call micro skill workers. So therefore in the 34:11 education institution we started impacting knowledge and in our industry 34:18 including the IT industry which I come from we started taking those knowledge workers and training them in what I call 34:25 subscale small scale. Now because of which in Indian IT we have 7 8 million 34:31 people and in different industries different sets of people were hired because of subskll. 34:37 Now you have to understand AI what it is doing is it is automating the subskll 34:43 and this has been happening for many years now right in remote infrastructure management in BO the subsklls have been 34:50 automated. The only thing it has happened is the acceleration of automation is increased. Now it has two 34:56 impact. Number one the subskll because it is automated people can do the same 35:03 work faster, cheaper, better and therefore they are not going to depend on the human being. There is where the 35:09 crisis is coming up. Now the response to that crisis is would we lose jobs. 35:25 Crunchy almonds or chopped pistachios. 35:45 What it means is there is a significant change in skills exactly the way it happened in industrial age that we 35:52 needed what we call subsklls in the indust in the AI age we need what we 35:57 call the macro skills. What does that mean? That means that having a knowledge of something or being able to do a 36:04 subskll is par. It is not needed to be able to re-imagine or I call them 36:10 re-imagineer to problem solve and to imagine a solution of a problem which nobody else 36:16 has been able to solve is the new skill which is required. Third point because there is significant 36:24 profitdriven subscriptiondriven uh vocabulary which is in the market 36:30 everybody is assuming that AI is technology. It is not. There are a lot of companies which are worth billions 36:36 and trillions of dollars just because of a technology and they're making us believe that AI is technology. It is not 36:43 AI and the real use of AI is a use case of AI. How do you use AI for which you 36:49 need different set of skills now? Therefore to have AI as a skill or 36:54 technology being taught in school will help subscription of a lot of companies and they'll make a lot of money 37:00 including my old company. But the issue is that the skill which we need to teach 37:06 our children today is to reimagine to solve problems to think of alternate 37:11 solutions and to do what humans do best because the machines will never be able to do it. And therefore the question is 37:18 how do we do it? I think India has an opportunity. First India has a threat 37:23 that the amount of people we have in what we call the subskll task is very large including IT industry. ",
            "col4": "DATA",
            "col8": "English"
        }
    ],
    "projectDeleted": false,
    "codeFrameColumn": [
        {
            "col1": "col1",
            "col2": "col3",
            "researchObjective": "",
            "filterColumn": [],
            "netSetting": true,
            "subNetSetting": true,
            "categorySetting": true,
            "sentimentSetting": true,
            "thinkingBudgetSetting": 2000,
            "questionType": "OPENEND",
            "otherBrand": false,
            "existingCodeFrame": [],
            "isList": false,
            "codeFrameWithDataList": [],
            "codeFrameList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 10294,
                        "CandidatesTokenCount": 3027,
                        "TotalTokenCount": 14880,
                        "thoughtsTokenCount": 1559,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Net": "AI's Impact on Employment",
                                "Subnets": [
                                    {
                                        "Subnet": "Job Displacement vs. Creation",
                                        "Categories": [
                                            {
                                                "Category": "Fears of job loss (historical context)",
                                                "Summary": "Discussion on historical fears of job loss with new technologies (e.g., computers in banks) and how they often led to increased productivity and new types of jobs."
                                            },
                                            {
                                                "Category": "Productivity Increase",
                                                "Summary": "AI is being used to increase productivity, allowing the same number of people to do more and better things, serving previously underserved markets."
                                            },
                                            {
                                                "Category": "New Job Roles",
                                                "Summary": "AI is expected to create entirely new job roles, such as 'care navigators' in healthcare, which did not exist before."
                                            }
                                        ],
                                        "Summary": "Examines the debate around whether AI leads to job losses or creates new opportunities, often drawing parallels from past technological shifts."
                                    },
                                    {
                                        "Subnet": "Changing Nature of Work",
                                        "Categories": [
                                            {
                                                "Category": "Humans Managing AI Agents",
                                                "Summary": "The future workforce will involve human beings managing and monitoring AI agents, with teams potentially comprising a mix of human and AI workers."
                                            },
                                            {
                                                "Category": "Focus on Human Accountability",
                                                "Summary": "Despite technological advancements, human beings will remain accountable for the outcomes generated by AI technologies."
                                            },
                                            {
                                                "Category": "Automation of Micro-Skills",
                                                "Summary": "AI automates repetitive and specialized 'micro-skills,' freeing humans to focus on higher-level tasks and problem-solving."
                                            }
                                        ],
                                        "Summary": "Describes how AI will transform daily tasks and responsibilities, shifting the focus towards human oversight and management of intelligent systems."
                                    },
                                    {
                                        "Subnet": "Economic Outcomes",
                                        "Categories": [
                                            {
                                                "Category": "Serving Underserved Segments",
                                                "Summary": "AI enables businesses to serve market segments that were previously unprofitable to reach with human resources, e.g., chatbots for low-value clients."
                                            },
                                            {
                                                "Category": "Value Creation",
                                                "Summary": "AI contributes to greater value creation and faster value creation across enterprises by enhancing efficiency and enabling new services."
                                            },
                                            {
                                                "Category": "ROI from AI Investment",
                                                "Summary": "While trillions are invested in AI, the significant Return on Investment (ROI) is anticipated to come from specific use cases and practical applications."
                                            }
                                        ],
                                        "Summary": "Discusses the financial and market implications of AI adoption, including market expansion and new avenues for profitability."
                                    }
                                ],
                                "Summary": "Covers how AI is transforming the workforce, including debates on job creation versus displacement, the evolving roles of human workers, and the economic benefits."
                            },
                            {
                                "Net": "AI in Industry-Specific Transformation",
                                "Subnets": [
                                    {
                                        "Subnet": "IT Services & Software Development",
                                        "Categories": [
                                            {
                                                "Category": "Increased Productivity",
                                                "Summary": "AI has generated significant productivity gains (e.g., 70-80%) in software product companies, allowing for faster innovation and higher quality products."
                                            },
                                            {
                                                "Category": "Agile Teams with AI Agents",
                                                "Summary": "Future agile development teams may consist of fewer human engineers working alongside multiple AI agents to define problems and manage tasks."
                                            },
                                            {
                                                "Category": "Enterprise Automation",
                                                "Summary": "AI is driving greater automation within enterprises, leading to more technology-driven and digitally native operating models."
                                            }
                                        ],
                                        "Summary": "Explores the specific ways AI is revolutionizing the IT and software development industry, from enhancing developer productivity to reshaping team structures."
                                    },
                                    {
                                        "Subnet": "Healthcare Sector",
                                        "Categories": [
                                            {
                                                "Category": "Reimagining Healthcare Access",
                                                "Summary": "AI is crucial for reimagining healthcare systems to provide meaningful access to the 5 billion people globally who currently lack it, moving care closer to communities."
                                            },
                                            {
                                                "Category": "Cost-Effective Care",
                                                "Summary": "Technology innovation, including AI, is essential to make healthcare more cost-effective, especially for emerging economies and aging populations."
                                            },
                                            {
                                                "Category": "AI as Net Job Creator",
                                                "Summary": "In healthcare, AI is viewed as a net job creator, enabling the training of millions of professionals for new roles like 'care navigators' to address growing health needs."
                                            }
                                        ],
                                        "Summary": "Details how AI is addressing critical challenges in healthcare, such as access, cost, and the potential for creating new care-delivery roles."
                                    },
                                    {
                                        "Subnet": "Media & Journalism",
                                        "Categories": [
                                            {
                                                "Category": "Threat to Traditional Roles",
                                                "Summary": "AI poses an existential threat to traditional media roles like reporters, anchors, and camera persons, as content can be generated or summarized automatically."
                                            },
                                            {
                                                "Category": "AI-Generated Content",
                                                "Summary": "AI tools are being used to create resumes, summarize news, and even generate different versions of anchors, impacting content creation and consumption."
                                            },
                                            {
                                                "Category": "IPR Concerns",
                                                "Summary": "A significant concern is that large AI companies are scraping content from Indian media organizations without compensation, threatening intellectual property rights."
                                            }
                                        ],
                                        "Summary": "Highlights the profound disruptive impact of AI on the media industry, including job security, content creation methods, and intellectual property challenges."
                                    }
                                ],
                                "Summary": "Analyzes how AI is specifically transforming key industries such as IT, healthcare, and media, detailing unique challenges and opportunities within each sector."
                            },
                            {
                                "Net": "Future of Skills & Education",
                                "Subnets": [
                                    {
                                        "Subnet": "Essential Skills for the AI Era",
                                        "Categories": [
                                            {
                                                "Category": "Lifelong Learnability",
                                                "Summary": "The ability to continuously learn and adapt is identified as the most crucial capability for individuals in an era of rapid technological change driven by AI."
                                            },
                                            {
                                                "Category": "Macro Skills (Reimagining, Problem-Solving)",
                                                "Summary": "The demand is shifting from 'micro-skills' to 'macro-skills' such as reimagining solutions, complex problem-solving, and critical thinking that machines cannot replicate."
                                            },
                                            {
                                                "Category": "Learning AI Tools (Personal Goal)",
                                                "Summary": "Individuals are encouraged to proactively learn and apply multiple AI tools to secure their jobs and enhance their careers, as older generations are slower to adapt."
                                            }
                                        ],
                                        "Summary": "Focuses on the core competencies and mindset required for individuals to thrive in an AI-driven future, emphasizing continuous learning and advanced cognitive skills."
                                    },
                                    {
                                        "Subnet": "Academic Readiness",
                                        "Categories": [
                                            {
                                                "Category": "Institutions Slow to Adapt",
                                                "Summary": "Educational institutions, even leading ones, are not adequately preparing the next generation for AI-driven job changes, showing a lack of vision."
                                            },
                                            {
                                                "Category": "Lack of Vision in Education",
                                                "Summary": "There's a perceived pushback or slowness by academic institutions to integrate AI skills into curricula from an early stage, such as middle school."
                                            },
                                            {
                                                "Category": "Students Using AI for Lazy Work",
                                                "Summary": "A concern is that students are using AI tools (e.g., ChatGPT) to generate generic resumes and content, indicating a lazy attitude rather than genuine skill development."
                                            }
                                        ],
                                        "Summary": "Examines the current state of educational systems and their ability (or lack thereof) to prepare students for the demands of an AI-transformed job market."
                                    },
                                    {
                                        "Subnet": "Reskilling Imperative",
                                        "Categories": [
                                            {
                                                "Category": "Obsolescence of Current Skills",
                                                "Summary": "Existing skills will rapidly become obsolete, necessitating a shift from 'upgrading' to 'reskilling' to acquire new competencies relevant for tomorrow."
                                            },
                                            {
                                                "Category": "Opportunity to Reskill",
                                                "Summary": "The current era presents a significant opportunity for individuals across all age groups to reskill themselves for future jobs."
                                            },
                                            {
                                                "Category": "Focus on Human Unique Capabilities",
                                                "Summary": "Education should focus on teaching children to do what humans do best—reimagine, solve problems, and think creatively—as machines cannot replicate these."
                                            }
                                        ],
                                        "Summary": "Highlights the critical need for individuals and institutions to proactively engage in reskilling initiatives to remain relevant in a rapidly evolving job landscape."
                                    }
                                ],
                                "Summary": "Addresses the critical transformation needed in skills development and education to prepare individuals for the demands of an AI-powered economy."
                            },
                            {
                                "Net": "Strategic Implications for India",
                                "Subnets": [
                                    {
                                        "Subnet": "Global Competitiveness",
                                        "Categories": [
                                            {
                                                "Category": "India's Opportunity to Leapfrog",
                                                "Summary": "India has a unique opportunity to leapfrog in AI adoption and application, rather than just catching up, especially in areas like healthcare."
                                            },
                                            {
                                                "Category": "Developing a Unique Proposition",
                                                "Summary": "India needs to develop a distinct competitive proposition to attract and create AI-related jobs on a global scale, similar to how it dominated IT services."
                                            },
                                            {
                                                "Category": "Competing for Global Jobs",
                                                "Summary": "The global landscape will involve countries competing to leverage AI to create jobs and capture a larger share of the new employment opportunities."
                                            }
                                        ],
                                        "Summary": "Discusses India's potential to assert its position on the global stage by strategically embracing and innovating with AI."
                                    },
                                    {
                                        "Subnet": "Policy & IPR",
                                        "Categories": [
                                            {
                                                "Category": "Protection of Intellectual Property Rights",
                                                "Summary": "There's an urgent need for policies and actions to protect the intellectual property rights of Indian media and content creators from being scraped by large AI models without compensation."
                                            },
                                            {
                                                "Category": "Government Incentives for LLMs (Missed Race)",
                                                "Summary": "Concerns are raised about India potentially having lost the race in developing proprietary Large Language Models (LLMs) and the need for strategic intervention."
                                            },
                                            {
                                                "Category": "Systemic Problems (Not Individual Focus)",
                                                "Summary": "While individuals focus on personal skill development, there's a broader need for systemic and policy-level interventions to manage AI's impact."
                                            }
                                        ],
                                        "Summary": "Explores the governmental and regulatory aspects necessary for India to navigate the AI revolution effectively, including intellectual property and policy development."
                                    },
                                    {
                                        "Subnet": "Historical Parallels & Future Outlook",
                                        "Categories": [
                                            {
                                                "Category": "Slow Institutional Response",
                                                "Summary": "Historical examples (e.g., computer adoption in banks) show that institutions respond slowly to new concepts and technologies, and AI is no different."
                                            },
                                            {
                                                "Category": "Technology Adoption (Banks, Google)",
                                                "Summary": "Recounts stories of initial resistance to technologies like computers in banks or Google by journalists, which eventually became indispensable, highlighting AI's similar trajectory."
                                            },
                                            {
                                                "Category": "India's IT Migration History",
                                                "Summary": "Draws a parallel between AI's potential and India's success in migrating IT jobs globally by offering a unique value proposition, suggesting a similar strategy for AI."
                                            }
                                        ],
                                        "Summary": "Reflects on past technological shifts to forecast AI's future trajectory and its potential to reshape India's economic and employment landscape."
                                    }
                                ],
                                "Summary": "Outlines the specific strategic challenges and opportunities AI presents for India, encompassing global competitiveness, policy-making, and historical learning."
                            },
                            {
                                "Net": "Nature & Adoption of AI Technology",
                                "Subnets": [
                                    {
                                        "Subnet": "AI as a Capability Multiplier",
                                        "Categories": [
                                            {
                                                "Category": "Deliver Faster/Better Outcomes",
                                                "Summary": "AI enables workers to achieve results faster, better, and with incredible outcomes in their existing jobs by multiplying their capabilities."
                                            },
                                            {
                                                "Category": "Amplify Human Abilities",
                                                "Summary": "AI is presented not as a replacement for human jobs but as a tool to amplify human abilities, making individuals more productive and effective."
                                            },
                                            {
                                                "Category": "Not Job Replacement Tool",
                                                "Summary": "A core message is that AI should be seen as a capability multiplier and an enabler, rather than primarily a tool for replacing human jobs."
                                            }
                                        ],
                                        "Summary": "Defines AI primarily as a tool that enhances human potential and productivity across various tasks and roles."
                                    },
                                    {
                                        "Subnet": "Phases of AI Evolution",
                                        "Categories": [
                                            {
                                                "Category": "Automation (Immediate)",
                                                "Summary": "The immediate phase (next 2-3 years) will see incredible automation across various processes, building on existing trends."
                                            },
                                            {
                                                "Category": "Workforce Restructuring (3-5 Years)",
                                                "Summary": "A subsequent phase (3-5 years) will involve significant workforce restructuring as jobs evolve and new roles emerge due to automation."
                                            },
                                            {
                                                "Category": "Business Model Transformation (Long Term)",
                                                "Summary": "The final and most profound phase will involve business model transformation, mirroring digital-native operating models and creating more value."
                                            }
                                        ],
                                        "Summary": "Describes a staged progression of AI's impact, from initial automation to fundamental shifts in business structures over time."
                                    },
                                    {
                                        "Subnet": "Technology Integration",
                                        "Categories": [
                                            {
                                                "Category": "More Technology in Enterprises",
                                                "Summary": "AI will lead to an increased integration of technology within enterprises, making tech more entrenched rather than less."
                                            },
                                            {
                                                "Category": "Digital Native Operating Models",
                                                "Summary": "AI will push enterprises to become more like 'digital native' companies, where software and AI are at the heart of running the business."
                                            },
                                            {
                                                "Category": "AI as a Use Case (Not Just Technology)",
                                                "Summary": "It's crucial to view AI as a use case and a solution to problems, rather than merely a technology, emphasizing its application-driven value."
                                            }
                                        ],
                                        "Summary": "Addresses how AI will be integrated into organizational structures and processes, leading to more technologically advanced and efficient operations."
                                    }
                                ],
                                "Summary": "Explains the fundamental characteristics of AI, its role as a capability enhancer, and the anticipated stages of its integration into industries."
                            },
                            {
                                "Net": "996 Non specific TOPIC",
                                "Subnets": [
                                    {
                                        "Subnet": "Irrelevant/Uncodable",
                                        "Categories": [
                                            {
                                                "Category": "General Non-Specific Responses",
                                                "Summary": "Responses that are not directly related to any identified theme or are too vague to be coded elsewhere."
                                            }
                                        ],
                                        "Summary": "Captures responses that are irrelevant to the core topics or cannot be clearly categorized within the defined codeframe."
                                    }
                                ],
                                "Summary": "For responses that are irrelevant, but could not be coded in the themes within the codeframe easily."
                            },
                            {
                                "Net": "997 Don’t Know Can’t Say",
                                "Subnets": [
                                    {
                                        "Subnet": "Non-Committal/Vague",
                                        "Categories": [
                                            {
                                                "Category": "Ambiguous or Indecisive Responses",
                                                "Summary": "Responses indicating uncertainty, lack of knowledge, or a refusal to commit to an opinion or statement."
                                            }
                                        ],
                                        "Summary": "For responses which are non-committal and vague."
                                    }
                                ],
                                "Summary": "For responses which are non-committal and vague."
                            },
                            {
                                "Net": "998 Not Applicable",
                                "Subnets": [
                                    {
                                        "Subnet": "Out of Scope",
                                        "Categories": [
                                            {
                                                "Category": "Responses Not Relevant to Discussion",
                                                "Summary": "Responses that are clearly outside the scope of the survey question or the overall project."
                                            }
                                        ],
                                        "Summary": "Captures responses that are not applicable to the question or project at hand."
                                    }
                                ],
                                "Summary": "For responses which are irrelevant, not applicable to the question."
                            },
                            {
                                "Net": "999 None-Nothing",
                                "Subnets": [
                                    {
                                        "Subnet": "Gibberish/Meaningless",
                                        "Categories": [
                                            {
                                                "Category": "Unintelligible or Nonsensical Input",
                                                "Summary": "Responses consisting of random characters, incomprehensible language, or expressions that convey no discernible meaning."
                                            }
                                        ],
                                        "Summary": "For responses which are plain gibberish non-language characters or mean nothing."
                                    }
                                ],
                                "Summary": "For responses which are plain gibberish non-language characters or mean nothing."
                            }
                        ],
                        "Summary": "The provided transcript is a panel discussion focusing on the multifaceted impact of Artificial Intelligence (AI) on employment, industries, and societal structures, particularly within the Indian context. Speakers from IT, healthcare, and media sectors share their perspectives on AI as a capability multiplier, a driver of productivity, and a catalyst for job transformation rather than outright job destruction. Key themes include the necessity of lifelong learning and reskilling, the slow adaptation of academic institutions, the distinct challenges and opportunities AI presents for India (e.g., leapfrogging vs. IPR protection), and the historical parallels of technology adoption. The discussion emphasizes that while AI will automate 'micro-skills,' it will also create new 'macro-skills' and jobs, fundamentally changing the nature of work across various sectors."
                    }
                }
            ],
            "onlyCodeFrame": true,
            "errorList": [],
            "codeFrameGenerateType": "CODE_FRAME_ONLY",
            "requestType": "CODEFRAME",
            "requestMode": "DEVELOPMENT",
            "sourceLanguage": "English",
            "targetLanguage": "English",
            "response": {
                "status": "OK",
                "data": {
                    "status": "OK",
                    "PromptTokenCount": 10294,
                    "CandidatesTokenCount": 3027,
                    "TotalTokenCount": 14880,
                    "thoughtsTokenCount": 1559,
                    "finishReason": "STOP",
                    "data": [
                        {
                            "Net": "AI's Impact on Employment",
                            "Subnets": [
                                {
                                    "Subnet": "Job Displacement vs. Creation",
                                    "Categories": [
                                        {
                                            "Category": "Fears of job loss (historical context)",
                                            "Summary": "Discussion on historical fears of job loss with new technologies (e.g., computers in banks) and how they often led to increased productivity and new types of jobs.",
                                            "code": 3
                                        },
                                        {
                                            "Category": "Productivity Increase",
                                            "Summary": "AI is being used to increase productivity, allowing the same number of people to do more and better things, serving previously underserved markets.",
                                            "code": 4
                                        },
                                        {
                                            "Category": "New Job Roles",
                                            "Summary": "AI is expected to create entirely new job roles, such as 'care navigators' in healthcare, which did not exist before.",
                                            "code": 5
                                        }
                                    ],
                                    "Summary": "Examines the debate around whether AI leads to job losses or creates new opportunities, often drawing parallels from past technological shifts.",
                                    "code": 2
                                },
                                {
                                    "Subnet": "Changing Nature of Work",
                                    "Categories": [
                                        {
                                            "Category": "Humans Managing AI Agents",
                                            "Summary": "The future workforce will involve human beings managing and monitoring AI agents, with teams potentially comprising a mix of human and AI workers.",
                                            "code": 7
                                        },
                                        {
                                            "Category": "Focus on Human Accountability",
                                            "Summary": "Despite technological advancements, human beings will remain accountable for the outcomes generated by AI technologies.",
                                            "code": 8
                                        },
                                        {
                                            "Category": "Automation of Micro-Skills",
                                            "Summary": "AI automates repetitive and specialized 'micro-skills,' freeing humans to focus on higher-level tasks and problem-solving.",
                                            "code": 9
                                        }
                                    ],
                                    "Summary": "Describes how AI will transform daily tasks and responsibilities, shifting the focus towards human oversight and management of intelligent systems.",
                                    "code": 6
                                },
                                {
                                    "Subnet": "Economic Outcomes",
                                    "Categories": [
                                        {
                                            "Category": "Serving Underserved Segments",
                                            "Summary": "AI enables businesses to serve market segments that were previously unprofitable to reach with human resources, e.g., chatbots for low-value clients.",
                                            "code": 11
                                        },
                                        {
                                            "Category": "Value Creation",
                                            "Summary": "AI contributes to greater value creation and faster value creation across enterprises by enhancing efficiency and enabling new services.",
                                            "code": 12
                                        },
                                        {
                                            "Category": "ROI from AI Investment",
                                            "Summary": "While trillions are invested in AI, the significant Return on Investment (ROI) is anticipated to come from specific use cases and practical applications.",
                                            "code": 13
                                        }
                                    ],
                                    "Summary": "Discusses the financial and market implications of AI adoption, including market expansion and new avenues for profitability.",
                                    "code": 10
                                }
                            ],
                            "Summary": "Covers how AI is transforming the workforce, including debates on job creation versus displacement, the evolving roles of human workers, and the economic benefits.",
                            "code": 1
                        },
                        {
                            "Net": "AI in Industry-Specific Transformation",
                            "Subnets": [
                                {
                                    "Subnet": "IT Services & Software Development",
                                    "Categories": [
                                        {
                                            "Category": "Increased Productivity",
                                            "Summary": "AI has generated significant productivity gains (e.g., 70-80%) in software product companies, allowing for faster innovation and higher quality products.",
                                            "code": 16
                                        },
                                        {
                                            "Category": "Agile Teams with AI Agents",
                                            "Summary": "Future agile development teams may consist of fewer human engineers working alongside multiple AI agents to define problems and manage tasks.",
                                            "code": 17
                                        },
                                        {
                                            "Category": "Enterprise Automation",
                                            "Summary": "AI is driving greater automation within enterprises, leading to more technology-driven and digitally native operating models.",
                                            "code": 18
                                        }
                                    ],
                                    "Summary": "Explores the specific ways AI is revolutionizing the IT and software development industry, from enhancing developer productivity to reshaping team structures.",
                                    "code": 15
                                },
                                {
                                    "Subnet": "Healthcare Sector",
                                    "Categories": [
                                        {
                                            "Category": "Reimagining Healthcare Access",
                                            "Summary": "AI is crucial for reimagining healthcare systems to provide meaningful access to the 5 billion people globally who currently lack it, moving care closer to communities.",
                                            "code": 20
                                        },
                                        {
                                            "Category": "Cost-Effective Care",
                                            "Summary": "Technology innovation, including AI, is essential to make healthcare more cost-effective, especially for emerging economies and aging populations.",
                                            "code": 21
                                        },
                                        {
                                            "Category": "AI as Net Job Creator",
                                            "Summary": "In healthcare, AI is viewed as a net job creator, enabling the training of millions of professionals for new roles like 'care navigators' to address growing health needs.",
                                            "code": 22
                                        }
                                    ],
                                    "Summary": "Details how AI is addressing critical challenges in healthcare, such as access, cost, and the potential for creating new care-delivery roles.",
                                    "code": 19
                                },
                                {
                                    "Subnet": "Media & Journalism",
                                    "Categories": [
                                        {
                                            "Category": "Threat to Traditional Roles",
                                            "Summary": "AI poses an existential threat to traditional media roles like reporters, anchors, and camera persons, as content can be generated or summarized automatically.",
                                            "code": 24
                                        },
                                        {
                                            "Category": "AI-Generated Content",
                                            "Summary": "AI tools are being used to create resumes, summarize news, and even generate different versions of anchors, impacting content creation and consumption.",
                                            "code": 25
                                        },
                                        {
                                            "Category": "IPR Concerns",
                                            "Summary": "A significant concern is that large AI companies are scraping content from Indian media organizations without compensation, threatening intellectual property rights.",
                                            "code": 26
                                        }
                                    ],
                                    "Summary": "Highlights the profound disruptive impact of AI on the media industry, including job security, content creation methods, and intellectual property challenges.",
                                    "code": 23
                                }
                            ],
                            "Summary": "Analyzes how AI is specifically transforming key industries such as IT, healthcare, and media, detailing unique challenges and opportunities within each sector.",
                            "code": 14
                        },
                        {
                            "Net": "Future of Skills & Education",
                            "Subnets": [
                                {
                                    "Subnet": "Essential Skills for the AI Era",
                                    "Categories": [
                                        {
                                            "Category": "Lifelong Learnability",
                                            "Summary": "The ability to continuously learn and adapt is identified as the most crucial capability for individuals in an era of rapid technological change driven by AI.",
                                            "code": 29
                                        },
                                        {
                                            "Category": "Macro Skills (Reimagining, Problem-Solving)",
                                            "Summary": "The demand is shifting from 'micro-skills' to 'macro-skills' such as reimagining solutions, complex problem-solving, and critical thinking that machines cannot replicate.",
                                            "code": 30
                                        },
                                        {
                                            "Category": "Learning AI Tools (Personal Goal)",
                                            "Summary": "Individuals are encouraged to proactively learn and apply multiple AI tools to secure their jobs and enhance their careers, as older generations are slower to adapt.",
                                            "code": 31
                                        }
                                    ],
                                    "Summary": "Focuses on the core competencies and mindset required for individuals to thrive in an AI-driven future, emphasizing continuous learning and advanced cognitive skills.",
                                    "code": 28
                                },
                                {
                                    "Subnet": "Academic Readiness",
                                    "Categories": [
                                        {
                                            "Category": "Institutions Slow to Adapt",
                                            "Summary": "Educational institutions, even leading ones, are not adequately preparing the next generation for AI-driven job changes, showing a lack of vision.",
                                            "code": 33
                                        },
                                        {
                                            "Category": "Lack of Vision in Education",
                                            "Summary": "There's a perceived pushback or slowness by academic institutions to integrate AI skills into curricula from an early stage, such as middle school.",
                                            "code": 34
                                        },
                                        {
                                            "Category": "Students Using AI for Lazy Work",
                                            "Summary": "A concern is that students are using AI tools (e.g., ChatGPT) to generate generic resumes and content, indicating a lazy attitude rather than genuine skill development.",
                                            "code": 35
                                        }
                                    ],
                                    "Summary": "Examines the current state of educational systems and their ability (or lack thereof) to prepare students for the demands of an AI-transformed job market.",
                                    "code": 32
                                },
                                {
                                    "Subnet": "Reskilling Imperative",
                                    "Categories": [
                                        {
                                            "Category": "Obsolescence of Current Skills",
                                            "Summary": "Existing skills will rapidly become obsolete, necessitating a shift from 'upgrading' to 'reskilling' to acquire new competencies relevant for tomorrow.",
                                            "code": 37
                                        },
                                        {
                                            "Category": "Opportunity to Reskill",
                                            "Summary": "The current era presents a significant opportunity for individuals across all age groups to reskill themselves for future jobs.",
                                            "code": 38
                                        },
                                        {
                                            "Category": "Focus on Human Unique Capabilities",
                                            "Summary": "Education should focus on teaching children to do what humans do best—reimagine, solve problems, and think creatively—as machines cannot replicate these.",
                                            "code": 39
                                        }
                                    ],
                                    "Summary": "Highlights the critical need for individuals and institutions to proactively engage in reskilling initiatives to remain relevant in a rapidly evolving job landscape.",
                                    "code": 36
                                }
                            ],
                            "Summary": "Addresses the critical transformation needed in skills development and education to prepare individuals for the demands of an AI-powered economy.",
                            "code": 27
                        },
                        {
                            "Net": "Strategic Implications for India",
                            "Subnets": [
                                {
                                    "Subnet": "Global Competitiveness",
                                    "Categories": [
                                        {
                                            "Category": "India's Opportunity to Leapfrog",
                                            "Summary": "India has a unique opportunity to leapfrog in AI adoption and application, rather than just catching up, especially in areas like healthcare.",
                                            "code": 42
                                        },
                                        {
                                            "Category": "Developing a Unique Proposition",
                                            "Summary": "India needs to develop a distinct competitive proposition to attract and create AI-related jobs on a global scale, similar to how it dominated IT services.",
                                            "code": 43
                                        },
                                        {
                                            "Category": "Competing for Global Jobs",
                                            "Summary": "The global landscape will involve countries competing to leverage AI to create jobs and capture a larger share of the new employment opportunities.",
                                            "code": 44
                                        }
                                    ],
                                    "Summary": "Discusses India's potential to assert its position on the global stage by strategically embracing and innovating with AI.",
                                    "code": 41
                                },
                                {
                                    "Subnet": "Policy & IPR",
                                    "Categories": [
                                        {
                                            "Category": "Protection of Intellectual Property Rights",
                                            "Summary": "There's an urgent need for policies and actions to protect the intellectual property rights of Indian media and content creators from being scraped by large AI models without compensation.",
                                            "code": 46
                                        },
                                        {
                                            "Category": "Government Incentives for LLMs (Missed Race)",
                                            "Summary": "Concerns are raised about India potentially having lost the race in developing proprietary Large Language Models (LLMs) and the need for strategic intervention.",
                                            "code": 47
                                        },
                                        {
                                            "Category": "Systemic Problems (Not Individual Focus)",
                                            "Summary": "While individuals focus on personal skill development, there's a broader need for systemic and policy-level interventions to manage AI's impact.",
                                            "code": 48
                                        }
                                    ],
                                    "Summary": "Explores the governmental and regulatory aspects necessary for India to navigate the AI revolution effectively, including intellectual property and policy development.",
                                    "code": 45
                                },
                                {
                                    "Subnet": "Historical Parallels & Future Outlook",
                                    "Categories": [
                                        {
                                            "Category": "Slow Institutional Response",
                                            "Summary": "Historical examples (e.g., computer adoption in banks) show that institutions respond slowly to new concepts and technologies, and AI is no different.",
                                            "code": 50
                                        },
                                        {
                                            "Category": "Technology Adoption (Banks, Google)",
                                            "Summary": "Recounts stories of initial resistance to technologies like computers in banks or Google by journalists, which eventually became indispensable, highlighting AI's similar trajectory.",
                                            "code": 51
                                        },
                                        {
                                            "Category": "India's IT Migration History",
                                            "Summary": "Draws a parallel between AI's potential and India's success in migrating IT jobs globally by offering a unique value proposition, suggesting a similar strategy for AI.",
                                            "code": 52
                                        }
                                    ],
                                    "Summary": "Reflects on past technological shifts to forecast AI's future trajectory and its potential to reshape India's economic and employment landscape.",
                                    "code": 49
                                }
                            ],
                            "Summary": "Outlines the specific strategic challenges and opportunities AI presents for India, encompassing global competitiveness, policy-making, and historical learning.",
                            "code": 40
                        },
                        {
                            "Net": "Nature & Adoption of AI Technology",
                            "Subnets": [
                                {
                                    "Subnet": "AI as a Capability Multiplier",
                                    "Categories": [
                                        {
                                            "Category": "Deliver Faster/Better Outcomes",
                                            "Summary": "AI enables workers to achieve results faster, better, and with incredible outcomes in their existing jobs by multiplying their capabilities.",
                                            "code": 55
                                        },
                                        {
                                            "Category": "Amplify Human Abilities",
                                            "Summary": "AI is presented not as a replacement for human jobs but as a tool to amplify human abilities, making individuals more productive and effective.",
                                            "code": 56
                                        },
                                        {
                                            "Category": "Not Job Replacement Tool",
                                            "Summary": "A core message is that AI should be seen as a capability multiplier and an enabler, rather than primarily a tool for replacing human jobs.",
                                            "code": 57
                                        }
                                    ],
                                    "Summary": "Defines AI primarily as a tool that enhances human potential and productivity across various tasks and roles.",
                                    "code": 54
                                },
                                {
                                    "Subnet": "Phases of AI Evolution",
                                    "Categories": [
                                        {
                                            "Category": "Automation (Immediate)",
                                            "Summary": "The immediate phase (next 2-3 years) will see incredible automation across various processes, building on existing trends.",
                                            "code": 59
                                        },
                                        {
                                            "Category": "Workforce Restructuring (3-5 Years)",
                                            "Summary": "A subsequent phase (3-5 years) will involve significant workforce restructuring as jobs evolve and new roles emerge due to automation.",
                                            "code": 60
                                        },
                                        {
                                            "Category": "Business Model Transformation (Long Term)",
                                            "Summary": "The final and most profound phase will involve business model transformation, mirroring digital-native operating models and creating more value.",
                                            "code": 61
                                        }
                                    ],
                                    "Summary": "Describes a staged progression of AI's impact, from initial automation to fundamental shifts in business structures over time.",
                                    "code": 58
                                },
                                {
                                    "Subnet": "Technology Integration",
                                    "Categories": [
                                        {
                                            "Category": "More Technology in Enterprises",
                                            "Summary": "AI will lead to an increased integration of technology within enterprises, making tech more entrenched rather than less.",
                                            "code": 63
                                        },
                                        {
                                            "Category": "Digital Native Operating Models",
                                            "Summary": "AI will push enterprises to become more like 'digital native' companies, where software and AI are at the heart of running the business.",
                                            "code": 64
                                        },
                                        {
                                            "Category": "AI as a Use Case (Not Just Technology)",
                                            "Summary": "It's crucial to view AI as a use case and a solution to problems, rather than merely a technology, emphasizing its application-driven value.",
                                            "code": 65
                                        }
                                    ],
                                    "Summary": "Addresses how AI will be integrated into organizational structures and processes, leading to more technologically advanced and efficient operations.",
                                    "code": 62
                                }
                            ],
                            "Summary": "Explains the fundamental characteristics of AI, its role as a capability enhancer, and the anticipated stages of its integration into industries.",
                            "code": 53
                        }
                    ],
                    "Summary": "The provided transcript is a panel discussion focusing on the multifaceted impact of Artificial Intelligence (AI) on employment, industries, and societal structures, particularly within the Indian context. Speakers from IT, healthcare, and media sectors share their perspectives on AI as a capability multiplier, a driver of productivity, and a catalyst for job transformation rather than outright job destruction. Key themes include the necessity of lifelong learning and reskilling, the slow adaptation of academic institutions, the distinct challenges and opportunities AI presents for India (e.g., leapfrogging vs. IPR protection), and the historical parallels of technology adoption. The discussion emphasizes that while AI will automate 'micro-skills,' it will also create new 'macro-skills' and jobs, fundamentally changing the nature of work across various sectors."
                }
            },
            "summaryList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 9955,
                        "CandidatesTokenCount": 2988,
                        "TotalTokenCount": 14212,
                        "thoughtsTokenCount": 1269,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Summary": "<h2>Comprehensive Synthesis Report: AI's Impact on Employment and Future Skills</h2><p>This report synthesizes key discussions and findings regarding the impact of Artificial Intelligence (AI) on employment, skills, and organizational structures, drawing from various expert perspectives. The central theme revolves around the balance between job displacement fears and the potential for productivity enhancement, job creation, and societal transformation through AI.</p><h3>1. Historical Perspective on Technological Disruption and Job Fears</h3><p>The discussion opened by addressing the historical pattern of institutional response to new technologies and the pervasive fear of job loss. Historically, institutions, especially those not accountable to markets, respond slowly to new concepts [Transcript 0:05]. A compelling example from 1985 recounts the introduction of computers in Indian banks, met with strong resistance from the All India Bank Employees Association due to fears of job displacement [Transcript 0:37]. However, the government pushed forward, and the outcome was unexpected: nobody lost jobs; instead, people became more productive, serving customers better, often doing more and different, more useful things with the same number of people [Transcript 1:00, 1:24]. This historical context suggests that while fear is common, the reality often involves increased productivity and evolving job roles, rather than outright elimination [Transcript 1:12]. While acknowledging that 'anything's possible' and the current wave 'can be different this time,' the necessity of a different outcome is not assured [Transcript 1:31].</p><h3>2. AI as a Capability Multiplier and Productivity Enhancer</h3><p>A predominant view is that AI should be perceived as a 'capability multiplier' rather than solely a job replacement tool [Transcript 8:28]. It enables individuals to deliver faster, better, and more incredible outcomes in their existing roles [Transcript 8:40].</p><h4>Supporting Evidence and Examples:</h4><ul><li><strong>Underserved Market Segments:</strong> An example from a company highlights using AI-powered chatbots and automated calls to serve the bottom 50% of clients who were previously uneconomical to reach via field or tele-sales [Transcript 2:17]. This expanded service to an 'underserved segment' without immediate job losses, increasing productivity and efficiency [Transcript 2:44, 2:55].</li><li><strong>Content Summarization:</strong> AI tools like 'Claude' are used to summarize long video or podcast content (e.g., a 30-40 minute interview summarized in 3 minutes), allowing quick absorption of information that otherwise might not have been consumed due to time constraints [Transcript 3:32]. This demonstrates AI enabling access to more content, fostering knowledge absorption even if traditional engagement metrics (like full watch time) are affected [Transcript 4:05].</li><li><strong>Personal Skill Advantage:</strong> A personal anecdote from 1989 illustrates the advantage of early adoption of new technology. Knowing how to use Harvard Graphics software for presentations (when others used hand-drawn transparencies) made an individual the 'go-to guy' and the 'most useful guy at the ground level,' highlighting that learning new, easy-to-use technologies creates immediate value and career opportunities [Transcript 6:31, 6:50]. This is directly analogous to learning AI tools today [Transcript 7:03].</li><li><strong>Software Development Productivity:</strong> In a software product company, AI has generated 'incredible productivity,' boosting it by 70-80% in the last 2-3 years, leading to faster innovation and higher quality products without necessarily reducing headcount [Transcript 12:28, 12:35].</li></ul><h3>3. Transforming Enterprises and Workforce Dynamics</h3><p>AI will lead to significant transformations in enterprises, making them 'more and more digitally native' [Transcript 10:44]. This implies not less technology, but 'more technology' and 'different kinds of jobs' [Transcript 9:54, 10:50].</p><h4>Workforce Restructuring Phases:</h4><p>The evolution of AI's impact on the workforce can be seen in three phases [Transcript 11:44]:</p><ul><li><strong>Automation:</strong> Immediate automation of tasks, leading to 'incredible automation' in the short term (2-3 years) [Transcript 11:44].</li><li><strong>Workforce Restructuring:</strong> Over the medium term (3-5 years), significant workforce restructuring will occur due to automation [Transcript 11:50].</li><li><strong>Business Model Transformation:</strong> The long-term impact involves business model transformation, mirroring digital native operating models and creating more value faster [Transcript 11:57].</li></ul><p>The concept of agile development teams might evolve from seven or eight human members to a model with 'three human beings and five agents,' where humans define problems, manage, and monitor AI agents, retaining ultimate accountability [Transcript 13:00, 13:19].</p><h3>4. AI's Role in Healthcare: A Case for Job Creation and Reimagination</h3><p>Healthcare is presented as a strong use case for AI, arguing for job creation rather than displacement [Transcript 15:34].</p><h4>Key Points for Healthcare:</h4><ul><li><strong>Unmet Global Need:</strong> 5 billion people globally lack meaningful access to healthcare, exacerbated by a 'sick-centric model' focused on tertiary care [Transcript 14:15, 14:21].</li><li><strong>Cost of Care:</strong> The cost of care will continue to rise with the current model, necessitating 'innovation, technology innovation... process innovation' to make healthcare cost-effective and accessible, especially for LMIC settings [Transcript 14:55, 15:15].</li><li><strong>AI as a Net Job Creator:</strong> In the US, 15% of the population is involved in healthcare; translating this to India means 150 million+ potential jobs [Transcript 15:41]. AI can enable the creation of new job types like 'care navigators for high-risk cohorts' (e.g., diabetes, TB, pregnancy, heart failure) which do not exist today [Transcript 16:30, 16:40]. As populations age and healthcare extends to smaller towns and villages, more people will be needed, enabled by technology like AI [Transcript 16:04, 16:11].</li><li><strong>India's Opportunity:</strong> India should aim to 'leapfrog' with AI in healthcare and other use cases, rather than merely catching up [Transcript 17:07, 17:13].</li></ul><h3>5. Challenges and Concerns: Job Displacement and Intellectual Property</h3><p>While the overall sentiment leans towards AI enabling new roles, significant challenges and concerns were raised.</p><h4>Specific Concerns:</h4><ul><li><strong>Direct Job Losses:</strong> Historical examples like the Telex operator being replaced by the fax machine, or the gopher by chip technology, underscore that some specific jobs do get eliminated by technological advancements [Transcript 20:47]. Individuals who fail to upskill may indeed lose their positions [Transcript 21:04].</li><li><strong>Media Industry Threat:</strong> The media sector faces an 'existential threat' due to AI [Transcript 29:34, 30:41]. News organizations struggle with budgets to send reporters to the ground, relying instead on AI tools [Transcript 29:25]. AI can generate content that mimics human output, raising concerns about journalistic integrity and the future of traditional media [Transcript 28:01].</li><li><strong>Intellectual Property Rights (IPR):</strong> A critical concern is that large technology companies are 'scraping content from Indian media organizations and not paying us for IPR,' while paying millions abroad [Transcript 31:42, 31:50]. This threatens the financial viability of media in its long form, as short-form, AI-generated content or summaries divert clicks and advertising revenue [Transcript 32:19, 32:38].</li><li><strong>Educational System Preparedness:</strong> Academic institutions, even those at the 'hub of technology innovation,' are not adequately preparing the next generation for AI-driven jobs [Transcript 17:46, 17:58]. Students are concerned about job market challenges, often misusing AI tools for applications (e.g., ChatGPT for resumes) without possessing the underlying skills [Transcript 27:42, 28:19].</li></ul><h3>6. The Urgency of Skill Transformation (Reskilling)</h3><p>The core issue is not the number of jobs, but the obsolescence of existing skills [Transcript 23:29, 23:40]. The industrial age focused on 'sub-skills' (breaking processes into smaller, scalable tasks), which AI is now automating [Transcript 33:22, 34:37].</p><h4>Key Aspects of Skill Transformation:</h4><ul><li><strong>Macro-Skills vs. Sub-Skills:</strong> The shift is from 'micro skill workers' to individuals possessing 'macro skills' [Transcript 34:06, 35:52]. This means moving beyond mere knowledge or sub-skill execution to 're-imagineer,' problem-solve, and 'imagine a solution of a problem which nobody else has been able to solve' [Transcript 36:04, 36:10].</li><li><strong>AI as a Use Case:</strong> AI is not just a technology but a 'use case,' requiring different skills to apply it effectively [Transcript 36:30, 36:43]. Teaching AI in schools should focus on problem-solving and reimagining, not just subscription-driven technology use [Transcript 36:54, 37:06].</li><li><strong>Lifelong Learnability:</strong> This is an 'essential ingredient' for all workers, as the pace of change will continuously accelerate [Transcript 19:18, 19:30, 19:36].</li><li><strong>India's Opportunity and Threat:</strong> India faces a threat due to its large workforce in sub-skill tasks [Transcript 37:23]. However, it also has an opportunity to develop a 'proposition' that makes it globally competitive in the AI-driven job market [Transcript 24:34, 24:40].</li></ul><h3>7. Conclusions</h3><p>The pervasive fear of job loss due to AI is a recurring theme with technological advancements, but historical data and current trends suggest a more nuanced outcome. AI functions primarily as a capability multiplier, enhancing productivity and enabling tasks that were previously unfeasible or uneconomical. While certain sub-skills and repetitive jobs are susceptible to automation, the technology concurrently creates new job categories and necessitates a fundamental shift in required human capabilities. Industries like healthcare stand to benefit immensely from AI-driven job creation, addressing unmet societal needs. However, significant challenges remain, particularly concerning intellectual property rights in content creation and the urgent need for educational institutions to adapt their curricula to prepare the future workforce for 'macro skills' and lifelong learning.</p><h3>8. Actionable Recommendations</h3><ul><li><strong>For Individuals (Especially Young Professionals and Students):</strong><ul><li><strong>Learn AI Tools:</strong> Set a personal target to learn 5-15 useful AI tools within a short timeframe (e.g., 3 months) and apply them in work [Transcript 7:22, 19:57, 20:04]. This will enhance job security and career enablement.</li><li><strong>Cultivate Lifelong Learnability:</strong> Recognize that continuous learning is paramount as technology evolves rapidly [Transcript 19:18, 19:30].</li><li><strong>Develop Macro-Skills:</strong> Focus on acquiring skills in problem-solving, imagining solutions, and reimagining processes, which AI cannot replicate [Transcript 36:04, 37:06].</li></ul></li><li><strong>For Educational Institutions:</strong><ul><li><strong>Curriculum Reform:</strong> Urgently integrate AI skills training, focusing on critical thinking, problem-solving, and the imaginative application of AI from middle school onwards, rather than just technology subscription [Transcript 27:49, 36:54, 37:06].</li><li><strong>Prepare for New Job Types:</strong> Proactively identify and train students for emerging AI-enabled roles, such as 'care navigators' [Transcript 16:30, 17:58].</li></ul></li><li><strong>For Businesses and Policy Makers:</strong><ul><li><strong>Embrace AI for Productivity and Underserved Markets:</strong> Leverage AI to increase efficiency, innovate faster, and serve previously inaccessible customer segments, potentially creating new value streams [Transcript 2:44, 12:42].</li><li><strong>Prioritize Reskilling Initiatives:</strong> Invest in training and reskilling programs for the existing workforce to transition from sub-skills to macro-skills, ensuring adaptability to changing job requirements [Transcript 23:46, 35:45].</li><li><strong>Address Intellectual Property Rights:</strong> Implement policies and frameworks to protect the IPR of content creators, especially media organizations, against unauthorized AI scraping and use [Transcript 31:42, 32:06].</li><li><strong>Foster AI Leapfrogging:</strong> For countries like India, focus on developing unique AI propositions and solutions that allow them to lead rather than merely catch up in global AI competition, potentially attracting a larger share of future jobs [Transcript 17:07, 24:34, 24:47].</li><li><strong>Combine AI with Human Creativity:</strong> Recognize that the most impactful applications of AI often involve human creativity and oversight, as demonstrated by the contrast between purely AI-generated and human-refined AI designs [Transcript 25:03, 25:10].</li></ul></li></ul>"
                            }
                        ]
                    }
                }
            ]
        }
    ],
    "headerCols": {
        "col1": "Part Id",
        "col2": "FileName",
        "col3": "Content"
    },
    "fileName": "",
    "projectType": "DOCUMENT",
    "sourceLanguage": "English",
    "targetLanguage": "English",
    "__v": 0
},
{
    "_id": "6992e3bf4c5e8f0974c8738f",
    "projectId": "835116",
    "projectName": "AI in Public Audit: Driving Transparency and Accountability",
    "projectDescription": "AI in Public Audit: Driving Transparency and Accountability",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-16T09:30:38.371Z",
    "projectUpdatedDate": "2026-02-16T09:30:38.371Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-16T09:30:38.371Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO_LINK",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames"
        },
        {
            "col1": "P1",
            "col2": "https://www.youtube.com/watch?v=lg2HbmlYtW4",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEOLINK",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:05",
                    "end": "00:00:08",
                    "text": "Thank you sir for the comprehensive and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:07",
                    "end": "00:00:10",
                    "text": "forward looking at this. You have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:08",
                    "end": "00:00:12",
                    "text": "clearly set the institutional context"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:10",
                    "end": "00:00:13",
                    "text": "for today's deliberations highlighting"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:12",
                    "end": "00:00:15",
                    "text": "both the opportunities and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:13",
                    "end": "00:00:18",
                    "text": "responsibilities that AI brings to the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:15",
                    "end": "00:00:20",
                    "text": "public world. We now move to the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:18",
                    "end": "00:00:23",
                    "text": "technical presentation on the technology"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:20",
                    "end": "00:00:25",
                    "text": "and AI initiatives of the CG of India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:23",
                    "end": "00:00:27",
                    "text": "This segment will provide insights into"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:25",
                    "end": "00:00:29",
                    "text": "the evolution of information systems"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:27",
                    "end": "00:00:32",
                    "text": "audit within IND and its expanding"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:29",
                    "end": "00:00:34",
                    "text": "scope. The institutional safeguards that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:32",
                    "end": "00:00:36",
                    "text": "underpin digital transformation. The"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:34",
                    "end": "00:00:38",
                    "text": "enabling infrastructure supporting AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:36",
                    "end": "00:00:41",
                    "text": "adoption and capacity building"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:38",
                    "end": "00:00:43",
                    "text": "initiatives AI strategy frameworks and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:41",
                    "end": "00:00:46",
                    "text": "roadmap for embedding AI responsibility"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:43",
                    "end": "00:00:49",
                    "text": "within the audit process. I invite Sri K"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:46",
                    "end": "00:00:51",
                    "text": "Sujit, director office of the CH of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:49",
                    "end": "00:00:54",
                    "text": "India who will present the institutional"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:51",
                    "end": "00:00:57",
                    "text": "perspective on AI enabled public"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:54",
                    "end": "00:00:60",
                    "text": "assurance. I also invite professor Madus"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:57",
                    "end": "00:01:02",
                    "text": "I met Madras who will share insights on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:60",
                    "end": "00:01:05",
                    "text": "the enabling infrastructure supporting a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:02",
                    "end": "00:01:07",
                    "text": "adoption for sovereign LM initiative."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:05",
                    "end": "00:01:11",
                    "text": "May I request both the speakers kindly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:07",
                    "end": "00:01:11",
                    "text": "take the stage."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:20",
                    "end": "00:01:25",
                    "text": "Good afternoon distinguished guest"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:22",
                    "end": "00:01:28",
                    "text": "senior management office"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:25",
                    "end": "00:01:29",
                    "text": ">> and my"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:28",
                    "end": "00:01:31",
                    "text": "so we are here to give a technical"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:29",
                    "end": "00:01:35",
                    "text": "presentation it's kind of a precursor to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:31",
                    "end": "00:01:38",
                    "text": "the upcoming panel discussion basic"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:35",
                    "end": "00:01:42",
                    "text": ">> idea behind this presentation is to uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:38",
                    "end": "00:01:47",
                    "text": "inform you all as to what"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:42",
                    "end": "00:01:47",
                    "text": "vision and where we are heading in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:47",
                    "end": "00:01:54",
                    "text": "extensively incorporating AI and ML in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:49",
                    "end": "00:01:55",
                    "text": "our processes processes. So we'd like to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:54",
                    "end": "00:01:57",
                    "text": "tell you all on as to what are the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:55",
                    "end": "00:02:00",
                    "text": "present projects that are going on using"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:57",
                    "end": "00:02:03",
                    "text": "AML and also on the NBC projects the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:00",
                    "end": "00:02:06",
                    "text": "vision that we are planning to do and uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:03",
                    "end": "00:02:08",
                    "text": "nothing uh you know uh best like this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:06",
                    "end": "00:02:11",
                    "text": "platform has to call all the academia"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:08",
                    "end": "00:02:13",
                    "text": "industry partners other governmental"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:11",
                    "end": "00:02:16",
                    "text": "departmental officers where we can uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:13",
                    "end": "00:02:19",
                    "text": "tell give an assurance as to how uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:16",
                    "end": "00:02:21",
                    "text": "securely we deal with the data and uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:19",
                    "end": "00:02:23",
                    "text": "how we intend to use technology for the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:21",
                    "end": "00:02:26",
                    "text": "betterment. ment of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:23",
                    "end": "00:02:29",
                    "text": "this I commence my presentation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:26",
                    "end": "00:02:32",
                    "text": "our approach to ember AI. So last year"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:29",
                    "end": "00:02:34",
                    "text": "in April we had come out with an AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:32",
                    "end": "00:02:36",
                    "text": "strategy framework document. So this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:34",
                    "end": "00:02:38",
                    "text": "lays the groundwork the vision the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:36",
                    "end": "00:02:40",
                    "text": "strategy framework as to how to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:38",
                    "end": "00:02:44",
                    "text": "incorporate AI in our business process."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:40",
                    "end": "00:02:45",
                    "text": "Uh across the world there are many"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:44",
                    "end": "00:02:48",
                    "text": "supreme institutions as we call in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:45",
                    "end": "00:02:50",
                    "text": "international community"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:48",
                    "end": "00:02:52",
                    "text": "offices across the world. So very few"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:50",
                    "end": "00:02:55",
                    "text": "have got the strategy framework up"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:52",
                    "end": "00:02:59",
                    "text": "published in their respective countries."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:55",
                    "end": "00:03:02",
                    "text": "S India is one of them where we lay our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:59",
                    "end": "00:03:06",
                    "text": "uh we lay our trust on four foundational"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:02",
                    "end": "00:03:08",
                    "text": "pillars. One is using technology AI in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:06",
                    "end": "00:03:11",
                    "text": "our audit processes and other business"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:08",
                    "end": "00:03:13",
                    "text": "operations. The second would be to audit"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:11",
                    "end": "00:03:15",
                    "text": "the AI systems applications which are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:13",
                    "end": "00:03:17",
                    "text": "being used by various government"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:15",
                    "end": "00:03:19",
                    "text": "partners. The third is on the capacity"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:17",
                    "end": "00:03:22",
                    "text": "building aspect without which we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:19",
                    "end": "00:03:25",
                    "text": "wouldn't be able to do one and two and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:22",
                    "end": "00:03:29",
                    "text": "the fourth would be to uh would be to uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:25",
                    "end": "00:03:31",
                    "text": "do an extensive R&D in the area of AML"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:29",
                    "end": "00:03:34",
                    "text": "and then on to create an enabling"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:31",
                    "end": "00:03:37",
                    "text": "infrastructure to support all the other."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:34",
                    "end": "00:03:40",
                    "text": "So this foundational framework is going"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:37",
                    "end": "00:03:44",
                    "text": "to help us and guide us"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:40",
                    "end": "00:03:46",
                    "text": "our uh vision of embedding AI into our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:44",
                    "end": "00:03:48",
                    "text": "business processes."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:46",
                    "end": "00:03:50",
                    "text": "I'll be talking in detail on the tech"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:48",
                    "end": "00:03:53",
                    "text": "based audit that we presently do and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:50",
                    "end": "00:03:55",
                    "text": "then we plan to do as well. The highest"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:53",
                    "end": "00:03:57",
                    "text": "security this is going to showcase on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:55",
                    "end": "00:03:60",
                    "text": "what are the security measures that are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:57",
                    "end": "00:04:02",
                    "text": "in place by the organization of CAG in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:60",
                    "end": "00:04:04",
                    "text": "order to handle the data in a safe and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:02",
                    "end": "00:04:06",
                    "text": "secure manner. the capacity building"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:04",
                    "end": "00:04:08",
                    "text": "initiatives. We are a strong workforce"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:06",
                    "end": "00:04:11",
                    "text": "of 45,000"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:08",
                    "end": "00:04:13",
                    "text": "employees across 140 plus offices across"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:11",
                    "end": "00:04:14",
                    "text": "the country and in order to incorporate"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:13",
                    "end": "00:04:17",
                    "text": "AI into our business process. It"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:14",
                    "end": "00:04:20",
                    "text": "wouldn't be possible if not for the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:17",
                    "end": "00:04:23",
                    "text": "staff to incorporate the accumment of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:20",
                    "end": "00:04:25",
                    "text": "their"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:23",
                    "end": "00:04:28",
                    "text": "enabling infrastructure how we are going"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:25",
                    "end": "00:04:29",
                    "text": "to deal with this massive change"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:28",
                    "end": "00:04:33",
                    "text": "and on the information systems audit"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:29",
                    "end": "00:04:36",
                    "text": "where they are heading."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:33",
                    "end": "00:04:38",
                    "text": "So Sai India the CG of India has been"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:36",
                    "end": "00:04:40",
                    "text": "doing information systems audit. So"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:38",
                    "end": "00:04:43",
                    "text": "these are the various IT applications"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:40",
                    "end": "00:04:44",
                    "text": "being operated by the government in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:43",
                    "end": "00:04:46",
                    "text": "order to give an assurance on the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:44",
                    "end": "00:04:50",
                    "text": "governance control system controls and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:46",
                    "end": "00:04:52",
                    "text": "the validation controls. So this audit"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:50",
                    "end": "00:04:53",
                    "text": "is to ensure that the IT application"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:52",
                    "end": "00:04:56",
                    "text": "created and being operated by any"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:53",
                    "end": "00:04:60",
                    "text": "government department is a proper"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:56",
                    "end": "00:05:02",
                    "text": "fullest correct application which can be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:60",
                    "end": "00:05:04",
                    "text": "used by the public for effective service"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:02",
                    "end": "00:05:06",
                    "text": "delivery. We have been doing this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:04",
                    "end": "00:05:08",
                    "text": "particular audit since the 2000. So more"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:06",
                    "end": "00:05:10",
                    "text": "than 500 plus reports have been laid in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:08",
                    "end": "00:05:12",
                    "text": "the parliament very important and huge"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:10",
                    "end": "00:05:13",
                    "text": "IT applications like IT business"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:12",
                    "end": "00:05:16",
                    "text": "applications, impact business"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:13",
                    "end": "00:05:19",
                    "text": "applications, GST and IRCBC. Right now"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:16",
                    "end": "00:05:21",
                    "text": "we doing the gem application. So all"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:19",
                    "end": "00:05:23",
                    "text": "these applications have been audited and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:21",
                    "end": "00:05:26",
                    "text": "given an assurance by the compressor and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:23",
                    "end": "00:05:29",
                    "text": "auditor journals of India and we felt"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:26",
                    "end": "00:05:31",
                    "text": "the need that we must go beyond the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:29",
                    "end": "00:05:34",
                    "text": "scope of IT audit into the area of cyber"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:31",
                    "end": "00:05:37",
                    "text": "security audit as well. Given the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:34",
                    "end": "00:05:39",
                    "text": "scenario of uh lot of cyber attacks the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:37",
                    "end": "00:05:41",
                    "text": "threat actors who are acting in the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:39",
                    "end": "00:05:43",
                    "text": "cyber defense cyber"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:41",
                    "end": "00:05:45",
                    "text": ">> we felt that over and above the certain"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:43",
                    "end": "00:05:47",
                    "text": "empanled auditors were going to do an"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:45",
                    "end": "00:05:50",
                    "text": "annual audit on the identification and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:47",
                    "end": "00:05:51",
                    "text": "give an assurance on the cyber security"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:50",
                    "end": "00:05:53",
                    "text": "of an application."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:51",
                    "end": "00:05:56",
                    "text": ">> C must also be giving an independent"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:53",
                    "end": "00:05:58",
                    "text": "assurance on the cyber readiness of any"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:56",
                    "end": "00:06:00",
                    "text": "department. So this year we started to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:58",
                    "end": "00:06:03",
                    "text": "pilot cyber security audits in two major"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:00",
                    "end": "00:06:09",
                    "text": "applications. Uh as long as the report"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:03",
                    "end": "00:06:09",
                    "text": "is uh published uh we'll be going"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:49",
                    "end": "00:08:56",
                    "text": "We realize that more than enough until"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:52",
                    "end": "00:09:00",
                    "text": "now we are dealing with more than 800."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:56",
                    "end": "00:09:00",
                    "text": "We have not the unstructured"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:04",
                    "end": "00:09:08",
                    "text": "more than"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:06",
                    "end": "00:09:09",
                    "text": "so and then we have now moved towards"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:08",
                    "end": "00:09:10",
                    "text": "cloud infrastructure wherein we are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:09",
                    "end": "00:09:13",
                    "text": "going to get the data from the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:10",
                    "end": "00:09:17",
                    "text": "ministries do analytics create secure"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:13",
                    "end": "00:09:20",
                    "text": "ETL pipelines come up with insights"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:17",
                    "end": "00:09:23",
                    "text": "having said this we also have come up"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:20",
                    "end": "00:09:26",
                    "text": "with artificial intelligence it's a AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:23",
                    "end": "00:09:29",
                    "text": "enabled tool audit toolkit"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:26",
                    "end": "00:09:31",
                    "text": "wherein we have created an order toolkit"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:29",
                    "end": "00:09:33",
                    "text": "which is going to get the input as the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:31",
                    "end": "00:09:35",
                    "text": "images of the beneficiaries. Even a non"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:33",
                    "end": "00:09:36",
                    "text": "techch person will be able to use this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:35",
                    "end": "00:09:39",
                    "text": "particular order tool and the output"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:36",
                    "end": "00:09:41",
                    "text": "that is going to show me is going to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:39",
                    "end": "00:09:43",
                    "text": "show me whether there are duplicate"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:41",
                    "end": "00:09:47",
                    "text": "beneficiaries those beneficiaries etc."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:43",
                    "end": "00:09:47",
                    "text": "Now one scheme alone."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:49",
                    "end": "00:09:53",
                    "text": "So this intends to give"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:53",
                    "end": "00:10:00",
                    "text": "this particular scheme on any register"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:57",
                    "end": "00:10:00",
                    "text": "has to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:01",
                    "end": "00:10:04",
                    "text": "capture"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:05",
                    "end": "00:10:11",
                    "text": "and how to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:07",
                    "end": "00:10:11",
                    "text": "verify this."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:12",
                    "end": "00:10:17",
                    "text": "So coming out of the traditional method"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:14",
                    "end": "00:10:19",
                    "text": "of audit the big data analytics is the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:17",
                    "end": "00:10:21",
                    "text": "intend to get the entire data set from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:19",
                    "end": "00:10:26",
                    "text": "the government store it in a secure"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:21",
                    "end": "00:10:27",
                    "text": "manner and do a comprehensive analysis."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:26",
                    "end": "00:10:29",
                    "text": "The OCR and NLP have been used"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:27",
                    "end": "00:10:32",
                    "text": "extensively right now in order to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:29",
                    "end": "00:10:36",
                    "text": "automate our uh regular uh day-to-day"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:32",
                    "end": "00:10:38",
                    "text": "process. We also developing a sovereign"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:36",
                    "end": "00:10:40",
                    "text": "large language model exclusively for the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:38",
                    "end": "00:10:42",
                    "text": "purpose of audit uh which will be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:40",
                    "end": "00:10:45",
                    "text": "explained in brief by my colleague"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:42",
                    "end": "00:10:47",
                    "text": "professor here."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:45",
                    "end": "00:10:49",
                    "text": "We have also audited the procurement"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:47",
                    "end": "00:10:52",
                    "text": "data set using a machine learning"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:49",
                    "end": "00:10:54",
                    "text": "algorithm and graph analysis. Usually"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:52",
                    "end": "00:10:55",
                    "text": "the traditional way of doing an audit is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:54",
                    "end": "00:10:57",
                    "text": "to select the units based on some this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:55",
                    "end": "00:11:00",
                    "text": "parameter mostly is based on the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:57",
                    "end": "00:11:01",
                    "text": "expenditure unit has input."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:00",
                    "end": "00:11:04",
                    "text": "Right now what we do is we give the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:01",
                    "end": "00:11:06",
                    "text": "entire procurement data set do an"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:04",
                    "end": "00:11:11",
                    "text": "analysis on that. We select the units"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:06",
                    "end": "00:11:11",
                    "text": "based on the red flags raised upon the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:57",
                    "end": "00:12:01",
                    "text": "late"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:59",
                    "end": "00:12:04",
                    "text": "and then the drones and computer mission"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:01",
                    "end": "00:12:08",
                    "text": "we have started utilizing a much more"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:04",
                    "end": "00:12:10",
                    "text": "defined satellite image where we can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:08",
                    "end": "00:12:13",
                    "text": "where we can uh use the high resisting"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:10",
                    "end": "00:12:16",
                    "text": "images for audit analysis. We have done"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:13",
                    "end": "00:12:19",
                    "text": "the volumetric analysis of mining mining"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:16",
                    "end": "00:12:23",
                    "text": "releases and right now as a part of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:19",
                    "end": "00:12:25",
                    "text": "seven databases we have also uh drones"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:23",
                    "end": "00:12:26",
                    "text": "to cancel"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:25",
                    "end": "00:12:29",
                    "text": "the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:26",
                    "end": "00:12:29",
                    "text": "actually"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:48",
                    "end": "00:12:55",
                    "text": "as per the compliant"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:51",
                    "end": "00:12:55",
                    "text": "guidelines. Make sure that all"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:59",
                    "end": "00:13:03",
                    "text": "this is not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:04",
                    "end": "00:13:09",
                    "text": "infrastructure."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:07",
                    "end": "00:13:12",
                    "text": "I would say that we are intending to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:09",
                    "end": "00:13:12",
                    "text": "have data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:15",
                    "end": "00:13:18",
                    "text": "set"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:18",
                    "end": "00:13:25",
                    "text": "analysis. We also have a center for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:20",
                    "end": "00:13:25",
                    "text": "excellence on financial a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:12",
                    "end": "00:16:19",
                    "text": "It has come only from the C office and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:15",
                    "end": "00:16:22",
                    "text": "couple of other places which has been"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:19",
                    "end": "00:16:22",
                    "text": "follow"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:22",
                    "end": "00:16:27",
                    "text": "and so on and so forth."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:25",
                    "end": "00:16:30",
                    "text": "The third component, the third component"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:27",
                    "end": "00:16:34",
                    "text": "is the human capital."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:30",
                    "end": "00:16:34",
                    "text": "The capital has been"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:35",
                    "end": "00:16:39",
                    "text": "students from IIT"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:37",
                    "end": "00:16:40",
                    "text": "professionals from India, data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:39",
                    "end": "00:16:42",
                    "text": "scientists from India, artificial"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:40",
                    "end": "00:16:47",
                    "text": "intelligence experts from India. Thank"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:42",
                    "end": "00:16:47",
                    "text": "you. So when we talk about human capital"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:48",
                    "end": "00:16:55",
                    "text": "about current 800 officers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:52",
                    "end": "00:16:57",
                    "text": "7800 officers of the CNG are taking up a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:55",
                    "end": "00:17:02",
                    "text": "course in data science machine learning"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:57",
                    "end": "00:17:03",
                    "text": "artificial intelligence cyber security"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:02",
                    "end": "00:17:06",
                    "text": "and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:03",
                    "end": "00:17:08",
                    "text": "the idea is to train about 5,000"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:06",
                    "end": "00:17:10",
                    "text": "officers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:08",
                    "end": "00:17:14",
                    "text": "the first model whatever we are talking"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:10",
                    "end": "00:17:17",
                    "text": "of sovereign mall of CAD was built in I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:14",
                    "end": "00:17:19",
                    "text": "met and I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:17",
                    "end": "00:17:23",
                    "text": "the second model"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:19",
                    "end": "00:17:24",
                    "text": ">> we are focusing on will be built using"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:23",
                    "end": "00:17:27",
                    "text": "50%"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:24",
                    "end": "00:17:31",
                    "text": "students who have graduated the officers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:27",
                    "end": "00:17:33",
                    "text": "who have graduated from CAG that means"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:31",
                    "end": "00:17:36",
                    "text": "the model is going to be built by"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:33",
                    "end": "00:17:40",
                    "text": "officers of CAG 50% of then this is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:36",
                    "end": "00:17:42",
                    "text": "within 3 years this will become 90%"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:40",
                    "end": "00:17:45",
                    "text": "why"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:42",
                    "end": "00:17:48",
                    "text": "because Coding is not technical."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:45",
                    "end": "00:17:50",
                    "text": "Machines generate code. So the question"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:48",
                    "end": "00:17:52",
                    "text": "is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:50",
                    "end": "00:17:53",
                    "text": "artificial intelligence is what is the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:52",
                    "end": "00:17:55",
                    "text": "problem that you're trying to solve?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:53",
                    "end": "00:17:57",
                    "text": "That becomes very important. What is it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:55",
                    "end": "00:17:59",
                    "text": "you're trying to solve? And I'm not an"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:57",
                    "end": "00:18:01",
                    "text": "auditor."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:59",
                    "end": "00:18:03",
                    "text": "I'm a computer scientist. So I cannot"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:01",
                    "end": "00:18:07",
                    "text": "understand your problem. Today you tell"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:03",
                    "end": "00:18:09",
                    "text": "me the problem and I try to solve it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:07",
                    "end": "00:18:11",
                    "text": ">> This is the data trying to solve the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:09",
                    "end": "00:18:14",
                    "text": "problem. You have the data. You have the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:11",
                    "end": "00:18:16",
                    "text": "tribal knowledge. You have the tribal"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:14",
                    "end": "00:18:19",
                    "text": "knowledge."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:16",
                    "end": "00:18:21",
                    "text": ">> The third is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:19",
                    "end": "00:18:23",
                    "text": "what is the metric? How do you say the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:21",
                    "end": "00:18:26",
                    "text": "model is true? How is how is the model"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:23",
                    "end": "00:18:30",
                    "text": "helping? I cannot give some accuracy"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:26",
                    "end": "00:18:32",
                    "text": "metric or I cannot do some metric."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:30",
                    "end": "00:18:35",
                    "text": "These three components"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:32",
                    "end": "00:18:38",
                    "text": "become extremely critical and we are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:35",
                    "end": "00:18:42",
                    "text": "trying to see that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:38",
                    "end": "00:18:45",
                    "text": "maybe in three years 90% of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:42",
                    "end": "00:18:45",
                    "text": "applications"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:49",
                    "end": "00:18:56",
                    "text": "that knowledge is important right and we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:52",
                    "end": "00:18:57",
                    "text": "talking about right today"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:56",
                    "end": "00:19:00",
                    "text": "what are the challenges because"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:57",
                    "end": "00:19:03",
                    "text": "challenge is the volume of data that we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:00",
                    "end": "00:19:04",
                    "text": "And that is the biggest advantage and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:03",
                    "end": "00:19:05",
                    "text": "the disadvantage. Every country is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:04",
                    "end": "00:19:08",
                    "text": "trying to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:05",
                    "end": "00:19:10",
                    "text": ">> get our knowledge to get the data that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:08",
                    "end": "00:19:12",
                    "text": "we have which is which is in a bounded"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:10",
                    "end": "00:19:17",
                    "text": "box. If you are able to build it off the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:12",
                    "end": "00:19:19",
                    "text": "battle one right and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:17",
                    "end": "00:19:23",
                    "text": "more importantly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:19",
                    "end": "00:19:25",
                    "text": ">> when we talk about audit today how is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:23",
                    "end": "00:19:27",
                    "text": "LLM going to help me? Yes. It is not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:25",
                    "end": "00:19:29",
                    "text": "about"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:27",
                    "end": "00:19:32",
                    "text": "magic which is going to solve that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:29",
                    "end": "00:19:32",
                    "text": "problem."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:37",
                    "end": "00:19:45",
                    "text": "One is the exploitation that I look at"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:41",
                    "end": "00:19:48",
                    "text": "previous try to understand what is it I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:45",
                    "end": "00:19:51",
                    "text": "should"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:48",
                    "end": "00:19:57",
                    "text": "then take go for but more importantly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:51",
                    "end": "00:19:57",
                    "text": "whats are capable is exploration. Today"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:02",
                    "end": "00:20:08",
                    "text": "I do not"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:04",
                    "end": "00:20:11",
                    "text": "cover it and that is where I feel"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:08",
                    "end": "00:20:16",
                    "text": "can really help other than the hype of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:11",
                    "end": "00:20:16",
                    "text": "this is what we trying to do Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:42",
                    "end": "00:20:47",
                    "text": "We have been acknledging the fact that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:49",
                    "end": "00:20:56",
                    "text": "and so we have an entire depend on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:53",
                    "end": "00:20:58",
                    "text": "technology process and in order to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:56",
                    "end": "00:21:00",
                    "text": "successfully do it we will not be able"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:58",
                    "end": "00:21:02",
                    "text": "to do it alone and so this platform"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:00",
                    "end": "00:21:05",
                    "text": "through this platform we request"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:02",
                    "end": "00:21:06",
                    "text": "extensive cooperation from the academia"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:05",
                    "end": "00:21:11",
                    "text": "industry partners and other government"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:06",
                    "end": "00:21:13",
                    "text": "officers to successfully uh uh take CA"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:11",
                    "end": "00:21:17",
                    "text": "uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:13",
                    "end": "00:21:17",
                    "text": "in terms of good governance."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:18",
                    "end": "00:21:22",
                    "text": "In order to know what we are right now,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:42",
                    "end": "00:21:46",
                    "text": "And to know what we are doing. Hence for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:45",
                    "end": "00:21:50",
                    "text": "you may please like follow us. This is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:46",
                    "end": "00:21:50",
                    "text": "our social media."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:54",
                    "end": "00:21:57",
                    "text": "Thank you all for coming."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:04",
                    "end": "00:22:08",
                    "text": "Thank you Mr. and Professor Madan for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:06",
                    "end": "00:22:10",
                    "text": "those insightful presentations. We now"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:08",
                    "end": "00:22:12",
                    "text": "move to the panel discussion on use of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:10",
                    "end": "00:22:14",
                    "text": "AI in public audit. While the technical"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:12",
                    "end": "00:22:16",
                    "text": "presentation outline the institutional"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:14",
                    "end": "00:22:19",
                    "text": "road map, this discussion will examine"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:16",
                    "end": "00:22:21",
                    "text": "the broader dimensions of a adoption and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:19",
                    "end": "00:22:22",
                    "text": "public audit including data readiness,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:21",
                    "end": "00:22:24",
                    "text": "government safeguards, infrastructure"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:22",
                    "end": "00:22:26",
                    "text": "requirements, skill transformation and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:24",
                    "end": "00:22:27",
                    "text": "responsible scale. I'm pleased to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:26",
                    "end": "00:22:30",
                    "text": "introduce our distinguished panel"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:27",
                    "end": "00:22:34",
                    "text": "members. Dr. Sanjit, Chief AI and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:30",
                    "end": "00:22:34",
                    "text": "digital officer, Vadwani Foundation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:37",
                    "end": "00:22:42",
                    "text": "Singh, Principal Director, Officer of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:38",
                    "end": "00:22:42",
                    "text": "the Cri"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:43",
                    "end": "00:22:48",
                    "text": "Srn Chakraati, Senior Vice President,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:46",
                    "end": "00:22:51",
                    "text": "National Institute of Smart Governance"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:48",
                    "end": "00:22:53",
                    "text": "and Professor Aupta, IIT. The discussion"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:51",
                    "end": "00:22:54",
                    "text": "will be moderated by Miss Priyanka"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:53",
                    "end": "00:22:57",
                    "text": "Sharma, partner public finance and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:54",
                    "end": "00:23:02",
                    "text": "digital government, KPMG. May I request"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:57",
                    "end": "00:23:02",
                    "text": "the panel to kindly take their seats and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:03",
                    "end": "00:23:07",
                    "text": "start"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:04",
                    "end": "00:23:07",
                    "text": "photograph"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:34",
                    "end": "00:23:37",
                    "text": "Master"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:57",
                    "end": "00:24:00",
                    "text": "Everyone"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:02",
                    "end": "00:24:07",
                    "text": "in public"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:04",
                    "end": "00:24:09",
                    "text": "the two conversations before this panel"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:07",
                    "end": "00:24:12",
                    "text": "have actually set the tone for this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:09",
                    "end": "00:24:15",
                    "text": "panel discussion. We have heard about"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:12",
                    "end": "00:24:19",
                    "text": "the key initiatives that CG is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:15",
                    "end": "00:24:21",
                    "text": "undertaking in the areas of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:19",
                    "end": "00:24:23",
                    "text": "you know AI in their own internal"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:21",
                    "end": "00:24:25",
                    "text": "functions. I think through this panel"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:23",
                    "end": "00:24:28",
                    "text": "discussion we will try and get a more"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:25",
                    "end": "00:24:31",
                    "text": "global perspective on how AI could be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:28",
                    "end": "00:24:34",
                    "text": "more effectively uh integrated into the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:31",
                    "end": "00:24:37",
                    "text": "functions of CG. So just very quickly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:34",
                    "end": "00:24:39",
                    "text": "because we have very little time left"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:37",
                    "end": "00:24:41",
                    "text": "right into the panel discussion I'll"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:39",
                    "end": "00:24:45",
                    "text": "quickly introduce the panelist starting"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:41",
                    "end": "00:24:48",
                    "text": "with uh Dr. Sanjie Kumar chief AI and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:45",
                    "end": "00:24:51",
                    "text": "digital officer at Wani Foundation. Dr."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:48",
                    "end": "00:24:53",
                    "text": "Kumar is a technologist researcher and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:51",
                    "end": "00:24:56",
                    "text": "entrepreneur with extensive experience"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:53",
                    "end": "00:25:01",
                    "text": "in building gen AI and analytics"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:56",
                    "end": "00:25:05",
                    "text": "solutions. We have Sri Navin Singhi who"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:01",
                    "end": "00:25:09",
                    "text": "is an officer of CG. He's a principal"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:05",
                    "end": "00:25:10",
                    "text": "director in commercial and CAG. He has"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:09",
                    "end": "00:25:14",
                    "text": "experience in audit across many"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:10",
                    "end": "00:25:17",
                    "text": "organizations and also global ones like"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:14",
                    "end": "00:25:19",
                    "text": "UN board auditors at New York also"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:17",
                    "end": "00:25:23",
                    "text": "coordinating audit of United Nations,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:19",
                    "end": "00:25:26",
                    "text": "UNICEF and other such organizations."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:23",
                    "end": "00:25:29",
                    "text": "We have professor Aam Gupta. He's an"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:26",
                    "end": "00:25:32",
                    "text": "associate professor at IIT Delhi and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:29",
                    "end": "00:25:35",
                    "text": "he's a key researcher in the areas of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:32",
                    "end": "00:25:37",
                    "text": "technology and you know interfacing"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:35",
                    "end": "00:25:40",
                    "text": "technology interfacing the you know"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:37",
                    "end": "00:25:42",
                    "text": "societal problems. We have Mr. Shrinat"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:40",
                    "end": "00:25:44",
                    "text": "Chakraati who is working as a senior VP"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:42",
                    "end": "00:25:46",
                    "text": "with national institute of smart"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:44",
                    "end": "00:25:50",
                    "text": "government with more than 30 years of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:46",
                    "end": "00:25:52",
                    "text": "experience in technology advisory. So uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:50",
                    "end": "00:25:55",
                    "text": "let's start the panel discussion. Dr."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:52",
                    "end": "00:25:59",
                    "text": "Sanj my question to you is that we've"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:55",
                    "end": "00:26:02",
                    "text": "seen in the earlier presentation all the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:59",
                    "end": "00:26:05",
                    "text": "initiatives that CG started to undertake"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:02",
                    "end": "00:26:07",
                    "text": "and since you are also now engaging"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:05",
                    "end": "00:26:11",
                    "text": "extensively with CAG I wanted to get"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:07",
                    "end": "00:26:14",
                    "text": "your views on their uh you know overall"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:11",
                    "end": "00:26:16",
                    "text": "preparedness with respect to you know"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:14",
                    "end": "00:26:18",
                    "text": "widespread institutional adoption"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:16",
                    "end": "00:26:20",
                    "text": "especially in the context of you know"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:18",
                    "end": "00:26:23",
                    "text": "data quality and the completeness"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:20",
                    "end": "00:26:25",
                    "text": "especially as we know that CG interfaces"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:23",
                    "end": "00:26:29",
                    "text": "with other departments and ministries."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:25",
                    "end": "00:26:29",
                    "text": "So what's your view around that?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:33",
                    "end": "00:26:43",
                    "text": ">> Thank you. So India given CA and given"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:39",
                    "end": "00:26:45",
                    "text": "how vast and varied India is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:43",
                    "end": "00:26:51",
                    "text": "the problem of data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:45",
                    "end": "00:26:55",
                    "text": ">> for CG systems is in fact a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:51",
                    "end": "00:26:60",
                    "text": "a very uh important big data problem"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:55",
                    "end": "00:27:05",
                    "text": "because both in terms of volume, vality,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:60",
                    "end": "00:27:09",
                    "text": "veracity and uh uh velocity the three"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:05",
                    "end": "00:27:10",
                    "text": "the three pillars on which big data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:09",
                    "end": "00:27:15",
                    "text": "lies."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:10",
                    "end": "00:27:18",
                    "text": "CG has to you know it is probably at the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:15",
                    "end": "00:27:21",
                    "text": "most complex"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:18",
                    "end": "00:27:22",
                    "text": "ladder of being a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:21",
                    "end": "00:27:26",
                    "text": "problems."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:22",
                    "end": "00:27:29",
                    "text": "So uh apart from what Sujit said about"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:26",
                    "end": "00:27:31",
                    "text": "structured data, there is so much of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:29",
                    "end": "00:27:36",
                    "text": "unstructured information which is uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:31",
                    "end": "00:27:41",
                    "text": "going to be processed uh whether it be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:36",
                    "end": "00:27:44",
                    "text": "uh you know vouchers and PDF files and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:41",
                    "end": "00:27:47",
                    "text": "images and given the language where you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:44",
                    "end": "00:27:50",
                    "text": "know differences between states"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:47",
                    "end": "00:27:51",
                    "text": ">> etc. the the problem is pretty complex"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:50",
                    "end": "00:27:56",
                    "text": "and it's increasingly getting more"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:51",
                    "end": "00:27:59",
                    "text": "complex because maybe new types of data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:56",
                    "end": "00:28:05",
                    "text": "will get added like speech and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:59",
                    "end": "00:28:07",
                    "text": "and GIS and uh videos. So it's going to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:05",
                    "end": "00:28:11",
                    "text": "get more and more complex."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:07",
                    "end": "00:28:14",
                    "text": "Now the idea of creating a data lake and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:11",
                    "end": "00:28:16",
                    "text": "uh then uh getting information out of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:14",
                    "end": "00:28:20",
                    "text": "unstructured information which comes"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:16",
                    "end": "00:28:23",
                    "text": "into the the data lake uh is uh is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:20",
                    "end": "00:28:25",
                    "text": "important and we'll have to gain"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:23",
                    "end": "00:28:27",
                    "text": "structured information from all the all"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:25",
                    "end": "00:28:30",
                    "text": "these large volumes of unstructured"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:27",
                    "end": "00:28:34",
                    "text": "information and create the relevant data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:30",
                    "end": "00:28:37",
                    "text": "marks for uh our to support our AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:34",
                    "end": "00:28:39",
                    "text": "initiatives. Now the good thing is that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:37",
                    "end": "00:28:43",
                    "text": "you know there is a lot of government"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:39",
                    "end": "00:28:47",
                    "text": "initiatives for example GSTN the centers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:43",
                    "end": "00:28:50",
                    "text": "of excellence and they have in fact uh"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:47",
                    "end": "00:28:51",
                    "text": "done a lot of work in this to actually"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:50",
                    "end": "00:28:54",
                    "text": "process this kind of this kind of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:51",
                    "end": "00:28:57",
                    "text": "information. So I don't think it is a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:54",
                    "end": "00:29:00",
                    "text": "very difficult problem. It might be a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:57",
                    "end": "00:29:03",
                    "text": "problem of reusing a lot of components"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:00",
                    "end": "00:29:06",
                    "text": "which are already tested in different"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:03",
                    "end": "00:29:08",
                    "text": "parts of the different projects. Uh but"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:06",
                    "end": "00:29:13",
                    "text": "it should be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:08",
                    "end": "00:29:13",
                    "text": "doable. I don't think"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:13",
                    "end": "00:29:18",
                    "text": ">> Thank you so much. So what we heard is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:15",
                    "end": "00:29:18",
                    "text": "Okay."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6992e2704c5e8f0974c87375",
    "projectId": "419865",
    "projectName": "India AI Impact Buildathon 2026 | AI for Social Good & Digital Fraud Prevention",
    "projectDescription": "India AI Impact Buildathon 2026 | AI for Social Good & Digital Fraud Prevention",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-16T09:25:00.722Z",
    "projectUpdatedDate": "2026-02-16T09:25:00.722Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-16T09:25:00.722Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO_LINK",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames"
        },
        {
            "col1": "P1",
            "col2": "https://www.youtube.com/watch?v=kxziGGmq7bY",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEOLINK",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:31",
                    "end": "00:00:35",
                    "text": "distinguished dignitaries who don't just"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:33",
                    "end": "00:00:38",
                    "text": "understand the landscape of technology"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:35",
                    "end": "00:00:40",
                    "text": "but are actively shaping it. As we take"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:38",
                    "end": "00:00:42",
                    "text": "a moment to acknowledge the presence are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:40",
                    "end": "00:00:44",
                    "text": "distinguished dignitaries joining us"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:42",
                    "end": "00:00:46",
                    "text": "today. I invite you all to keep the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:44",
                    "end": "00:00:48",
                    "text": "applause going as I introduced our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:46",
                    "end": "00:00:51",
                    "text": "esteemed guest honorable guest Shria"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:48",
                    "end": "00:00:54",
                    "text": "Bishek Singh GI Additional Secretary"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:51",
                    "end": "00:00:58",
                    "text": "Ministry of Electronics and IT CEO India"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:54",
                    "end": "00:01:00",
                    "text": "AI and director general NIC and and our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:58",
                    "end": "00:01:03",
                    "text": "distinguished guest Mr. Dvashish Mishra"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:00",
                    "end": "00:01:07",
                    "text": "Chief General Manager Delhi Circle SBI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:03",
                    "end": "00:01:11",
                    "text": "Dr. Buddhachar Chief Coordinator Officer"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:07",
                    "end": "00:01:13",
                    "text": "AICTTE ministry of education Mr. Sepani,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:11",
                    "end": "00:01:16",
                    "text": "senior vice president and head of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:13",
                    "end": "00:01:18",
                    "text": "strategic initiative at CLTech."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:16",
                    "end": "00:01:20",
                    "text": ">> Joining them, Mr. Ankit Kakar, senior"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:18",
                    "end": "00:01:22",
                    "text": "director, technical services"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:20",
                    "end": "00:01:25",
                    "text": "engineering, MongoDB."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:22",
                    "end": "00:01:28",
                    "text": "Mr. Nirajwalia, India's top tech content"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:25",
                    "end": "00:01:31",
                    "text": "creator with over 3.3 million followers."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:28",
                    "end": "00:01:34",
                    "text": "Mr. Abhinav Go, senior vice president"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:31",
                    "end": "00:01:36",
                    "text": "strategy at Ccl. And Mr. Spalamorgan,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:34",
                    "end": "00:01:38",
                    "text": "co-founder GUI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:36",
                    "end": "00:01:39",
                    "text": ">> A huge round of applause for our valued"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:38",
                    "end": "00:01:42",
                    "text": "guest. Thank you for sparing your"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:39",
                    "end": "00:01:45",
                    "text": "valuable time with our AI builders."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:42",
                    "end": "00:01:47",
                    "text": "Now we officially kick off. Let's look"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:45",
                    "end": "00:01:51",
                    "text": "at the agenda for the day."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:47",
                    "end": "00:01:55",
                    "text": ">> So we have broken down today into four"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:51",
                    "end": "00:01:58",
                    "text": "power pack stages. Enhance, evaluate,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:55",
                    "end": "00:02:01",
                    "text": "pitch, and celebrate winners. Stage one"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:58",
                    "end": "00:02:03",
                    "text": "is enhance. You will have a 30inut"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:01",
                    "end": "00:02:06",
                    "text": "sprint to fine-tune your solutions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:03",
                    "end": "00:02:10",
                    "text": "against specific test cases. Stage two"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:06",
                    "end": "00:02:12",
                    "text": "is evaluate where the scoreboard will do"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:10",
                    "end": "00:02:15",
                    "text": "the talking based on live performances."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:12",
                    "end": "00:02:19",
                    "text": "Top six teams will be selected to move"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:15",
                    "end": "00:02:22",
                    "text": "forward. Stage three is pitch where top"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:19",
                    "end": "00:02:25",
                    "text": "six teams will present the solution on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:22",
                    "end": "00:02:28",
                    "text": "this national stage and the and answer"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:25",
                    "end": "00:02:30",
                    "text": "the jury's tough questions. Remember"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:28",
                    "end": "00:02:32",
                    "text": "this is the most exciting and most"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:30",
                    "end": "00:02:35",
                    "text": "impactful stage where ideas are tested"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:32",
                    "end": "00:02:38",
                    "text": "for clarity, confidence and real world"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:35",
                    "end": "00:02:41",
                    "text": "impact. And finally we'll jump to stage"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:38",
                    "end": "00:02:43",
                    "text": "four that is celebrate. We will crown"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:41",
                    "end": "00:02:46",
                    "text": "our champions and celebrate them on this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:43",
                    "end": "00:02:48",
                    "text": "national stage with our esteemed jury."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:46",
                    "end": "00:02:50",
                    "text": ">> It is going to be fast and it is going"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:48",
                    "end": "00:02:52",
                    "text": "to be intense. But first to set the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:50",
                    "end": "00:02:54",
                    "text": "context on why we are here and the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:52",
                    "end": "00:02:58",
                    "text": "vision behind this buildathon, please"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:54",
                    "end": "00:03:01",
                    "text": "welcome Mr. SP Balaman co-founder Gurui"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:58",
                    "end": "00:03:01",
                    "text": "on the stage."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:26",
                    "end": "00:03:30",
                    "text": "Good afternoon everyone."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:28",
                    "end": "00:03:33",
                    "text": "It gives me an immense pleasure to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:30",
                    "end": "00:03:36",
                    "text": "welcome you all for the India AI impact"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:33",
                    "end": "00:03:39",
                    "text": "buildathon grand finale at this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:36",
                    "end": "00:03:41",
                    "text": "prestigious Barat Mandabam. This"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:39",
                    "end": "00:03:44",
                    "text": "prestigious Barat Madam has witnessed"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:41",
                    "end": "00:03:48",
                    "text": "lot of transformative moments of our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:44",
                    "end": "00:03:50",
                    "text": "nation and today I'm sure what it is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:48",
                    "end": "00:03:52",
                    "text": "going to witness again is one of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:50",
                    "end": "00:03:56",
                    "text": "nation's historic moment that will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:52",
                    "end": "00:03:59",
                    "text": "kindle the a spirit of the youth."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:56",
                    "end": "00:04:03",
                    "text": "Having said that any venue is enriched"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:59",
                    "end": "00:04:07",
                    "text": "by the people in it. Okay. So today we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:03",
                    "end": "00:04:10",
                    "text": "have our nation's policy builders, our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:07",
                    "end": "00:04:13",
                    "text": "technologist, industry experts and the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:10",
                    "end": "00:04:15",
                    "text": "youth who is aspirational to create the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:13",
                    "end": "00:04:18",
                    "text": "to create the a revolution for our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:15",
                    "end": "00:04:21",
                    "text": "country. So it's a great pleasure for me"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:18",
                    "end": "00:04:24",
                    "text": "to in invite you all to be part of this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:21",
                    "end": "00:04:26",
                    "text": "mission of our nation and my heartfelt"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:24",
                    "end": "00:04:30",
                    "text": "gratitude to each and every one of you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:26",
                    "end": "00:04:30",
                    "text": "for being here."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:30",
                    "end": "00:04:35",
                    "text": "Having said this, I'll just do a quick"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:32",
                    "end": "00:04:37",
                    "text": "recap of what has gone in the last four"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:35",
                    "end": "00:04:41",
                    "text": "months of efforts to make this day"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:37",
                    "end": "00:04:44",
                    "text": "happen. Okay. So when this hackathon was"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:41",
                    "end": "00:04:47",
                    "text": "rolled out across the country to invite"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:44",
                    "end": "00:04:48",
                    "text": "people to come part and join this AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:47",
                    "end": "00:04:51",
                    "text": "revolution,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:48",
                    "end": "00:04:54",
                    "text": "more than 40,000 plus people registered"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:51",
                    "end": "00:04:56",
                    "text": "for it with a common goal of how they"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:54",
                    "end": "00:04:59",
                    "text": "can come along and solve the problem of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:56",
                    "end": "00:05:01",
                    "text": "our nation. Okay. From this 40,000"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:59",
                    "end": "00:05:03",
                    "text": "40,000 people when we see the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:01",
                    "end": "00:05:06",
                    "text": "demographic of the register registration"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:03",
                    "end": "00:05:09",
                    "text": "it is across the places of our country"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:06",
                    "end": "00:05:12",
                    "text": "more than 600 plus cities were covered"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:09",
                    "end": "00:05:14",
                    "text": "by the registered participants and if"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:12",
                    "end": "00:05:16",
                    "text": "you look at the diversification of the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:14",
                    "end": "00:05:18",
                    "text": "people who registered to solve this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:16",
                    "end": "00:05:20",
                    "text": "problem. There were students there were"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:18",
                    "end": "00:05:23",
                    "text": "graduates and they were working"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:20",
                    "end": "00:05:24",
                    "text": "professionals which showed how much"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:23",
                    "end": "00:05:27",
                    "text": "people are willing to come and be part"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:24",
                    "end": "00:05:30",
                    "text": "of this revolution. But interesting part"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:27",
                    "end": "00:05:32",
                    "text": "is there were people from non- tech"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:30",
                    "end": "00:05:34",
                    "text": "background as well who joined this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:32",
                    "end": "00:05:36",
                    "text": "movement and wanted to give their"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:34",
                    "end": "00:05:38",
                    "text": "contribution"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:36",
                    "end": "00:05:40",
                    "text": "be an inclusive platform. How could we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:38",
                    "end": "00:05:43",
                    "text": "exclude them? So we've opened up our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:40",
                    "end": "00:05:46",
                    "text": "learning contents for people to start"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:43",
                    "end": "00:05:48",
                    "text": "learning AI from the basics even if they"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:46",
                    "end": "00:05:50",
                    "text": "don't have AI knowledge right from the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:48",
                    "end": "00:05:52",
                    "text": "basics and how they can build a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:50",
                    "end": "00:05:55",
                    "text": "solutions which can actually create an"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:52",
                    "end": "00:05:58",
                    "text": "impact. Okay. So that is how these"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:55",
                    "end": "00:06:00",
                    "text": "40,000 people are enriched to learn and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:58",
                    "end": "00:06:02",
                    "text": "then build and then compete in this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:00",
                    "end": "00:06:06",
                    "text": "contest. So there's no one is left"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:02",
                    "end": "00:06:10",
                    "text": "behind. Okay."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:06",
                    "end": "00:06:11",
                    "text": "Then once you educate them and once you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:10",
                    "end": "00:06:14",
                    "text": "empower them with the skills right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:11",
                    "end": "00:06:16",
                    "text": "that's when the real challenge comes how"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:14",
                    "end": "00:06:18",
                    "text": "to pick the top performers from them."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:16",
                    "end": "00:06:20",
                    "text": "Okay. Imagine 40,000 people are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:18",
                    "end": "00:06:23",
                    "text": "competing against a challenge."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:20",
                    "end": "00:06:26",
                    "text": "That's when Guey's tech platform came"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:23",
                    "end": "00:06:28",
                    "text": "into the picture. Okay. So, which is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:26",
                    "end": "00:06:31",
                    "text": "actually the backbone of Guey powered by"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:28",
                    "end": "00:06:35",
                    "text": "technologies help us to shortlist the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:31",
                    "end": "00:06:38",
                    "text": "top winner or top 200 teams from this 47"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:35",
                    "end": "00:06:40",
                    "text": "part 42,000 participants. Imagine how"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:38",
                    "end": "00:06:42",
                    "text": "how humongous the task is, right? But"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:40",
                    "end": "00:06:44",
                    "text": "the platform made it in such a way that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:42",
                    "end": "00:06:48",
                    "text": "executed their code base against the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:44",
                    "end": "00:06:52",
                    "text": "test cases and then filtered the top 200"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:48",
                    "end": "00:06:54",
                    "text": "teams which comprises of 850 people from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:52",
                    "end": "00:06:57",
                    "text": "this 40,000"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:54",
                    "end": "00:07:01",
                    "text": "and these 850 850 people are just"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:57",
                    "end": "00:07:04",
                    "text": "sitting behind all guys. You are among"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:01",
                    "end": "00:07:06",
                    "text": "the top two percentage of the finest AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:04",
                    "end": "00:07:08",
                    "text": "builders of this country. Imagine this."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:06",
                    "end": "00:07:12",
                    "text": "Wow."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:08",
                    "end": "00:07:15",
                    "text": "Right. So once you have all this you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:12",
                    "end": "00:07:18",
                    "text": "have made this all I spoke about is what"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:15",
                    "end": "00:07:21",
                    "text": "we have done but why this builder"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:18",
                    "end": "00:07:25",
                    "text": "matters right? We are here for a common"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:21",
                    "end": "00:07:27",
                    "text": "cause of solving a problem of a nation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:25",
                    "end": "00:07:30",
                    "text": "together with AI out the tools. But"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:27",
                    "end": "00:07:33",
                    "text": "imagine today we are living in a world"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:30",
                    "end": "00:07:36",
                    "text": "where every nation is trying to compete"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:33",
                    "end": "00:07:38",
                    "text": "and lead the AI race. Every day lot of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:36",
                    "end": "00:07:39",
                    "text": "LLMs are getting introduced. You are int"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:38",
                    "end": "00:07:45",
                    "text": "every common man is introduced to lot of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:39",
                    "end": "00:07:47",
                    "text": "AI tools. Okay. But imagine whoever"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:45",
                    "end": "00:07:49",
                    "text": "conquers this AI race will lead the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:47",
                    "end": "00:07:53",
                    "text": "global economy"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:49",
                    "end": "00:07:56",
                    "text": "and we have the top leaders or top"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:53",
                    "end": "00:07:59",
                    "text": "forerunners of our country who are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:56",
                    "end": "00:08:01",
                    "text": "shaping this a revolution along with us"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:59",
                    "end": "00:08:04",
                    "text": "who is running this mission for India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:01",
                    "end": "00:08:07",
                    "text": "But this responsibility just doesn't lie"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:04",
                    "end": "00:08:10",
                    "text": "with them. It's on the shoulders of each"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:07",
                    "end": "00:08:15",
                    "text": "and every one of us over here right as a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:10",
                    "end": "00:08:17",
                    "text": "as a Indian citizen. Every one of us are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:15",
                    "end": "00:08:20",
                    "text": "empowered to be part of this revolution"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:17",
                    "end": "00:08:23",
                    "text": "and to step forward with our efforts to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:20",
                    "end": "00:08:25",
                    "text": "see how our country can move forward and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:23",
                    "end": "00:08:28",
                    "text": "that is the one and only intention or"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:25",
                    "end": "00:08:31",
                    "text": "outcome expected of this buildathon,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:28",
                    "end": "00:08:34",
                    "text": "right? To lead this world with our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:31",
                    "end": "00:08:37",
                    "text": "skills economy, right? Because if you if"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:34",
                    "end": "00:08:39",
                    "text": "you make sure that the skills are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:37",
                    "end": "00:08:41",
                    "text": "enriching the people of our country then"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:39",
                    "end": "00:08:44",
                    "text": "it will have a direct implication on the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:41",
                    "end": "00:08:46",
                    "text": "jobs on the solutions on the products"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:44",
                    "end": "00:08:50",
                    "text": "and make our country the global leader"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:46",
                    "end": "00:08:53",
                    "text": "in the upcoming world. Right? So that is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:50",
                    "end": "00:08:56",
                    "text": "the one and only target of this and we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:53",
                    "end": "00:08:59",
                    "text": "want to have you all as the AI builders"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:56",
                    "end": "00:09:01",
                    "text": "and not just the consumers."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:59",
                    "end": "00:09:03",
                    "text": "Okay, that is a statement by us like"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:01",
                    "end": "00:09:06",
                    "text": "Indians, we wanted to be the AI builders"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:03",
                    "end": "00:09:09",
                    "text": "and not just the consumers, right? When"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:06",
                    "end": "00:09:11",
                    "text": "we wanted to build that kind of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:09",
                    "end": "00:09:13",
                    "text": "aspirational thing for our country, the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:11",
                    "end": "00:09:16",
                    "text": "first thing to do is democratize the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:13",
                    "end": "00:09:19",
                    "text": "learning. How could we do that? That is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:16",
                    "end": "00:09:21",
                    "text": "when this India AI impact presummit"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:19",
                    "end": "00:09:24",
                    "text": "happened, right? We joined hands with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:21",
                    "end": "00:09:28",
                    "text": "the India team of the mighty and then"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:24",
                    "end": "00:09:31",
                    "text": "went on ground to the places where even"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:28",
                    "end": "00:09:32",
                    "text": "there is a which are far to reach to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:31",
                    "end": "00:09:35",
                    "text": "make sure everyone gets that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:32",
                    "end": "00:09:39",
                    "text": "opportunity. Okay. The team traveled"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:35",
                    "end": "00:09:42",
                    "text": "from Punjab to Tamil Nadu from Gujarat"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:39",
                    "end": "00:09:44",
                    "text": "to Tripura to make sure everyone gets a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:42",
                    "end": "00:09:49",
                    "text": "fair opportunity to skill themselves and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:44",
                    "end": "00:09:49",
                    "text": "be part of this movement. More than"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:51",
                    "end": "00:09:57",
                    "text": "More than 10,000 students are directly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:54",
                    "end": "00:10:00",
                    "text": "given a hands-on workshop on the field"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:57",
                    "end": "00:10:03",
                    "text": "by industry partners like HCL Techch."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:00",
                    "end": "00:10:05",
                    "text": "Imagine industry coming together with us"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:03",
                    "end": "00:10:07",
                    "text": "and then going into the academia"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:05",
                    "end": "00:10:10",
                    "text": "institutions and enriching our future"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:07",
                    "end": "00:10:13",
                    "text": "generation. What else the best can have?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:10",
                    "end": "00:10:16",
                    "text": "And this happened in around 48 venues"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:13",
                    "end": "00:10:18",
                    "text": "over the past four months. Okay, we are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:16",
                    "end": "00:10:19",
                    "text": "saying how much effort it takes for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:18",
                    "end": "00:10:22",
                    "text": "doing it in one venue, right? This has"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:19",
                    "end": "00:10:25",
                    "text": "been done in 48 venues over the last"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:22",
                    "end": "00:10:27",
                    "text": "four months to make this happen. Okay,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:25",
                    "end": "00:10:29",
                    "text": "and this hands-on workshop made sure"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:27",
                    "end": "00:10:31",
                    "text": "that people learned the technology in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:29",
                    "end": "00:10:34",
                    "text": "their native language and then practiced"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:31",
                    "end": "00:10:36",
                    "text": "in the platform and got things done."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:34",
                    "end": "00:10:38",
                    "text": "Okay. Next is we have made them skilled"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:36",
                    "end": "00:10:41",
                    "text": "but what is the outcome? Give them real"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:38",
                    "end": "00:10:42",
                    "text": "world problem. Let them go and solve it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:41",
                    "end": "00:10:45",
                    "text": "That's when the final transformation of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:42",
                    "end": "00:10:47",
                    "text": "this India AI buildathon has been"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:45",
                    "end": "00:10:51",
                    "text": "brought into picture along with India AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:47",
                    "end": "00:10:54",
                    "text": "team and HCL. So what's happening now is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:51",
                    "end": "00:10:58",
                    "text": "today as we speak right we are creating"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:54",
                    "end": "00:11:01",
                    "text": "the or we are nurturing the founders and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:58",
                    "end": "00:11:02",
                    "text": "the builders for the AI movement. What"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:01",
                    "end": "00:11:05",
                    "text": "we are going to do today as you finish"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:02",
                    "end": "00:11:08",
                    "text": "this event right we are just sowed the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:05",
                    "end": "00:11:10",
                    "text": "seeds of everyone taking forward their"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:08",
                    "end": "00:11:12",
                    "text": "own step to make sure how they can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:10",
                    "end": "00:11:13",
                    "text": "contribute to the AI revolution of India"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:12",
                    "end": "00:11:16",
                    "text": "and we will see that in the going"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:13",
                    "end": "00:11:18",
                    "text": "forward season. Okay. So having said"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:16",
                    "end": "00:11:21",
                    "text": "this I would like to take an opportunity"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:18",
                    "end": "00:11:23",
                    "text": "to say when Google has taken this on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:21",
                    "end": "00:11:25",
                    "text": "shoulders there are two things that we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:23",
                    "end": "00:11:27",
                    "text": "need to bet on. One is the government of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:25",
                    "end": "00:11:29",
                    "text": "India. Who else can support us other"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:27",
                    "end": "00:11:31",
                    "text": "than them? That is when the India AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:29",
                    "end": "00:11:34",
                    "text": "team and the major team came into us"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:31",
                    "end": "00:11:36",
                    "text": "came along with us who said go and make"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:34",
                    "end": "00:11:38",
                    "text": "this happen we are there to support you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:36",
                    "end": "00:11:40",
                    "text": "and from the industry side we had the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:38",
                    "end": "00:11:41",
                    "text": "HCL tech who said we'll be there to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:40",
                    "end": "00:11:44",
                    "text": "provide the resources and infrastructure"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:41",
                    "end": "00:11:46",
                    "text": "to make this happen. That's how the GUI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:44",
                    "end": "00:11:49",
                    "text": "made this transition of doing this and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:46",
                    "end": "00:11:53",
                    "text": "we are all seeing the outcome here. So"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:49",
                    "end": "00:11:54",
                    "text": "with this moment once I take back many"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:53",
                    "end": "00:11:56",
                    "text": "people would have known heard about what"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:54",
                    "end": "00:11:58",
                    "text": "GUI is or heard about the name GUI for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:56",
                    "end": "00:12:01",
                    "text": "the first time right to just give a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:58",
                    "end": "00:12:03",
                    "text": "background it's a a tech skill online"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:01",
                    "end": "00:12:04",
                    "text": "skilling platform where you can learn"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:03",
                    "end": "00:12:07",
                    "text": "technology skills in your native"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:04",
                    "end": "00:12:10",
                    "text": "language and then practice hands-on so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:07",
                    "end": "00:12:13",
                    "text": "we are part of the IIT Met incubation"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:10",
                    "end": "00:12:16",
                    "text": "and started by started in 2014 as a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:13",
                    "end": "00:12:19",
                    "text": "YouTube channel now being part of HCL"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:16",
                    "end": "00:12:22",
                    "text": "family. Okay. Contributing and proudly"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:19",
                    "end": "00:12:24",
                    "text": "supporting 4.5 million learners across"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:22",
                    "end": "00:12:26",
                    "text": "the globe by teaching them technology in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:24",
                    "end": "00:12:29",
                    "text": "20 different languages."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:26",
                    "end": "00:12:32",
                    "text": "Okay. So founded by tech nodes Arun and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:29",
                    "end": "00:12:34",
                    "text": "Shvi contributing this growth. This"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:32",
                    "end": "00:12:38",
                    "text": "proud moment makes us feel that the best"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:34",
                    "end": "00:12:40",
                    "text": "is yet to come. Okay. Having said this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:38",
                    "end": "00:12:43",
                    "text": "and uh we also proud recipient of two"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:40",
                    "end": "00:12:46",
                    "text": "Guinness World Records for skiing more"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:43",
                    "end": "00:12:48",
                    "text": "number of people in AI. And the proud"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:46",
                    "end": "00:12:51",
                    "text": "moment here is the first record called I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:48",
                    "end": "00:12:53",
                    "text": "hold along we hold it along with ACT"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:51",
                    "end": "00:12:54",
                    "text": "represented by Bddha GI who are the one"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:53",
                    "end": "00:12:57",
                    "text": "who made this possible as a thank you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:54",
                    "end": "00:12:59",
                    "text": "buddhaj"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:57",
                    "end": "00:13:01",
                    "text": "and whatever we are witnessing here is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:59",
                    "end": "00:13:02",
                    "text": "just the beginning we want this moment"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:01",
                    "end": "00:13:07",
                    "text": "to create the spark in each and every"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:02",
                    "end": "00:13:10",
                    "text": "one of you by end of this event right"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:07",
                    "end": "00:13:12",
                    "text": "that the ask from us don't just stop it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:10",
                    "end": "00:13:14",
                    "text": "this is just the start we will be there"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:12",
                    "end": "00:13:16",
                    "text": "to guide you we will be there to support"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:14",
                    "end": "00:13:19",
                    "text": "you to make this movement go across the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:16",
                    "end": "00:13:21",
                    "text": "nation and kindle the spark of how India"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:19",
                    "end": "00:13:25",
                    "text": "can lead the a revolution as along with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:21",
                    "end": "00:13:25",
                    "text": "our leaders. Thank you so much"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:30",
                    "end": "00:13:36",
                    "text": ">> stage for the next part of the ceremony."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:34",
                    "end": "00:13:38",
                    "text": ">> I would also like to invite Mr. Abinav"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:36",
                    "end": "00:13:41",
                    "text": "gosh, senior vice president strategy at"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:38",
                    "end": "00:13:43",
                    "text": "seal group to join him on stage for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:41",
                    "end": "00:13:46",
                    "text": "felicitating our honorable guest"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:43",
                    "end": "00:13:47",
                    "text": "Shiaishek Singh G."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:46",
                    "end": "00:13:49",
                    "text": "Ladies and gentlemen, now it is our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:47",
                    "end": "00:13:51",
                    "text": "privilege to formally welcome our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:49",
                    "end": "00:13:53",
                    "text": "honorable guest Shabishek Singh Gi,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:51",
                    "end": "00:13:56",
                    "text": "Additional Secretary, Ministry of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:53",
                    "end": "00:13:58",
                    "text": "Electronics, NIT, CEO, India, Director"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:56",
                    "end": "00:14:01",
                    "text": "General, NIC. Shabbishek Singh GI is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:58",
                    "end": "00:14:02",
                    "text": "presently leading the India AI mission,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:01",
                    "end": "00:14:05",
                    "text": "a flagship initiative aimed at"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:02",
                    "end": "00:14:07",
                    "text": "strengthening India's AI ecosystem,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:05",
                    "end": "00:14:09",
                    "text": "promoting responsible AI adoption, and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:07",
                    "end": "00:14:11",
                    "text": "advancing innovation at scale for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:09",
                    "end": "00:14:16",
                    "text": "national development. We are deeply"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:11",
                    "end": "00:14:16",
                    "text": "honored by his presence with us today."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:58",
                    "end": "00:15:01",
                    "text": "Sir, we would be deeply honored to hear"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:60",
                    "end": "00:15:03",
                    "text": "your thoughts and guidance for this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:01",
                    "end": "00:15:05",
                    "text": "gathering."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:03",
                    "end": "00:15:07",
                    "text": ">> Thank you. Thank you. and uh the Swiss"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:05",
                    "end": "00:15:10",
                    "text": "leaders from Guei and SC and all the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:07",
                    "end": "00:15:12",
                    "text": "jury members and all experts here there"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:10",
                    "end": "00:15:13",
                    "text": "and all the young minds who are here who"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:12",
                    "end": "00:15:15",
                    "text": "are wanting to take part in the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:13",
                    "end": "00:15:18",
                    "text": "buildathan trying to build something"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:15",
                    "end": "00:15:20",
                    "text": "which is impactful which can create a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:18",
                    "end": "00:15:23",
                    "text": "difference which can impact people at"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:20",
                    "end": "00:15:26",
                    "text": "large. So today of course we are at the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:23",
                    "end": "00:15:29",
                    "text": "AI impact summit and just looking at the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:26",
                    "end": "00:15:31",
                    "text": "kind of participation you we have the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:29",
                    "end": "00:15:33",
                    "text": "whole global leaders the whole media and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:31",
                    "end": "00:15:35",
                    "text": "all is talking about the global leaders"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:33",
                    "end": "00:15:36",
                    "text": "who are coming and the researchers who"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:35",
                    "end": "00:15:39",
                    "text": "are coming and the political leaders who"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:36",
                    "end": "00:15:41",
                    "text": "are coming but the real power and the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:39",
                    "end": "00:15:43",
                    "text": "real strength of India is in all of you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:41",
                    "end": "00:15:46",
                    "text": "who are here to take part in s such"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:43",
                    "end": "00:15:48",
                    "text": "hackathons and and the kind of response"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:46",
                    "end": "00:15:49",
                    "text": "we have got like outside as per numbers"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:48",
                    "end": "00:15:52",
                    "text": "that have been told more than 25,000"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:49",
                    "end": "00:15:55",
                    "text": "people are So that shows the interest of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:52",
                    "end": "00:15:57",
                    "text": "India in such such events and such"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:55",
                    "end": "00:15:59",
                    "text": "activities. It's not that it's a cold"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:57",
                    "end": "00:16:02",
                    "text": "play concert where people are going."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:59",
                    "end": "00:16:04",
                    "text": "People are going here to create value to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:02",
                    "end": "00:16:05",
                    "text": "show what they can how they can learn"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:04",
                    "end": "00:16:08",
                    "text": "about technology. How can they can adopt"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:05",
                    "end": "00:16:10",
                    "text": "to technology? How they can build things"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:08",
                    "end": "00:16:12",
                    "text": "that can improve lives. How AI can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:10",
                    "end": "00:16:15",
                    "text": "improve the life of a farmer. How AI can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:12",
                    "end": "00:16:18",
                    "text": "improve health care. How AI can impact"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:15",
                    "end": "00:16:20",
                    "text": "education. How we can help people with"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:18",
                    "end": "00:16:22",
                    "text": "with disabilities. So there are multiple"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:20",
                    "end": "00:16:24",
                    "text": "such possibilities which are there. In"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:22",
                    "end": "00:16:27",
                    "text": "fact, I was just at the expo this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:24",
                    "end": "00:16:29",
                    "text": "morning and uh I could see so many"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:27",
                    "end": "00:16:32",
                    "text": "solutions, so many solutions built by"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:29",
                    "end": "00:16:34",
                    "text": "not big companies but by young startups"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:32",
                    "end": "00:16:36",
                    "text": "who are very 20somes kids who are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:34",
                    "end": "00:16:38",
                    "text": "building solutions that are making a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:36",
                    "end": "00:16:41",
                    "text": "change and I'm sure many of you who are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:38",
                    "end": "00:16:43",
                    "text": "here who have the ability to do that. If"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:41",
                    "end": "00:16:46",
                    "text": "you look at any metric, if you look at"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:43",
                    "end": "00:16:48",
                    "text": "like the Corsera was conducting a event"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:46",
                    "end": "00:16:49",
                    "text": "this morning and their head was telling"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:48",
                    "end": "00:16:52",
                    "text": "me the maximum number of people who"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:49",
                    "end": "00:16:54",
                    "text": "subscribe to the AI courses come from"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:52",
                    "end": "00:16:56",
                    "text": "India. So this shows the urge of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:54",
                    "end": "00:16:58",
                    "text": "learning. This shows the urge of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:56",
                    "end": "00:17:00",
                    "text": "enterprise. This shows the capability"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:58",
                    "end": "00:17:03",
                    "text": "that we have in building things which"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:00",
                    "end": "00:17:04",
                    "text": "can really change the world. So I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:03",
                    "end": "00:17:07",
                    "text": "congratulate all of you who are taking"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:04",
                    "end": "00:17:09",
                    "text": "part in this buildathon. I'm sure a very"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:07",
                    "end": "00:17:11",
                    "text": "interesting pitch session lies ahead and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:09",
                    "end": "00:17:13",
                    "text": "we will be hearing much more about what"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:11",
                    "end": "00:17:15",
                    "text": "you want to do what you how you can do"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:13",
                    "end": "00:17:17",
                    "text": "and then it will be our responsibility"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:15",
                    "end": "00:17:20",
                    "text": "that how do we help you realize your"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:17",
                    "end": "00:17:22",
                    "text": "dreams whether it's through ensuring for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:20",
                    "end": "00:17:24",
                    "text": "startups ensuring funding through VCs"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:22",
                    "end": "00:17:28",
                    "text": "and funds whether it's through access to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:24",
                    "end": "00:17:30",
                    "text": "lowc cost compute whether it's access to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:28",
                    "end": "00:17:33",
                    "text": "data sets and more importantly access to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:30",
                    "end": "00:17:35",
                    "text": "mentoring there's so many experts who"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:33",
                    "end": "00:17:36",
                    "text": "have gone done this who have gone"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:35",
                    "end": "00:17:39",
                    "text": "through this if If they can mentor young"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:36",
                    "end": "00:17:40",
                    "text": "startups, if they mentor young minds,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:39",
                    "end": "00:17:42",
                    "text": "then we can really really create an"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:40",
                    "end": "00:17:44",
                    "text": "impact and the whole world knows what"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:42",
                    "end": "00:17:46",
                    "text": "India can do when it comes to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:44",
                    "end": "00:17:48",
                    "text": "technology. Not only the big tech"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:46",
                    "end": "00:17:50",
                    "text": "companies who are offering uh IT"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:48",
                    "end": "00:17:52",
                    "text": "services to the whole world but even our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:50",
                    "end": "00:17:55",
                    "text": "whole digital public infrastructure"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:52",
                    "end": "00:17:57",
                    "text": "story like how we used technology for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:55",
                    "end": "00:17:59",
                    "text": "transforming governance how do we build"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:57",
                    "end": "00:18:02",
                    "text": "the identity platform of Aadhaar how we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:59",
                    "end": "00:18:05",
                    "text": "do did UPIs and UPI incidentally for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:02",
                    "end": "00:18:07",
                    "text": "this summit we got RBI's approval to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:05",
                    "end": "00:18:09",
                    "text": "allow UPI for the foreign delegates so"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:07",
                    "end": "00:18:12",
                    "text": "those of you who have come from outside"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:09",
                    "end": "00:18:15",
                    "text": "you can also download the app the check"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:12",
                    "end": "00:18:17",
                    "text": "pay app check UPI app and load load your"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:15",
                    "end": "00:18:19",
                    "text": "wallet, the one world wallet with your"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:17",
                    "end": "00:18:21",
                    "text": "credit card and then go and scan the QR"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:19",
                    "end": "00:18:23",
                    "text": "codes at the various food outlets and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:21",
                    "end": "00:18:27",
                    "text": "other places and experience the magic of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:23",
                    "end": "00:18:30",
                    "text": "what it takes to pay things seamlessly."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:27",
                    "end": "00:18:32",
                    "text": "So our DPI story has a lesson for us in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:30",
                    "end": "00:18:34",
                    "text": "our AI story. How we can build the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:32",
                    "end": "00:18:36",
                    "text": "common common layers, the common"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:34",
                    "end": "00:18:38",
                    "text": "platforms of compute, data sets,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:36",
                    "end": "00:18:41",
                    "text": "foundational models and then allow"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:38",
                    "end": "00:18:43",
                    "text": "innovation at top of it. So that's the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:41",
                    "end": "00:18:44",
                    "text": "value that we are trying to create and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:43",
                    "end": "00:18:47",
                    "text": "I'm sure with the young minds here and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:44",
                    "end": "00:18:48",
                    "text": "with this initiative like buildathan and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:47",
                    "end": "00:18:50",
                    "text": "others we'll be able to really really"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:48",
                    "end": "00:18:52",
                    "text": "move forward. I'll just conclude by"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:50",
                    "end": "00:18:54",
                    "text": "making one request one appeal. I don't"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:52",
                    "end": "00:18:56",
                    "text": "know how many of you have already taken"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:54",
                    "end": "00:18:59",
                    "text": "the AI pledge. Has anyone taken the AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:56",
                    "end": "00:19:01",
                    "text": "pledge? Yes, a few people have taken. So"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:59",
                    "end": "00:19:03",
                    "text": "what we have since this morning 8:00 we"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:01",
                    "end": "00:19:05",
                    "text": "are trying to attempt a Guinness record"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:03",
                    "end": "00:19:07",
                    "text": "of maximum number of responsible AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:05",
                    "end": "00:19:09",
                    "text": "pledges in 24 hours and India will show"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:07",
                    "end": "00:19:12",
                    "text": "at the start of the summit what we can"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:09",
                    "end": "00:19:15",
                    "text": "do. So already we have crossed 50,000"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:12",
                    "end": "00:19:17",
                    "text": "plus but I'm sure if by the end of this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:15",
                    "end": "00:19:19",
                    "text": "session all of you will go to aip"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:17",
                    "end": "00:19:22",
                    "text": "pledge.india.goav.in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:19",
                    "end": "00:19:24",
                    "text": "and take that brief pledge which will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:22",
                    "end": "00:19:25",
                    "text": "also add to what we are trying to do and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:24",
                    "end": "00:19:27",
                    "text": "maybe on the 18th we'll get a Guinness"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:25",
                    "end": "00:19:30",
                    "text": "record. So look forward to getting your"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:27",
                    "end": "00:19:32",
                    "text": "contribution on this and in fact once"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:30",
                    "end": "00:19:33",
                    "text": "again congratulate the organizers for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:32",
                    "end": "00:19:35",
                    "text": "doing this buildathan and with your"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:33",
                    "end": "00:19:37",
                    "text": "energy I'm sure the outcomes that will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:35",
                    "end": "00:19:39",
                    "text": "come the solutions that you will"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:37",
                    "end": "00:19:41",
                    "text": "identify we'll be able to continue it it"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:39",
                    "end": "00:19:43",
                    "text": "should not end at this session today but"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:41",
                    "end": "00:19:45",
                    "text": "we should be able to follow up so that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:43",
                    "end": "00:19:47",
                    "text": "when next time we do a summit or we do a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:45",
                    "end": "00:19:49",
                    "text": "event they are the ones who will be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:47",
                    "end": "00:19:51",
                    "text": "showcasing their solutions to honorable"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:49",
                    "end": "00:19:53",
                    "text": "prime minister. So wish you all the best"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:51",
                    "end": "00:19:55",
                    "text": "and look forward to seeing impactful"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:53",
                    "end": "00:19:57",
                    "text": "solutions. Thank you. Thank you sir for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:55",
                    "end": "00:19:59",
                    "text": "those inspiring words. I would request"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:57",
                    "end": "00:20:01",
                    "text": "you for a minute for the flag off. Now I"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:59",
                    "end": "00:20:05",
                    "text": "would request our all dignitaries Mr."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:01",
                    "end": "00:20:08",
                    "text": "Dashish Mishra, Dr. Buddha, Mr. Sepi,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:05",
                    "end": "00:20:10",
                    "text": "Mr. Nirwalia, Mr. Ankit Kakar and her"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:08",
                    "end": "00:20:12",
                    "text": "Gurui leadership team to join Shiaishek"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:10",
                    "end": "00:20:15",
                    "text": "Singh G for the official flag of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:12",
                    "end": "00:20:15",
                    "text": "ceremony."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:08",
                    "end": "00:21:11",
                    "text": "Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:16",
                    "end": "00:21:21",
                    "text": "Thank you respected dignitaries for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:18",
                    "end": "00:21:25",
                    "text": "officially launching the grand finale of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:21",
                    "end": "00:21:27",
                    "text": "India AI impact buildathon."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:25",
                    "end": "00:21:30",
                    "text": ">> Guys, this is your moment. Give it your"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:27",
                    "end": "00:21:32",
                    "text": "all"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:30",
                    "end": "00:21:34",
                    "text": "guys. This is your moment. Give it your"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:32",
                    "end": "00:21:39",
                    "text": "all. I don't expect such low energies"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:34",
                    "end": "00:21:41",
                    "text": "among the AI builders. Right. Okay. Now"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:39",
                    "end": "00:21:44",
                    "text": "it's time to jump to into stage one that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:41",
                    "end": "00:21:47",
                    "text": "is enhance. To brief you about how to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:44",
                    "end": "00:21:50",
                    "text": "enhance your solution, we need an AI"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:47",
                    "end": "00:21:53",
                    "text": "expert. Please welcome Dr. Amit Janma,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:50",
                    "end": "00:21:57",
                    "text": "head of data science at GUI for the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:53",
                    "end": "00:21:57",
                    "text": "solution enhancement brief."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:30",
                    "end": "00:22:34",
                    "text": "Hi everyone."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:32",
                    "end": "00:22:37",
                    "text": "Hi everyone."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:34",
                    "end": "00:22:41",
                    "text": "Okay, so welcome to the final sprint of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:37",
                    "end": "00:22:43",
                    "text": "India AI impact buildathon."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:41",
                    "end": "00:22:45",
                    "text": "So congratulations to all the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:43",
                    "end": "00:22:48",
                    "text": "participants joining us right now."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:45",
                    "end": "00:22:52",
                    "text": "You are among the top 200 teams selected"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:48",
                    "end": "00:22:52",
                    "text": "for the final sprint."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:54",
                    "end": "00:22:59",
                    "text": "So for the past few weeks you have been"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:56",
                    "end": "00:23:01",
                    "text": "developing solutions for two problem"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:59",
                    "end": "00:23:03",
                    "text": "statements. You know those two problem"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:01",
                    "end": "00:23:05",
                    "text": "statements."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:03",
                    "end": "00:23:07",
                    "text": "The first problem statement in the first"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:05",
                    "end": "00:23:11",
                    "text": "problem statement you had to build a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:07",
                    "end": "00:23:15",
                    "text": "system which detects whether a voice is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:11",
                    "end": "00:23:18",
                    "text": "AI generated or not. For instance, if an"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:15",
                    "end": "00:23:20",
                    "text": "AI clone voice is mimicking your"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:18",
                    "end": "00:23:24",
                    "text": "relative and asking for money, your"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:20",
                    "end": "00:23:26",
                    "text": "system should be able to detect it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:24",
                    "end": "00:23:29",
                    "text": "Similar way, the motivation behind the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:26",
                    "end": "00:23:32",
                    "text": "second problem statement is to set a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:29",
                    "end": "00:23:35",
                    "text": "honeypot trap that engages with the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:32",
                    "end": "00:23:37",
                    "text": "scammer and intelligently extracts the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:35",
                    "end": "00:23:42",
                    "text": "information from them such as bank"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:37",
                    "end": "00:23:45",
                    "text": "accounts, UPI IDs or email addresses."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:42",
                    "end": "00:23:47",
                    "text": "This final sprint will decide the fate"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:45",
                    "end": "00:23:49",
                    "text": "of your solutions."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:47",
                    "end": "00:23:52",
                    "text": "So let me go through what exactly you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:49",
                    "end": "00:23:56",
                    "text": "have to do in the next 30 minutes."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:52",
                    "end": "00:23:58",
                    "text": "We have added hidden test cases for both"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:56",
                    "end": "00:24:01",
                    "text": "the problems that resembles real world"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:58",
                    "end": "00:24:03",
                    "text": "scenarios."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:01",
                    "end": "00:24:06",
                    "text": "Your task is to enhance your current"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:03",
                    "end": "00:24:10",
                    "text": "solutions so that it generalizes across"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:06",
                    "end": "00:24:11",
                    "text": "all the possible scenarios."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:10",
                    "end": "00:24:15",
                    "text": "To help you optimize your current"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:11",
                    "end": "00:24:18",
                    "text": "solution, we have also added sample test"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:15",
                    "end": "00:24:20",
                    "text": "cases that you can download from the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:18",
                    "end": "00:24:22",
                    "text": "hackathon platform."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:20",
                    "end": "00:24:25",
                    "text": "Now, let me quickly go through the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:22",
                    "end": "00:24:25",
                    "text": "hackathon platform."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:27",
                    "end": "00:24:31",
                    "text": "So, I mean it's the same hackathon"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:29",
                    "end": "00:24:33",
                    "text": "platform that you have been using. The"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:31",
                    "end": "00:24:37",
                    "text": "URL is hackathon.goovi.in"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:33",
                    "end": "00:24:39",
                    "text": "in and you can use your credentials that"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:37",
                    "end": "00:24:41",
                    "text": "you have been using for the past few"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:39",
                    "end": "00:24:44",
                    "text": "weeks. Once you add your credentials,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:41",
                    "end": "00:24:46",
                    "text": "you can sign in."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:44",
                    "end": "00:24:48",
                    "text": "So in the dashboard, you will see the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:46",
                    "end": "00:24:50",
                    "text": "title card of India AI impact"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:48",
                    "end": "00:24:53",
                    "text": "buildathon."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:50",
                    "end": "00:24:56",
                    "text": "You can click on it again. It's the same"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:53",
                    "end": "00:24:59",
                    "text": "UI. At the end, you will see a final"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:56",
                    "end": "00:25:03",
                    "text": "stage has been added for you. Now,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:59",
                    "end": "00:25:06",
                    "text": "please read the instructions carefully."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:03",
                    "end": "00:25:08",
                    "text": "I mean take some time and go through all"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:06",
                    "end": "00:25:11",
                    "text": "the instructions."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:08",
                    "end": "00:25:14",
                    "text": "If you see for both the problems we have"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:11",
                    "end": "00:25:15",
                    "text": "added the sample test cases. So for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:14",
                    "end": "00:25:18",
                    "text": "example for the first problem which is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:15",
                    "end": "00:25:20",
                    "text": "AI generated void detection. If you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:18",
                    "end": "00:25:22",
                    "text": "click on it you will see the sample"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:20",
                    "end": "00:25:26",
                    "text": "audio files that we have added which"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:22",
                    "end": "00:25:28",
                    "text": "generalizes across different scenarios."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:26",
                    "end": "00:25:31",
                    "text": "Same way for the second problem"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:28",
                    "end": "00:25:35",
                    "text": "statement we have added the chat history"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:31",
                    "end": "00:25:36",
                    "text": "in JSON format."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:35",
                    "end": "00:25:38",
                    "text": "Now once you have read all the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:36",
                    "end": "00:25:43",
                    "text": "instructions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:38",
                    "end": "00:25:48",
                    "text": "at the end you can submit the solution"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:43",
                    "end": "00:25:48",
                    "text": "u one second let me log in again"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:49",
                    "end": "00:25:55",
                    "text": "right so it's will once the buildathon"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:52",
                    "end": "00:25:56",
                    "text": "starts you can see this I mean it's the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:55",
                    "end": "00:25:59",
                    "text": "button is frozen but you can see the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:56",
                    "end": "00:26:03",
                    "text": "button once you click on this button you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:59",
                    "end": "00:26:08",
                    "text": "will see three fields the first field is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:03",
                    "end": "00:26:10",
                    "text": "about the uh uh it's a deployment link."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:08",
                    "end": "00:26:13",
                    "text": "So you have to add your updated"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:10",
                    "end": "00:26:15",
                    "text": "deployment link. The second field is the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:13",
                    "end": "00:26:18",
                    "text": "API key. So make sure your API key is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:15",
                    "end": "00:26:19",
                    "text": "working and it has enough credits. So"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:18",
                    "end": "00:26:22",
                    "text": "you have to add your API key in the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:19",
                    "end": "00:26:24",
                    "text": "second field. And we have added a new"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:22",
                    "end": "00:26:26",
                    "text": "field as a third field which is the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:24",
                    "end": "00:26:29",
                    "text": "GitHub repository link. So you have to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:26",
                    "end": "00:26:31",
                    "text": "add your GitHub repository link there."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:29",
                    "end": "00:26:34",
                    "text": "Once you have done that, you can submit"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:31",
                    "end": "00:26:37",
                    "text": "your solution. Now you can only submit"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:34",
                    "end": "00:26:41",
                    "text": "twice. So people again you can only"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:37",
                    "end": "00:26:43",
                    "text": "submit two times. So make sure that you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:41",
                    "end": "00:26:47",
                    "text": "are confident with your solution before"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:43",
                    "end": "00:26:49",
                    "text": "submitting. Okay. Once you submit you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:47",
                    "end": "00:26:52",
                    "text": "will see in a few minutes score will be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:49",
                    "end": "00:26:56",
                    "text": "updated at the top and with a a short"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:52",
                    "end": "00:26:56",
                    "text": "remark also."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:57",
                    "end": "00:27:02",
                    "text": "Also we have we have designed a live"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:60",
                    "end": "00:27:04",
                    "text": "leaderboard for you which we'll be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:02",
                    "end": "00:27:07",
                    "text": "showing you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:04",
                    "end": "00:27:10",
                    "text": "Uh so the leaderboard will look"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:07",
                    "end": "00:27:14",
                    "text": "something like this where we'll be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:10",
                    "end": "00:27:17",
                    "text": "continuously showing you the top uh 20"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:14",
                    "end": "00:27:19",
                    "text": "teams from student category as well as"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:17",
                    "end": "00:27:21",
                    "text": "the professional working professional"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:19",
                    "end": "00:27:24",
                    "text": "category."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:21",
                    "end": "00:27:26",
                    "text": "Now throughout the sprint we'll be"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:24",
                    "end": "00:27:28",
                    "text": "helping you. If you have any questions"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:26",
                    "end": "00:27:32",
                    "text": "our team will I our team will be moving"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:28",
                    "end": "00:27:36",
                    "text": "around. So all the best and start"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:32",
                    "end": "00:27:36",
                    "text": "building. Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:08",
                    "end": "00:28:17",
                    "text": "So the portal activates as exa at"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:11",
                    "end": "00:28:21",
                    "text": "exactly 1 p.m. So it's 12:59. Uh anyways"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:17",
                    "end": "00:28:21",
                    "text": "uh can you switch this?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:22",
                    "end": "00:28:27",
                    "text": "Yes. So"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:24",
                    "end": "00:28:30",
                    "text": "now you can see the submit solution is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:27",
                    "end": "00:28:34",
                    "text": "activated. So the buildathon is"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:30",
                    "end": "00:28:37",
                    "text": "officially on. You have 30 minutes and"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:34",
                    "end": "00:28:41",
                    "text": "yeah again start building."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:37",
                    "end": "00:28:41",
                    "text": "Thank you guys."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:50",
                    "end": "00:28:56",
                    "text": "Thank you Amit. Now before we move on to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:53",
                    "end": "00:28:58",
                    "text": "the next stage guys, we expect you all"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:56",
                    "end": "00:29:00",
                    "text": "to maintain full decorum throughout this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:58",
                    "end": "00:29:03",
                    "text": "round. Respect the space, respect the"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:00",
                    "end": "00:29:06",
                    "text": "time and respect the competition. There"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:03",
                    "end": "00:29:11",
                    "text": "it is. The clock is live. 30 minutes on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:06",
                    "end": "00:29:14",
                    "text": "board. The countdown starts now. Our"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:11",
                    "end": "00:29:16",
                    "text": "mentors are here to guide you. If you"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:14",
                    "end": "00:29:17",
                    "text": "need clarity, raise your hand. We are"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:16",
                    "end": "00:29:22",
                    "text": "here to support your enhancement"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:17",
                    "end": "00:29:22",
                    "text": "process. Your time starts now."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:57",
                    "end": "00:31:02",
                    "text": "You need to speak the Wi-Fi."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:49",
                    "end": "00:32:55",
                    "text": "you like to talk"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:51",
                    "end": "00:32:55",
                    "text": ">> just smaller."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:56",
                    "end": "00:32:59",
                    "text": ">> Okay. So, how is your experience of this"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:58",
                    "end": "00:33:02",
                    "text": "whole buildaththon journey?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:59",
                    "end": "00:33:04",
                    "text": ">> It is very nice. I'm very excite excited"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:02",
                    "end": "00:33:07",
                    "text": "to do uh the hackathon and want to"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:04",
                    "end": "00:33:07",
                    "text": "participate."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6992f1114c5e8f0974c874f9",
    "projectId": "369979",
    "projectName": "Harnessing AI to Transform India's Judicial Ecosystem | AI impact summit india 2026",
    "projectDescription": "How AI Can Transform Justice _ The Future of India_s Judicial System _ Panel Discussion",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-16T10:27:28.151Z",
    "projectUpdatedDate": "2026-02-16T10:27:28.151Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-16T10:27:28.151Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "How AI Can Transform Justice _ The Future of India_s Judicial System _ Panel Discussion(MP3_128K).mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771237629647.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:27",
                    "text": " तुम नमर्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मार्मा"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:27",
                    "end": "00:00:57",
                    "text": " ប ប ប ប ប ប"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:57",
                    "end": "00:01:28",
                    "text": " 16 16"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:28",
                    "end": "00:01:56",
                    "text": " जो जो हाल का नहीं आ रहा है, उस गूप में एक सीट बेजे थे न, उस में हार हाल का लंबर लिखा रहा है। और कुछ के जिसके जिसके पाक है हाल, उनको बोडिये कि हम कुछ कर के पुछे कि आप की नहीं हैं। क्योकि अभी हम एक हाल में जायें के लो टाइन लगे हैं।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:56",
                    "end": "00:02:22",
                    "text": " But should we not bother? What is needed is for all of us to learn from those kind of examples that the machine, this capital of the machine is human. And algorithms are made for storing everything. Whatever goes in, is retained there."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:22",
                    "end": "00:02:52",
                    "text": " It's different than the human mind. The human mind is able to differentiate between right and wrong, correct or incorrect. So the fault line between any over-enthusiastic, gen Z person who simply relied on the artificial intelligence output and presented such and so-called citations for his senior, it was the embarrassment of the senior."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:02:52",
                    "end": "00:03:27",
                    "text": " So friends, when we are going to discuss together, we will discuss about the second aspect of constitution. Thank you very much."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:30",
                    "end": "00:03:58",
                    "text": " So we start with a very positive note that probably the artificial intelligence is ready to adopt and adapt to this technology and ensure that it becomes a part of the artificial intelligence system provided. There is no unintelligent use of artificial intelligence. That's a very direct thing you mentioned sir. Moving on to Aarya. So although we are ready to adopt this technology Aarya, but there are going to be many practical implementation issues and concerns, especially structural..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:03:58",
                    "end": "00:04:28",
                    "text": " So my question to you is that what are those key challenges in implementing AI-based tools within the judicial system and a follow up question that how can algorithmic bias arise in judicial AI applications and what safeguards are necessary to protect them?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:28",
                    "end": "00:04:58",
                    "text": " But just to straight up time, I would like to do two topics together. One is JVM, which is judicial decision-maker, and the other one is KVM, which is automated decision-making. So bear with me when I am running past in between the terms JVM and KVM. First and foremost, in any sort of KVM process, which is automated decision-making, how does automated decision-making work?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:58",
                    "end": "00:05:29",
                    "text": " There is a complement of artificial intelligence, there is a complement of machine learning, there is a complement of natural language processing in the communities, commonly known as large language models. So all of these combine over a stack of artificial neural networks, collaboratively in a system where a machine is in a position, basis three things to be delivering automated decisions without human intervention. What are these? First is intelligent perception."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:29",
                    "end": "00:05:59",
                    "text": " very basic what you are currently using in a beautiful context is the intelligent perception you have large volumes of data in natural language and the machine interacts with the set information in the normal prompt and output sending perspective the second thing is called as intelligent cognition where it takes the local knowledge of the world where there is something called as a feedback loop where there is something called as a self-learning loop which means that even if your true new data was"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:59",
                    "end": "00:06:28",
                    "text": " in intelligent cognition, the machine is in a position to learn and adapt basis what is already being done in the past is what we call as intelligent decision making which is automated decision making now across the globe various jurisdictions have been debating on whether ADM can qualify as J&K it is very encouraging"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:28",
                    "end": "00:06:58",
                    "text": " The Supreme Court has initiated the initiative which helps in research and translation, overcoming language barriers. Similar to Estonia, as another country has adopted, we are number one for this kind of work. This is the general policy that we have to do, the AI transparent group, and we then anonymize it as a result of trials, to prevent, this is our action initiative. Where I see a bunch of these meetings, in fact one has been very far, between a little China or some people, it's probably two leaders."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:06:58",
                    "end": "00:07:29",
                    "text": " प्राजिक 206 तान्द भाणिच्यों सपूर्थ अपने कुछ प्राजिक प्राजिक प्राजिक प्राजिक प्राजिक प्राजिक प्राजिक प्राजिक"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:29",
                    "end": "00:07:58",
                    "text": " បបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបបប�"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:58",
                    "end": "00:08:25",
                    "text": " प्रश्ट् त्री प्रश्ट् प्रश्ट् प्रश्ट् प्रश्ट् प्रश्ट् प्रश्ट् प्रश्ट् प्रश्ट् प्रश्ट् प्रश्ट्"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:25",
                    "end": "00:08:53",
                    "text": " So, if you see, go back to 2017, when India did not have a right to privacy, it will only use JDM to decide whether Indians deserve a right to privacy. I don't think we will all be discussing about privacy anymore. Why so? Because all the passwords made before 2017 did not similarly recognize informational privacy and privacy as a fundamental right. They are always attached with certain factual context."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:53",
                    "end": "00:09:21",
                    "text": " ប ប ប ប"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:21",
                    "end": "00:09:50",
                    "text": " is commonly considered as hindsight bias, which is very prevalent in India. So, if it were as a matter of whether it is really a crime happening in India, as AI does, it could support some hindsight bias as to how actually the same machine has worked, so on and so forth. You might end up saying that a big time improvement is largely populated by a certain kind of color, and therefore the rate of crime is significantly higher. This is one of the illustrations of what we call as hindsight bias."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:50",
                    "end": "00:10:20",
                    "text": " There is also recent virus. So a number of times if a particular incident has occurred, then the AI starts believing it as a fact. It will not go down to verify whether it was a fact or not. So these are significant viruses that can impair how efficient this issue will do much. The second piece is black box problem. What is black box problem? Technical experts will have introduced. Black box is still undefinable. We can't really explain how"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:20",
                    "end": "00:10:50",
                    "text": " machine learning is worth it. And this is because of the various layers of human network within which the data is getting translated. So as a result of which there is a need of capacity, which translates into lack of transparency. So applying an AI tool called JDM might significantly suffer from black box as happening in the case of compiles. Whereas the petitioner actually stated that I need to know what is the algorithmic logic that has been used by this particular tool."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:10:50",
                    "end": "00:11:35",
                    "text": " We'll have to pause a bit here for a very important mandate of group photograph because that's how we are supposed to begin it. So for two minutes quickly you can come forward and request all our distinguished panelists for a group photograph to come forward and then we can continue with the discussion."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:36",
                    "end": "00:12:05",
                    "text": " Yes, so we can continue with the session now. We have started with the question of constitutional modality, constitutional values and how do we align them with AI. We have also looked at some of the challenges, implementation challenges."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:05",
                    "end": "00:12:33",
                    "text": " I have a very important question here that while we've been discussing about integrating AI into the judicial system, there's another important question here which is that it's all about reducing the burden on the court and can AI help resolve disputes outside the judicial adjudication system remains to be a very, very important point to be deliberated upon. So on this I would like to ask you how can AI..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:33",
                    "end": "00:13:14",
                    "text": " it has efficiency predictability cost effectiveness when it comes to other dispute resolution mechanisms and what are your insights from a practitioner's perspective what challenges do law firms face with respect to dispute resolution structure and by what position they are?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:14",
                    "end": "00:13:46",
                    "text": " आपस्ट्रिज्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्वार्"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:13:46",
                    "end": "00:14:16",
                    "text": " The question of whether the legitimate is delivered by the mechanical means or is there a human being at the end of the process and it is the human being which has rendered the decision. If the decision is passed over to the machine, it would be bad law because no one has delegated that power to the machine."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:16",
                    "end": "00:14:50",
                    "text": " Even in selection of an arbitrator or selection of a mediator, the decision is such that the party has, the decision is that do you have to, has the delegation of authority been made to the person individually and has not been made to the machine. So this is the typical characteristic or differentiation between the process as it is."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:50",
                    "end": "00:15:35",
                    "text": " In fact, it could be a blound for setting aside an arbitrary reward. The transparent party content, the original VI generated and the non-delegation of decision-making process are keys as far as when you take the law for example, there are multiple purposes affecting the aspect of say, aspecting the evidence, question of preparing a list of things."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:35",
                    "end": "00:16:12",
                    "text": " There is a machine related hallucination which typically happens when you spread your data and you allow external things into the data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:12",
                    "end": "00:16:43",
                    "text": " So the question is that do you get wrong parties, do you get wrong judges, process of decision making. This is a common complaint in relation to processes where the machine has been deftly mannered in which it is decided. So there are several judgments even of our course."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:43",
                    "end": "00:17:17",
                    "text": " The question of whether there are false judgments in the sense that the name was familiar, but the context of the decision was just not correct. In fact, lawyers were found to be resorting to this particular technique and therefore the human supervision over EI has to be ensured. It's not a hands-free kind of situation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:17",
                    "end": "00:17:45",
                    "text": " In two cases in the Yamon vs Chennai Metro Rail Corporation, the Chennai High Court actually developed a procedure whereby engineers worked with the court and they prepared four principles. The system must be confined to the record, so no access to external stuff. It must not invent, infer or give legal opinions. Only present public re-express retrieved experts."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:45",
                    "end": "00:18:16",
                    "text": " The question of transparency was also discussed in the sense that if you put in more data the parties are accessing stuff which is not in the case and external material is resorted to..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:16",
                    "end": "00:18:44",
                    "text": " You have to advise the other side and these are matters which are decided when you really set down the principles which will be applied to the case or in the deviation before you actually prepare your material. So the agreement between parties and the expanse that you really are going to cover is a matter of agreement especially when you have brought it to EI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:48",
                    "end": "00:19:17",
                    "text": " So in the case of Deepak v. Heart & Soul Entertainment Ltd., the question of hate generated cases came up and the court came down very heavily. In US, for example, there are sanctions against a lawyer where he does not supervise adequately and the case is led to be open-ended and the results question has to be looked at."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:17",
                    "end": "00:19:49",
                    "text": " In the problem of AMG wires private limited versus natural basement assessment center, the government was accused of, you know, fixing the process in a manner where it relied in a particular outcome. And that was punished for the reason that the outcomes were pre-fixed and therefore there was no freedom in decision making. And the process effectively therefore proceeds."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:50",
                    "end": "00:20:21",
                    "text": " Thank you. Thank you so much sir for your insights. And all of us in the room here would agree that we need to shed our hesitation in AI when it comes to adopting this technology. But while saying so, it's also important that there has to be responsible adoption of AI, right? It means calibrate and deployment. And coming to that, we need to learn more about how to exactly implement and deploy AI. So we need to learn about use cases. And I think that Nishi could be the best person to inform us and give us some useful use cases in judiciary."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:21",
                    "end": "00:20:54",
                    "text": " When it comes to a locked-in AI, it is a responsible man."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:54",
                    "end": "00:21:24",
                    "text": " . . . . . . . . . . . . ."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:24",
                    "end": "00:21:54",
                    "text": " Let you access data, let 30,000 people do the data. So one of the great challenges that any judge would face is, what about my drugs? So there has to be a complete data erasure mechanism. When the judge removes the machine from its place, the machine should have absolutely no chance of what would then run the judge. All this is possible, and it's a sub 10,000,000 solution. The only challenge that comes is, as Madame itself, how do you build a good tool which can assist the judge?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:55",
                    "end": "00:22:24",
                    "text": " So it's not so much of a machine, but one thing that can come is that there are mechanisms for guardrails. That is, the problem with elements is it never stops. It will never tell you, I don't know. So that is the thing that lawyers, technologists and experts have to come back and sit and say, please stop here, and say, this is all I know. Last 50 years, I have checked, this is the reference, and the legal assistance of the judge or the educator."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:24",
                    "end": "00:22:55",
                    "text": " आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:55",
                    "end": "00:23:25",
                    "text": " បបបបបបប"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:25",
                    "end": "00:23:54",
                    "text": " And the legal experts come together, and with a sovereign NLM, which is a lot of people we are killing, it can be of real assistance to the close to 650 district courts, to NIMHA, the high courts, and even to a number of adjudicators, not just the judges. The NCLAT, the NCLT is income tax accepted tribunals. In those cases, it was not so difficult also, because most of the arguments or cases are very fact-based."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:54",
                    "end": "00:24:19",
                    "text": " And that was similar to the corporate case. So, what I might suggest to the panelists, just start with the adjudicators, the tribunals, we have a learning, and then that can very easily come to civil and corporate cases. Our criminal cases are not very short because we have to study that after. But as far as civil and corporate cases are concerned, in the real case of a civil court machine, we have to bring to the educator a little bit of network data storage."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:19",
                    "end": "00:24:42",
                    "text": " And that, I don't have to tell you a very important thing. Anyone in the world who has access to it will have access, as was said, and no one else has access. In fact, it is also true. All connections require no Wi-Fi, no computers, no land connection, and still no one has access to it. That's all I have to say. Thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:42",
                    "end": "00:25:12",
                    "text": " To all the audience here, I think you would agree with me. We have got some very, very important key takeaways from the three important conversations which we have already had. Moving on to Vikas, you have been very active in judicial automation tools and also towards discrete resolution. The question still is AI-driven automation, how would that influence judicial efficiency case automation? So can we have quick interest from you as well?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:16",
                    "end": "00:27:15",
                    "text": " We are already, as some of you might know, we are using Transparency and Transcribe, Supreme Court Proceedings and Constructions, Dubai International Financial Court, the Afghani Court, the Supreme Court. The background to say, what we are building in India is not just the Indian courts as one of the biggest consumers that we want to address, but I think it's also about taking and setting the steps where economic zones, where those are the kind of things that we are already being responsible, AI, that, but give them agreement between them to identify per roster."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:15",
                    "end": "00:30:25",
                    "text": " you might have 17 minutes 5 seconds between when the lawyer is walking from the scene to argue with you 5 seconds you will try and understand what the case is and then you probably have got this building too what does the judge want? all we need is a crystallized position we are talking about transparency, we are talking about more hallucinations today imagine a hearing where the lawyer is starting with saying this is the fact you start with saying the judge already knows what the agreed position is so back up a little bit straight to the list but until that which judges? black box"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:25",
                    "end": "00:32:59",
                    "text": " आपका आपका आपका आपका आपका।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:18",
                    "end": "00:33:52",
                    "text": " आपका आपका आपका आपका आपका आपका"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:52",
                    "end": "00:34:22",
                    "text": " Who is the judge and how the processes are effectively evolved. So that's the solution which has been found and that's what is likely to happen in relation to future management. Thank you so much. And now I will request Justice Manishariji if you can give your concluding remarks. We'll start from the same question we are on for concluding remarks also. How can we integrate AI into the judicial system by ensuring constitutional values at the same time?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:22",
                    "end": "00:34:49",
                    "text": " Thank you so very much."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:49",
                    "end": "00:35:17",
                    "text": " So the possibilities are available, possibilities are plenty, we keep on"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:17",
                    "end": "00:35:45",
                    "text": " ប ប ប ប ប ប"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:45",
                    "end": "00:36:09",
                    "text": " and delivering of justice in our courts. And the effort of our system is delivering justice in our courts. So, so far as the background or tendency etc., they are of concern and the efforts are being made. They are machines to the extent they help us work in safety through..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:09",
                    "end": "00:36:39",
                    "text": " បបបបបបប"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:39",
                    "end": "00:37:08",
                    "text": " The other day, we have been discussing the idea of sentencing policy in the internal justice system. We know that in that internal justice system, in the end of the day, after conviction, the love interest would be looking at all those aggravating..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:08",
                    "end": "00:37:38",
                    "text": " We have those concepts called lovers."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:38",
                    "end": "00:38:11",
                    "text": " These are those mind, you know, emotional sentiments, starting from love, then love, then your first words, then your first laughter, sorrow, anger, energy, fear, despair, wonder. Now these are human sentiments, and they are integral part of a decision-making, and we are dealing with these other factors where the human brain..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:13",
                    "end": "00:38:55",
                    "text": " आपका आपका आपका आपका आपका आपका आपका।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:56",
                    "end": "00:39:24",
                    "text": " Thank you. Thank you so much, folks. And as we conclude with this session today, there are three important reflections which stand out very, very clearly. The first reflection is that technology is no longer at the periphery of judicial system. It's already becoming a part of the main architecture of this entire judicial ecosystem. And all kinds of initiatives which Supreme Court has already taken, they very well demonstrate that we are on the way towards institutional reforms for adopting and adapting to this technology."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:24",
                    "end": "00:39:51",
                    "text": " The second important reflection here is that no matter what, every development, every technological advancement, every framework which we create should be constitutionally anchored and human-sensitive in its approach because it's we human beings who are at the end of it. And the third very important reflection which emerges today is that the future lies not in replacing human beings, which is the exception of the previous one, not in replacing the judicial system but in strengthening the entire system."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:39:51",
                    "end": "00:40:19",
                    "text": " when it comes to equipping courts and judges with tools, technologies that enhance research, streamline case management and reduce tendency, and of course, expand access to justice. So with this, the roadmap ahead, which is a very important key takeaway from this important family discussion, the roadmap ahead requires clear ethical guardrails, it requires transparent implementation, it requires institutional oversight,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:19",
                    "end": "00:40:43",
                    "text": " capacity building within the judiciary, and finally a very important one, continuous public trust. So with that we conclude the session. Thank you very much everyone. On behalf of National University Delhi, I extend my sincere gratitude to all the panellists here, Justice Maheshwari Ji, Shardul Sir, Vikram, Arya, Rajnish Ji, to each one of you for taking out time to join us for this panel discussion."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:40:43",
                    "end": "00:41:15",
                    "text": " I also extend my sincere gratitude to Maithi, to India AI Impact team, Dr. Sonu Gupta and her team for giving us this important platform for this deliberation today and to of course all of you, one of you all the participants for your thoughtful attention and engagement today. So with this I would request my colleague Dr. Akshana Chandra to present a token of gratitude to the panelists here. This is the momentum from Maithi as well as National University of Delhi."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:41:17",
                    "end": "00:42:00",
                    "text": " आपकुन से जादे हैं। आपकुन से जादे हैं। आपकुन से जादे हैं। आपकुन से जादे हैं।"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:00",
                    "end": "00:42:32",
                    "text": " Thank you very much everyone, thank you."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:42:32",
                    "end": "00:42:39",
                    "text": " यह लगेगा निना स्लॉट भी तो इसका ऐसा है यह भी निलगेगा"
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
{
    "_id": "6992c2734c5e8f0974c86fce",
    "projectId": "644837",
    "projectName": "The Benefits of AI in Road Safety: Reducing Accident and Saving Lives- AI impact Summit india 2026",
    "projectDescription": "The Benefits of AI in Road Safety: Reducing Accident and Saving Lives- AI impact Summit india 2026",
    "projectCategory": "AI-impact-summit-2026",
    "projectCreatedDate": "2026-02-16T07:08:33.908Z",
    "projectUpdatedDate": "2026-02-16T07:08:33.908Z",
    "projectStatus": "Active",
    "companyId": "6992b828c247e02fa5f2bbc8",
    "projectCreatedBy": "6992b828c247e02fa5f2bbc9",
    "projectUpdatedBy": "6992b828c247e02fa5f2bbc9",
    "projectLastAccessDate": "2026-02-16T07:28:48.099Z",
    "projectDeleted": false,
    "projectDataLength": 2,
    "projectType": "VIDEO",
    "cfDataLength": 1,
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "",
            "col4": "Status",
            "col5": "TimeStamps",
            "col6": "Diarization",
            "col7": "Type",
            "col8": "Language",
            "col9": "Model",
            "col10": "Speakers",
            "col11": "SpeakersNames",
            "col12": "FileURL"
        },
        {
            "col1": "P1",
            "col2": "AI for Road Safety_ Saving Lives Through Intelligent Systems(M4A_128K).mp3",
            "col3": "",
            "col4": "DATA",
            "col5": "YES",
            "col6": "NO",
            "col7": "VIDEO",
            "col8": "English",
            "col9": "BASE",
            "col10": "0",
            "col11": "",
            "col12": "uploads/file-1771225703240.mp3",
            "jsonArray": [
                {
                    "speaker": "Unknown",
                    "start": "00:00:00",
                    "end": "00:00:24",
                    "text": " It's WHO, therefore they are kind of writing a bad data on us. B, is there somewhere by which the data is clear or not clear and what we can do? That's where Ministry of Road Transport, next slide please, Road Transport kind of picked up a thought process from IIT Madras. And they started laying the foundations of"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:24",
                    "end": "00:00:53",
                    "text": " a good data-driven road safety initiative. All your AI are on top of these as layers or stacks that you've built. So many congratulations for a pretty future-looking decisions that were taken at Marth at that point of time. They decided that, hey, let's have a national database. Go back, please. Imagine like India."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:00:53",
                    "end": "00:01:23",
                    "text": " You have how many people? Maybe 140 crores. How many police stations? People may not know. And how many crashes? Broadly, we know. All of it in a single umbrella, we get as a data. Nowhere in the world it has been attempted. Go there, please. This database is called as IRAD and EDAR, one of the sister agencies or daughter agencies of METI."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:01:23",
                    "end": "00:04:11",
                    "text": " NIC was the implementation agency. It went very simple. This got kicked off where we are co-opting districts, which is the lowest level of governance and government decision-making, using of Sanjaya for engineering interventions, administrative interventions, and trauma care has started."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:11",
                    "end": "00:04:39",
                    "text": " tools which were all of them have an AI basis for making it, but we don't want to call it as a, of course, my student did put AI there. I wish it was not there, but the reality is there is all these things use a lot of analytics and data crunching and decision supports that are coming about. There is some amount of intelligence that gets built into the analytics that are there."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:04:39",
                    "end": "00:05:11",
                    "text": " These are data-driven governance. So governance is not by thought of one person or the person who sits in the seat. But governance now gets done as a process. It's moving from people-centric to process-centric. Now the next higher level is just not the government, but people also included. Anybody from Orissa here? OK. So Orissa is the first state to go about taking the idea that we had that let's have a"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:05:11",
                    "end": "00:06:57",
                    "text": " connected or people-focused, people-centric government decision-making that happens. That's the policy planning implementation and monitoring all under one umbrella called RAT, which"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:04",
                    "end": "00:07:28",
                    "text": " Come home safe is the tag. Please go on. So a variety of implementation, variety of tools, variety of things that are there, everything in all the four E's that are there. But in there is a dark hole that needs to be addressed. That is a dark hole that I would like to appeal to all of you, which is an empathy."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:28",
                    "end": "00:07:57",
                    "text": " Please factor the human, whichever role that you are in, whether you are on a road, whether you are making a device, whether you are going to apply it, have an empathy towards everybody. And there was also a purpose, so when we were asked to do one of the things Shiatul Singh pointed out, let's not look at many things, let's just think about bringing young students as they call Gen Z's. The idea was to focus"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:07:57",
                    "end": "00:08:28",
                    "text": " and kindle the thought with the Gen Z. So there will be a couple of things that we will share. We would want you to participate. The COERs will be happy to work with you. The ideas are unlimited coming from kids. It gets filtered and trained minds usually start looking the other way for bright ideas. So next slide, please. So the platform. So we have Shiv Pankaj Ji from Morth."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:28",
                    "end": "00:08:57",
                    "text": " who had been literally behind us about creating a platform that can be a very good learning environment. Remember I told when you make it idiot proof, you make better idiots. So how can we create a learning tool? So we created, theoretically, there should be three gates for licensing, a knowledge gate, skill gate, and a practice gate. The knowledge gate,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:08:57",
                    "end": "00:09:27",
                    "text": " requires a mechanism by which the knowledge is better understood and you can use it better. From that perspective, we said that the etiquette that is also to come, the other things that are there, if you knew what things were, you will probably be more empathetic. It is just that we do not know how we end up doing whatever we want to do, right? So to address this, what did we do in the traditional Indic society? At least in everywhere, this was the same."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:27",
                    "end": "00:09:51",
                    "text": " same thing. In the South India, there used to be a small courtyard in the front of a house which used to be called Tinnai, which is where the elders used to sit and teach the younger kids how it is to be in a society. I'm sure the same thing happened in the courtyards in everywhere, whichever part of the country. So but Tinnai has AI in it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:09:51",
                    "end": "00:11:51",
                    "text": " So we decided that we will label and do it as ThinI. ThinI is at this point of time not launched, but it is just being introduced. Please, it's a beta version. Next slide. So the QR code is there for a few seconds. Please do scan in."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:11:52",
                    "end": "00:12:20",
                    "text": " Play around with it, tell why all it fails, what should be done to make it better before we formally launch it sometime very soon. Next slide, please. Anything else to be added to that? No, sir. QR code, if you can display once again. Yeah, I'd request all of you to please scan this QR code and join us in making Tenay the next big thing in road safety. And thank you, sir. Thank you. Next slide. Yeah."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:20",
                    "end": "00:12:54",
                    "text": " This is IIT Madras. IIT Madras is for all. Please do come to IIT Madras, and I extend that, and I end the talk. Thank you. Thank you so much, sir. Right. So that was Professor Venkatesh Balasubramanian for you guys, giving you a brief glimpse of what Center of Excellence for Road Safety actually does in the road safety sector. And now we embark on the next big thing."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:12:54",
                    "end": "00:14:18",
                    "text": " in road safety sector, not only in India, but across the global south. For that, I would like to invite Mr. Atul Singh, COO, CORS, IIT Madras on the stage. The stage is yours, sir. Thanks a lot, and we are running out of time, so we'll keep it very concise. Can we start with the teaser first? Thanks a lot, team."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:18",
                    "end": "00:14:49",
                    "text": " as Professor Venkatesh invited you to IIT Madras. As a CEO, my work is to find a way to get you into IIT Madras. And this is a hackathon that we have launched. And we request you to utilize it and be a part of it. Before I go into it, I just want to hear from such a young and vibrant audience, as she calls Genji. This is going to be for 30 seconds. And I want you to be real quick. If you can have mic around."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:14:49",
                    "end": "00:15:20",
                    "text": " And I'm 100% sure that Genji does not need a mic, do they? Right? So let's see. Let's have a scenario where there is a road accident in front of your college. And this is also Valentine's week. So let's take it like that. Your loved and dear ones have got into an accident. Can you think of one technology or AI tool or digital app or something like that which will help you save that life?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:15:22",
                    "end": "00:16:06",
                    "text": " OK. Any more idea? Gemini. What are you going to ask on Gemini? You'll send a picture and do what? Say what? Will you post it on Instagram or something? Fall detection. OK. OK. Wonderful. Wonderful. So just two more thoughts. Anyone? Nearest hospital. Brilliant idea. Right? But how do I know that which is the nearest hospital? How do I locate it? Google Maps. OK. Wonderful."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:06",
                    "end": "00:16:35",
                    "text": " Now let me give you an event. Let's say your loved one has a brain injury. Now tell me, how will you locate the nearest hospital, the relevant hospital? Ambulance. OK. So this is one of the problems statement that we have as a part of this hackathon. Four days back, I met an accident. I was going from Chennai to Rameswaram."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:16:35",
                    "end": "00:17:04",
                    "text": " When I got into accident, it was not severe, but someone came and hit me from the back I could not find the nearest or my jurisdiction police station, right? I could not find let's say I had the head injury I don't know which hospital to take Google Maps with a saw hospital are engaged It's very difficult to make that decision at that point of time to my console hospital man lay down right so this is one of the following statement that we have and Can we have the QR code, please the poster?"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:05",
                    "end": "00:17:36",
                    "text": " And this is what we are announcing. We are inviting all of you to be a part of it. And the next 10 seconds or 20 seconds, we also have one more offer for you. The first 50 registrations, we are going to mentor you right from IIT Madras to create a solution for it. The first 50 registrations. Can we have it now only another 20 seconds? Wonderful. So first 50 registrations."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:17:36",
                    "end": "00:18:04",
                    "text": " You'll get an invite from us, and we'll be mentoring you to create a solution for it. Thanks a lot. So that's National Road Safety Hackathon 2026. We have already made it one of the biggest in India, and we plan to make it the biggest in Global South, the very foundation of this AI summit. And we really hope to see you all at the campus, which Professor just invited you, and we'll be very happy to have you at the campus."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:04",
                    "end": "00:18:33",
                    "text": " Thanks a lot, Imani. That's all from myself. Yeah. Thank you. Thank you, Atul. Now, without much further ado, as we're running short of time, we move on to the first panel discussion for this morning, which is themed at policies enabling innovation and road safety. And for that, I'd like to request our panelists for the session to please come on the stage. Sri R.P. Shukla Sir, Director, Road Safety Cell, Ministry of Road Transport and Highways. A huge round of applause, please."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:18:33",
                    "end": "00:19:23",
                    "text": " Sri Pankaj Agarwal Sir, Chief Engineer, Ministry of Road Transport and Highways and the moderator for the session will be our very own Atul Singh. Please, stage is yours. So both of them are from Ministry of Road Transport and Highways and in terms of road safety, like Professor Venkatesh gave them a lot of credit and they truly deserve it. Whatever we have achieved in road safety,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:23",
                    "end": "00:19:57",
                    "text": " Like, these are the people, I think they deserve a round of applause from all of you. And sir, they really want to hear from you, so we'll start with the questions. And the first thing, right, the MV Act, the Motor Vehicle Act that we had recently, it laid the foundation for electronic monitoring and enforcement. And we also saw recently, you know, you also released the data sharing policy. So, so much proactive steps are being taken by the Ministry of Road Transport and highways. But..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:19:57",
                    "end": "00:20:29",
                    "text": " AI is catching much faster. There are so many emerging technologies coming up. So any further policy being considered to support this evolution, you would want to highlight, sir. First of all, I would like to say all of the people that road safety is not a thing which is a social problem, it is a personal also. So government role is always as a facilitator, but you are the driver and you yourself are the passenger. So just like..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:29",
                    "end": "00:20:58",
                    "text": " So you have a need for water. Mobility, consumption is also a factor. If you compare yourself, what is the role of AI? We have a lot of data available with us, but the problem is how to act on that. Enforcement system is available. We have a lot of challenges issued, but nobody is coming and doing it. We are unable to enforce the challenge through e-monitoring systems. So AI is providing all information to us. A lot of technologies are available."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:20:58",
                    "end": "00:21:24",
                    "text": " Government is providing the enabling environment to every state because policing is more in the domain of the state governments. So we are in touch with the stakeholders, like state governments, and giving them all the support through technologies, grants, and everything we are providing. So it is the first step. The main data indicates that the main problem is the speeding and the traffic rules violation."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:24",
                    "end": "00:21:50",
                    "text": " So capturing it and providing as an evidence before and without human intervention, here the AI can definitely play a big role. Nowadays, whatever data we have, only a police person would record the data. That is not the actual data. There are many other offenses which contribute to the accidents. And the real picture can only come, we have all the cameras installed on the high speed."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:21:50",
                    "end": "00:22:18",
                    "text": " high speed corridors and which happen to carry maximum transport need or fulfilling the maximum transport. So what is the problem? If we are not able to analyze the captured data, the way your legislation system work, then it will be difficult to enforce the law and we will not be able to bring down the fatality. That is the first step."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:22:18",
                    "end": "00:23:17",
                    "text": " The second we want to integrate with this thin AI in the future."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:17",
                    "end": "00:23:44",
                    "text": " So in that case, how to prevent those for which we do not have any data. So we have to do some vehicle engineering sort of a thing, give some feedback to the driver. So for that, mandating V2D communication."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:23:44",
                    "end": "00:24:00",
                    "text": " We hope it's a proper, because we believe that nobody wants to commit the error to so great that he should injure himself. So with that intention, if YouTube technology can provide some feedback to him, he can correct in the, I mean, before."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:00",
                    "end": "00:24:31",
                    "text": " it can lead to any collision. So vehicle to vehicle communication technology has been basically mandated and we have given right now the domain frequency of 30 megahertz. So there is a lot of scope for all the AI, we want to work on the AI to how to come out with the software and sort of a thing. AI-based road safety projects another part. Pre-licensing system. State government is mandated to give the license."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:24:31",
                    "end": "00:25:02",
                    "text": " But driving is basically a life skill. We are taught computers. It has been mandated. What about the driving? No school teaches driving. I mean the normal school. I'm not talking about the government licensing school. It is a need. So what you do, you go to a different system. Why not it should be integrated in the school itself at the age of 16? So that is another goal of thin AI, which this IIT Madras is working."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:02",
                    "end": "00:25:28",
                    "text": " It should be mandated as a part of curriculum while teaching computer, particularly the girls also. Everybody should learn driving. Once they will learn driving, they will be able to judge the person who is driving, even as a passenger, how he is driving. So it will definitely lead some awareness what are the driving skills. And most of the people, we do not know driving, so they can't judge how other person is driving."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:28",
                    "end": "00:25:57",
                    "text": " So that will create a consciousness among the people that these are the things we need to learn and a certification sort of a thing, it will generate and give you the knowledge and give some practice. So that consciousness and awareness, definitely we feel that we're able to reduce the accidents and help in capturing the base data. So similarly, further policy reforms connected mobility, frameworks for lining mobility and emission norms."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:25:57",
                    "end": "00:26:25",
                    "text": " Right, sir. Another point is how to control the pollution. So in that also, we are basically trying to find a way out so that people should not cheat by having some context environments because they are issuing on their own without the vehicle going to the center. So we are developing an AI tool for that also. The environment is a major concern in the urban center, not in the rural that much."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:25",
                    "end": "00:26:54",
                    "text": " Another problem which is a very good area to work upon, we are, all our industry is car-centric, not two-wheeler-centric. All safety features are for car, whatever the two-wheeler. It is a very, very big area because by 45% fatalities are of two-wheelers, 20% pedestrian. So all, everything is about car, but nobody thinks of scooter. So that is in a big area which ministry in future."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:26:54",
                    "end": "00:27:24",
                    "text": " planning how to go about the launch of invite the industry and invite the young people because it's simple strike to a two-wheeler because of his own or anybody else it will it will get destabilized and people fall and have a head injury and 35,000 death occurred in a collision of a two-wheeler to two-wheeler not by car out of 150,000 so it is that and those are all low-speed data"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:24",
                    "end": "00:27:53",
                    "text": " So how is it possible? So they should be given some feedback if somebody is coming or your behavior is something you have to come out of the box thinking for the purpose of two-wheeler. So with that, I will... Yes, I think, sir, you answered the next two, three questions also. So it made my life very easy, sir. And as sir rightly said, underreporting of data is a big problem. For example, many of you might be doing..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:27:53",
                    "end": "00:28:25",
                    "text": " Around 150,000 people die due to road accidents. The numbers can be much more higher because people don't report. And that's why we are running a program called data-driven hyperlocal intervention, where we are doing field perception survey to capture those uncaptured data. So many of you would want to know more about it, and please, I will request you to read about it. Sir also mentioned one more point. Technology has some limitations, and humans are very important."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:25",
                    "end": "00:28:54",
                    "text": " My next question is to R.P. Shukla sir and he's pioneering a program called Sarak Suraksha Mitra, which is very relevant to all of you, the Genji crowd which is sitting here. So sir, please tell the audience a little bit about what that program is and how these people can be a part of it. Thank you Atul. This is the Sarak Suraksha Mitra program. It's a program in collaboration with the Ministry of Youth Affairs and Sports."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:28:55",
                    "end": "00:29:24",
                    "text": " aim to have the youths involved on the volunteer basis which are self-motivated and dedicated towards the road safety concerns. Basic concept behind involving the youth is that the so many problems in the road safety sectors are the area specific. So making a policy from Delhi level or a central level or some from the state level headquarters may not most of the times applicable at the district levels or the rural levels."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:24",
                    "end": "00:29:52",
                    "text": " Now the rural connectivity is also very good. And the rural roads made under the Pradhan Mantri Gramit Siddhu Krishna are also carrying a substantial amount of traffic, which the users are very, you can say that they're new to the road safety environments. So this is a bottom-up approach, where the local youth of your group, of Janji group, what the madam said that, is..."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:29:52",
                    "end": "00:30:22",
                    "text": " to be hired as a, not hired as a recruited as a volunteer. You will be given at 15 days training, five days training for the online training and then there is the offline training. Now this program is running in the top 100 fatality districts, not in a pan India basis, but it will be going to be launched soon on the pan India basis. The second concept behind this is the concept of road safety auditor is a very new to our country."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:23",
                    "end": "00:30:53",
                    "text": " Normally we do not find any institution where the road safety auditors are being trained. So we are not just run this program to have a one program, we want to run it as an opportunity availability also. Where you can learn the basics of the road safety auditing and helping the local authorities, you will work with the district road safety committees headed by the respective district magistrates and commissioners. So you can assist them in helping them"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:30:53",
                    "end": "00:31:25",
                    "text": " what the JNG thinks about the road safety, what can be the future challenges of the road safety with the new vehicle design, new road designs. The second most important thing which we want from you volunteers and JNG, the empathy what the Professor Bala Subramanian was talking all about. What happened when a road accident happened? A question was raised by Atul during your Valentine week. Which day is today? It is over."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:25",
                    "end": "00:31:60",
                    "text": " So 16 was the last day. So many people said so many things. Then he zeroed down that he has a head injury. Who has told them that first we will take him to the hospital? Does anybody think of taking him to the hospital? Somebody is talking about the Gemini, somebody is talking about something else, somebody is talking about something else. But nobody thinks of saving his or her life."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:31:60",
                    "end": "00:32:26",
                    "text": " Kumar Sambhav, which tells the story of how the Lord Kartikeya was born. In that he told the Parvati that your body is the source of all good things or all bad things. So first love your body, first protect your body. So as your Janji, you are least bothered about yourself. You are bothered about so many other things, but you are least bothered about yourself."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:32:26",
                    "end": "00:32:59",
                    "text": " How many girls sitting here are putting helmet when they are driving the two wheelers? Be honest to yourself. And now how many girls are in the room? Who are not wearing. So this is the problem where we want to involve you as a volunteer to understand what is the need of the helmet. It is not just to prevent the challan, it is to prevent you. Most common answer of you girls are that it clumsies the hair. But dear you need the head for hair."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:02",
                    "end": "00:33:35",
                    "text": " So put it. So we want to involve you at that basic grassroot level where you can understand why these legislations are made. They are not made for revenue purposes to collect the challans and enhance the government's pocket and as checker. They are here to make you safe. Roads are for you, not for the challans. Use them judiciously. And use whatever information you will be gaining through the Sadak Suraksha Mithra program. We are trying to escalate it."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:33:35",
                    "end": "00:34:05",
                    "text": " is that it will become, at the completion of the program, you will get a certificate co-branded with the IIT Madras. And any certificate from any IIT has its own value, need not to say that. So, you will become a future, our road safety advocate, you will become our future road safety auditor, you will become our future road safety AI influencer. So many things you can do."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:05",
                    "end": "00:34:35",
                    "text": " when you join this program. So it is, I earnestly request all of you, whosoever is driving, everyone is road user, pedestrian to the one who is driving the Volvo bus. Everybody is road user and everybody is at threat. So whenever you saw an accident, there are the four lines from the very noted shayar, Nawaz Devbandi,"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:34:35",
                    "end": "00:35:04",
                    "text": " Whenever you saw a road accident, don't think about anything else. Just think to take him to the nearest hospital. Be empathetic towards him."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:04",
                    "end": "00:35:36",
                    "text": " Don't make reels. Don't put it on Instagram. Nobody is watching Instagram except you people. Most of the time, your reels are kept on swiping. I just met a doctor from the NIMHANS who said that most of the youths have having partial paralyzing of the thumbs. So this is all about the madam is knocking horns. Because I'm constantly getting the push. One thing I want to say, just ma'am."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:35:36",
                    "end": "00:36:05",
                    "text": " So most of you are involving in the AI. So I want to give one or two concepts to you. So help the Professor Bala Subrahmanyam and our Chief Engineer Mr. Pankaj Agarwal in developing some Distal Road Safety Academy, some Distal Road Safety Museum where the newcomers or the old people can come and see where they committed the mistakes, develop a road safety diary for"
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:05",
                    "end": "00:36:38",
                    "text": " you people where you can daily make an entry where the mistakes you have committed what you not to repeat. Thank you, ma'am. Sorry, ma'am. No, sir. We wanted to continue it. Because we're short on time. Right. Thank you so much for this panel discussion. Also, we had panel two with the industry leaders from Volvo and Lightmetrics. Thank you so much, sir. Thank you so much."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:36:38",
                    "end": "00:37:18",
                    "text": " I'd like to request Professor to kindly come on stage and felicitate the guests from ministry. And the speakers from Volvo. I'd like to request Mr. Deepan Rajasur from Volvo Group to kindly come on the stage. And Dr. Pushkar Patwardhan, co-founder of Lightmetrics Technologies, to kindly come on the stage. A big round of applause, please. OK. Go, go, go."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:37:49",
                    "end": "00:38:04",
                    "text": " Since we are running short of time, we are not able to continue with our next panel discussion. And with this, I really request all the youngsters out here to come up with better solutions."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:04",
                    "end": "00:38:52",
                    "text": " based on AI or not AI, come up with better solutions for road safety, join our road safety hackathon. And I finally thank you all for being such a great audience. Thank you for the speakers for joining us. Thank you everyone for coming and joining us early in the morning. With this, we bring this session to a close. I, Himani Suri, thanking you all bottom of my heart on behalf of Center of Excellence for Road Safety, IIT Madras. Thank you all."
                },
                {
                    "speaker": "Unknown",
                    "start": "00:38:53",
                    "end": "00:39:08",
                    "text": " for the heart, for the heart, for the heart, for the heart."
                }
            ]
        }
    ],
    "sourceLanguage": "English",
    "targetLanguage": "English"
},
    {
    "_id": "697eff001c7796fce90d2548",
    "projectId": "613084",
    "projectName": "Budget 2026-2027 - Speech of Nirmala Sitharaman, Minister of Finance, February 1, 2026",
    "projectDescription": "",
    "projectCategory": "Finance",
    "img":"assets/img/india Budget 2026.webp",
    "img-title":"Budget 2026-2027 - Speech of Nirmala Sitharaman, Minister of Finance, February 1, 2026",
    "projectCreatedDate": "2026-02-01T07:21:35.158Z",
    "projectUpdatedDate": "2026-02-01T07:21:35.158Z",
    "projectStatus": "Active",
    "companyId": "69453c81e603a0a401b6f37b",
    "projectCreatedBy": "69453c81e603a0a401b6f37c",
    "projectUpdatedBy": "69453c81e603a0a401b6f37c",
    "projectLastAccessDate": "2026-02-02T05:51:13.457Z",
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "Content",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "budget_speech.docx",
            "col3": "Budget 202 6 -202 7 Speech of Nirmala Sitharaman Minister of Finance February 1 , 202 6 Hon’ble Speaker, On the sacred occasion of Magha Purnima and the birth anniversary of Guru Ravidas, I present the Budget for the year 2026-2027. Introduction Since we assumed office 12 years ago, India’s economic trajectory has been marked by stability, fiscal discipline, sustained growth and moderate inflation. This is the result of conscious choices we have made, even in times of heightened uncertainty and disruption. Our Government , led by Hon’ble Prime Minister Modi, has decisively and consistently chosen action over ambivalence, reform over rhetoric and people over populism. We have pursued far reaching structural reforms, fiscal prudence and monetary stability whilst maintaining a strong thrust on public investment. Keeping atmanirbharta as a lodestar, we have built domestic manufacturing capacity, energy security and reduced critical import dependencies. Simultaneously, we have ensured that citizens benefit from every action of the Government, undertaking reforms to support employment generation, agricultural productivity, household purchasing power and universal services to people. These measures have delivered a high growth rate of around 7% and helped us make substantial strides in poverty reduction and improvement in the lives of our people. Today, we face an external environment in which trade and multilateralism are imperilled and access to resources and supply chains are disrupted. New technologies are transforming production systems while sharply increasing demands on water, energy and critical minerals. India will continue to take confident steps towards Viksit Bharat, balancing ambition with inclusion. As a growing economy with expanding trade and capital needs, India must also remain deeply integrated with global markets, exporting more and attracting stable long-term investment. Part A As I begin Part A, I want to express my gratitude to the people for standing firmly with us as we forge our way together towards becoming one of the largest economies of the world. Our aim is to transform aspiration into achievement and potential into performance, as we ensure that the dividends of growth reach every farmer, the scheduled caste, the scheduled tribes, the nomads, the youth, the poor and the women. In the Viksit Bharat Young Leaders Dialogue 2026, several innovative ideas were shared with our Prime Minister, which have inspired many of the proposals, making this a unique Yuva Shakti-driven Budget . Our Government’s ‘Sankalp’ is to focus on our poor, underprivileged and the disadvantaged . To deliver on this Sankalp, and given that this is the first Budget prepared in Kartavya Bhawan, we are inspired by 3 kartavya : Our first kartavya is to accelerate and sustain economic growth, by enhancing productivity and competitiveness, and building resilience to volatile global dynamics. Our second kartavya is to fulfil aspirations of our people and build their capacity, making them strong partners in India’s path to prosperity. Our third kartavya , aligned with our vision of Sabka Sath, Sabka Vikas, is to ensure that every family, community, region and sector has access to resources, amenities and opportunities for meaningful participation. This threefold approach requires a supportive ecosystem. The first requirement is to sustain the momentum of structural reforms — continuous, adaptive, and forward-looking. Second, a robust and resilient financial sector is central to mobilising savings, allocating capital efficiently and managing risks. Third, cutting-edge technologies, including AI applications, can serve as force multipliers for better governance. Reform Express Our Government has undertaken comprehensive economic reforms towards creating employment, boosting productivity and accelerating growth. After the Prime Minister’s announcement on Independence Day in 2025, over 350 reforms have been rolled out. Th ese include GST simplification, notification of Labour Codes, and rationalisation of mandatory Quality Control Orders. High Level Committees have been formed and in parallel, the Central Government is working with the State Governments on deregulation and reducing compliance requirements. The Reform Express is well on its way and will maintain its momentum to help us fulfil our kartavya . I now move to the specific proposals. Under our first kartavya to accelerate and sustain economic growth , I propose interventions in six areas: i ) Scaling up manufacturing in 7 strategic and frontier sectors; ii) Rejuvenating legacy industrial sectors; iii) Creating “Champion MSMEs”; iv) Delivering a powerful push to Infrastructure; v) Ensuring long-term energy security and stability; and vi) Developing City Economic Regions. Scaling up manufacturing in 7 strategic and frontier sectors: Biopharma SHAKTI (Strategy for Healthcare Advancement through Knowledge, Technology and Innovation) – India’s disease burden is observed to be shifting towards non-communicable diseases, like diabetes, cancer and autoimmune disorders. Biologic medicines are key to longevity and quality of life at affordable costs. To develop India as a global Biopharma manufacturing hub, I propose the Biopharma SHAKTI with an outlay of ₹ 10,000 crores over the next 5 years. This will build the ecosystem for domestic production of biologics and biosimilars. The Strategy will include a Biopharma-focused network with 3 new National Institutes of Pharmaceutical Education and Research (NIPER) and upgrading 7 existing ones. It will also create a network of over 1000 accredited India Clinical Trials sites. We propose to strengthen the Central Drugs Standard Control Organisation to meet global standards and approval timeframes through a dedicated scientific review cadre and specialists. India Semiconductor Mission (ISM) 1.0 expanded India’s semiconductor sector capabilities. Building on this, we will launch ISM 2.0 to produce equipment and materials, design full-stack Indian IP, and fortify supply chains. We will also focus on industry-led research and training centres to develop technology and skilled workforce. The Electronics Components Manufacturing Scheme, launched in April 2025 with an outlay of ₹22,919 crore, already has investment commitments at double the target. We propose to increase the outlay to ₹40,000 crore to capitalise on the momentum. A Scheme for Rare Earth Permanent Magnets was launched in November 2025. We now propose to support the mineral-rich States of Odisha, Kerala, Andhra Pradesh and Tamil Nadu to establish dedicated Rare Earth Corridors to promote mining, processing, research and manufacturing. To enhance domestic chemical production and reduce import-dependency, we will launch a Scheme to support States in establishing 3 dedicated Chemical Parks, through challenge route, on a cluster-based plug-and-play model. Strong capital goods capability is a determinant of productivity and quality across different sectors. Towards building this capacity, I propose the following: Hi-Tech Tool Rooms will be established by CPSEs at 2 locations as digitally enabled automated service bureaus that locally design, test, and manufacture high-precision components at scale and at lower cost. A Scheme for Enhancement of Construction and Infrastructure Equipment (CIE) will be introduced to strengthen domestic manufacturing of high-value and technologically-advanced CIE. This can range from lifts in a multi-story apartment, fire-fighting equipment, large and small, to tunnel-boring equipment for building metros and high-altitude roads. I also propose a Scheme for Container Manufacturing to create a globally competitive container manufacturing ecosystem, with a budgetary allocation of ₹ 10,000 crore over a 5 year period. For the labour-intensive Textile Sector , I propose an Integrated Programme with 5 sub-parts: The National Fibre Scheme for self-reliance in natural fibres such as silk, wool and jute, man-made fibres , and new-age fibres ; Textile Expansion and Employment Scheme to modernise traditional clusters with capital support for machinery, technology upgradation and common testing and certification centres; A National Handloom and Handicraft programme to integrate and strengthen existing schemes and ensure targeted support for weavers and artisans; Tex-Eco Initiative to promote globally competitive and sustainable textiles and apparels; Samarth 2.0 to modernize and upgrade the textile skilling ecosystem through collaboration with industry and academic institutions. Further, I propose to set up Mega Textile Parks in challenge mode. They can also focus on bringing value addition to technical textiles. I propose to launch the Mahatma Gandhi Gram Swaraj initiative to strengthen khadi, handloom and handicrafts. This will help in global market linkage and branding. It will streamline and support training, skilling, quality of process and production. This will benefit our weavers, village industries, One - District – One - Product initiative and rural youth. India has the potential to emerge as a global hub for high quality, affordable sports goods . I propose a dedicated initiative for sports goods that will promote manufacturing, research and innovation in equipment design as well as material sciences. Rejuvenation of Legacy Industrial Clusters I propose to introduce a Scheme to revive 200 legacy industrial clusters to improve their cost competitiveness and efficiency through infrastructure and technology upgradation. Creating “Champion SMEs” and supporting micro enterprises: Recognising MSMEs as a vital engine of growth, I propose a three-pronged approach to help them grow as ‘Champions’: Equity Support I propose to introduce a dedicated ₹10,000 crore SME Growth Fund , to create future Champions, incentivizing enterprises based on select criteria. I also propose to top up the Self-Reliant India Fund set up in 2021, with ₹2,000 crore to continue support to micro enterprises and maintain their access to risk capital. Liquidity Support With TReDS , more than ₹ 7 lakh crore has been made available to MSMEs. T o leverage its full potential, I propose 4 measures: ( i ) mandate TReDS as the transaction settlement platform for all purchases from MSMEs by CPSEs, serving as a benchmark for other corporates; (ii) introduce a credit guarantee support mechanism through CGTMSE for invoice discounting on TReDS platform; (iii) link GeM with TReDS for sharing information with financiers about government purchases from MSMEs, encouraging cheaper and quicker financing; (iv) introduce TReDS receivables as asset-backed securities, helping develop a secondary market, enhancing liquidity and settlement of transactions. Professional Support Government will facilitate Professional Institutions such as ICAI, ICSI, ICMAI to design short-term, modular courses and practical tools to develop a cadre of ‘Corporate Mitras’, especially in Tier-II and Tier-III towns. These accredited para-professionals will help MSMEs meet compliance requirements at affordable costs. Infrastructure During this past decade our Government has undertaken several initiatives for large-scale enhancement of public infrastructure including through new financing instruments such as Infrastructure Investment Trusts ( InVITs ) and Real Estate Investment Trusts (REITs) and institutions like NIIF and NABFID. We shall continue to focus on developing infrastructure in cities with over 5 lakh population (Tier II and Tier III), which have expanded to become growth centres. Public capex has increased manifold from ₹2 lakh crore in FY2014-15 to an allocation of ₹11.2 lakh crore in BE 2025-26. In FY2026-27, I propose to increase it to ₹12.2 lakh crore to continue the momentum. To strengthen the confidence of private developers regarding risks during infrastructure development and construction phase, I propose to set up an Infrastructure Risk Guarantee Fund to provide prudently calibrated partial credit guarantees to lenders. Over the years, REITs have emerged as a successful instrument for asset monetisation . I propose to accelerate recycling of significant real estate assets of CPSEs through the setting up of dedicated REITs . To promote environmentally sustainable movement of cargo, I propose to: a) Establish new Dedicated Freight Corridors connecting Dankuni in the East, to Surat in the West; b) o perationalise 20 new National Waterways (NW) over next 5 years, starting with NW-5 in Odisha to connect mineral rich areas of Talcher and Angul and industrial centres like Kalinga Nagar to the Ports of Paradeep and Dhamra . Training Institutes will be set up as Regional Centres of Excellence for development of the required manpower. This will benefit youth in the entire stretch of the waterways to train and acquire skills. Further, a ship repair ecosystem catering to inland waterways will also be set up at Varanasi and Patna; c) launch a Coastal Cargo Promotion Scheme for incentivising a modal shift from rail and road, to increase the share of inland waterways and coastal shipping from 6 % to 12 % by 2047. To enhance last-mile and remote connectivity, and promote tourism, I propose to give incentives to indigenize manufacturing of seaplanes. A Seaplane VGF Scheme will be also be introduced to provide support for operations. Carbon Capture Utilization and Storage (CCUS) Aligning with the roadmap launched in December 2025, CCUS technologies at scale will achieve higher readiness levels in end-use applications across five industrial sectors, including, power, steel, cement, refineries and chemicals . An outlay of ₹ 20,000 crore is proposed over the next 5 years. City Economic Regions Cities are India’s engines of growth, innovation, and opportunities. We shall now focus on Tier II and Tier III cities, and even temple-towns, which need modern infrastructure and basic amenities. This Budget aims to further amplify the potential of cities to deliver the economic power of agglomerations by mapping city economic regions (CER), based on their specific growth drivers. An allocation of ₹ 5000 crore per CER over 5 years is proposed for implementing their plans through a challenge mode with a reform-cum-results based financing mechanism. In order to promote environmentally sustainable passenger systems, we will develop seven High-Speed Rail corridors between cities as ‘growth connectors’ , namely i ) Mumbai-Pune, ii) Pune-Hyderabad, iii) Hyderabad-Bengaluru, iv) Hyderabad-Chennai, v) Chennai-Bengaluru, vi) Delhi-Varanasi, vii) Varanasi-Siliguri. Financial Sector The Indian banking sector today is characterised by strong balance sheets, historic highs in profitability, improved asset quality and coverage exceeding 98% of villages in the country. At this juncture, we are well-placed to futuristically evaluate the measures needed to continue on the path of reform-led growth of this sector. I propose setting up a “High Level Committee on Banking for Viksit Bharat” , to comprehensively review the sector and align it with India’s next phase of growth, while safeguarding financial stability, inclusion and consumer protection. The vision for NBFCs for Viksit Bharat has been outlined with clear targets for credit disbursement and technology adoption. In order to achieve scale and improve efficiency in the Public Sector NBFCs, as a first step, it is proposed to restructure the Power Finance Corporation and Rural Electrification Corporation. I propose a comprehensive review of the Foreign Exchange Management (Non-debt Instruments) Rules to create a more contemporary, user-friendly framework for foreign investments, consistent with India’s evolving economic priorities. Corporate Bond Market I propose to introduce a market making framework with suitable access to funds and derivatives on corporate bond indices. I also propose to introduce total return swaps on corporate bonds. Municipal Bonds To encourage the issuance of municipal bonds of higher value by large cities, I propose an incentive of ₹100 crore for a single bond issuance of more than ₹1000 crore. The current scheme under AMRUT which incentivises issuances up to ₹200 crore, will also continue to support smaller and medium towns. Ease of Doing Business Individual Persons Resident Outside India (PROI) will be permitted to invest in equity instruments of listed Indian companies through the Portfolio Investment Scheme. It is also proposed to increase the investment limit for an individual PROI under this scheme from 5% to 10%, with an overall investment limit for all individual PROIs to 24%, from the current 10%. Emerging technologies, including AI 21 st Century is technology driven. Adoption of technology is for the benefit of all people - farmers in the field, women in STEM, youth keen to upskill and Divyangjan to access newer opportunities. The Government has taken several steps to support new technologies through AI Mission, National Quantum Mission, Anusandhan National Research Fund, and Research, Development and Innovation Fund. Our second kartavya is to fulfil aspirations and build capacity. Close to 25 crore individuals have come out of multidimensional poverty through a decade of our Government’s sustained and reform-oriented efforts. Our Government has therefore decided to place a renewed emphasis on the Services Sector to provide a pathway to fulfilling aspirations of a youthful India, with the following measures. High-Powered ‘Education to Employment and Enterprise’ Standing Committee I propose to set up a High-Powered ‘Education to Employment and Enterprise’ Standing Committee to recommend measures that focus on the Services Sector as a core driver of Viksit Bharat. This will make us a global leader in services, with a 10% global share by 2047. The Committee will prioritise areas to optimise the potential for growth, employment and exports. They will also assess t he impact of emerging technologies, including AI, on jobs and skill requirements and propose measures thereof. Creation of Professionals for Viksit Bharat To create a new range of skilled career pathways for our youth, I propose interventions in the following sectors: Health Existing institutions for Allied Health Professionals (AHPs) will be upgraded and new AHP Institutions established in private and Government sectors. This will cover 10 selected disciplines, including optometry, radiology, anesthesia, OT Technology, Applied Psychology and Behavioural Health and add 100,000 AHPs over the next 5 years. A strong Care Ecosystem , covering geriatric and allied care services will be built. A variety of NSQF-aligned programmes will be developed to train multiskilled caregivers combining core care and allied skills, such as, wellness, yoga and operation of medical and assistive devices. In the coming year, 1.5 lakh caregivers will be trained. Hubs for Medical Value Tourism To promote India as a hub for medical tourism services, I propose to launch a Scheme to support States in establishing five Regional Medical Hubs , in partnership with the private sector. These Hubs will serve as integrated healthcare complexes that combine medical, educational and research facilities. They will have AYUSH Centres , Medical Value Tourism Facilitation Centres and infrastructure for diagnostics, post-care and rehabilitation. These Hubs will provide diverse job opportunities for health professionals including doctors and AHPs. AYUSH Ancient Indian yoga, already respected in several parts of the world, was given mass global recognition when Hon’ble PM took it to the UN. Post-COVID, Ayurveda gained a similar global acceptance and recognition. Exporting quality Ayurvedic products helps farmers who grow the herbs and the youth who process the products. To meet growing global demand, a few more steps are being taken. I propose to ( i ) set up 3 new All India Institutes of Ayurveda; (ii) upgrade AYUSH pharmacies and Drug Testing Labs for higher standards of certification ecosystem, and make available more skilled personnel; (iii) upgrade the WHO Global Traditional Medicine Centre in Jamnagar to bolster evidence-based research, training and awareness for traditional medicine. Animal Husbandry Livestock contributes close to 16% of farm income, including of poor and marginal households. To scale up availability of veterinary professionals by more than 20,000, I propose to roll out a loan-linked capital subsidy support scheme for establishment of veterinary and para-vet colleges, veterinary hospitals, diagnostic laboratories and breeding facilities in the private sector. Collaboration between Indian and foreign institutions will also be facilitated. Orange Economy India’s Animation, Visual Effects, Gaming and Comics (AVGC) sector is a growing industry, projected to require 2 million professionals by 2030. I propose to support the Indian Institute of Creative Technologies, Mumbai in setting up AVGC Content Creator Labs in 15,000 secondary schools and 500 colleges. Design The Indian design industry is expanding rapidly and yet there is a shortage of Indian designers. I propose to establish through challenge route, a new National Institute of Design to boost design education and development in the eastern region of India. Education Our Government will support States, through challenge route, in creating 5 University Townships in the vicinity of major industrial and logistic corridors. These planned academic zones will host multiple universities, colleges, research institutions, skill centres and residential complexes. In Higher Education STEM institutions, prolonged hours of study and laboratory work pose some challenges for girl students. Through VGF/capital support, 1 girls’ hostel will be established in every district . To promote Astrophysics and Astronomy via immersive experiences, 4 Telescope Infrastructure facilities will be set up or upgraded - the National Large Solar Telescope, the National Large Optical-infrared Telescope, the Himalayan Chandra Telescope and the COSMOS-2 Planetarium. Tourism The Tourism sector has the potential to play a large role in employment generation, forex earnings and expanding the local economy. I propose to set up a National Institute of Hospitality by upgrading the existing National Council for Hotel Management and Catering Technology. It will function as a bridge between academia, industry and the Government. I also propose a pilot scheme for upskilling 10,000 guides in 20 iconic tourist sites through a standardized, high-quality 12-week training course in hybrid mode, in collaboration with an Indian Institute of Management. A National Destination Digital Knowledge Grid will be established to digitally document all places of significance—cultural, spiritual and heritage. This initiative will create a new ecosystem of jobs for local researchers, historians, content creators and technology partners. India has the potential and opportunity to offer world-class trekking and hiking experience. We will develop ecologically sustainable ( i ) Mountain trails in Himachal Pradesh, Uttarakhand and Jammu and Kashmir; Araku Valley in the Eastern Ghats and Podhigai Malai in the Western Ghats. (ii) Turtle Trails along key nesting sites in the coastal areas of Odisha, Karnataka and Kerala; and (iii) Bird watching trails along the Pulikat lake in Andhra Pradesh and Tamil Nadu. Under the visionary leadership of Honorable Prime Minister, we established the International Big Cat Alliance in 2024. This year, India is hosting the first ever Global Big Cat Summit, where heads of governments and ministers from 95 range countries will deliberate on collective strategies for conservation. Heritage and Culture Tourism I propose to develop 15 archeological sites including Lothal, Dholavira , Rakhigarhi , Adichanallur , Sarnath, Hastinapur, and Leh Palace into vibrant, experiential cultural destinations. Excavated landscapes will be opened to the public through curated walkways . Immersive storytelling skills and technologies will be introduced to help conservation labs, interpretation centres, and guides. Sports The Sports Sector provides multiple means of employment, skilling and job opportunities. Taking forward the systematic nurturing of sports talent which is set in motion through the Khelo India programme, I propose to launch a Khelo India Mission to transform the Sports sector over the next decade. The Mission will facilitate: a) An integrated talent development pathway, supported by training centres (foundational, intermediate and elite levels); b) systematic development of coaches and support staff; c) integration of sports science and technology; d) competitions and leagues to promote sports culture and provide platforms; and, e) development of sports infrastructure for training and competition. Our third kartavya aligns with our vision of Sabka Sath, Sabka Vikas towards a Viksit Bharat. This requires targeted efforts for a) Increasing farmer incomes through productivity enhancement and entrepreneurship, with special attention to small and marginal farmers ; b) Empowering Divyangjan through access to livelihood opportunities, training and high-quality assistive devices ; c) Empowering the vulnerable to access mental health and trauma care; d) Focus on the Purvodaya States and the North-East Region to accelerate development and employment opportunities. Increasing Farmer Incomes Fisheries: We will undertake initiatives ( i ) for integrated development of 500 reservoirs and Amrit Sarovars (ii) strengthen the fisheries value chain in coastal areas and enable market linkages involving start-ups and women-led groups together with Fish Farmers Producer Organisations . Animal Husbandry: To provide quality employment opportunities in rural and peri-urban areas, we will support the Animal Husbandry Sector in entrepreneurship development through: (a) a Credit-Linked Subsidy Programme (b) scaling-up and modernisation of livestock enterprises (c) enhance creation of livestock, dairy and poultry-focused integrated-value chains and (d) encourage creation of Livestock Farmer Producers Organisations . High Value Agriculture : To diversify farm outputs, increase productivity, enhance farmers’ incomes, and create new employment opportunities, we will support high value crops such as coconut, sandalwood , cocoa and cashew in our coastal areas. Agar trees in North East and nuts such as, almonds, walnuts and pine nuts in our hilly regions will also be supported. India is the world’s largest producer of coconuts. About 30 million people, including nearly 10 million farmers, depend on coconuts for their livelihood. To further enhance competitiveness in coconut production, I propose a Coconut Promotion Scheme to increase production and enhance productivity through various interventions including replacing old and non-productive trees with new saplings/plants/varieties in major coconut growing States. A dedicated programme is proposed for Indian cashew and cocoa to make India self-reliant in raw cashew and cocoa production and processing, enhance export competitiveness and transform Indian Cashew and Indian Cocoa into premium global brands by 2030. Sandalwood is closely linked to India’s social and cultural heritage. Our Government will partner with State Governments to promote focused cultivation and post-harvest processing to restore the glory of the Indian Sandalwood ecosystem. To rejuvenate old, low-yielding orchards and expand high-density cultivation of walnuts, almonds and pine nuts , we will support a dedicated programme to enhance farmer incomes and in bringing value addition by engaging youth. Bharat-VISTAAR (Virtually Integrated System to Access Agricultural Resources) I propose to launch Bharat-VISTAAR—a multilingual AI tool that shall integrate the AgriStack portals and the ICAR package on agricultural practices with AI systems. This will enhance farm productivity, enable better decisions for farmers and reduce risk by providing customised advisory support. SHE-Marts for Rural Women-led Enterprises Building on the success of the Lakhpati Didi Programme, I propose to help women take the next step from credit-led livelihoods to being owners of enterprises. Self-Help Entrepreneur (SHE) Marts will be set up as community-owned retail outlets within the cluster level federations through enhanced and innovative financing instruments. Empowering Divyangjan Divyangjan Kaushal Yojana : IT, AVGC sectors, Hospitality and Food and Beverages sectors offer task-oriented and process-driven roles, which are suitable for Divyangjans . We will ensure dignified livelihood opportunities through industry-relevant and customized training specific to each divyang group. Divyang Sahara Yojana : T imely access to high-quality assistive devices for all eligible Divyangjans is a fundamental need. I propose to ( i ) support the Artificial Limbs Manufacturing Corporation of India ( ALIMCO) to scale up production of assistive devices, invest in R&D and AI integration, (ii) strengthen PM Divyasha Kendras and support setting up of Assistive Technology Marts as modern retail-style centres where Divyangjans and senior citizens can see, try and purchase assistive products. Reaffirming our commitment to Mental Health and Trauma Care There are no national institutes for mental healthcare in north India. We will therefore set up a NIMHANS-2 and also upgrade National Mental Health Institutes in Ranchi and Tezpur as Regional Apex Institutions. Emergencies expose families, particularly the poor and vulnerable, to unexpected expenditure. We will strengthen and increase these capacities by 50% in District Hospitals by establishing Emergency and Trauma Care Centres . Focus on the Purvodaya States and the North-Eastern Region Purvodaya : I propose the development of an integrated East Coast Industrial Corridor with a well-connected node at Durgapur, creation of 5 tourism destinations in the 5 Purvodaya States, and the provision of 4,000 e-buses. Buddhist Sites in North-Eastern Region: The North-Eastern Region is a civilizational confluence of Theravada and Mahayana/Vajrayana traditions. I propose to launch a Scheme for Development of Buddhist Circuits in Arunachal Pradesh, Sikkim, Assam, Manipur, Mizoram and Tripura. The Scheme will cover preservation of temples and monasteries, pilgrimage interpretation centers , connectivity and pilgrim amenities. 16 th Finance Commission On 17 th November 2025, the 16 th Finance Commission submitted its report to the President. As mandated under Article 281 of the Constitution, the Government is to lay the Report along with the Explanatory Memorandum on the Action Taken Report on the recommendations of the Commission in Parliament. The Government has accepted the recommendation of the Commission to retain the vertical share of devolution at 41%. As recommended by the Commission, I have provided ₹1.4 lakh crore to the States for the FY 2026-27 as Finance Commission Grants. These include Rural and Urban Local Body and Disaster Management Grants. Fiscal Consolidation Government has been delivering on our fiscal commitments consistently without compromising on social needs. To strive towards accepted standards of fiscal management, in Budget 2025-26, I had indicated that the Central Government would target reaching a debt-to-GDP ratio of 50±1 percent by 2030-31. In line with this, the debt-to-GDP ratio is estimated to be 55.6 percent of GDP in BE 2026-27, compared to 56.1 percent of GDP in RE 2025-26. A declining debt-to-GDP ratio will gradually free up resources for priority sector expenditure by reducing the outgo on interest payments. One of the main operational instruments for debt targeting is the fiscal deficit. I am happy to inform this august House that I have fulfilled my commitment made in FY 2021-22 to reduce fiscal deficit below 4.5 percent of GDP by 2025-26. In RE 2025-26, the fiscal deficit has been estimated at par with BE of 2025-26 at 4.4 percent of GDP. In line with the new fiscal prudence path of debt consolidation, the fiscal deficit in BE 2026-27 is estimated to be 4.3 percent of GDP. Revised Estimates 2025-26 The Revised Estimates of the non-debt receipts are ₹34 lakh crore of which the Centre’s net tax receipts are ₹26.7 lakh crore. The Revised Estimate of the total expenditure is ₹49.6 lakh crore, of which the capital expenditure is about ₹11 lakh crore. Budget Estimates 2026-27 Coming to 2026-27, the non-debt receipts and the total expenditure are estimated as ₹36.5 lakh crore and ₹53.5 lakh crore respectively. The Centre’s net tax receipts are estimated at ₹28.7 lakh crore. To finance the fiscal deficit, the net market borrowings from dated securities are estimated at ₹11.7 lakh crore. The balance financing is expected to come from small savings and other sources. The gross market borrowings are estimated at ₹17.2 lakh crore. I will now move to Part B. PART B Direct Taxes Speaker Sir, Now I present my proposals on Direct Taxes. New Income Tax Act In July 2024, I announced a comprehensive review of the Income Tax Act, 1961. This was completed in a record time and the Income Tax Act, 2025 will come into effect from 1 st April, 2026. The simplified Income Tax Rules and Forms will be notified shortly, giving adequate time to taxpayers to acquaint themselves with its requirements . The forms have been redesigned such that ordinary citizens can comply without difficulty. Ease of Living I propose that any interest awarded by the Motor Accident Claims Tribunal to a natural person will be exempt from Income Tax, and any TDS on this account will be done away with . I propose to reduce TCS rate on the sale of overseas tour program package from the current 5 percent and 20 percent to 2 percent without any stipulation of amount. I propose to reduce TCS rate for pursuing education and for medical purposes under the Liberalized Remittance Scheme (LRS) from 5 percent to 2 percent. Supply of manpower services is proposed to be specifically brought within the ambit of payment to contractors for the purpose of TDS to avoid ambiguity. Thus, TDS on these services will be at the rate of either 1 percent or 2 percent only. I propose a scheme for small taxpayers wherein a rule-based automated process will enable obtaining a lower or nil deduction certificate instead of filing an application with the assessing officer. For the ease of taxpayers holding securities in multiple companies, I propose to enable depositories to accept Form 15G or Form 15H from the investor and provide it directly to various relevant companies. I propose to extend time available for revising returns from 31 st December to up to 31 st March with the payment of a nominal fee. I also propose to stagger the timeline for filing of tax returns. Individuals with ITR 1 and ITR 2 returns will continue to file till 31 st July and non-audit business cases or trusts are proposed to be allowed timetill 31 st August. TDS on the sale of immovable property by a non-resident is proposed to be deducted and deposited through resident buyer’s PAN based challan instead of requiring TAN. To address practical issues of small taxpayers like students, young professionals, tech employees, relocated NRIs, and such others, I propose to introduce a one-time 6-month foreign asset disclosure scheme for these taxpayers to disclose income or assets below a certain size. This scheme would be applicable for two categories of taxpayers namely, (A) who did not disclose their overseas income or asset and (B) who disclosed their overseas income and/or paid due tax, but could not declare the asset acquired. For category (A), the limit of undisclosed income/asset is proposed to be up to 1 crore rupees. They need to pay 30 percent of Fair Market Value of asset or 30 percent of undisclosed income as tax and 30 percent as additional income tax in lieu of penalty and would thereby get immunity from prosecution. For category (B), asset value is proposed to be up to 5 crore rupees. Here, immunity from both penalty and prosecution will be available with the payment of fee of 1 lakh rupees. Rationalizing Penalty and Prosecution Multiplicity of proceedings are a hindrance to the ease of doing business. I propose to integrate assessment & penalty proceedings by way of a common order for both. There will be no interest liability on the taxpayer on the penalty amount for the period of appeal before the first appellate authority irrespective of the outcome of appeal process. Further, quantum of pre-payment is being reduced from 20 percent to 10 percent and will continue to be calculated only on core tax demand. As an additional measure for reducing litigation, I propose to allow taxpayers to update their returns even after reassessment proceedings have been initiated, at an additional 10 percent tax rate over and above the rate applicable for the relevant year. The assessing officer will then use only this updated return in his proceedings. There is already a framework for immunity from penalty and prosecution in the cases of underreporting. I propose to apply this framework of immunity to misreporting too. However, in such a case the taxpayer will need to pay 100 percent of the tax amount as an additional income tax over and above the tax and interest due. Penalties for certain technical defaults such as failure to get accounts audited, non-furnishing of transfer pricing audit report and default in furnishing statement for financial transactions, are proposed to be converted into fee. I propose to rationalise prosecution framework under the Income Tax Act while maintaining a careful balance for deterrence in some serious offences. Non-production of books of account and documents, and requirement of TDS payment, where payment is made in kind, are being decriminalised. Further, minor offences will attract fine only. The remaining prosecutions will be graded commensurate with the quantum of offence. They will entail only simple imprisonment, with maximum imprisonment reduced to two years, and power to courts to convert even those into fine. There is no penalty presently for non-disclosure of non-immovable foreign assets with aggregate value less than 20 lakh rupees. I propose to also provide them with immunity from prosecution with retrospective effect from 1.10.2024. Cooperatives Deduction is already allowed to a primary cooperative society engaged in supplying milk, oilseeds, fruits or vegetables raised or grown by its members. I propose to extend this deduction to also include supply of cattle feed and cotton seed produced by its members. I propose to allow inter-cooperative society dividend income as deduction under the new tax regime to the extent it is further distributed to its members. I further propose to allow exemption for a period of 3 years, to dividend income received by a notified national co-operative federation, on their investments made in companies up to 31.1.2026. This exemption would be allowed only for dividends further distributed to its member co-operatives. Supporting IT sector as India’s growth engine India is a global leader in software development services, IT enabled services, knowledge process outsourcing services and contract R&D services relating to software development. These business segments are quite inter-connected with each other. All these services are proposed to be clubbed under a single category of Information Technology Services with a common safe harbour margin of 15.5 percent applicable to all. The threshold for availing safe harbour for IT services is being enhanced substantially from 300 crore rupees to 2,000 crore rupees. Safe harbour for IT services shall be approved by an automated rule-driven process without any need for tax officer to examine and accept the application. Once applied by an IT Services company, the same safe harbour can be continued for a period of 5 years at a stretch at its choice. For IT services companies who want to conclude Advance Pricing Agreement (APA), I propose to fast track Unilateral APA process for IT services and endeavour to conclude it within a period of 2 years. The period of 2 years can be extended by a further period of 6 months on taxpayer’s request. I propose to extend the facility of modified returns available to the entity entering APA to its associated entities also. Attracting global business and investment R ecognising the need to enable critical infrastructure and boost investment in data centres, I propose to provide tax holiday till 2047 to any foreign company that provides cloud services to customers globally by using data centre services from India. It will, however, need to provide services to Indian customers through an Indian reseller entity. I also propose to provide a safe harbour of 15 percent on cost in case the company providing data centre services from India is a related entity. To harness the efficiency of just-in-time logistics for electronic manufacturing, I propose to provide safe harbour to non-residents for component warehousing in a bonded warehouse at a profit margin of 2 percent of the invoice value. The resultant tax of about 0.7 percent will be much lower than in competing jurisdictions. To provide fillip to toll manufacturing in India, I propose to provide exemption from income tax for 5 years, to any non-resident who provides capital goods, equipment or tooling, to any toll manufacturer in a bonded zone. To encourage vast pool of global talent to work in India for a longer period of time, I propose to provide exemption to global (non-India sourced) income of a non-resident expert, for a stay period of 5 years under notified schemes. I propose to provide exemption from Minimum Alternate Tax (MAT) to all non-residents who pay tax on presumptive basis. Tax administration I propose to constitute a Joint Committee of Ministry of Corporate Affairs and Central Board of Direct Taxes for incorporating the requirements of Income Computation and Disclosure Standards (ICDS) in the Indian Accounting Standards ( IndAS ) itself. Separate accounting requirement based on ICDS will be done away with from the tax year 2027-28. To support PM Modi’s vision of home-grown accounting and advisory firms to become global leaders, I propose to rationalise the definition of accountant for the purposes of Safe Harbour Rules. Other Tax proposals Change in taxation of buyback was brought in to address the improper use of buyback route by promoters. In the interest of minority shareholders, I propose to tax buyback for all types of shareholders as Capital Gains. However, to disincentivize misuse of tax arbitrage, promoters will pay an additional buyback tax. This will make effective tax 22 percent for corporate promoters. For non-corporate promoters the effective tax will be 30 percent. TCS rate for sellers of specific goods namely alcoholic liquor, scrap and minerals will be rationalized to 2 percent and that on tendu leaves will be reduced from 5 percent to 2 percent. I propose to raise the STT on Futures to 0.05 percent from present 0.02 percent. STT on options premium and exercise of options are both proposed to be raised to 0.15 percent from the present rate of 0.1 percent and 0.125 percent respectively . We reformed the taxation landscape for corporates in 2019 by providing them a simplified regime with lower tax rate so that they could productively focus on business rather than on claim of deductions and exemptions. To encourage companies to shift to the new regime, set-off of brought forward MAT credit is proposed to be allowed to companies only in the new regime. Set-off using available MAT credit is proposed to be allowed to an extent of 1/4 th of the tax liability in the new regime. MAT is proposed to be made final tax. So, there will be no further credit accumulation from 1 st April 2026. In line with this change, the rate of final tax is being reduced to 14 percent from the current MAT rate of 15 percent. The brought forward MAT credit of taxpayers accumulated till 31 st March 2026, will continue to be available to them for set-off as above. Indirect Taxes I shall now take up proposals related to Indirect Taxes. My proposals for Customs and Central Excise aim to further simplify the tariff structure, support domestic manufacturing, promote export competitiveness, and correct inversion in duty. Review of exemptions and tariff simplification To continue weeding out long continuing customs duty exemptions, I propose to remove certain exemptions on items which are being manufactured in India or where the imports are negligible. Similarly, to further simplify the process of ascertaining the rate of duty applicable on a particular item, I propose to incorporate certain effective rates in various customs notifications to the tariff schedule itself. I shall now take up sector specific proposals. Promotion of exports of marine, leather, and textile products I propose to increase the limit for duty-free imports of specified inputs used for processing seafood products for export, from the current 1 per cent to 3 per cent of the FOB value of the previous year’s export turnover. I also propose to allow duty-free imports of specified inputs, which is currently available for exports of leather or synthetic footwear, to exports of Shoe Uppers as well. I propose to e xtend the time period for export of final product from the existing 6 months to 1 year, for exporters of leather or textile garments, leather or synthetic footwear and other leather products. Energy transition and security I propose to extend the basic customs duty exemption given to capital goods used for manufacturing Lithium-Ion Cells for batteries, to those used for manufacturing Lithium-Ion Cells for battery energy storage systems too. I propose to exempt basic customs duty on import of sodium antimonate for use in manufacture of solar glass. Nuclear Power I propose to e x tend the existing basic customs duty exemption on imports of goods required for Nuclear Power Projects till the year 2035 and expand it for all nuclear plants irrespective of their capacity. Critical Minerals It is proposed to provide basic customs duty exemption to the import of capital goods required for processing of critical minerals in India. Biogas blended CNG I propose to exclude the entire value of biogas while calculating the Central Excise duty payable on biogas blended CNG. Civil and Defence Aviation I propose to exempt basic customs duty on components and parts required for the manufacture of civilian, training and other aircrafts. It is proposed to exempt basic customs duty on raw materials imported for manufacture of parts of aircraft to be used in maintenance, repair, or overhaul requirements by Units in the Defence sector. Electronics To deepen value addition in the consumer electronics sector, I propose to exempt basic customs duty on specified parts used in the manufacture of microwave",
            "col4": "DATA"
        },
        {
            "col1": "P1_Part_2",
            "col2": "budget_speech.docx",
            "col3": "ovens. Special Economic Zone To address the concerns arising about utilization of capacities by manufacturing units in the Special Economic Zones due to global trade disruptions, I propose, as a special one-time measure, to facilitate sales by eligible manufacturing units in SEZs to the Domestic Tariff Area (DTA) at concessional rates of duty. The quantity of such sales will be limited to a prescribed proportion of their exports. Necessary regulatory changes will be undertaken to operationalise these measures while ensuring level-playing field for the units working in the DTA. Ease of Living To rationalize the customs duty structure for goods imported for personal use, I propose to reduce the tariff rate on all dutiable goods imported for personal use from 20 per cent to 10 per cent. To provide relief to patients, particularly those suffering from cancer, I propose to exempt basic customs duty on 17 drugs or medicines. I also propose to ⁠ add 7 more rare diseases for the purposes of exempting import duties on personal imports of drugs, medicines and Food for Special Medical Purposes (FSMP) used in their treatment. Customs Process India's role and share in global trade is poised for a major leap, in line with our ambition and journey towards 'Viksit Bharat' . In this regard, I propose many measures for custom processes to have minimal intervention for smoother and faster movement of goods and greater certainty to the trade. Trust-based systems I propose to enhance duty deferral period for Tier 2 and Tier 3 Authorised Economic Operators, known as AEOs, from 15 days to 30 days. I propose to provide eligible manufacturer-importers the same duty deferral facility. This should encourage them to get themselves accredited as a full-fledged Tier 3- AEO in due course. To provide greater certainty and for better business planning, I propose to extend validity period of advance ruling, binding on Customs, from the present 3 years to 5 years. In the spirit of whole-of-the-government approach, Government agencies will be encouraged to leverage AEO accreditation for preferential treatment in clearing their cargo. Regular importers with trusted longstanding supply chains will be recognized in the risk system, so that the need for verification of their cargo every time can be minimized. Export cargo using electronic sealing will be provided through clearance from the factory premises to the ship. For import of goods not needing any compliance, filing of bill of entry by a trusted importer, and arrival of goods will automatically notify Customs for completing their clearance formalities. This will enable goods to be released immediately on arrival. The Customs warehousing framework will be transformed into a warehouse operator-centric system with self-declarations, electronic tracking and risk-based audit. These reforms will move away from the current system of officer-dependent approvals, and reduce transaction delays and compliance costs. Ease of Doing Business Approvals required for cargo clearance from various Government agencies will be seamlessly processed through a single and interconnected digital window by the end of the financial year. Processes involved in clearance of food, drugs, plant, animal & wild life products, accounting for around 70 percent of interdicted cargo, will be operationalised on this system by April 2026 itself. For goods not having any compliance requirement, clearance will be done by Customs immediately after online registration is completed by the importer, subject to the payment of duty. Customs Integrated System (CIS) will be rolled out in 2 years as a single, integrated and scalable platform for all the customs processes. Utilization of non-intrusive scanning with advanced imaging and AI technology for risk assessment will be expanded in a phased manner with the objective to scan every container across all the major ports. New export opportunities To support Indian fishermen to fully harness the economic value of marine resources beyond our territorial waters, the following measures will be taken. Fish catch by an Indian fishing vessel in Exclusive Economic Zone (EEZ) or on the High Seas will be made free of duty. Landing of such fish on foreign port will be treated as export of goods. Safeguards will be put in place to prevent misuse during fish catch, transit and transshipment. To support aspirations of India’s small businesses, artisans and start-ups to access global markets through e-commerce , I am pleased to announce complete removal of the current value cap of ₹10 lakh per consignment on courier exports. In addition, handling of rejected and returned consignments will be improved with effective use of technology for identifying such consignments. Ease of Living I propose to revise provisions governing baggage clearance during international travel to address genuine concerns of passengers. The revised rules will enhance duty-free allowances in line with the present-day travel realities and provide clarity in temporary carriage of goods brought in or taken out. There are honest taxpayers who are willing to settle disputes by paying all their dues. But they get deterred due to negative connotation associated with penalty. They will now be able close cases by paying an additional amount in lieu of penalty. Honourable Speaker Sir, with this, I commend the Budget to this august House. Jai Hind! Annex ure to Part A Indicative Terms of Reference of the High-Level Education-to-Employment and Enterprise Standing Committee i dentify services sub-sectors with potential for growth, employment and exports , identify sector-specific gaps and measures to unlock employment potential; identify c ross-sectoral policy and regulatory issues, including standards-setting and accreditation; examine areas for services export; assess t he impact of emerging technologies, including AI, on jobs and skill requirements; propose specific measures for embedding AI in the education curriculum from school level onwards and upgrading State Councils of Educational Research and Training institutes for teacher training; propose measures for upskilling and re-skilling of technology professionals/engineers in AI and emerging technologies; and Propose measures for AI enabled matching of worders, jobs and training opportunities; and propose measures to make the informal workflow visible, verifiable and future-ready, to enhance upward mobility prospects; and propose steps to be taken to attract skilled diaspora and foreign talent into the country. Annexure to Part B Amendments relating to Direct Taxes EASE OF LIVING ( i ) Exemption of interest on compensation amount awarded by Motor Accident Claims Tribunal (MACT) In order to alleviate the sufferings of victims of motor vehicle accident and their family, it is proposed that any interest awarded on compensation amount in the case of individual awarded by MACT shall be exempt. It is also proposed that in case of an individual, no tax shall be deducted at source for such interest, irrespective of the amount of interest awarded by MACT. (ii) Removal of ambiguity on the application of rate of TDS on account of supply of manpower It is proposed to include supply of manpower within the definition of “work” under section 402(47) of the Income-tax Act, 2025 so as to provide that tax on such supply of manpower shall be deducted at source as “payment to contractors” under the provisions of section 393(1) [Table: Sl. No. 6( i ) and (ii)] and not under the provisions of “fee for professional services” under section 393(1) [Table: Sl. No. (iii)]. (iii) Enabling electronic verification and issuance of certificate for deduction of income-tax at lower rate or no deduction of income-tax I t is proposed to ease the compliance burden of small taxpayers by providing an online option to the payee, to apply for issuance of certificate for lower or nil deduction of income-tax which is proposed to be issued online after electronic verification. (iv) Enabling filing of declaration for no deduction of tax at source under section 393(6) of the Income-tax Act, 2025 to the depository It is proposed to allow filing of the declaration by a taxpayer for no deduction of tax at source, to a depository, where income is of the nature: ( i ) income from units of a mutual fund (ii) interest income from securities and (iii) dividends . This will address the present requirements of an investor to file separate declaration to different payers. The depositories shall in turn report such declarations to the person responsible for payment of such income It is also proposed that the person responsible for paying such income shall furnish the declaration received by it from the taxpayer to the Department on quarterly basis rather on monthly basis as at present. (v) Extending time to file revised return or belated return. Presently revised return can be filed upto 31 st December following the tax year. Return filing period extends upto October 31 st for persons engaged in international transactions under section 92E. In this regard, it is proposed to allow extending the time of filing revised return upto 31 st March following the tax year. This revised return can be of original return or belated return. A nominal fee of Rs. 1000 or 5000 is also proposed where the revision of original or belated return is made after 31 st December depending upon whether the income is upto or more than Rs. 5 lakh . (vi) Change in due date of filing Income-tax Return for non-auditable business and trusts It is proposed to provide staggered time line for filing of tax returns due on the 31 st of July. Individuals filing ITR 1 and ITR 2 shall continue to file tax returns by the 31 st July and for non-audit business cases or trusts, 31 st August shall be the due date. (vii) Reduction of compliance on sale of immovable property by non-resident to resident individual or HUF It is proposed to provide that resident individual or HUF, shall not be required to obtain tax deduction and collection account number (TAN) to deduct tax at source in respect of any consideration on transfer of any immovable property by non-resident under section 393(2) [Table Sl. No. 17]. Instead, the deduction shall be reported by quoting the PAN in same manner as transaction of similar nature between two residents. (viii) Rationalising the due date to deposit employee contribution by the employer to claim such contribution as deduction It is proposed that deduction of any amount of contribution received by the assessee being an employer, from an employee, shall be allowed as deduction in the hands of the assessee if such amount is credited by the assessee to the account of the employee, in any provident fund or superannuation fund or any fund set up under the provisions of the Employees’ State Insurance Act, on or before the due date of filing of his return of income under section 263(1) of the Act. (ix) Rationalising the provision related to computation of profits and gains of an insurance business other than life insurance business It is proposed that any amount which had earlier been added to the income of non-life insurance business, as tax was not deducted or paid as per the provisions of section 35(b)( i ) or (ii) of the Act, shall be allowed as deduction in the tax year in which tax is deducted or paid as per the provisions of section 35(b)( i ) or (ii) of the Act. (x) Introduction of Foreign Assets of Small Taxpayers – Disclosure Scheme (FAST – DS ) , 2026 It is proposed to introduce a time-bound scheme for declaration of foreign assets and foreign sourced income for taxpayers involving amounts below certain threshold. RATIONALISING PENALTY AND PROSECUTION ( i ) Reduction in multiplicity of proceedings Presently, penalty proceedings are initiated after completion of the assessment proceedings. This takes a long time for finalising issues emerging out of any assessment. Multiplicity of proceedings increases number of pending appeals, cost of litigation and compliance. With a view to provide fast-track settlement of disputes, it is proposed to integrate assessment & penalty proceedings by way of a common order after providing reasonable opportunity to the taxpayer to explain the issue. To provide relief to taxpayers on account of increase in quantum of demand which may arise as a result, it is proposed that the interest on penalty would be kept in abeyance during the pendency of appeal before first appellate authority. (ii) Immunity from penalty from underreporting in consequence of misreporting of income and prosecution in such cases There are broadly two kind of penalties – underreporting of income due to mistakes or oversight where penalty is 50% on tax amount and framework to underreporting of income in consequence of misreporting of income underreporting in consequence of misreporting of income on account of giving wrong or faulty information or misrepresenting the type of income where the penalty is 200% on tax amount. There is already a framework for immunity from penalty and prosecution, under section 478 and 479 of the Act, if the penalty is initiated for underreporting of income. In this regard, it is proposed to extend the same on payment of 100% of tax amount as additional income-tax. However, the misreporting of income in respect of unexplained cash credit, etc. is proposed to be settled with a payment of 120% of the tax. In such cases immunity shall not be granted where prosecution is initiated as per provision of chapter-XXII of the Act. (iii) Conversion of penalty to fee It is proposed to convert ( i ) penalty for failure to get accounts audited, (ii) penalty for non-furnishing of TP report and (iii) penalty for default in furnishing statement for financial transactions or reportable accounts into fee to be charged per day of the default subject to a maximum ceiling. (iv) Rationalization of the prosecution framework It is proposed that production of books of account and documents, and requirement of ensuring payment of TDS from the deductee where payment is made in kind, be completely decriminalised. It is further proposed that all prosecutions shall be rationalised to simple imprisonment instead of rigorous imprisonment. Maximum punishment for any offence (except for repeated offence) is proposed to be reduced to 2 years instead of 7 years. In cases where presently the maximum punishment is two years, the punishment has been reduced to 6 months with or without fine and with no minimum imprisonment. It is further proposed that prosecution for the offences under Income-tax Act, 2025 shall be based on the amount of tax evaded and the punishment shall be proportionate to the gravity of crime. In such cases, the requirement of maximum punishment of imprisonment has been done away with apart from relaxing the requirement of mandatory fine to optional. It is further proposed, for minor offence, only fine shall be provided as a punishment. (v) Rationalising the tax rate for special income charged under section 195 of the Act Presently there is special tax rate on certain incomes like income in the nature of cash credits, unexplained investments, etc. The tax rate is 60% and penalty is 10% of tax. It is proposed to rationalise the tax rate to 30% on these incomes. Penalty on such amount would be merged with penalty for underreporting of income in consequence of misreporting of income that is 200% of tax amount. (vi) Relaxation of search assessment in case of person other than the searched person in certain situation Provisions for assessment in search cases was introduced by Finance (No. 2) Act, 2024. In the new scheme, where incriminating material pertaining to other person, relates only to a single tax year, the other person is also required to undergo the full block assessment procedure, resulting in an increased compliance burden on such person against whom no search or requisition was initiated. It is proposed to limit the period of block in case of other person, where the undisclosed income of the other person pertains only to one tax year. The definition of block period is accordingly proposed to be amended in such cases. (vii) Time limit to complete search assessment It is also proposed to amend section 296 of the Income-tax Act, 2025 so as to take the date of initiation of search as the reference point to decide the date of limitation for block assessment and consequently, the period of twelve months is proposed to be extended to eighteen months in the case of specified person. (viii) Extending the scope of filing of updated return There is facility for updating tax returns where the taxpayer wants to show any additional income. This facility is available for a period of 4 years for an additional tax liability of 25%, 50%, 60%, 70% from the first to the fourth year after the relevant tax year when the return of income required to be filed for the first time. To provide additional measure for reducing litigation, it is proposed to allow the taxpayer to update the return even after reassessment proceedings have been initiated. The updation is proposed to be enabled at an additional 10% tax rate over and above the rate applicable for relevant year. It is further proposed to allow filing of updated return in cases where tax payer reduces the amount of loss filed in original return under section 263(1). It is further proposed that where the taxpayer files updated return and reports additional income then penalty shall not be leviable on such additional income. (ix) Immunity from prosecution under the Black Money Act Under the Black Money (Undisclosed Foreign Income and Assets) and Imposition of Tax Act, 2015, there is no penalty for non-disclosure of non-immovable assets with aggregate value less than twenty lakh rupees. It is proposed to extend this immunity for prosecution in such cases with retrospective effect from 1.10.2024. (x) Penalty provision for non-furnishing of statement or furnishing inaccurate information in a statement on transaction of crypto-assets To ensure compliance to the provisions of section 509 of the Income-tax Act, 2025 and create a deterrence for non-furnishing of statement or for furnishing inaccurate information in respect of crypto assets in such statement, it is proposed to introduce penalty provision. Penalty of Rs. 200 per day for non-furnishing of statement and Rs. 50,000 for furnishing inaccurate particulars and failure to correct such inaccuracy is proposed to be levied. COOPERATIVES ( i ) Deduction of profit and gains to a primary co-operative where they supply cattle feed and cotton seed to a federal co-operative. Deduction of profit and gains is presently allowed to a primary cooperative society engaged in supplying milk, oilseeds, fruits or vegetables raised or grown by its members to a federal cooperative society and others engaged in the same activities. It is proposed to extend this deduction to a primary co-operative engaged in supplying of cattle feed and cotton seed to, inter alia, a federal co-operative or government organizations. (ii) Deduction of inter-cooperative society dividend income under the new tax regime. The dividend received by a cooperative society from another cooperative society is allowed as a deduction in the old tax regime. Non-allowance of this deduction in the new tax regime may result in double taxation as it may be taxed in the hands of the members on further distribution by the cooperative societies. Therefore, it is proposed to allow the inter-cooperative society dividend income as deduction under the new tax regime to the extent it is further distributed to the members. (iii) Deduction of dividend income received by a notified national co-operative federation in the new tax regime. It is proposed to allow exemption to dividend income received by a notified national federal cooperative from a company for a period of three years. This deduction is limited to the dividend received on investments made till 31.1.2026. Further, this exemption would be allowed only to the extent that the dividends are further distributed to the members of the co-operatives. ATTRACTING GLOBAL BUSINESS AND INVESTMENT ( i ) Tax holiday up to 2047 to any foreign company who provides services by procuring data centre services in India R ecognising the need to enable critical infrastructure and boost investment in data centres, it is proposed to provide a tax holiday up to 2047 to any foreign company who provides services to any part of the world outside India by procuring data centre services in India. Sale of such services to Indian users shall be made through an Indian reseller entity and taxed appropriately. It is also proposed to provide a safe harbour of 15% to the resident entity providing data centre services to a related foreign company (who is providing cloud services to any part of the world outside India). (ii) Fillip to toll manufacturing engaged in manufacturing of electronic goods To provide fillip to toll manufacturing in India, it is proposed to provide exemption to any foreign company who provides capital goods, equipment and tooling to any toll manufacturer in a bonded zone who is engaged in manufacturing of electronic goods. The exemption is proposed for a period of five tax years beginning on 1 st April, 2026. (iii) Exemption to the global income (other than Indian sourced income) to an expert who visits India and stays for a longer period To enable vast pool of global talent to come and work in India for a longer period of time, there is a need to provide tax certainty to them that only their Indian sourced income will be taxed in India despite their long period of stay. Accordingly, it is proposed to provide exemption to the global income (other than Indian sourced income) to an expert who visits India and stays for a period of five years. The expert visiting India should have been a non-resident in the previous five years when he visits India and should be providing services under notified Government scheme. (iv) Exemption from MAT to non-residents availing presumptive taxation scheme Non-residents who avail presumptive scheme of taxation are exempt from applicability of Minimum Alternate Tax (MAT) provisions. It is proposed to extend such exemption from MAT to all non-residents who pay tax on a presumptive basis. (v) Incentivizing prospecting and exploration of critical minerals In order to incentivise the prospecting and exploration of the critical minerals, it is proposed to include certain critical minerals in the list of minerals in Schedule XII of the Act, thereby making expenditure on prospecting and exploring of such critical minerals eligible for deduction as per the provision of section 51 of the Act. (vi) Extension of period of deduction for units in IFSC and rationalization of tax rate To increase the competitiveness of IFSC, it is proposed to increase the period of deduction under section 147 to 20 consecutive years out of 25 years for units in IFSC and 20 consecutive years for OBUs. It is also proposed that the business income of these units from IFSC after the expiry of period of deduction will be taxed at rate of 15%. (vii) Rationalization of certain terms for treasury centres in IFSC It is proposed to rationalize the provisions of deemed dividend applicable to treasury centre in IFSC by providing that provisions of deemed dividend shall not be applicable if ( i ) the parent entity or the principal of the group shall be listed in a country or territory outside India; and (ii) such parent or principal entity and other group entity to the transaction is located in a country or territory outside India as may specified by the Central Government, by notification in the Official Gazette. RATIONALISATION OF CORPORATE TAX REGIME ( i ) Reduction of rate of Minimum Alternate Tax (MAT) and allowance of set-off of brought forward MAT credit to companies shifting to the new tax regime To enable companies to shift to the new regime, MAT is proposed to be made as a final tax and the corresponding rate is reduced from 15% to 14%. There shall be no allowance of credit in future tax years in respect of such payment. Further, the set-off of any brought forward MAT credit available from prior to tax year 2026-27 will only be allowed to domestic companies which shift henceforth to the new regime. This set-off of MAT credit brought forward as on 1/4/2026 is proposed to be allowed in the new tax regime to domestic companies to the extent of 25% of their tax liability. The brought forward MAT credit shall be available only up to fifteenth year from the year when the corresponding credit was first available. In the case of foreign companies, set off is proposed to be allowed to the extent of the difference between the tax on the total income and the minimum alternate tax, for the tax year in which normal tax is more than MAT. RATIONALISATION OF OTHER DIRECT TAX PROVISIONS ( i ) Rationalisation of share buyback It is proposed to provide that consideration received by a shareholder on buy-back shall be chargeable to tax under the head “Capital Gains” instead of being treated as dividend income. It is also proposed to provide for a differential rate for promoters wherein the effective rate on gains in buyback will be 22% for promoters which are domestic companies and 30% for promoters other than domestic companies. (ii) Rationalisation of tax collection at source (TCS) rates It is proposed to reduce multiplicity of TCS rates. Also, certain TCS rates are rationalised to address the cash flow issues on this account. Sl. No. Nature of receipt Current Rate Proposed Rate 1 Sale of alcoholic liquor for human consumption. 1%. 2%. 2 Sale of tendu leaves. 5%. 2%. 3 Sale of scrap. 1%. 2%. 4 Sale of minerals, being coal or lignite or iron ore. 1%. 2%. 5 Remittance under the Liberalised Remittance Scheme of an amount or aggregate of the amounts exceeding ten lakh rupees— ( a ) 5% for purposes of education or medical treatment; ( b ) 20% for purposes other than education or medical treatment. ( a ) 2% for purposes of education or medical treatment; ( b ) 20% for purposes other than education or medical treatment. 6 Sale of “overseas tour programme package” including expenses for travel or hotel stay or boarding or lodging or any such similar or related expenditure. ( a ) 5% of amount or aggregate of amounts up to ten lakh rupees; ( b ) 20% of amount or aggregate of amounts exceeding ten lakh rupees. 2% (iii) STT rate increase To provide reasonable course correction in F&O segment in the capital market and generate additional revenues for the Government, it is proposed to raise the STT on Futures to 0.05% from present 0.02%. STT on options premium and exercise of options is proposed to be raised to 0.15% from the present rate of 0.1% and 0.125% respectively. (iv) Capital Gains Exemption for Sovereign Gold Bonds It is proposed to provide that the exemption from capital gains tax in respect of Sovereign Gold Bonds shall be available only where such bonds are subscribed to by an individual at the time of original issue and are held continuously until redemption on maturity, It is also proposed to provide that this exemption applies uniformly to all issuances of Sovereign Gold Bonds by the Reserve Bank of India. (v) Rationalisation of Schedule XI relating to Recognised Provident Funds It is proposed to amend Schedule XI to rationalise the provisions relating to recognised provident funds by deleting parity-based and percentage-based limits on employer contributions, removing salary-linked relaxations and shareholder-based distinctions, aligning eligibility for recognition with exemption under section 17 of the Employees’ Provident Funds and Miscellaneous Provisions Act, 1952, and modifying investment-related provisions to remove rigid statutory caps inconsistent with prevailing EPFO norms. (vi) Removal of Interest Deduction against Dividend and Mutual Fund Income It is proposed to provide that no deduction shall be allowed in respect of any interest expenditure incurred in relation to dividend income or income from units of mutual funds, and to omit the existing provision permitting such deduction subject to a specified ceiling. (vii) Enabling provision to provide clarity on situations where an amount which has been claimed as deduction or which has not been added in the total income will be deemed income It is proposed that where any sum has been allowed as deduction or has not been included in the total income under the repealed Income-tax Act, 1961, such sum will be deemed to be income under Income-tax Act, 2025 , even without violations of any conditions, if it was to be included in the total income under the provisions of Income-tax Act, 1961 had it not been repealed. (viii) Rationalising the provisions related to tonnage tax scheme It is proposed to rationalise the tonnage tax scheme provisions to align it with the Inland Vessels Act, 2021 and rules made thereunder. (ix) Disability Pension for Armed Forces It is proposed to provide a specific exemption for disability pension granted to members of the Armed Forces including paramilitary personnel, covering both the service element and the disability element, where the individual has been invalided out of service on account of a bodily disability attributable to, or aggravated by, military, naval or air force service, and to exclude cases of retirement on superannuation or otherwise. (x) Exemption on income in respect of compulsory acquisition of any land under RFCTLARR Act In order to specifically provide exemption for acquisition of land under the provisions of Right to Fair Compensation and Transparency in Land Acquisition, Rehabilitation and Resettlement Act, 2013, it is proposed to provide exemption to an individual or a Hindu undivided family on any income in respect of any award or agreement made on account of compulsory acquisition of any land under the said Act (other than the award or agreement made under section 46 of said Act). (xi) Facility to the associated entity of the person entering into Advance Pricing Agreement (APA) to file modified return In Advance Pricing Agreement (APA), there is already a facility for the entity entering APA to file modified returns according to the agreement. It is proposed to extend the same facility to the associated entity of the person entering into agreement where it’s income also changes on account of the agreement. (xii) Amendments in the nature of clarifications There are certain legal issues in which there are differing judgement of courts. These relate to time-limit for assessment after Dispute Resolution Panel proceedings, time-limit for Transfer Pricing Officer order, Document Identification Number and issuance of notice for re-assessment by the Jurisdictional Assessing Officer. In this regard, it is proposed to clarify these issues in the Income-tax Act, 1961 and Income-tax Act, 2025 to provide certainty to the provisions. (xiii) Other minor modification in the Income-tax Act, 2025 It is proposed to provide definition of “commodity derivative” in section 66 of the Act. It is proposed to provide definition of “authorised person” in section 402 of the Act. It is proposed to correct referencing error in Note 3 of section 393(1) [Table: Sl. No. 3( i )] from [Table: Sl. No. 3(iii)] to [Table: Sl. No. 3( i )]. It is proposed to correct referencing error in section 99(2) from 99(1)(a)( i ) to section 99(1)(a)(ii). It is proposed to amend section 400(2) of the Income-tax Act, 2025 to align it with the intent of the provisions of Income-tax Act, 1961 and to provide that the guidelines issued by the Board under this section shall apply to income-tax authorities as well as the person liable to deduct or collect, as the case may be, income-tax. It is proposed to amend sections 58, 162, 164, 165, 202 and 270 of the Income-tax Act, 2025 to remove duplicate reference to both the section 144 and Chapter VIII, as chapter VIII already includes section 144. It is proposed to amend schedule VI [Note 1(g)] to align the definition of the specified fund as provided in Schedule VI [Note 1(g)] of the Income-tax Act, 2025 with the provisions of section 10(4D) of the Income-tax Act, 1961. It is proposed to amend section 352(4) [Table: Sl. No. 8] to align the provisions relating to accreted income in the case of merger of registered NPOs with any other entity other than a registered non-profit organisation (NPO) or with any other registered NPO having the same or similar objects but the said merger does not fulfil such conditions as may be prescribed or with a registered NPO that does not have same or similar objects. It is proposed to insert a new section on the lines of section 12AC of the Income-tax Act, 1961 to allow the merger of the registered NPO with any other registered NPO having the same or similar objects if the said merger fulfils such conditions, as may be prescribed. It is proposed to amend section 351 of the Income-tax Act, 2025 to remove the reference of section 346 of the Income-tax Act, 2025 in said section to align the provision with the Income-tax Act, 1961 It is proposed to amend section 332(1)(f) of the Income-tax Act, 2025 to remove the reference of Schedule VII [Table: Sl. No. 10 to 16] in said section so that such funds may not be required to register themselves as the registered NPO. It is proposed to amend section 349 to enable the filing of belated return by the registered NPO. It is proposed to change the annual value of property or part thereof to be treated as nil from “for two years” to “ upto two years”. It is proposed to amend section 22(2) of the Income-tax Act, 2025 so as to provide that aggregate amount of deduction for interest on borrowed capital shall be inclusive of prior-period interest payable. It is proposed to amend section 262(10)(c) to enable Central Board of Direct Taxes (CBDT) to make rules for quoting of Permanent Account Number in such documents which does not relate to business or profession. Annexure to Part B Amendments relating to Indirect Taxes A. LEGISLATIVE CHANGES IN CUSTOMS LAWS A.1 Amendments in the Customs Act, 1962 Sub-section (2) of section 1 of the Customs Act, 1962 is being amended to extend the jurisdiction of the said Act beyond the territorial waters of India, for the purpose of fishing and fishing related activities. In section 2, a new clause is being inserted to define the expression ‘Indian-flagged fishing vessel’. Sub-section (6) of section 28 is being amended so as to provide that the penalty paid under sub-section (5) of section 28, on determination under sub-section (6) thereof, shall be deemed to be a charge for non-payment of duty. Sub-section (2) of section 28J is being amended so as to provide that advance ruling under sub-section (1) of that section shall remain valid for a period of five years or till there is a change in law or facts on the basis of which the advance ruling has been pronounced, whichever is earlier. The proviso to the said sub-section is also being substituted so as to provide that in respect of any advance ruling in force on the date on which the Finance Bill, 2026 receives the assent of the President, the Authority shall, upon a request by the applicant, extend the validity of the ruling for five years from the date of the ruling. A new section 56A is being inserted to provide special provisions for fishing and fishing related activities by an Indian-flagged fishing vessel beyond territorial waters of India. It also provides that fish harvested beyond the territorial waters of India may be brought into India free of duty and to treat fish that has landed at foreign port as export of goods in such manner as may be provided by rules. It also provides to make regulations to provide for the form and manner of making an entry in respect of fish harvested by an Indian-flagged fishing vessel including its declaration, custody, examination, assessment of duty, clearance, transit or transhipment . (vii) In the Customs Act, for section 67, the following section shall be substituted, namely: - “67. The owner of any warehoused goods may remove them from one warehouse to another, subject to such conditions as may be prescribed.”. The proposed section seeks to do away with the requirement of prior permission of the proper officer under the said section for removal of warehoused goods from one custom bonded warehouse to another. (viii) In section 84 of the Customs Act, in clause (b), for the words “ the examination ”, the words “ the custody, examination ” shall be substituted. The amendment seeks to enable the Board to make provisions for the custody of goods imported or to be exported under the regulations framed under this section. These changes shall come into effect from date of assent to the Finance Bill, 2026. A.2 Amendments in the Customs Tariff Act, 1975 ( i ) The First Schedule to the Customs Tariff Act, 1975 is being amended to, add 148 new tariff entries in chapter 03, 08, 12, 13, 20, 21, 22, 25,26, 28, 29, 33, 39, 41, 47, 48, 73, 81, 84, 85, 86, substitute/delete 54 tariff entries and amend 2 tariff entries in chapter 29 and 1 tariff entry in chapter 85. (ii) Further the tariff rate on certain tariff items is being modified with effect from 02.02.2026 and the rates on certain other tariff items are also being modified as part of rate rationalisation with effect 01.04.2026. In respect of various goods, to operationalize effective basic customs duty rates from First Schedule instead of exemption notifications, new tariff lines have been created and the rate of duty for certain existing tariff lines has been amended. This is a simplification exercise and there is no change in the applicable rate of duties on such goods. New tariff lines are also being introduced to help in better identification of the coated pipes for exports, getting actual transaction data of precursor chemicals and help in their effective monitoring, facilitating, tracking exports and deciding policy measures for Plant based extract products, and providing boost to eco-friendly industries For details of the above, the Explanatory Memorandum to the Finance Bill, 2026, may be referred. These changes will come into effect from 01.05.2026, unless otherwise specified. A.3 Amendment to Rules and Regulations under Customs Act, 1962 In addition to the above, new Baggage Rules are being issued for international passengers. Deferred duty payment is being made monthly and a new class of eligible importers is being created by amending the existing Deferred Payment of Import Duty Rules, 2016. Cap on the value of goods exported through courier is being removed by amending Courier Regulations. Further, to ease e-commerce, the returns and rejects procedures in courier is being relaxed. To decongest the courier terminals and to improve the overall efficiencies and ease of import and export, the Courier Regulations are being amended to allow for return to origin. B. LEGISLATIVE CHANGES IN GST LAWS [ Save as otherwise provided, these changes will be brought into effect from a date to be notified in coordination with States, as per recommendations of the GST council] B.1 Amendments in provisions related to post-sale discounts under section 15 of the CGST Act, 2017 Sub-section (3) of section 15 of the Central Goods and Services Tax Act, 2017 is being amended to do away with the requirement of linking the post-sale discount with an agreement and to refer to issuance of credit note under section 34 where the input tax credit is to be reversed by the recipient. B .2 Amendments in Section 34 of the CGST Act, 2017 Section 34 of the Central Goods and Services Tax Act, 2017 is being amended so as to include the reference of section 15 in the said section. B.3 Amendments in Section 54 of the CGST Act, 2017 to provide for provisional refunds on account of inverted duty structure Sub-section (6) of Section 54 of the Central Goods and Services Tax Act, 2017 is being amended to extend the provisions of provisional refund to refunds arising out of inverted duty structure. B.4 Amendments in Section 54 of the CGST Act, 2017 to provide for refund of tax below Rs. 1000/- Sub-section (14) of Section 54 of the Central Goods and Services Tax Act, 2017 is being amended to remove the threshold limit for sanction of refund claims in case of goods exported out of India with payment of tax. B.5 Amendments in Section 101A of the CG ST Act, 2017 Sub-section (1A) is being inserted in Section 101A of the Central Goods and Services Tax Act, 2017 to provide that the Central Government may, pending the constitution of the National Appellate Authority, by notification empower an existing Tribunal for hearing appeals under section 101B of the CGST Act, 2017; and to provide that the provisions of sub-sections (2) to (13) shall not be applicable where a Tribunal has been so empowered under sub-section (1A). This change shall come into effect from the 1 st day of April, 2026. B.6 Amendments in Section 13 of the I GST Act, 2017 to provide for place of supply for intermediary services Clause (b) of sub-section (8) of section 13 of the Integrated Goods and Services Tax Act, 2017 is being omitted so as to provide that the place of supply for \"intermediary services\" will be determined as per the default provision under section 13(2) of the IGST Act. C. CUSTOMS DUTY RATE CHANGES C.1 To reduce input costs, boost domestic manufacturing and promote export competitiveness, customs duty on the following items are being reduced [with effect from 02.02.2026, unless otherwise specified] S. No. Commodity Rate of Basic Customs Duty From (per cent) To (per cent) I Critical Minerals 1. Monazite 2.5% Nil II. Renewable Energy 2. Sodium antimonate for use in manufacture of solar glass 7.5% Nil 3. Specified capital goods for use in manufacture of lithium-ion cells for batteries of Battery Energy Storage System As applicable Nil III. Nuclear Energy 4. All goods for generation of nuclear power falling under tariff item 8401 30 00 7.5% Nil 5. Control and Protector Absorber Rods, Burnable Absorber Rods for generation of nuclear power falling under tariff item 8401 40 00 7.5% Nil 6. Goods required for the setting up of specified Nuclear Power Projects, irrespective of their capacity, where the projects have been registered with the Customs Houses concerned on or before 30 th September 2035 in compliance with the Project Import Regulations, 1986 As applicable Nil IV. Electronics 7. Specified goods for use in the manufacture of Microwave Ovens falling under tariff item 8516 50 00 As applicable Nil V. Civil Aviation 8. Components or parts including engines, of aircraft, for manufacture of aircraft and parts of the aircraft As applicable Nil VI. Defence Sector 9. Raw materials for manufacture of parts of aircraft for maintenance, repair, or overhauling of aircraft or components or parts of aircraft, including engines, when imported by Public Sector Units under the Ministry of Defence As applicable Nil VII. Drugs/ Medicines 10. 17 new drugs/medicines to be added in List 3 appended to TABLE I of notification No. 45/2025-Customs dated 24.10.2025 5%/10% Nil 11. 7 rare diseases that are part of National Policy for Rare Disease (NPRD), 2021 to be added in List 22 appended to TABLE I of notification No. 45/2025-Customs dated 24.10.2025 for customs duty exemption on drugs, medicines and food for special medical purposes, when imported for personal use As applicable Nil VIII. Personal Imports (with",
            "col4": "DATA"
        },
        {
            "col1": "P1_Part_3",
            "col2": "budget_speech.docx",
            "col3": "effect from 01.04.2026) 12. All dutiable goods, imported for personal use under Chapter heading 9804 10%/20% 10% Note: Description of entries is indicative. Notification/Tariff may be referred for complete description. C.2 Increase/Modification in Customs duty [with effect from 02.02.2026] S. No. Commodity Rate of Basic Customs Duty From (per cent) To (per cent) I. Chemicals 1. Potassium hydroxide Nil 7.5% II. Umbrella and its parts 2. Umbrellas (other than garden umbrellas) covered under tariff items 6601 9100 and 6601 9900 20% 20% or Rs. 60 per piece, whichever is higher 3. Parts, trimmings and accessories of articles of heading 6601 to 6602 10% 10% or Rs. 25 per kg., whichever is higher Note: Description of entries is indicative. Notification/Tariff may be referred for complete description. C.3 Review of exemptions A review of the existing customs notifications providing concessional BCD rates to various goods was undertaken. Consequently, upon review, the following exemptions are being lapsed. C.3.1 To lapse with effect from 02.02.2026 S. No. S. No. of Table I in 45/2025-Customs Description 1 Animals and birds imported by Zoo 113 Alpha pinene 123 Artificial plasma 128 Ammonium phosphate or ammonium nitro-phosphate, for use as manure or for the production of complex fertilisers 132 Potassium sulphate, containing not more than 52% by weight of potassium oxide* 137 Other diagnostic or laboratory reagents falling under tariff heading 3822 90 90* 213 INVAR 258 Coffee roasting, brewing or vending machines for use in the manufacture or processing of coffee 285 Parts of radio trunking terminals 287 CD-ROMs containing books of an educational nature, journals, periodicals (magazines) or newspapers 310 Loco simulators Note: Description of entries is indicative. Notification may be referred for complete description. * no change in the effective rate of duty on the goods C.3.2 To lapse with effect from 01.04.2026 S. No. S. No. of notification No. 45/2025-Customs Description In Table I of notification No. 45/2025-Customs 93 Naphtha, for use in the manufacture of fertilisers 95 LPG, consumed in the manufacture of polyisobutylene by DTA unit received from SEZ unit and returned by DTA to SEZ from where such LPG were received. 107 Silicon for the manufacture of un-diffused silicon wafers, and un-diffused silicon wafers for the manufacture of solar cells or solar cell modules** 117 Maltol, for use in the manufacture of deferiprone 145 Specified goods in manufacture of Copper-T contraceptives 154 Ethylene – Propylene – Non-Conjugated Diene Rubber (EPDM) for use in the manufacture of insulated wires and cables 172 Hydrophilic and hydrophobic non- woven fabrics, imported for use in the manufacture of adult diapers 201 Spent catalyst or ash containing precious metals 218 Metal parts for use in the manufacture of electrical insulators 219 Pipes and tubes for use in manufacture of boilers 231 Permanent magnets for manufacture of synchronous generators above 500KW for use in wind operated electricity generators 236 Zeolite for use in the manufacture of wash coat for catalytic converters 243 High speed cold-set or high speed heat set web offset printing machines along with mail room equipment 271 Cash dispenser or automatic banknote dispenser and its parts and components 275 Television equipment, cameras and other equipment for taking films, imported by a foreign film unit or television team 276 Photographic, filming, sound recording of foreign origin, if imported into India after having been exported there from 291 Parts and components of digital still image video cameras 309 Raw materials or parts for use in manufacture of e-Readers 370 X-Ray tubes used in manufacture of X-ray machines for medical, surgical or veterinary use 372 Flat panel detector for use in manufacture of X-Ray machine for medical, surgical or veterinary use 397 Parts of video games for the manufacture of video games In Table IV of notification No. 45/2025-Customs 1 Motion pictures, music, gaming software for use on gaming consoles printed or recorded on media In notification No. 113/2003-Customs dated 22 July 2003 Exemption to castor oil cake and castor de-oiled cake manufactured from indigenous castor oil seeds on indigenous plant and machinery by unit in SEZ and brought to DTA Note: Description of entries is indicative. Notification may be referred for complete description. ** no change in the effective rate of duty on the goods D. Export Promotion Measures D.1 Increase in value limit of duty-free import of specified goods for use in processing of sea-food The value limit of duty-free imports of specified goods imported for use in processing of sea-food has been increased from 1% to 3% of the FOB value of seafood products exported during the preceding financial year. D.2 Increase in time period of export for exporters of specified goods manufactured from duty-free imported inputs The time period of export of textile/leather garments, leather/synthetic footwear or any other leather product by exporters is being extended from six months to twelve months. D.3 Extension of duty exemption on inputs for manufacture of specified goods for export The benefit of duty exemption on inputs for manufacture of leather/synthetic footwear for export is being extended to exporters of shoe-uppers also. E. Central Excise Duty Changes E.1 Exemption of Central Excise Duty on value of Biogas/Compressed Biogas (CBG) contained in Blended Compressed Natural Gas (CNG) The value of Biogas/Compressed Biogas (CBG) and the appropriate Central Tax, State Tax, Union Territory Tax or Integrated Tax, as the case may be, paid on such Biogas or CBG contained in blended CNG, is being excluded from the transaction value for the purpose of computation of central excise duty on such blended CNG. E.2 Amendment to Seventh Schedule to the Finance Act, 2001 to revise the National Calamity Contingent Duty (NCCD) Rate with effect from 01.05.2026 with no change in the effective duty rate*** S.No . Description NCCD Rate From (per cent) To (per cent) 1. Chewing tobacco 25% 60% 2. Jarda scented tobacco 25% 60% 3 . Other 25% 60% Note: Description of entries is indicative. Notification/Tariff may be referred for complete description. ***effective duty rate will be maintained at 25% by notification F. Others There are a few other changes. For details of the budget proposals, the Explanatory Memorandum and other relevant budget documents may be referred to.",
            "col4": "DATA"
        }
    ],
    "projectDeleted": false,
    "codeFrameColumn": [
        {
            "col1": "col1",
            "col2": "col3",
            "researchObjective": "",
            "filterColumn": [],
            "netSetting": true,
            "subNetSetting": true,
            "categorySetting": true,
            "sentimentSetting": true,
            "thinkingBudgetSetting": 2000,
            "questionType": "OPENEND",
            "otherBrand": false,
            "existingCodeFrame": [],
            "isList": false,
            "codeFrameWithDataList": [],
            "codeFrameList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 21105,
                        "CandidatesTokenCount": 5681,
                        "TotalTokenCount": 28344,
                        "thoughtsTokenCount": 1558,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Net": "Economic Strategy & Fiscal Management",
                                "Subnets": [
                                    {
                                        "Subnet": "Overall Vision & Principles",
                                        "Categories": [
                                            {
                                                "Category": "Viksit Bharat Initiative",
                                                "Summary": "The government's overarching vision to transform India into a developed nation, balancing ambition with inclusion."
                                            },
                                            {
                                                "Category": "Governance Philosophy",
                                                "Summary": "Emphasizes action over ambivalence, reform over rhetoric, and people over populism, guided by 'Atmanirbharta'."
                                            },
                                            {
                                                "Category": "Sankalp & Kartavya Approach",
                                                "Summary": "Focuses on the poor, underprivileged, and disadvantaged, driven by three 'kartavya' (duties): accelerate growth, fulfill aspirations, and ensure equitable access."
                                            }
                                        ],
                                        "Summary": "Outlines the government's foundational economic philosophy, long-term vision for 'Viksit Bharat', and guiding principles for inclusive growth and self-reliance."
                                    },
                                    {
                                        "Subnet": "Macroeconomic Stability & Growth",
                                        "Categories": [
                                            {
                                                "Category": "Sustained Growth & Poverty Reduction",
                                                "Summary": "Highlights India's economic trajectory marked by stability, fiscal discipline, and sustained growth around 7%, leading to substantial poverty reduction."
                                            },
                                            {
                                                "Category": "Managing External Environment",
                                                "Summary": "Addresses challenges like disrupted trade, multilateralism, and supply chains, while striving for global market integration."
                                            },
                                            {
                                                "Category": "Conscious Economic Choices",
                                                "Summary": "Emphasizes deliberate decisions for stability, fiscal prudence, and monetary stability amidst global uncertainties."
                                            }
                                        ],
                                        "Summary": "Details India's positive economic performance over the past decade, including high growth rates, poverty alleviation, and strategic responses to global economic challenges."
                                    },
                                    {
                                        "Subnet": "Structural Reforms & Deregulation",
                                        "Categories": [
                                            {
                                                "Category": "Reform Express Initiatives",
                                                "Summary": "Encompasses over 350 reforms including GST simplification, Labour Codes, and rationalization of Quality Control Orders."
                                            },
                                            {
                                                "Category": "Government Collaboration",
                                                "Summary": "Central and State Governments are working in parallel on deregulation and reducing compliance requirements for businesses."
                                            },
                                            {
                                                "Category": "Continuous & Adaptive Reforms",
                                                "Summary": "Focus on sustaining momentum with forward-looking reforms to enhance productivity and competitiveness."
                                            }
                                        ],
                                        "Summary": "Describes the ongoing and comprehensive economic reforms aimed at boosting productivity, creating employment, and simplifying regulatory frameworks across various sectors."
                                    },
                                    {
                                        "Subnet": "Fiscal Policy & Budgetary Allocation",
                                        "Categories": [
                                            {
                                                "Category": "Fiscal Consolidation Path",
                                                "Summary": "Commitment to reduce fiscal deficit below 4.5% of GDP by 2025-26 and targeting a debt-to-GDP ratio of 50±1% by 2030-31."
                                            },
                                            {
                                                "Category": "Increased Public Capital Expenditure",
                                                "Summary": "Significant increase in public capex allocation from ₹2 lakh crore in FY2014-15 to ₹12.2 lakh crore in BE 2026-27 to sustain growth momentum."
                                            },
                                            {
                                                "Category": "16th Finance Commission Recommendations",
                                                "Summary": "Acceptance of the Commission's recommendation to retain vertical share of devolution at 41% and allocation of ₹1.4 lakh crore to States."
                                            }
                                        ],
                                        "Summary": "Covers the government's strategy for maintaining fiscal discipline, managing national debt, and strategically allocating funds for public investment and state support."
                                    }
                                ],
                                "Summary": "This topic covers the overarching economic philosophy, macroeconomic performance, structural reform agenda, and fiscal management strategies outlined in the budget speech, emphasizing stability, growth, and inclusive development."
                            },
                            {
                                "Net": "Manufacturing & Industrial Growth",
                                "Subnets": [
                                    {
                                        "Subnet": "Strategic Sector Expansion",
                                        "Categories": [
                                            {
                                                "Category": "Biopharma SHAKTI",
                                                "Summary": "₹10,000 crore outlay over 5 years to develop India as a global Biopharma manufacturing hub, including NIPER upgrades and clinical trial sites."
                                            },
                                            {
                                                "Category": "India Semiconductor Mission (ISM) 2.0",
                                                "Summary": "Expansion to produce equipment/materials, design full-stack Indian IP, and fortify supply chains, with increased outlay to ₹40,000 crore for electronics components."
                                            },
                                            {
                                                "Category": "Rare Earths & Chemical Parks",
                                                "Summary": "Support for mineral-rich states to establish Rare Earth Corridors and 3 dedicated Chemical Parks to enhance domestic production."
                                            }
                                        ],
                                        "Summary": "Focuses on boosting domestic capabilities and global competitiveness in high-tech and strategic manufacturing sectors, reducing import dependencies."
                                    },
                                    {
                                        "Subnet": "Core Industrial Capability & Capital Goods",
                                        "Categories": [
                                            {
                                                "Category": "Hi-Tech Tool Rooms",
                                                "Summary": "Establishment of digitally enabled automated service bureaus by CPSEs for high-precision component design and manufacturing."
                                            },
                                            {
                                                "Category": "Construction & Infrastructure Equipment (CIE)",
                                                "Summary": "Scheme to strengthen domestic manufacturing of high-value and technologically-advanced CIE, from lifts to tunnel-boring equipment."
                                            },
                                            {
                                                "Category": "Container Manufacturing Ecosystem",
                                                "Summary": "₹10,000 crore budgetary allocation over 5 years to create a globally competitive container manufacturing industry."
                                            }
                                        ],
                                        "Summary": "Initiatives to enhance the manufacturing base for critical capital goods and equipment essential for various industries and infrastructure projects."
                                    },
                                    {
                                        "Subnet": "Key Traditional Industries & Value Addition",
                                        "Categories": [
                                            {
                                                "Category": "Integrated Textile Programme",
                                                "Summary": "A five-part program including National Fibre Scheme, Textile Expansion, Handloom/Handicraft promotion, Tex-Eco Initiative, and Samarth 2.0, with Mega Textile Parks."
                                            },
                                            {
                                                "Category": "Mahatma Gandhi Gram Swaraj Initiative",
                                                "Summary": "To strengthen khadi, handloom, and handicrafts by supporting global market linkage, branding, training, and quality improvement."
                                            },
                                            {
                                                "Category": "Sports Goods Manufacturing",
                                                "Summary": "Dedicated initiative to promote manufacturing, research, and innovation in high-quality, affordable sports goods."
                                            }
                                        ],
                                        "Summary": "Programs aimed at modernizing, expanding, and adding value to labor-intensive and traditional industries, supporting artisans and rural youth."
                                    },
                                    {
                                        "Subnet": "MSME & Legacy Cluster Support",
                                        "Categories": [
                                            {
                                                "Category": "Champion MSME Development",
                                                "Summary": "A three-pronged approach: ₹10,000 crore SME Growth Fund, ₹2,000 crore top-up for Self-Reliant India Fund, and enhanced TReDS for liquidity support."
                                            },
                                            {
                                                "Category": "Legacy Industrial Cluster Revival",
                                                "Summary": "Scheme to revive 200 legacy industrial clusters through infrastructure and technology upgradation to improve cost competitiveness."
                                            },
                                            {
                                                "Category": "Professional Support for MSMEs",
                                                "Summary": "Facilitation of 'Corporate Mitras' by professional institutions to help MSMEs meet compliance requirements affordably in Tier-II/III towns."
                                            }
                                        ],
                                        "Summary": "Measures designed to empower Micro, Small, and Medium Enterprises (MSMEs) to become 'Champions' through equity, liquidity, and professional assistance, along with rejuvenating older industrial areas."
                                    }
                                ],
                                "Summary": "This topic covers government initiatives and schemes aimed at bolstering various manufacturing sectors, from high-tech to traditional, and supporting micro, small, and medium enterprises (MSMEs) and legacy industrial clusters to drive economic growth and employment."
                            },
                            {
                                "Net": "Infrastructure Development & Urban Planning",
                                "Subnets": [
                                    {
                                        "Subnet": "National Infrastructure Push",
                                        "Categories": [
                                            {
                                                "Category": "Enhanced Public Capital Expenditure",
                                                "Summary": "Increase in public capex to ₹12.2 lakh crore in FY2026-27 to sustain momentum in infrastructure development."
                                            },
                                            {
                                                "Category": "Infrastructure Risk Guarantee Fund",
                                                "Summary": "Establishment of a fund to provide partial credit guarantees to lenders, strengthening confidence for private developers."
                                            },
                                            {
                                                "Category": "Real Estate Asset Monetization",
                                                "Summary": "Acceleration of recycling significant real estate assets of CPSEs through dedicated REITs."
                                            }
                                        ],
                                        "Summary": "Strategies for continued massive public investment in infrastructure, de-risking private participation, and leveraging existing assets for further development."
                                    },
                                    {
                                        "Subnet": "Transport & Logistics Connectivity",
                                        "Categories": [
                                            {
                                                "Category": "Dedicated Freight Corridors",
                                                "Summary": "Establishment of new corridors connecting Dankuni to Surat to promote environmentally sustainable cargo movement."
                                            },
                                            {
                                                "Category": "National Waterways & Coastal Shipping",
                                                "Summary": "Operationalization of 20 new National Waterways, training institutes, ship repair ecosystems, and a Coastal Cargo Promotion Scheme to increase water transport share."
                                            },
                                            {
                                                "Category": "High-Speed Rail & Remote Connectivity",
                                                "Summary": "Development of seven High-Speed Rail corridors as 'growth connectors' and incentives for seaplane manufacturing and operations for last-mile connectivity."
                                            }
                                        ],
                                        "Summary": "Plans for modernizing and expanding India's transportation networks, including rail, waterways, coastal shipping, and air travel, to enhance logistics and connectivity."
                                    },
                                    {
                                        "Subnet": "Energy Security & Green Initiatives",
                                        "Categories": [
                                            {
                                                "Category": "Carbon Capture Utilization and Storage (CCUS)",
                                                "Summary": "₹20,000 crore outlay over 5 years to achieve higher readiness levels for CCUS technologies across industrial sectors like power, steel, cement."
                                            },
                                            {
                                                "Category": "Renewable Energy Manufacturing",
                                                "Summary": "Extension of basic customs duty exemption for capital goods used in manufacturing Lithium-Ion Cells for batteries and exemption for sodium antimonate for solar glass."
                                            },
                                            {
                                                "Category": "Nuclear Power Expansion",
                                                "Summary": "Extension of basic customs duty exemption on goods for Nuclear Power Projects till 2035 and expansion to all plants, irrespective of capacity."
                                            }
                                        ],
                                        "Summary": "Measures to ensure long-term energy security, promote green technologies, and support the development of sustainable energy sources like nuclear and renewables."
                                    },
                                    {
                                        "Subnet": "City-Centric Development & Smart Urbanization",
                                        "Categories": [
                                            {
                                                "Category": "City Economic Regions (CER) Development",
                                                "Summary": "Focus on Tier II, Tier III cities, and temple-towns; allocation of ₹5000 crore per CER over 5 years in a challenge mode for growth drivers."
                                            },
                                            {
                                                "Category": "Modern Urban Infrastructure",
                                                "Summary": "Aims to provide modern infrastructure and basic amenities to cities to amplify their potential as engines of growth and innovation."
                                            },
                                            {
                                                "Category": "Environmentally Sustainable Passenger Systems",
                                                "Summary": "Development of seven High-Speed Rail corridors between cities as 'growth connectors' for eco-friendly urban transport."
                                            }
                                        ],
                                        "Summary": "Strategies for enhancing the economic potential and livability of urban centers, particularly smaller cities, through targeted infrastructure development and regional planning."
                                    }
                                ],
                                "Summary": "This topic outlines the government's ambitious plans for physical infrastructure development, including transportation, logistics, and energy, alongside a dedicated focus on transforming Tier II and Tier III cities into economic growth hubs through modern planning and amenities."
                            },
                            {
                                "Net": "Human Capital & Social Empowerment",
                                "Subnets": [
                                    {
                                        "Subnet": "Education & Skill Development",
                                        "Categories": [
                                            {
                                                "Category": "Education to Employment & Enterprise Committee",
                                                "Summary": "High-Powered Standing Committee to recommend measures focusing on the Services Sector as a core driver, assessing AI impact on jobs and skills."
                                            },
                                            {
                                                "Category": "Health Professionals & Care Ecosystem",
                                                "Summary": "Upgrading existing and establishing new Allied Health Professional (AHP) Institutions (100,000 AHPs over 5 years) and developing NSQF-aligned caregiver programs (1.5 lakh trained)."
                                            },
                                            {
                                                "Category": "Creative Technologies & Design Education",
                                                "Summary": "Support for AVGC Content Creator Labs in schools/colleges and establishment of a new National Institute of Design in Eastern India."
                                            },
                                            {
                                                "Category": "Higher Education & Astronomy Promotion",
                                                "Summary": "Creation of 5 University Townships, 1 girls' hostel per district in STEM institutions, and upgrading 4 Telescope Infrastructure facilities."
                                            }
                                        ],
                                        "Summary": "Initiatives focused on enhancing the skill sets of the youth, developing a robust workforce for the services sector, and strengthening education infrastructure from school to higher learning."
                                    },
                                    {
                                        "Subnet": "Health & Wellness Promotion",
                                        "Categories": [
                                            {
                                                "Category": "Medical Value Tourism Hubs",
                                                "Summary": "Scheme to support States in establishing five Regional Medical Hubs with private partnerships, integrating healthcare, education, and research facilities."
                                            },
                                            {
                                                "Category": "AYUSH System Strengthening",
                                                "Summary": "Setting up 3 new All India Institutes of Ayurveda, upgrading pharmacies/labs, and bolstering the WHO Global Traditional Medicine Centre for evidence-based research."
                                            },
                                            {
                                                "Category": "Mental Health & Trauma Care",
                                                "Summary": "Establishment of NIMHANS-2, upgrading existing National Mental Health Institutes, and increasing emergency/trauma care capacities in District Hospitals by 50%."
                                            }
                                        ],
                                        "Summary": "Focuses on improving healthcare access, promoting India as a medical tourism destination, integrating traditional medicine, and strengthening mental health and emergency services."
                                    },
                                    {
                                        "Subnet": "Rural Economy & Agricultural Support",
                                        "Categories": [
                                            {
                                                "Category": "Increasing Farmer Incomes",
                                                "Summary": "Initiatives for integrated development of reservoirs, strengthening fisheries value chain, and supporting entrepreneurship in Animal Husbandry."
                                            },
                                            {
                                                "Category": "High Value Agriculture Promotion",
                                                "Summary": "Support for crops like coconut, sandalwood, cocoa, cashew, agar trees, almonds, walnuts, and pine nuts to diversify farm outputs and increase productivity."
                                            },
                                            {
                                                "Category": "Bharat-VISTAAR AI Tool",
                                                "Summary": "Multilingual AI tool integrating AgriStack and ICAR practices with AI systems to enhance farm productivity, decision-making, and risk reduction."
                                            },
                                            {
                                                "Category": "SHE-Marts for Rural Women",
                                                "Summary": "Establishment of community-owned retail outlets for women-led enterprises, building on the Lakhpati Didi Programme, to facilitate entrepreneurship."
                                            }
                                        ],
                                        "Summary": "Programs aimed at boosting farmer incomes, diversifying agricultural outputs, modernizing rural enterprises, and empowering rural women through technology and financial support."
                                    },
                                    {
                                        "Subnet": "Tourism, Culture & Sports Development",
                                        "Categories": [
                                            {
                                                "Category": "Tourism Sector Growth",
                                                "Summary": "Upgrading National Council for Hotel Management to National Institute of Hospitality, pilot scheme for upskilling 10,000 guides, and National Destination Digital Knowledge Grid."
                                            },
                                            {
                                                "Category": "Experiential Heritage & Culture Destinations",
                                                "Summary": "Development of 15 archaeological sites (e.g., Lothal, Dholavira) into vibrant cultural destinations with curated walkways and immersive storytelling."
                                            },
                                            {
                                                "Category": "Adventure & Nature Trails",
                                                "Summary": "Development of ecologically sustainable Mountain, Turtle, and Bird watching trails across various states to promote niche tourism."
                                            },
                                            {
                                                "Category": "Khelo India Mission",
                                                "Summary": "Launch of a mission to transform the Sports sector over the next decade through talent development, coach training, sports science, competitions, and infrastructure."
                                            }
                                        ],
                                        "Summary": "Initiatives to enhance India's tourism potential by promoting heritage, culture, and nature-based experiences, alongside a comprehensive mission to develop the sports sector."
                                    },
                                    {
                                        "Subnet": "Inclusive Regional & Vulnerable Group Development",
                                        "Categories": [
                                            {
                                                "Category": "Empowering Divyangjan",
                                                "Summary": "Divyangjan Kaushal Yojana for industry-relevant training in IT, AVGC, Hospitality; Divyang Sahara Yojana for scaling assistive device production by ALIMCO and PM Divyasha Kendras."
                                            },
                                            {
                                                "Category": "Purvodaya & North-Eastern Region Focus",
                                                "Summary": "Development of an integrated East Coast Industrial Corridor, 5 tourism destinations, and 4,000 e-buses for Purvodaya States; scheme for Buddhist Circuits in North-Eastern Region."
                                            },
                                            {
                                                "Category": "International Big Cat Alliance",
                                                "Summary": "India hosting the first Global Big Cat Summit to deliberate on collective conservation strategies for big cats."
                                            }
                                        ],
                                        "Summary": "Targeted efforts to empower persons with disabilities through skill development and assistive devices, accelerate development in eastern and north-eastern regions, and promote wildlife conservation."
                                    }
                                ],
                                "Summary": "This topic encompasses government efforts to develop human capital, improve social welfare, boost rural economies, promote tourism and culture, and ensure inclusive growth for all sections of society, including vulnerable groups and specific regions."
                            },
                            {
                                "Net": "Financial Sector & Investment Reforms",
                                "Subnets": [
                                    {
                                        "Subnet": "Banking Sector Strengthening",
                                        "Categories": [
                                            {
                                                "Category": "High Level Committee on Banking",
                                                "Summary": "Proposed committee to comprehensively review the banking sector and align it with India’s next phase of growth, safeguarding financial stability and inclusion."
                                            },
                                            {
                                                "Category": "Strong Balance Sheets & Profitability",
                                                "Summary": "Highlights the Indian banking sector's current state of strong balance sheets, historic highs in profitability, and improved asset quality."
                                            },
                                            {
                                                "Category": "Rural Banking Coverage",
                                                "Summary": "Ensures banking coverage exceeding 98% of villages in the country."
                                            }
                                        ],
                                        "Summary": "Measures aimed at reviewing and strengthening the Indian banking sector to support future growth while ensuring stability and widespread access."
                                    },
                                    {
                                        "Subnet": "Non-Banking Financial Companies (NBFCs)",
                                        "Categories": [
                                            {
                                                "Category": "Vision for Viksit Bharat NBFCs",
                                                "Summary": "Outlined with clear targets for credit disbursement and technology adoption to contribute to India's development."
                                            },
                                            {
                                                "Category": "Restructuring Public Sector NBFCs",
                                                "Summary": "Proposed restructuring of Power Finance Corporation and Rural Electrification Corporation as a first step to achieve scale and improve efficiency."
                                            }
                                        ],
                                        "Summary": "Strategies to enhance the role and efficiency of Non-Banking Financial Companies (NBFCs) in achieving national development goals."
                                    },
                                    {
                                        "Subnet": "Capital Markets Development",
                                        "Categories": [
                                            {
                                                "Category": "Corporate Bond Market Reforms",
                                                "Summary": "Introduction of a market making framework, access to funds and derivatives on corporate bond indices, and total return swaps on corporate bonds."
                                            },
                                            {
                                                "Category": "Municipal Bonds Incentives",
                                                "Summary": "₹100 crore incentive for single municipal bond issuances over ₹1000 crore by large cities, with continued support for smaller towns."
                                            },
                                            {
                                                "Category": "Enhanced Portfolio Investment Scheme",
                                                "Summary": "Permitting Individual Persons Resident Outside India (PROI) to invest in listed Indian equities, increasing individual limit from 5% to 10% and overall limit to 24%."
                                            }
                                        ],
                                        "Summary": "Proposals to deepen and broaden India's capital markets, making them more attractive for both domestic and foreign investors, and supporting municipal financing."
                                    },
                                    {
                                        "Subnet": "Foreign Investment & Ease of Doing Business",
                                        "Categories": [
                                            {
                                                "Category": "Foreign Exchange Management Review",
                                                "Summary": "Comprehensive review of FEMA (Non-debt Instruments) Rules to create a more contemporary and user-friendly framework for foreign investments."
                                            },
                                            {
                                                "Category": "Incentivizing Data Centers & Toll Manufacturing",
                                                "Summary": "Tax holiday till 2047 for foreign companies providing cloud services via Indian data centers and income tax exemption for 5 years for non-residents providing capital goods for toll manufacturing."
                                            },
                                            {
                                                "Category": "Attracting Global Talent",
                                                "Summary": "Exemption of global (non-India sourced) income for non-resident experts staying for 5 years under notified schemes."
                                            }
                                        ],
                                        "Summary": "Reforms to liberalize and simplify the foreign investment framework, attract global businesses and talent, and improve the overall ease of doing business in India."
                                    }
                                ],
                                "Summary": "This topic details government initiatives to strengthen the financial sector, including banking and NBFCs, develop capital markets (corporate and municipal bonds), and enhance foreign investment by streamlining regulations and offering incentives for critical sectors like data centers and manufacturing."
                            },
                            {
                                "Net": "Taxation Modernization & Simplification",
                                "Subnets": [
                                    {
                                        "Subnet": "Direct Tax Reforms",
                                        "Categories": [
                                            {
                                                "Category": "New Income Tax Act, 2025",
                                                "Summary": "Comprehensive review and enactment of a simplified act effective April 1, 2026, with redesigned forms for ease of compliance."
                                            },
                                            {
                                                "Category": "Ease of Living for Taxpayers",
                                                "Summary": "Exemption of MACT interest, reduced TCS rates for overseas tour packages (2%), education/medical LRS (2%), and simplified TDS processes."
                                            },
                                            {
                                                "Category": "Rationalizing Penalty & Prosecution",
                                                "Summary": "Integration of assessment & penalty proceedings, immunity from penalty for misreporting (with conditions), conversion of minor penalties to fees, and decriminalization of certain offenses."
                                            },
                                            {
                                                "Category": "Supporting IT Sector & Global Business",
                                                "Summary": "Clubbing IT services under a single category with a 15.5% safe harbor margin (threshold up to ₹2,000 crore), fast-tracking APA processes, and tax holidays for foreign cloud/data center providers."
                                            }
                                        ],
                                        "Summary": "Outlines a major overhaul of the direct tax system, including a new Income Tax Act, simplified compliance for individuals and businesses, rationalization of penalties, and incentives for key growth sectors."
                                    },
                                    {
                                        "Subnet": "Indirect Tax Reforms (Customs & Excise)",
                                        "Categories": [
                                            {
                                                "Category": "Customs Duty Review & Simplification",
                                                "Summary": "Removal of long-continuing exemptions, incorporation of effective rates into tariff schedule, and creation of new tariff lines for better identification and monitoring."
                                            },
                                            {
                                                "Category": "Export Promotion Measures",
                                                "Summary": "Increased limit for duty-free imports for seafood exports, extension of time for export of final products for leather/textile, and extension of duty exemption for shoe-uppers."
                                            },
                                            {
                                                "Category": "Strategic Sector Support",
                                                "Summary": "Customs duty exemptions for capital goods for Lithium-Ion cells, sodium antimonate for solar glass, nuclear power projects, critical minerals processing, and civil/defense aviation manufacturing components."
                                            },
                                            {
                                                "Category": "Ease of Living & Patient Relief (Customs)",
                                                "Summary": "Reduction of tariff rate on all dutiable goods for personal use from 20% to 10%, and exemption of basic customs duty on 17 specific drugs/medicines and 7 rare disease treatments."
                                            }
                                        ],
                                        "Summary": "Details changes in customs and excise duties aimed at simplifying tariffs, boosting domestic manufacturing and exports, supporting strategic sectors, and providing relief for citizens and patients."
                                    },
                                    {
                                        "Subnet": "GST Reforms",
                                        "Categories": [
                                            {
                                                "Category": "Post-Sale Discounts & Provisional Refunds",
                                                "Summary": "Amendments to section 15 and 34 of the CGST Act to simplify post-sale discount provisions and extend provisional refunds to inverted duty structure cases."
                                            },
                                            {
                                                "Category": "Refund Thresholds & Appellate Authority",
                                                "Summary": "Removal of the ₹1000 refund limit for exported goods and provisions for empowering an existing Tribunal for hearing appeals under section 101A of the CGST Act."
                                            },
                                            {
                                                "Category": "Place of Supply for Intermediary Services",
                                                "Summary": "Omission of clause (b) of sub-section (8) of section 13 of the IGST Act to determine place of supply for intermediary services as per default provisions."
                                            }
                                        ],
                                        "Summary": "Proposed amendments to the CGST and IGST Acts to refine provisions related to discounts, refunds, and the appellate mechanism, while also clarifying rules for intermediary services."
                                    },
                                    {
                                        "Subnet": "Tax Administration & Compliance",
                                        "Categories": [
                                            {
                                                "Category": "Depositary-Led Declaration Filing",
                                                "Summary": "Allowing taxpayers to file declarations for no TDS to depositories for mutual fund, interest, and dividend income, streamlining compliance."
                                            },
                                            {
                                                "Category": "Staggered IT Return Filing & Revised Returns",
                                                "Summary": "Extension of time for filing revised returns up to March 31st and staggered due dates for ITR 1/2 (July 31st) and non-audit business (August 31st)."
                                            },
                                            {
                                                "Category": "Foreign Assets of Small Taxpayers Disclosure Scheme (FAST-DS)",
                                                "Summary": "Introduction of a time-bound scheme for declaration of foreign assets and foreign-sourced income below certain thresholds, offering immunity for compliance."
                                            },
                                            {
                                                "Category": "Customs Process Modernization",
                                                "Summary": "Enhancing duty deferral, extending validity of advance rulings, leveraging AEO accreditation, electronic sealing for export cargo, and rolling out Customs Integrated System (CIS) for digital clearance."
                                            }
                                        ],
                                        "Summary": "Initiatives focused on enhancing the efficiency of tax administration, simplifying compliance procedures for taxpayers, and leveraging technology for faster and more transparent processes."
                                    },
                                    {
                                        "Subnet": "Other Tax Adjustments & Rationalizations",
                                        "Categories": [
                                            {
                                                "Category": "Rationalization of Share Buyback",
                                                "Summary": "Consideration received by shareholders on buyback chargeable as Capital Gains, with differential rates for corporate (22%) and non-corporate (30%) promoters."
                                            },
                                            {
                                                "Category": "TCS & STT Rate Adjustments",
                                                "Summary": "Rationalization of TCS rates for specific goods (e.g., alcoholic liquor, scrap, minerals to 2%, tendu leaves to 2%) and increase in STT rates for Futures and Options in the capital market."
                                            },
                                            {
                                                "Category": "Minimum Alternate Tax (MAT) Changes",
                                                "Summary": "MAT to be a final tax, reduced rate from 15% to 14%, and allowance of set-off of brought forward MAT credit (25% of tax liability) for companies shifting to the new tax regime."
                                            },
                                            {
                                                "Category": "Cooperative Sector Support",
                                                "Summary": "Deduction of profits for primary cooperatives supplying cattle feed/cotton seed, inter-cooperative society dividend income deduction under new regime, and exemption for notified national cooperative federations."
                                            }
                                        ],
                                        "Summary": "Covers various specific tax proposals including adjustments to capital market taxes, rationalization of corporate tax regimes, and targeted benefits for the cooperative sector."
                                    }
                                ],
                                "Summary": "This topic covers the comprehensive reforms and proposals related to direct and indirect taxation, including a new Income Tax Act, simplification measures for taxpayers, rationalization of penalties, and strategic adjustments to customs, excise, and GST to promote economic activities and ensure compliance."
                            },
                            {
                                "Net": "996 Non specific TOPIC",
                                "Subnets": [
                                    {
                                        "Subnet": "Uncategorized/Irrelevant Responses",
                                        "Categories": [
                                            {
                                                "Category": "Unclassified Content",
                                                "Summary": "Responses that are not directly align with any defined themes and require further review for potential upcoding or new categories."
                                            },
                                            {
                                                "Category": "Content Needing Upcoding",
                                                "Summary": "Responses that are somewhat relevant but cannot be precisely categorized within the existing codeframe hierarchy."
                                            }
                                        ],
                                        "Summary": "This subnet captures responses that are relevant to the survey but do not fit neatly into the predefined hierarchical structure of topics, subtopics, and categories, indicating a need for potential expansion or refinement of the codeframe for these responses."
                                    }
                                ],
                                "Summary": "This topic is for responses that are broadly relevant to the survey's context but do not fit into the established themes or could not be easily coded within the existing codeframe, often requiring further analysis or upcoding."
                            },
                            {
                                "Net": "997 Don’t Know Can’t Say",
                                "Subnets": [
                                    {
                                        "Subnet": "Non-Committal Responses",
                                        "Categories": [
                                            {
                                                "Category": "Lack of Information",
                                                "Summary": "Responses indicating the respondent does not have enough information to provide a meaningful answer."
                                            },
                                            {
                                                "Category": "Uncertainty/Hesitation",
                                                "Summary": "Responses expressing indecision, doubt, or an unwillingness to commit to a specific opinion or answer."
                                            }
                                        ],
                                        "Summary": "This subnet collects responses where individuals express a lack of knowledge, uncertainty, or an inability to provide a definitive answer due to various reasons."
                                    }
                                ],
                                "Summary": "This topic is designated for responses that are non-committal, vague, or indicate that the respondent does not know or cannot provide a clear answer."
                            },
                            {
                                "Net": "998 Not Applicable",
                                "Subnets": [
                                    {
                                        "Subnet": "Irrelevant to Question/Project",
                                        "Categories": [
                                            {
                                                "Category": "Out of Scope",
                                                "Summary": "Responses that are factually correct but do not pertain to the specific question asked or the scope of the market research project."
                                            },
                                            {
                                                "Category": "Personal Opinions (Off-Topic)",
                                                "Summary": "Responses that provide personal views or information not requested by the survey question."
                                            }
                                        ],
                                        "Summary": "This subnet includes responses that are clearly not applicable or relevant to the specific question or the broader objectives of the market research project."
                                    }
                                ],
                                "Summary": "This topic is for responses that are deemed irrelevant or not applicable to the particular question or the overall context of the market research study."
                            },
                            {
                                "Net": "999 None-Nothing",
                                "Subnets": [
                                    {
                                        "Subnet": "Gibberish/Meaningless Content",
                                        "Categories": [
                                            {
                                                "Category": "Random Characters/Symbols",
                                                "Summary": "Responses consisting of random letters, numbers, symbols, or keyboard mashing, indicating no coherent linguistic content."
                                            },
                                            {
                                                "Category": "Empty/Blank Responses",
                                                "Summary": "Responses that are intentionally left blank or contain placeholders like 'N/A', '---', or similar indicators of no content."
                                            }
                                        ],
                                        "Summary": "This subnet captures responses that are unintelligible, nonsensical, or contain no meaningful information, often due to gibberish, non-language characters, or being left blank."
                                    }
                                ],
                                "Summary": "This topic is for responses that are unintelligible, nonsensical, or provide no meaningful information, including gibberish, non-language characters, or responses that effectively convey 'nothing'."
                            }
                        ],
                        "Summary": "This document outlines the Budget Speech for 2026-2027 by the Minister of Finance, Nirmala Sitharaman, focusing on India's economic trajectory, structural reforms, and vision for 'Viksit Bharat' (Developed India). It details comprehensive proposals across key sectors including manufacturing, infrastructure, human capital, financial services, and social welfare, alongside significant direct and indirect tax reforms aimed at simplifying compliance, promoting investment, and ensuring fiscal prudence. The speech emphasizes a threefold approach: accelerating economic growth, fulfilling aspirations through capacity building, and ensuring equitable access to resources for all sections of society, while also addressing environmental sustainability and digital transformation."
                    }
                }
            ],
            "onlyCodeFrame": true,
            "errorList": [],
            "codeFrameGenerateType": "CODE_FRAME_ONLY",
            "requestType": "CODEFRAME",
            "requestMode": "DEVELOPMENT",
            "sourceLanguage": "English",
            "targetLanguage": "English",
            "response": {
                "status": "OK",
                "data": {
                    "status": "OK",
                    "PromptTokenCount": 21105,
                    "CandidatesTokenCount": 5681,
                    "TotalTokenCount": 28344,
                    "thoughtsTokenCount": 1558,
                    "finishReason": "STOP",
                    "data": [
                        {
                            "Net": "Economic Strategy & Fiscal Management",
                            "Subnets": [
                                {
                                    "Subnet": "Overall Vision & Principles",
                                    "Categories": [
                                        {
                                            "Category": "Viksit Bharat Initiative",
                                            "Summary": "The government's overarching vision to transform India into a developed nation, balancing ambition with inclusion.",
                                            "code": 3
                                        },
                                        {
                                            "Category": "Governance Philosophy",
                                            "Summary": "Emphasizes action over ambivalence, reform over rhetoric, and people over populism, guided by 'Atmanirbharta'.",
                                            "code": 4
                                        },
                                        {
                                            "Category": "Sankalp & Kartavya Approach",
                                            "Summary": "Focuses on the poor, underprivileged, and disadvantaged, driven by three 'kartavya' (duties): accelerate growth, fulfill aspirations, and ensure equitable access.",
                                            "code": 5
                                        }
                                    ],
                                    "Summary": "Outlines the government's foundational economic philosophy, long-term vision for 'Viksit Bharat', and guiding principles for inclusive growth and self-reliance.",
                                    "code": 2
                                },
                                {
                                    "Subnet": "Macroeconomic Stability & Growth",
                                    "Categories": [
                                        {
                                            "Category": "Sustained Growth & Poverty Reduction",
                                            "Summary": "Highlights India's economic trajectory marked by stability, fiscal discipline, and sustained growth around 7%, leading to substantial poverty reduction.",
                                            "code": 7
                                        },
                                        {
                                            "Category": "Managing External Environment",
                                            "Summary": "Addresses challenges like disrupted trade, multilateralism, and supply chains, while striving for global market integration.",
                                            "code": 8
                                        },
                                        {
                                            "Category": "Conscious Economic Choices",
                                            "Summary": "Emphasizes deliberate decisions for stability, fiscal prudence, and monetary stability amidst global uncertainties.",
                                            "code": 9
                                        }
                                    ],
                                    "Summary": "Details India's positive economic performance over the past decade, including high growth rates, poverty alleviation, and strategic responses to global economic challenges.",
                                    "code": 6
                                },
                                {
                                    "Subnet": "Structural Reforms & Deregulation",
                                    "Categories": [
                                        {
                                            "Category": "Reform Express Initiatives",
                                            "Summary": "Encompasses over 350 reforms including GST simplification, Labour Codes, and rationalization of Quality Control Orders.",
                                            "code": 11
                                        },
                                        {
                                            "Category": "Government Collaboration",
                                            "Summary": "Central and State Governments are working in parallel on deregulation and reducing compliance requirements for businesses.",
                                            "code": 12
                                        },
                                        {
                                            "Category": "Continuous & Adaptive Reforms",
                                            "Summary": "Focus on sustaining momentum with forward-looking reforms to enhance productivity and competitiveness.",
                                            "code": 13
                                        }
                                    ],
                                    "Summary": "Describes the ongoing and comprehensive economic reforms aimed at boosting productivity, creating employment, and simplifying regulatory frameworks across various sectors.",
                                    "code": 10
                                },
                                {
                                    "Subnet": "Fiscal Policy & Budgetary Allocation",
                                    "Categories": [
                                        {
                                            "Category": "Fiscal Consolidation Path",
                                            "Summary": "Commitment to reduce fiscal deficit below 4.5% of GDP by 2025-26 and targeting a debt-to-GDP ratio of 50±1% by 2030-31.",
                                            "code": 15
                                        },
                                        {
                                            "Category": "Increased Public Capital Expenditure",
                                            "Summary": "Significant increase in public capex allocation from ₹2 lakh crore in FY2014-15 to ₹12.2 lakh crore in BE 2026-27 to sustain growth momentum.",
                                            "code": 16
                                        },
                                        {
                                            "Category": "16th Finance Commission Recommendations",
                                            "Summary": "Acceptance of the Commission's recommendation to retain vertical share of devolution at 41% and allocation of ₹1.4 lakh crore to States.",
                                            "code": 17
                                        }
                                    ],
                                    "Summary": "Covers the government's strategy for maintaining fiscal discipline, managing national debt, and strategically allocating funds for public investment and state support.",
                                    "code": 14
                                }
                            ],
                            "Summary": "This topic covers the overarching economic philosophy, macroeconomic performance, structural reform agenda, and fiscal management strategies outlined in the budget speech, emphasizing stability, growth, and inclusive development.",
                            "code": 1
                        },
                        {
                            "Net": "Manufacturing & Industrial Growth",
                            "Subnets": [
                                {
                                    "Subnet": "Strategic Sector Expansion",
                                    "Categories": [
                                        {
                                            "Category": "Biopharma SHAKTI",
                                            "Summary": "₹10,000 crore outlay over 5 years to develop India as a global Biopharma manufacturing hub, including NIPER upgrades and clinical trial sites.",
                                            "code": 20
                                        },
                                        {
                                            "Category": "India Semiconductor Mission (ISM) 2.0",
                                            "Summary": "Expansion to produce equipment/materials, design full-stack Indian IP, and fortify supply chains, with increased outlay to ₹40,000 crore for electronics components.",
                                            "code": 21
                                        },
                                        {
                                            "Category": "Rare Earths & Chemical Parks",
                                            "Summary": "Support for mineral-rich states to establish Rare Earth Corridors and 3 dedicated Chemical Parks to enhance domestic production.",
                                            "code": 22
                                        }
                                    ],
                                    "Summary": "Focuses on boosting domestic capabilities and global competitiveness in high-tech and strategic manufacturing sectors, reducing import dependencies.",
                                    "code": 19
                                },
                                {
                                    "Subnet": "Core Industrial Capability & Capital Goods",
                                    "Categories": [
                                        {
                                            "Category": "Hi-Tech Tool Rooms",
                                            "Summary": "Establishment of digitally enabled automated service bureaus by CPSEs for high-precision component design and manufacturing.",
                                            "code": 24
                                        },
                                        {
                                            "Category": "Construction & Infrastructure Equipment (CIE)",
                                            "Summary": "Scheme to strengthen domestic manufacturing of high-value and technologically-advanced CIE, from lifts to tunnel-boring equipment.",
                                            "code": 25
                                        },
                                        {
                                            "Category": "Container Manufacturing Ecosystem",
                                            "Summary": "₹10,000 crore budgetary allocation over 5 years to create a globally competitive container manufacturing industry.",
                                            "code": 26
                                        }
                                    ],
                                    "Summary": "Initiatives to enhance the manufacturing base for critical capital goods and equipment essential for various industries and infrastructure projects.",
                                    "code": 23
                                },
                                {
                                    "Subnet": "Key Traditional Industries & Value Addition",
                                    "Categories": [
                                        {
                                            "Category": "Integrated Textile Programme",
                                            "Summary": "A five-part program including National Fibre Scheme, Textile Expansion, Handloom/Handicraft promotion, Tex-Eco Initiative, and Samarth 2.0, with Mega Textile Parks.",
                                            "code": 28
                                        },
                                        {
                                            "Category": "Mahatma Gandhi Gram Swaraj Initiative",
                                            "Summary": "To strengthen khadi, handloom, and handicrafts by supporting global market linkage, branding, training, and quality improvement.",
                                            "code": 29
                                        },
                                        {
                                            "Category": "Sports Goods Manufacturing",
                                            "Summary": "Dedicated initiative to promote manufacturing, research, and innovation in high-quality, affordable sports goods.",
                                            "code": 30
                                        }
                                    ],
                                    "Summary": "Programs aimed at modernizing, expanding, and adding value to labor-intensive and traditional industries, supporting artisans and rural youth.",
                                    "code": 27
                                },
                                {
                                    "Subnet": "MSME & Legacy Cluster Support",
                                    "Categories": [
                                        {
                                            "Category": "Champion MSME Development",
                                            "Summary": "A three-pronged approach: ₹10,000 crore SME Growth Fund, ₹2,000 crore top-up for Self-Reliant India Fund, and enhanced TReDS for liquidity support.",
                                            "code": 32
                                        },
                                        {
                                            "Category": "Legacy Industrial Cluster Revival",
                                            "Summary": "Scheme to revive 200 legacy industrial clusters through infrastructure and technology upgradation to improve cost competitiveness.",
                                            "code": 33
                                        },
                                        {
                                            "Category": "Professional Support for MSMEs",
                                            "Summary": "Facilitation of 'Corporate Mitras' by professional institutions to help MSMEs meet compliance requirements affordably in Tier-II/III towns.",
                                            "code": 34
                                        }
                                    ],
                                    "Summary": "Measures designed to empower Micro, Small, and Medium Enterprises (MSMEs) to become 'Champions' through equity, liquidity, and professional assistance, along with rejuvenating older industrial areas.",
                                    "code": 31
                                }
                            ],
                            "Summary": "This topic covers government initiatives and schemes aimed at bolstering various manufacturing sectors, from high-tech to traditional, and supporting micro, small, and medium enterprises (MSMEs) and legacy industrial clusters to drive economic growth and employment.",
                            "code": 18
                        },
                        {
                            "Net": "Infrastructure Development & Urban Planning",
                            "Subnets": [
                                {
                                    "Subnet": "National Infrastructure Push",
                                    "Categories": [
                                        {
                                            "Category": "Enhanced Public Capital Expenditure",
                                            "Summary": "Increase in public capex to ₹12.2 lakh crore in FY2026-27 to sustain momentum in infrastructure development.",
                                            "code": 37
                                        },
                                        {
                                            "Category": "Infrastructure Risk Guarantee Fund",
                                            "Summary": "Establishment of a fund to provide partial credit guarantees to lenders, strengthening confidence for private developers.",
                                            "code": 38
                                        },
                                        {
                                            "Category": "Real Estate Asset Monetization",
                                            "Summary": "Acceleration of recycling significant real estate assets of CPSEs through dedicated REITs.",
                                            "code": 39
                                        }
                                    ],
                                    "Summary": "Strategies for continued massive public investment in infrastructure, de-risking private participation, and leveraging existing assets for further development.",
                                    "code": 36
                                },
                                {
                                    "Subnet": "Transport & Logistics Connectivity",
                                    "Categories": [
                                        {
                                            "Category": "Dedicated Freight Corridors",
                                            "Summary": "Establishment of new corridors connecting Dankuni to Surat to promote environmentally sustainable cargo movement.",
                                            "code": 41
                                        },
                                        {
                                            "Category": "National Waterways & Coastal Shipping",
                                            "Summary": "Operationalization of 20 new National Waterways, training institutes, ship repair ecosystems, and a Coastal Cargo Promotion Scheme to increase water transport share.",
                                            "code": 42
                                        },
                                        {
                                            "Category": "High-Speed Rail & Remote Connectivity",
                                            "Summary": "Development of seven High-Speed Rail corridors as 'growth connectors' and incentives for seaplane manufacturing and operations for last-mile connectivity.",
                                            "code": 43
                                        }
                                    ],
                                    "Summary": "Plans for modernizing and expanding India's transportation networks, including rail, waterways, coastal shipping, and air travel, to enhance logistics and connectivity.",
                                    "code": 40
                                },
                                {
                                    "Subnet": "Energy Security & Green Initiatives",
                                    "Categories": [
                                        {
                                            "Category": "Carbon Capture Utilization and Storage (CCUS)",
                                            "Summary": "₹20,000 crore outlay over 5 years to achieve higher readiness levels for CCUS technologies across industrial sectors like power, steel, cement.",
                                            "code": 45
                                        },
                                        {
                                            "Category": "Renewable Energy Manufacturing",
                                            "Summary": "Extension of basic customs duty exemption for capital goods used in manufacturing Lithium-Ion Cells for batteries and exemption for sodium antimonate for solar glass.",
                                            "code": 46
                                        },
                                        {
                                            "Category": "Nuclear Power Expansion",
                                            "Summary": "Extension of basic customs duty exemption on goods for Nuclear Power Projects till 2035 and expansion to all plants, irrespective of capacity.",
                                            "code": 47
                                        }
                                    ],
                                    "Summary": "Measures to ensure long-term energy security, promote green technologies, and support the development of sustainable energy sources like nuclear and renewables.",
                                    "code": 44
                                },
                                {
                                    "Subnet": "City-Centric Development & Smart Urbanization",
                                    "Categories": [
                                        {
                                            "Category": "City Economic Regions (CER) Development",
                                            "Summary": "Focus on Tier II, Tier III cities, and temple-towns; allocation of ₹5000 crore per CER over 5 years in a challenge mode for growth drivers.",
                                            "code": 49
                                        },
                                        {
                                            "Category": "Modern Urban Infrastructure",
                                            "Summary": "Aims to provide modern infrastructure and basic amenities to cities to amplify their potential as engines of growth and innovation.",
                                            "code": 50
                                        },
                                        {
                                            "Category": "Environmentally Sustainable Passenger Systems",
                                            "Summary": "Development of seven High-Speed Rail corridors between cities as 'growth connectors' for eco-friendly urban transport.",
                                            "code": 51
                                        }
                                    ],
                                    "Summary": "Strategies for enhancing the economic potential and livability of urban centers, particularly smaller cities, through targeted infrastructure development and regional planning.",
                                    "code": 48
                                }
                            ],
                            "Summary": "This topic outlines the government's ambitious plans for physical infrastructure development, including transportation, logistics, and energy, alongside a dedicated focus on transforming Tier II and Tier III cities into economic growth hubs through modern planning and amenities.",
                            "code": 35
                        },
                        {
                            "Net": "Human Capital & Social Empowerment",
                            "Subnets": [
                                {
                                    "Subnet": "Education & Skill Development",
                                    "Categories": [
                                        {
                                            "Category": "Education to Employment & Enterprise Committee",
                                            "Summary": "High-Powered Standing Committee to recommend measures focusing on the Services Sector as a core driver, assessing AI impact on jobs and skills.",
                                            "code": 54
                                        },
                                        {
                                            "Category": "Health Professionals & Care Ecosystem",
                                            "Summary": "Upgrading existing and establishing new Allied Health Professional (AHP) Institutions (100,000 AHPs over 5 years) and developing NSQF-aligned caregiver programs (1.5 lakh trained).",
                                            "code": 55
                                        },
                                        {
                                            "Category": "Creative Technologies & Design Education",
                                            "Summary": "Support for AVGC Content Creator Labs in schools/colleges and establishment of a new National Institute of Design in Eastern India.",
                                            "code": 56
                                        },
                                        {
                                            "Category": "Higher Education & Astronomy Promotion",
                                            "Summary": "Creation of 5 University Townships, 1 girls' hostel per district in STEM institutions, and upgrading 4 Telescope Infrastructure facilities.",
                                            "code": 57
                                        }
                                    ],
                                    "Summary": "Initiatives focused on enhancing the skill sets of the youth, developing a robust workforce for the services sector, and strengthening education infrastructure from school to higher learning.",
                                    "code": 53
                                },
                                {
                                    "Subnet": "Health & Wellness Promotion",
                                    "Categories": [
                                        {
                                            "Category": "Medical Value Tourism Hubs",
                                            "Summary": "Scheme to support States in establishing five Regional Medical Hubs with private partnerships, integrating healthcare, education, and research facilities.",
                                            "code": 59
                                        },
                                        {
                                            "Category": "AYUSH System Strengthening",
                                            "Summary": "Setting up 3 new All India Institutes of Ayurveda, upgrading pharmacies/labs, and bolstering the WHO Global Traditional Medicine Centre for evidence-based research.",
                                            "code": 60
                                        },
                                        {
                                            "Category": "Mental Health & Trauma Care",
                                            "Summary": "Establishment of NIMHANS-2, upgrading existing National Mental Health Institutes, and increasing emergency/trauma care capacities in District Hospitals by 50%.",
                                            "code": 61
                                        }
                                    ],
                                    "Summary": "Focuses on improving healthcare access, promoting India as a medical tourism destination, integrating traditional medicine, and strengthening mental health and emergency services.",
                                    "code": 58
                                },
                                {
                                    "Subnet": "Rural Economy & Agricultural Support",
                                    "Categories": [
                                        {
                                            "Category": "Increasing Farmer Incomes",
                                            "Summary": "Initiatives for integrated development of reservoirs, strengthening fisheries value chain, and supporting entrepreneurship in Animal Husbandry.",
                                            "code": 63
                                        },
                                        {
                                            "Category": "High Value Agriculture Promotion",
                                            "Summary": "Support for crops like coconut, sandalwood, cocoa, cashew, agar trees, almonds, walnuts, and pine nuts to diversify farm outputs and increase productivity.",
                                            "code": 64
                                        },
                                        {
                                            "Category": "Bharat-VISTAAR AI Tool",
                                            "Summary": "Multilingual AI tool integrating AgriStack and ICAR practices with AI systems to enhance farm productivity, decision-making, and risk reduction.",
                                            "code": 65
                                        },
                                        {
                                            "Category": "SHE-Marts for Rural Women",
                                            "Summary": "Establishment of community-owned retail outlets for women-led enterprises, building on the Lakhpati Didi Programme, to facilitate entrepreneurship.",
                                            "code": 66
                                        }
                                    ],
                                    "Summary": "Programs aimed at boosting farmer incomes, diversifying agricultural outputs, modernizing rural enterprises, and empowering rural women through technology and financial support.",
                                    "code": 62
                                },
                                {
                                    "Subnet": "Tourism, Culture & Sports Development",
                                    "Categories": [
                                        {
                                            "Category": "Tourism Sector Growth",
                                            "Summary": "Upgrading National Council for Hotel Management to National Institute of Hospitality, pilot scheme for upskilling 10,000 guides, and National Destination Digital Knowledge Grid.",
                                            "code": 68
                                        },
                                        {
                                            "Category": "Experiential Heritage & Culture Destinations",
                                            "Summary": "Development of 15 archaeological sites (e.g., Lothal, Dholavira) into vibrant cultural destinations with curated walkways and immersive storytelling.",
                                            "code": 69
                                        },
                                        {
                                            "Category": "Adventure & Nature Trails",
                                            "Summary": "Development of ecologically sustainable Mountain, Turtle, and Bird watching trails across various states to promote niche tourism.",
                                            "code": 70
                                        },
                                        {
                                            "Category": "Khelo India Mission",
                                            "Summary": "Launch of a mission to transform the Sports sector over the next decade through talent development, coach training, sports science, competitions, and infrastructure.",
                                            "code": 71
                                        }
                                    ],
                                    "Summary": "Initiatives to enhance India's tourism potential by promoting heritage, culture, and nature-based experiences, alongside a comprehensive mission to develop the sports sector.",
                                    "code": 67
                                },
                                {
                                    "Subnet": "Inclusive Regional & Vulnerable Group Development",
                                    "Categories": [
                                        {
                                            "Category": "Empowering Divyangjan",
                                            "Summary": "Divyangjan Kaushal Yojana for industry-relevant training in IT, AVGC, Hospitality; Divyang Sahara Yojana for scaling assistive device production by ALIMCO and PM Divyasha Kendras.",
                                            "code": 73
                                        },
                                        {
                                            "Category": "Purvodaya & North-Eastern Region Focus",
                                            "Summary": "Development of an integrated East Coast Industrial Corridor, 5 tourism destinations, and 4,000 e-buses for Purvodaya States; scheme for Buddhist Circuits in North-Eastern Region.",
                                            "code": 74
                                        },
                                        {
                                            "Category": "International Big Cat Alliance",
                                            "Summary": "India hosting the first Global Big Cat Summit to deliberate on collective conservation strategies for big cats.",
                                            "code": 75
                                        }
                                    ],
                                    "Summary": "Targeted efforts to empower persons with disabilities through skill development and assistive devices, accelerate development in eastern and north-eastern regions, and promote wildlife conservation.",
                                    "code": 72
                                }
                            ],
                            "Summary": "This topic encompasses government efforts to develop human capital, improve social welfare, boost rural economies, promote tourism and culture, and ensure inclusive growth for all sections of society, including vulnerable groups and specific regions.",
                            "code": 52
                        },
                        {
                            "Net": "Financial Sector & Investment Reforms",
                            "Subnets": [
                                {
                                    "Subnet": "Banking Sector Strengthening",
                                    "Categories": [
                                        {
                                            "Category": "High Level Committee on Banking",
                                            "Summary": "Proposed committee to comprehensively review the banking sector and align it with India’s next phase of growth, safeguarding financial stability and inclusion.",
                                            "code": 78
                                        },
                                        {
                                            "Category": "Strong Balance Sheets & Profitability",
                                            "Summary": "Highlights the Indian banking sector's current state of strong balance sheets, historic highs in profitability, and improved asset quality.",
                                            "code": 79
                                        },
                                        {
                                            "Category": "Rural Banking Coverage",
                                            "Summary": "Ensures banking coverage exceeding 98% of villages in the country.",
                                            "code": 80
                                        }
                                    ],
                                    "Summary": "Measures aimed at reviewing and strengthening the Indian banking sector to support future growth while ensuring stability and widespread access.",
                                    "code": 77
                                },
                                {
                                    "Subnet": "Non-Banking Financial Companies (NBFCs)",
                                    "Categories": [
                                        {
                                            "Category": "Vision for Viksit Bharat NBFCs",
                                            "Summary": "Outlined with clear targets for credit disbursement and technology adoption to contribute to India's development.",
                                            "code": 82
                                        },
                                        {
                                            "Category": "Restructuring Public Sector NBFCs",
                                            "Summary": "Proposed restructuring of Power Finance Corporation and Rural Electrification Corporation as a first step to achieve scale and improve efficiency.",
                                            "code": 83
                                        }
                                    ],
                                    "Summary": "Strategies to enhance the role and efficiency of Non-Banking Financial Companies (NBFCs) in achieving national development goals.",
                                    "code": 81
                                },
                                {
                                    "Subnet": "Capital Markets Development",
                                    "Categories": [
                                        {
                                            "Category": "Corporate Bond Market Reforms",
                                            "Summary": "Introduction of a market making framework, access to funds and derivatives on corporate bond indices, and total return swaps on corporate bonds.",
                                            "code": 85
                                        },
                                        {
                                            "Category": "Municipal Bonds Incentives",
                                            "Summary": "₹100 crore incentive for single municipal bond issuances over ₹1000 crore by large cities, with continued support for smaller towns.",
                                            "code": 86
                                        },
                                        {
                                            "Category": "Enhanced Portfolio Investment Scheme",
                                            "Summary": "Permitting Individual Persons Resident Outside India (PROI) to invest in listed Indian equities, increasing individual limit from 5% to 10% and overall limit to 24%.",
                                            "code": 87
                                        }
                                    ],
                                    "Summary": "Proposals to deepen and broaden India's capital markets, making them more attractive for both domestic and foreign investors, and supporting municipal financing.",
                                    "code": 84
                                },
                                {
                                    "Subnet": "Foreign Investment & Ease of Doing Business",
                                    "Categories": [
                                        {
                                            "Category": "Foreign Exchange Management Review",
                                            "Summary": "Comprehensive review of FEMA (Non-debt Instruments) Rules to create a more contemporary and user-friendly framework for foreign investments.",
                                            "code": 89
                                        },
                                        {
                                            "Category": "Incentivizing Data Centers & Toll Manufacturing",
                                            "Summary": "Tax holiday till 2047 for foreign companies providing cloud services via Indian data centers and income tax exemption for 5 years for non-residents providing capital goods for toll manufacturing.",
                                            "code": 90
                                        },
                                        {
                                            "Category": "Attracting Global Talent",
                                            "Summary": "Exemption of global (non-India sourced) income for non-resident experts staying for 5 years under notified schemes.",
                                            "code": 91
                                        }
                                    ],
                                    "Summary": "Reforms to liberalize and simplify the foreign investment framework, attract global businesses and talent, and improve the overall ease of doing business in India.",
                                    "code": 88
                                }
                            ],
                            "Summary": "This topic details government initiatives to strengthen the financial sector, including banking and NBFCs, develop capital markets (corporate and municipal bonds), and enhance foreign investment by streamlining regulations and offering incentives for critical sectors like data centers and manufacturing.",
                            "code": 76
                        },
                        {
                            "Net": "Taxation Modernization & Simplification",
                            "Subnets": [
                                {
                                    "Subnet": "Direct Tax Reforms",
                                    "Categories": [
                                        {
                                            "Category": "New Income Tax Act, 2025",
                                            "Summary": "Comprehensive review and enactment of a simplified act effective April 1, 2026, with redesigned forms for ease of compliance.",
                                            "code": 94
                                        },
                                        {
                                            "Category": "Ease of Living for Taxpayers",
                                            "Summary": "Exemption of MACT interest, reduced TCS rates for overseas tour packages (2%), education/medical LRS (2%), and simplified TDS processes.",
                                            "code": 95
                                        },
                                        {
                                            "Category": "Rationalizing Penalty & Prosecution",
                                            "Summary": "Integration of assessment & penalty proceedings, immunity from penalty for misreporting (with conditions), conversion of minor penalties to fees, and decriminalization of certain offenses.",
                                            "code": 96
                                        },
                                        {
                                            "Category": "Supporting IT Sector & Global Business",
                                            "Summary": "Clubbing IT services under a single category with a 15.5% safe harbor margin (threshold up to ₹2,000 crore), fast-tracking APA processes, and tax holidays for foreign cloud/data center providers.",
                                            "code": 97
                                        }
                                    ],
                                    "Summary": "Outlines a major overhaul of the direct tax system, including a new Income Tax Act, simplified compliance for individuals and businesses, rationalization of penalties, and incentives for key growth sectors.",
                                    "code": 93
                                },
                                {
                                    "Subnet": "Indirect Tax Reforms (Customs & Excise)",
                                    "Categories": [
                                        {
                                            "Category": "Customs Duty Review & Simplification",
                                            "Summary": "Removal of long-continuing exemptions, incorporation of effective rates into tariff schedule, and creation of new tariff lines for better identification and monitoring.",
                                            "code": 99
                                        },
                                        {
                                            "Category": "Export Promotion Measures",
                                            "Summary": "Increased limit for duty-free imports for seafood exports, extension of time for export of final products for leather/textile, and extension of duty exemption for shoe-uppers.",
                                            "code": 100
                                        },
                                        {
                                            "Category": "Strategic Sector Support",
                                            "Summary": "Customs duty exemptions for capital goods for Lithium-Ion cells, sodium antimonate for solar glass, nuclear power projects, critical minerals processing, and civil/defense aviation manufacturing components.",
                                            "code": 101
                                        },
                                        {
                                            "Category": "Ease of Living & Patient Relief (Customs)",
                                            "Summary": "Reduction of tariff rate on all dutiable goods for personal use from 20% to 10%, and exemption of basic customs duty on 17 specific drugs/medicines and 7 rare disease treatments.",
                                            "code": 102
                                        }
                                    ],
                                    "Summary": "Details changes in customs and excise duties aimed at simplifying tariffs, boosting domestic manufacturing and exports, supporting strategic sectors, and providing relief for citizens and patients.",
                                    "code": 98
                                },
                                {
                                    "Subnet": "GST Reforms",
                                    "Categories": [
                                        {
                                            "Category": "Post-Sale Discounts & Provisional Refunds",
                                            "Summary": "Amendments to section 15 and 34 of the CGST Act to simplify post-sale discount provisions and extend provisional refunds to inverted duty structure cases.",
                                            "code": 104
                                        },
                                        {
                                            "Category": "Refund Thresholds & Appellate Authority",
                                            "Summary": "Removal of the ₹1000 refund limit for exported goods and provisions for empowering an existing Tribunal for hearing appeals under section 101A of the CGST Act.",
                                            "code": 105
                                        },
                                        {
                                            "Category": "Place of Supply for Intermediary Services",
                                            "Summary": "Omission of clause (b) of sub-section (8) of section 13 of the IGST Act to determine place of supply for intermediary services as per default provisions.",
                                            "code": 106
                                        }
                                    ],
                                    "Summary": "Proposed amendments to the CGST and IGST Acts to refine provisions related to discounts, refunds, and the appellate mechanism, while also clarifying rules for intermediary services.",
                                    "code": 103
                                },
                                {
                                    "Subnet": "Tax Administration & Compliance",
                                    "Categories": [
                                        {
                                            "Category": "Depositary-Led Declaration Filing",
                                            "Summary": "Allowing taxpayers to file declarations for no TDS to depositories for mutual fund, interest, and dividend income, streamlining compliance.",
                                            "code": 108
                                        },
                                        {
                                            "Category": "Staggered IT Return Filing & Revised Returns",
                                            "Summary": "Extension of time for filing revised returns up to March 31st and staggered due dates for ITR 1/2 (July 31st) and non-audit business (August 31st).",
                                            "code": 109
                                        },
                                        {
                                            "Category": "Foreign Assets of Small Taxpayers Disclosure Scheme (FAST-DS)",
                                            "Summary": "Introduction of a time-bound scheme for declaration of foreign assets and foreign-sourced income below certain thresholds, offering immunity for compliance.",
                                            "code": 110
                                        },
                                        {
                                            "Category": "Customs Process Modernization",
                                            "Summary": "Enhancing duty deferral, extending validity of advance rulings, leveraging AEO accreditation, electronic sealing for export cargo, and rolling out Customs Integrated System (CIS) for digital clearance.",
                                            "code": 111
                                        }
                                    ],
                                    "Summary": "Initiatives focused on enhancing the efficiency of tax administration, simplifying compliance procedures for taxpayers, and leveraging technology for faster and more transparent processes.",
                                    "code": 107
                                },
                                {
                                    "Subnet": "Other Tax Adjustments & Rationalizations",
                                    "Categories": [
                                        {
                                            "Category": "Rationalization of Share Buyback",
                                            "Summary": "Consideration received by shareholders on buyback chargeable as Capital Gains, with differential rates for corporate (22%) and non-corporate (30%) promoters.",
                                            "code": 113
                                        },
                                        {
                                            "Category": "TCS & STT Rate Adjustments",
                                            "Summary": "Rationalization of TCS rates for specific goods (e.g., alcoholic liquor, scrap, minerals to 2%, tendu leaves to 2%) and increase in STT rates for Futures and Options in the capital market.",
                                            "code": 114
                                        },
                                        {
                                            "Category": "Minimum Alternate Tax (MAT) Changes",
                                            "Summary": "MAT to be a final tax, reduced rate from 15% to 14%, and allowance of set-off of brought forward MAT credit (25% of tax liability) for companies shifting to the new tax regime.",
                                            "code": 115
                                        },
                                        {
                                            "Category": "Cooperative Sector Support",
                                            "Summary": "Deduction of profits for primary cooperatives supplying cattle feed/cotton seed, inter-cooperative society dividend income deduction under new regime, and exemption for notified national cooperative federations.",
                                            "code": 116
                                        }
                                    ],
                                    "Summary": "Covers various specific tax proposals including adjustments to capital market taxes, rationalization of corporate tax regimes, and targeted benefits for the cooperative sector.",
                                    "code": 112
                                }
                            ],
                            "Summary": "This topic covers the comprehensive reforms and proposals related to direct and indirect taxation, including a new Income Tax Act, simplification measures for taxpayers, rationalization of penalties, and strategic adjustments to customs, excise, and GST to promote economic activities and ensure compliance.",
                            "code": 92
                        }
                    ],
                    "Summary": "This document outlines the Budget Speech for 2026-2027 by the Minister of Finance, Nirmala Sitharaman, focusing on India's economic trajectory, structural reforms, and vision for 'Viksit Bharat' (Developed India). It details comprehensive proposals across key sectors including manufacturing, infrastructure, human capital, financial services, and social welfare, alongside significant direct and indirect tax reforms aimed at simplifying compliance, promoting investment, and ensuring fiscal prudence. The speech emphasizes a threefold approach: accelerating economic growth, fulfilling aspirations through capacity building, and ensuring equitable access to resources for all sections of society, while also addressing environmental sustainability and digital transformation."
                }
            },
            "summaryList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 20766,
                        "CandidatesTokenCount": 6061,
                        "TotalTokenCount": 28546,
                        "thoughtsTokenCount": 1719,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Summary": "<h2>Comprehensive Synthesis Report: Budget 2026-2027</h2><h3>Introduction and Economic Trajectory (Budget Speech 2026-2027)</h3><p>The Union Budget for 2026-2027, presented by the Minister of Finance, Nirmala Sitharaman, on February 1, 2026, marks a continuation of India's economic journey, which has been characterized by stability, fiscal discipline, sustained growth, and moderate inflation over the past 12 years. This trajectory is attributed to deliberate choices made by the government, prioritizing action over ambivalence, reform over rhetoric, and people over populism, even amidst global uncertainties. The administration, under Prime Minister Modi, has implemented extensive structural reforms, maintained fiscal prudence and monetary stability, and significantly boosted public investment. A guiding principle has been 'Atmanirbharta' (self-reliance), fostering domestic manufacturing, ensuring energy security, and reducing critical import dependencies. Simultaneously, reforms have targeted employment generation, agricultural productivity, household purchasing power, and universal service access, resulting in an impressive 7% growth rate and substantial poverty reduction.</p><p>Currently, India faces a challenging external environment marked by disrupted trade, imperiled multilateralism, and strained supply chains. Emerging technologies are transforming production systems, increasing demands on critical resources. Despite these challenges, India is committed to confidently progressing towards 'Viksit Bharat' (Developed India), balancing ambitious growth with inclusive development. This necessitates deep integration with global markets to boost exports and attract long-term investment.</p><h3>The 'Sankalp' and Three 'Kartavyas' (Duties)</h3><p>The Budget 2026-2027, notably the first prepared in Kartavya Bhawan, is inspired by the outcomes of the 'Viksit Bharat Young Leaders Dialogue 2026', making it a 'Yuva Shakti-driven Budget'. The government's 'Sankalp' (resolve) is to prioritize the poor, underprivileged, and disadvantaged. To fulfill this, three core 'Kartavyas' have been identified:</p><ol><li><b>Accelerate and sustain economic growth:</b> By enhancing productivity and competitiveness and building resilience to global volatility.</li><li><b>Fulfill aspirations and build capacity:</b> Empowering citizens to become strong partners in India's prosperity.</li><li><b>Ensure equitable access:</b> Aligning with 'Sabka Sath, Sabka Vikas' (Together with all, Development for all), ensuring every family, community, region, and sector has access to resources, amenities, and opportunities.</li></ol><p>These duties require a supportive ecosystem, emphasizing continuous and adaptive structural reforms, a robust financial sector for capital mobilization and risk management, and the leveraging of cutting-edge technologies, including AI, for improved governance.</p><h4>Reform Express (Budget Speech 2026-2027)</h4><p>The government has undertaken comprehensive economic reforms, with over 350 rolled out since August 2025 following the Prime Minister's announcement. Key reforms include GST simplification, notification of Labour Codes, and rationalization of Quality Control Orders. High-Level Committees are working with State Governments on deregulation and reducing compliance burdens, ensuring the 'Reform Express' maintains momentum.</p><h3>Part A: Accelerating and Sustaining Economic Growth (First Kartavya)</h3><p>Interventions are proposed across six key areas:</p><h4>Scaling up Manufacturing in 7 Strategic and Frontier Sectors (Budget Speech 2026-2027)</h4><ul><li><b>Biopharma SHAKTI:</b> An outlay of ₹10,000 crores over 5 years to establish India as a global Biopharma manufacturing hub. This includes 3 new National Institutes of Pharmaceutical Education and Research (NIPER) and upgrading 7 existing ones, alongside creating a network of over 1000 accredited India Clinical Trials sites. The Central Drugs Standard Control Organisation will be strengthened with a dedicated scientific review cadre.</li><li><b>India Semiconductor Mission (ISM) 2.0:</b> Expanding capabilities in equipment, materials, full-stack Indian IP design, and fortifying supply chains. Industry-led research and training centers will be established.</li><li><b>Electronics Components Manufacturing Scheme:</b> Outlay increased to ₹40,000 crore from ₹22,919 crore to capitalize on investment momentum.</li><li><b>Rare Earth Permanent Magnets Scheme:</b> Support mineral-rich states (Odisha, Kerala, Andhra Pradesh, Tamil Nadu) to establish dedicated Rare Earth Corridors for mining, processing, research, and manufacturing.</li><li><b>Chemical Parks:</b> A scheme to support states in establishing 3 dedicated Chemical Parks via a challenge route, utilizing a cluster-based plug-and-play model to boost domestic chemical production and reduce import dependency.</li><li><b>Capital Goods Capability:</b><ul><li>Hi-Tech Tool Rooms will be established by CPSEs at 2 locations as digitally-enabled automated service bureaus.</li><li>Scheme for Enhancement of Construction and Infrastructure Equipment (CIE) to strengthen domestic manufacturing of advanced CIE.</li><li>Scheme for Container Manufacturing with a ₹10,000 crore allocation over 5 years to create a globally competitive ecosystem.</li></ul></li><li><b>Textile Sector:</b> An Integrated Programme with 5 sub-parts: National Fibre Scheme, Textile Expansion and Employment Scheme, National Handloom and Handicraft programme, Tex-Eco Initiative, and Samarth 2.0. Mega Textile Parks will be set up in challenge mode, focusing on technical textiles. The 'Mahatma Gandhi Gram Swaraj' initiative will strengthen khadi, handloom, and handicrafts through global market linkages and branding.</li><li><b>Sports Goods:</b> A dedicated initiative to promote manufacturing, research, and innovation in equipment design and material sciences.</li></ul><h4>Rejuvenation of Legacy Industrial Clusters (Budget Speech 2026-2027)</h4><p>A scheme will be introduced to revive 200 legacy industrial clusters, focusing on infrastructure and technology upgradation to improve cost competitiveness and efficiency.</p><h4>Creating “Champion MSMEs” and Supporting Micro Enterprises (Budget Speech 2026-2027)</h4><p>A three-pronged approach:</p><ul><li><b>Equity Support:</b> A dedicated ₹10,000 crore SME Growth Fund will incentivize 'Champion' enterprises. The Self-Reliant India Fund will receive a ₹2,000 crore top-up for micro enterprises.</li><li><b>Liquidity Support:</b> Four measures to leverage TReDS: mandate for CPSE purchases, credit guarantee through CGTMSE for invoice discounting, linkage of GeM with TReDS for financier information, and introduction of TReDS receivables as asset-backed securities.</li><li><b>Professional Support:</b> Facilitation for Professional Institutions to design short-term courses for 'Corporate Mitras' in Tier-II and Tier-III towns to help MSMEs with compliance.</li></ul><h4>Delivering a Powerful Push to Infrastructure (Budget Speech 2026-2027)</h4><ul><li><b>Continued Focus:</b> Expanding public infrastructure, especially in Tier II and Tier III cities (over 5 lakh population). Public capital expenditure is proposed to increase from ₹11.2 lakh crore (BE 2025-26) to ₹12.2 lakh crore (BE 2026-27).</li><li><b>Infrastructure Risk Guarantee Fund:</b> To provide partial credit guarantees to lenders, boosting private developer confidence.</li><li><b>Dedicated REITs:</b> Accelerating the recycling of CPSE real estate assets.</li><li><b>Environmentally Sustainable Cargo Movement:</b><ul><li>New Dedicated Freight Corridors connecting Dankuni (East) to Surat (West).</li><li>Operationalization of 20 new National Waterways (e.g., NW-5 in Odisha). Regional Centres of Excellence for manpower training and a ship repair ecosystem in Varanasi and Patna.</li><li>Coastal Cargo Promotion Scheme to increase inland waterways and coastal shipping's share from 6% to 12% by 2047.</li></ul></li><li><b>Last-Mile and Remote Connectivity:</b> Incentives for indigenizing seaplane manufacturing and a Seaplane VGF Scheme for operations.</li></ul><h4>Ensuring Long-Term Energy Security and Stability (Budget Speech 2026-2027)</h4><ul><li><b>Carbon Capture Utilization and Storage (CCUS):</b> An outlay of ₹20,000 crore over 5 years to achieve higher readiness levels in power, steel, cement, refineries, and chemical sectors.</li></ul><h4>Developing City Economic Regions (CER) (Budget Speech 2026-2027)</h4><ul><li><b>Focus on Tier II, Tier III, and Temple Towns:</b> To amplify cities' potential as economic powerhouses.</li><li><b>CER Mapping:</b> Allocation of ₹5000 crore per CER over 5 years for implementation through a challenge mode, with reform-cum-results based financing.</li><li><b>Environmentally Sustainable Passenger Systems:</b> Development of seven High-Speed Rail corridors as 'growth connectors' (e.g., Mumbai-Pune, Delhi-Varanasi).</li></ul><h3>Financial Sector and Emerging Technologies (Budget Speech 2026-2027)</h3><h4>Financial Sector Review (Budget Speech 2026-2027)</h4><p>India's banking sector exhibits strong balance sheets, high profitability, and improved asset quality. A 'High Level Committee on Banking for Viksit Bharat' will review the sector to align it with future growth, safeguarding stability, inclusion, and consumer protection.</p><ul><li><b>NBFCs:</b> Public Sector NBFCs like Power Finance Corporation and Rural Electrification Corporation will be restructured.</li><li><b>Foreign Exchange Management (Non-debt Instruments) Rules:</b> Comprehensive review for a more contemporary, user-friendly framework.</li><li><b>Corporate Bond Market:</b> Introduction of a market-making framework with access to funds and derivatives on corporate bond indices, and total return swaps.</li><li><b>Municipal Bonds:</b> Incentive of ₹100 crore for single issuances over ₹1000 crore by large cities. The AMRUT scheme will continue for smaller towns.</li><li><b>Ease of Doing Business:</b> Individual Persons Resident Outside India (PROI) can invest in listed Indian equities via Portfolio Investment Scheme; limits increased from 5% to 10% (individual) and 10% to 24% (all PROIs).</li></ul><h4>Emerging Technologies, Including AI (Budget Speech 2026-2027)</h4><p>The 21st century is technology-driven, benefitting farmers, women in STEM, youth, and Divyangjan. Government support includes the AI Mission, National Quantum Mission, Anusandhan National Research Fund, and Research, Development and Innovation Fund.</p><h3>Part A: Fulfilling Aspirations and Building Capacity (Second Kartavya)</h3><h4>Poverty Reduction and Services Sector Focus (Budget Speech 2026-2027)</h4><p>Nearly 25 crore individuals have moved out of multidimensional poverty. The government now emphasizes the Services Sector as a core driver for a youthful India's aspirations, aiming for a 10% global share by 2047.</p><h4>High-Powered ‘Education to Employment and Enterprise’ Standing Committee (Budget Speech 2026-2027)</h4><p>This committee will recommend measures focusing on the Services Sector, prioritizing growth, employment, and exports. Its mandate includes assessing AI's impact on jobs and skills, proposing AI integration in education curricula, upskilling tech professionals, AI-enabled job matching, making informal workflow visible, and attracting skilled diaspora and foreign talent.</p><h4>Creation of Professionals for Viksit Bharat (Budget Speech 2026-2027)</h4><ul><li><b>Health:</b> Upgrading existing and establishing new Allied Health Professional (AHP) institutions in 10 disciplines (adding 100,000 AHPs). A strong Care Ecosystem will be built, training 1.5 lakh multiskilled caregivers.</li><li><b>Hubs for Medical Value Tourism:</b> Scheme to support states in establishing five Regional Medical Hubs (PPP model), integrating medical, educational, and research facilities, including AYUSH Centres.</li><li><b>AYUSH:</b> 3 new All India Institutes of Ayurveda, upgrading AYUSH pharmacies/Drug Testing Labs, and upgrading the WHO Global Traditional Medicine Centre in Jamnagar for evidence-based research.</li><li><b>Animal Husbandry:</b> Loan-linked capital subsidy scheme for veterinary/para-vet colleges, hospitals, labs, and breeding facilities in the private sector, targeting over 20,000 professionals.</li><li><b>Orange Economy (AVGC):</b> Supporting the Indian Institute of Creative Technologies, Mumbai, to set up AVGC Content Creator Labs in 15,000 secondary schools and 500 colleges, aiming for 2 million professionals by 2030.</li><li><b>Design:</b> A new National Institute of Design will be established in the eastern region via a challenge route.</li><li><b>Education:</b> Support for states to create 5 University Townships near industrial/logistic corridors. One girls' hostel will be established in every district. 4 Telescope Infrastructure facilities will be set up or upgraded for astrophysics and astronomy.</li><li><b>Tourism:</b> National Institute of Hospitality established (upgrading NCHMCT). A pilot scheme will upskill 10,000 guides at 20 iconic sites. A National Destination Digital Knowledge Grid will be established.</li><li><b>Ecologically Sustainable Trails:</b> Development of Mountain trails (e.g., Himachal Pradesh), Turtle Trails (e.g., Odisha coast), and Bird watching trails (e.g., Pulikat Lake).</li><li><b>International Big Cat Alliance:</b> Following its establishment in 2024, India will host the first Global Big Cat Summit in 2026.</li><li><b>Heritage and Culture Tourism:</b> 15 archaeological sites (e.g., Lothal, Dholavira) will be developed into vibrant, experiential cultural destinations using immersive storytelling technologies.</li><li><b>Sports:</b> Khelo India Mission will transform the sector over the next decade through integrated talent development, coach training, sports science integration, competitions, and infrastructure development.</li></ul><h3>Part A: Ensuring Equitable Access (Third Kartavya)</h3><p>This kartavya aligns with 'Sabka Sath, Sabka Vikas' for Viksit Bharat:</p><h4>Increasing Farmer Incomes (Budget Speech 2026-2027)</h4><ul><li><b>Fisheries:</b> Integrated development of 500 reservoirs/Amrit Sarovars, strengthening value chains, and market linkages with startups and women-led groups.</li><li><b>Animal Husbandry:</b> Entrepreneurship development through a Credit-Linked Subsidy Programme, modernization of livestock enterprises, integrated-value chains, and Livestock Farmer Producers Organisations.</li><li><b>High Value Agriculture:</b> Support for high-value crops (coconut, sandalwood, cocoa, cashew), Agar trees in the North East, and nuts (almonds, walnuts, pine nuts) in hilly regions. A Coconut Promotion Scheme and dedicated programs for Indian cashew and cocoa will enhance competitiveness and create premium global brands. Focused cultivation and processing of sandalwood.</li><li><b>Bharat-VISTAAR (Virtually Integrated System to Access Agricultural Resources):</b> A multilingual AI tool integrating AgriStack portals and ICAR practices for enhanced productivity, better decision-making, and risk reduction via customized advisory support.</li><li><b>SHE-Marts for Rural Women-led Enterprises:</b> Building on the Lakhpati Didi Programme, Self-Help Entrepreneur (SHE) Marts will be community-owned retail outlets through innovative financing.</li></ul><h4>Empowering Divyangjan (Budget Speech 2026-2027)</h4><ul><li><b>Divyangjan Kaushal Yojana:</b> Customized training for dignified livelihood opportunities in IT, AVGC, Hospitality, and Food & Beverages sectors.</li><li><b>Divyang Sahara Yojana:</b> Scaling up ALIMCO's production of assistive devices, R&D in AI integration, strengthening PM Divyasha Kendras, and establishing Assistive Technology Marts.</li></ul><h4>Reaffirming Commitment to Mental Health and Trauma Care (Budget Speech 2026-2027)</h4><p>NIMHANS-2 will be set up in North India, and National Mental Health Institutes in Ranchi and Tezpur will be upgraded as Regional Apex Institutions. Emergency and Trauma Care Centres in District Hospitals will be strengthened and increased by 50%.</p><h4>Focus on Purvodaya States and North-Eastern Region (Budget Speech 2026-2027)</h4><ul><li><b>Purvodaya:</b> Development of an integrated East Coast Industrial Corridor (with Durgapur node), creation of 5 tourism destinations in 5 Purvodaya States, and provision of 4,000 e-buses.</li><li><b>Buddhist Sites in North-Eastern Region:</b> Scheme for Development of Buddhist Circuits in Arunachal Pradesh, Sikkim, Assam, Manipur, Mizoram, and Tripura, covering preservation, interpretation centers, connectivity, and pilgrim amenities.</li></ul><h4>16th Finance Commission and Fiscal Consolidation (Budget Speech 2026-2027)</h4><p>The 16th Finance Commission's report has been submitted, and the government has accepted its recommendation to retain the vertical share of devolution at 41%. ₹1.4 lakh crore is provided to states for FY 2026-27 as Finance Commission Grants. The government maintains fiscal prudence, targeting a debt-to-GDP ratio of 50±1% by 2030-31 (estimated 55.6% in BE 2026-27). The fiscal deficit commitment of below 4.5% of GDP by 2025-26 has been met (RE 2025-26: 4.4%), with a target of 4.3% in BE 2026-27.</p><h4>Budget Estimates 2026-27 (Budget Speech 2026-2027)</h4><p>Non-debt receipts are estimated at ₹36.5 lakh crore, and total expenditure at ₹53.5 lakh crore. Net tax receipts are ₹28.7 lakh crore. Net market borrowings from dated securities are ₹11.7 lakh crore, with gross market borrowings at ₹17.2 lakh crore.</p><h3>Part B: Direct Taxes</h3><h4>New Income Tax Act and Ease of Living (Budget Speech 2026-2027)</h4><p>The Income Tax Act, 2025, will come into effect from April 1, 2026, simplifying rules and forms. Measures for 'Ease of Living' include:</p><ul><li>Exemption of Motor Accident Claims Tribunal interest for natural persons from Income Tax; no TDS.</li><li>Reduction of TCS rate on overseas tour packages from 5-20% to 2%.</li><li>Reduction of TCS rate for education and medical purposes under LRS from 5% to 2%.</li><li>Supply of manpower services specifically brought under TDS for contractors (1% or 2%).</li><li>Automated process for lower/nil deduction certificates for small taxpayers.</li><li>Depositories enabled to accept Form 15G/15H for securities.</li><li>Extension of time for revising returns from Dec 31 to Mar 31 (with nominal fee).</li><li>Staggered tax return filing deadlines: Individuals (ITR 1 & 2) by July 31; non-audit business/trusts by Aug 31.</li><li>TDS on immovable property sale by non-residents through resident buyer's PAN challan (no TAN required).</li><li>One-time 6-month Foreign Asset Disclosure Scheme for small taxpayers (up to ₹1 crore undisclosed income/asset with 30% tax + 30% additional tax for category A; up to ₹5 crore asset for category B with ₹1 lakh fee, immunity from penalty/prosecution).</li></ul><h4>Rationalizing Penalty and Prosecution (Budget Speech 2026-2027)</h4><p>Measures to reduce litigation and compliance burden:</p><ul><li>Integration of assessment & penalty proceedings into a common order.</li><li>No interest liability on penalty during appeal before first appellate authority.</li><li>Pre-payment quantum reduced from 20% to 10% on core tax demand.</li><li>Taxpayers can update returns even after reassessment initiation (additional 10% tax).</li><li>Immunity from penalty/prosecution extended to 'misreporting' of income (with 100% additional income tax).</li><li>Penalties for certain technical defaults (e.g., audit failure, non-furnishing TP report) converted into a fee.</li><li>Rationalized prosecution framework: decriminalization of non-production of books/documents and TDS payment in kind; simple imprisonment (max 2 years) for other offenses, with courts having power to convert to fine; minor offenses to attract only fine.</li><li>Immunity from prosecution for non-disclosure of non-immovable foreign assets less than ₹20 lakh (retrospective from Oct 1, 2024).</li></ul><h4>Cooperatives (Budget Speech 2026-2027)</h4><ul><li>Deduction for primary cooperative societies supplying cattle feed and cotton seed produced by members.</li><li>Inter-cooperative society dividend income allowed as deduction under new tax regime (if further distributed to members).</li><li>Exemption for 3 years to dividend income received by notified national co-operative federations on investments made till Jan 31, 2026 (if further distributed).</li></ul><h4>Supporting IT Sector as India’s Growth Engine (Budget Speech 2026-2027)</h4><ul><li>Clubbing software development, IT enabled services, KPO, and contract R&D services into a single category: Information Technology Services, with a common safe harbor margin of 15.5%.</li><li>Threshold for safe harbor for IT services increased from ₹300 crore to ₹2,000 crore.</li><li>Automated, rule-driven approval for IT services safe harbor, renewable for 5 years.</li><li>Fast-tracking Unilateral APA process for IT services (within 2 years, extendable by 6 months).</li><li>Modified returns facility for APA extended to associated entities.</li></ul><h4>Attracting Global Business and Investment (Budget Speech 2026-2027)</h4><ul><li>Tax holiday till 2047 for foreign companies providing global cloud services using Indian data centers (services to Indian customers through Indian reseller).</li><li>Safe harbor of 15% on cost for related entities providing data center services in India.</li><li>Safe harbor for non-residents for component warehousing in bonded warehouses (2% profit margin on invoice value).</li><li>5-year income tax exemption for non-residents providing capital goods/equipment/tooling to toll manufacturers in bonded zones.</li><li>Exemption for global (non-India sourced) income of non-resident experts for a 5-year stay under notified schemes.</li><li>Exemption from Minimum Alternate Tax (MAT) to all non-residents paying tax on a presumptive basis.</li></ul><h4>Tax Administration and Other Tax Proposals (Budget Speech 2026-2027)</h4><ul><li>Joint Committee of MCA and CBDT to integrate ICDS requirements into IndAS.</li><li>Taxation of buyback for all shareholders as Capital Gains. Promoters will pay additional buyback tax (22% for corporate, 30% for non-corporate).</li><li>Rationalization of TCS rates for alcoholic liquor, scrap, minerals to 2%; tendu leaves reduced from 5% to 2%.</li><li>STT on Futures raised to 0.05% (from 0.02%); on options premium and exercise raised to 0.15% (from 0.1% and 0.125%).</li><li>MAT made a final tax, rate reduced to 14% (from 15%). Set-off of brought forward MAT credit (till Mar 31, 2026) allowed to companies shifting to new regime, up to 25% of tax liability.</li></ul><h3>Part B: Indirect Taxes</h3><h4>Customs and Central Excise (Budget Speech 2026-2027)</h4><p>Proposals aim to simplify tariff structure, support domestic manufacturing, promote export competitiveness, and correct duty inversion.</p><ul><li><b>Review of exemptions and tariff simplification:</b> Removal of exemptions on domestically manufactured items or negligible imports. Incorporation of effective rates into the tariff schedule for simplification.</li><li><b>Promotion of exports:</b><ul><li>Duty-free import limit for seafood processing inputs increased from 1% to 3% of previous year's FOB value.</li><li>Duty-free imports of specified inputs extended to Shoe Uppers for export.</li><li>Time period for export of final products (leather/textile garments, footwear) extended from 6 months to 1 year.</li></ul></li><li><b>Energy transition and security:</b> Basic customs duty exemption for capital goods for manufacturing Lithium-Ion Cells extended to Battery Energy Storage Systems. Exemption on sodium antimonate for solar glass.</li><li><b>Nuclear Power:</b> Existing basic customs duty exemption on imports for Nuclear Power Projects extended till 2035 and expanded for all plants.</li><li><b>Critical Minerals:</b> Basic customs duty exemption for capital goods for processing critical minerals in India.</li><li><b>Biogas blended CNG:</b> Exclusion of entire biogas value from Central Excise duty calculation on blended CNG.</li><li><b>Civil and Defence Aviation:</b> Basic customs duty exemption on components for civilian/training aircraft manufacturing. Exemption on raw materials for aircraft parts used in MRO by Defence PSUs.</li><li><b>Electronics:</b> Basic customs duty exemption on specified parts for microwave oven manufacturing.</li><li><b>Special Economic Zone (SEZ):</b> One-time measure to facilitate sales from eligible manufacturing units in SEZs to Domestic Tariff Area (DTA) at concessional rates (limited proportion of exports), with regulatory changes to ensure a level playing field.</li></ul><h4>Ease of Living (Indirect Taxes) (Budget Speech 2026-2027)</h4><ul><li>Tariff rate on all dutiable goods imported for personal use reduced from 20% to 10%.</li><li>Basic customs duty exemption on 17 drugs/medicines and for 7 more rare diseases (for personal imports).</li></ul><h4>Customs Process and Ease of Doing Business (Budget Speech 2026-2027)</h4><p>Measures for minimal intervention, smoother goods movement, and greater certainty:</p><ul><li><b>Trust-based systems:</b> Duty deferral period for Tier 2 and Tier 3 AEOs enhanced from 15 to 30 days. Manufacturer-importers given same deferral. Validity of advance ruling extended from 3 to 5 years. Recognition of regular importers for minimized verification. Factory-to-ship clearance for electronically sealed export cargo. Immediate release of non-compliance goods on arrival for trusted importers.</li><li><b>Warehouse operator-centric system:</b> Self-declarations, electronic tracking, risk-based audit for customs warehousing.</li><li><b>Single digital window:</b> Seamless processing of cargo clearance approvals from various government agencies by end of FY, with food, drugs, plant/animal products operational by April 2026.</li><li><b>Customs Integrated System (CIS):</b> Rollout in 2 years as a single, scalable platform.</li><li><b>Non-intrusive scanning:</b> Expanded use of advanced imaging and AI technology for risk assessment, aiming to scan every container at major ports.</li></ul><h4>New Export Opportunities (Budget Speech 2026-2027)</h4><ul><li>Fish catch by Indian fishing vessels in EEZ or High Seas made duty-free; landing at foreign ports treated as export.</li><li>Complete removal of ₹10 lakh value cap on courier exports. Improved handling of rejected/returned consignments.</li></ul><h4>Baggage Clearance and Dispute Resolution (Budget Speech 2026-2027)</h4><ul><li>Revised baggage rules for international travel to enhance duty-free allowances and clarify temporary carriage of goods.</li><li>Honest taxpayers can close disputes by paying an additional amount in lieu of penalty.</li></ul>"
                            }
                        ]
                    }
                }
            ]
        }
    ],
    "fileName": "",
    "projectType": "DOCUMENT",
    "sourceLanguage": "English",
    "targetLanguage": "English",
    "__v": 0,
    "headerCols": {
        "col1": "Part Id",
        "col2": "FileName",
        "col3": "Content"
    }
},
{
        "_id": "696620f322d63318094b9abd",
        "projectId": "644541",
        "projectName": "Animal Farm by George Orwell",
         "projectDescription": "",
        "projectCategory": "Literature",
        "referenceUrl": "https://bfcpublications.com/blog/top-10-public-domain-books/",
        "projectCreatedDate": "2026-01-13T10:39:46.565Z",
        "projectUpdatedDate": "2026-01-13T10:39:46.565Z",
        "projectStatus": "Active",
        "companyId": "69453c81e603a0a401b6f37b",
        "projectCreatedBy": "69453c81e603a0a401b6f37c",
        "projectUpdatedBy": "69453c81e603a0a401b6f37c",
        "projectLastAccessDate": "2026-01-29T11:31:00.969Z",
        "projectDeleted": false,
        "projectDataLength": 6,
        "projectType": "DOCUMENT",
        "cfDataLength": 1,
        "projectData": [
            {
                "col1": "Part Id",
                "col2": "FileName",
                "col3": "",
                "col4": "Status"
            },
            {
                "col1": "P1_Part_1",
                "col2": "Animal Farm by George Orwell.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_2",
                "col2": "Animal Farm by George Orwell.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_3",
                "col2": "Animal Farm by George Orwell.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_4",
                "col2": "Animal Farm by George Orwell.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_5",
                "col2": "Animal Farm by George Orwell.pdf",
                "col3": "",
                "col4": "DATA"
            }
        ],
        "sourceLanguage": "English",
        "targetLanguage": "English"
    },
    {
        "_id": "69788ddf110067d978416d9f",
        "projectId": "158704",
        "projectName": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement",
         "projectDescription": "",
        "projectCategory": "legal",
        "referenceUrl": "https://www.sci.gov.in/document/his-holiness-kesavananda-bharati-v-state-of-kerala-1973-supp-scr-1/",
        "projectCreatedDate": "2026-01-27T10:05:18.753Z",
        "projectUpdatedDate": "2026-01-27T10:05:18.753Z",
        "projectStatus": "Active",
        "companyId": "69453c81e603a0a401b6f37b",
        "projectCreatedBy": "69453c81e603a0a401b6f37c",
        "projectUpdatedBy": "69453c81e603a0a401b6f37c",
        "projectLastAccessDate": "2026-01-29T09:43:20.721Z",
        "projectDeleted": false,
        "projectDataLength": 60,
        "projectType": "DOCUMENT",
        "cfDataLength": 1,
        "projectData": [
            {
                "col1": "Part Id",
                "col2": "FileName",
                "col3": "",
                "col4": "Status"
            },
            {
                "col1": "P1_Part_1",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_2",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_3",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_4",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_5",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_6",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_7",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_8",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_9",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_10",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_11",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_12",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_13",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_14",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_15",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_16",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_17",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_18",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_19",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_20",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_21",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_22",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_23",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_24",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_25",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_26",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_27",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_28",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_29",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_30",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_31",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_32",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_33",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_34",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_35",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_36",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_37",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_38",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_39",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_40",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_41",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_42",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_43",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_44",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_45",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_46",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_47",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_48",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_49",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_50",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_51",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_52",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_53",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_54",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_55",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_56",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_57",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_58",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_59",
                "col2": "Kesavananda-Bharati-v-State-of-Kerala-1973 - Supreme-Court-of India Judgement.pdf",
                "col3": "",
                "col4": "DATA"
            }
        ],
        "sourceLanguage": "English",
        "targetLanguage": "English"
    },
    {
        "_id": "696b731a85c26b752f77e588",
        "projectId": "980987",
        "projectName": "Management of Cancer During Pregnancy ASCO Guideline (loren-et-al-2025)",
         "projectDescription": "",
        "projectCategory": "Medical",
        "referenceUrl": "https://ascopubs.org/doi/pdf/10.1200/JCO-25-02115",
        "projectCreatedDate": "2026-01-17T11:31:36.403Z",
        "projectUpdatedDate": "2026-01-17T11:31:36.403Z",
        "projectStatus": "Active",
        "companyId": "69453c81e603a0a401b6f37b",
        "projectCreatedBy": "69453c81e603a0a401b6f37c",
        "projectUpdatedBy": "69453c81e603a0a401b6f37c",
        "projectLastAccessDate": "2026-01-29T07:47:37.241Z",
        "projectDeleted": false,
        "projectDataLength": 8,
        "projectType": "DOCUMENT",
        "cfDataLength": 1,
        "projectData": [
            {
                "col1": "Part Id",
                "col2": "FileName",
                "col3": "",
                "col4": "Status"
            },
            {
                "col1": "P1_Part_1",
                "col2": "Management of Cancer During Pregnancy ASCO Guideline (loren-et-al-2025).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_2",
                "col2": "Management of Cancer During Pregnancy ASCO Guideline (loren-et-al-2025).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_3",
                "col2": "Management of Cancer During Pregnancy ASCO Guideline (loren-et-al-2025).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_4",
                "col2": "Management of Cancer During Pregnancy ASCO Guideline (loren-et-al-2025).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_5",
                "col2": "Management of Cancer During Pregnancy ASCO Guideline (loren-et-al-2025).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_6",
                "col2": "Management of Cancer During Pregnancy ASCO Guideline (loren-et-al-2025).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_7",
                "col2": "Management of Cancer During Pregnancy ASCO Guideline (loren-et-al-2025).pdf",
                "col3": "",
                "col4": "DATA"
            }
        ],
        "sourceLanguage": "English",
        "targetLanguage": "English"
    },
    {
        "_id": "6966220822d63318094b9b1f",
        "projectId": "797454",
        "projectName": "General Data Protection Regulation (GDPR) - Regulation (EU) 2016679 of the EU Parliament",
         "projectDescription": "",
        "projectCategory": "legal",
        "referenceUrl": "https://www.justice.gov/archive/oip/courtdecisions/public-domain.html",

        "projectCreatedDate": "2026-01-13T10:44:23.274Z",
        "projectUpdatedDate": "2026-01-28T12:40:40.221Z",
        "projectStatus": "Active",
        "companyId": "69453c81e603a0a401b6f37b",
        "projectCreatedBy": "69453c81e603a0a401b6f37c",
        "projectUpdatedBy": "69453c81e603a0a401b6f37c",
        "projectLastAccessDate": "2026-01-29T07:36:12.028Z",
        "projectDeleted": false,
        "projectDataLength": 9,
        "projectType": "DOCUMENT",
        "cfDataLength": 1,
        "projectData": [
            {
                "col1": "Part Id",
                "col2": "FileName",
                "col3": "",
                "col4": "Status"
            },
            {
                "col1": "P1_Part_1",
                "col2": "General Data Protection Regulation (GDPR) - REGULATION (EU) 2016679 OF THE EUROPEAN PARLIAMENT  - CELEX_32016R0679_EN_TXT.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_2",
                "col2": "General Data Protection Regulation (GDPR) - REGULATION (EU) 2016679 OF THE EUROPEAN PARLIAMENT  - CELEX_32016R0679_EN_TXT.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_3",
                "col2": "General Data Protection Regulation (GDPR) - REGULATION (EU) 2016679 OF THE EUROPEAN PARLIAMENT  - CELEX_32016R0679_EN_TXT.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_4",
                "col2": "General Data Protection Regulation (GDPR) - REGULATION (EU) 2016679 OF THE EUROPEAN PARLIAMENT  - CELEX_32016R0679_EN_TXT.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_5",
                "col2": "General Data Protection Regulation (GDPR) - REGULATION (EU) 2016679 OF THE EUROPEAN PARLIAMENT  - CELEX_32016R0679_EN_TXT.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_6",
                "col2": "General Data Protection Regulation (GDPR) - REGULATION (EU) 2016679 OF THE EUROPEAN PARLIAMENT  - CELEX_32016R0679_EN_TXT.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_7",
                "col2": "General Data Protection Regulation (GDPR) - REGULATION (EU) 2016679 OF THE EUROPEAN PARLIAMENT  - CELEX_32016R0679_EN_TXT.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_8",
                "col2": "General Data Protection Regulation (GDPR) - REGULATION (EU) 2016679 OF THE EUROPEAN PARLIAMENT  - CELEX_32016R0679_EN_TXT.pdf",
                "col3": "",
                "col4": "DATA"
            }
        ],
        "sourceLanguage": "English",
        "targetLanguage": "English"
    },
    {
        "_id": "69662a5322d63318094b9d65",
        "projectId": "348821",
        "projectName": "श्रीमद्भगवद्गीता - The Bhagavad Gita",
         "projectDescription": "",
        "projectCategory": "spirituality",
        "referenceUrl": "https://sanskritacademy.delhi.gov.in/sites/default/files/2022-09/final_geeta.pdf​",
        "projectCreatedDate": "2026-01-13T11:19:46.313Z",
        "projectUpdatedDate": "2026-01-13T11:19:46.313Z",
        "projectStatus": "Active",
        "companyId": "69453c81e603a0a401b6f37b",
        "projectCreatedBy": "69453c81e603a0a401b6f37c",
        "projectUpdatedBy": "69453c81e603a0a401b6f37c",
        "projectLastAccessDate": "2026-01-29T07:10:38.549Z",
        "projectDeleted": false,
        "projectDataLength": 14,
        "projectType": "DOCUMENT",
        "cfDataLength": 1,
        "projectData": [
            {
                "col1": "Part Id",
                "col2": "FileName",
                "col3": "",
                "col4": "Status"
            },
            {
                "col1": "P1_Part_1",
                "col2": "श्रीमद्भगवद्गीता - The Bhagavad Gita.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_2",
                "col2": "श्रीमद्भगवद्गीता - The Bhagavad Gita.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_3",
                "col2": "श्रीमद्भगवद्गीता - The Bhagavad Gita.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_4",
                "col2": "श्रीमद्भगवद्गीता - The Bhagavad Gita.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_5",
                "col2": "श्रीमद्भगवद्गीता - The Bhagavad Gita.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_6",
                "col2": "श्रीमद्भगवद्गीता - The Bhagavad Gita.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_7",
                "col2": "श्रीमद्भगवद्गीता - The Bhagavad Gita.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_8",
                "col2": "श्रीमद्भगवद्गीता - The Bhagavad Gita.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_9",
                "col2": "श्रीमद्भगवद्गीता - The Bhagavad Gita.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_10",
                "col2": "श्रीमद्भगवद्गीता - The Bhagavad Gita.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_11",
                "col2": "श्रीमद्भगवद्गीता - The Bhagavad Gita.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_12",
                "col2": "श्रीमद्भगवद्गीता - The Bhagavad Gita.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_13",
                "col2": "श्रीमद्भगवद्गीता - The Bhagavad Gita.pdf",
                "col3": "",
                "col4": "DATA"
            }
        ],
        "sourceLanguage": "Sanskrit",
        "targetLanguage": "Sanskrit"
    },
    {
        "_id": "69662e7a22d63318094b9ebf",
        "projectId": "296019",
        "projectName": "Roe v. Wade, 410 U.S. 113 (1973)",
         "projectDescription": "",
        "projectCategory": "legal",
        "referenceUrl": "https://www.justice.gov/archive/oip/courtdecisions/public-domain.html",
        "projectCreatedDate": "2026-01-13T11:37:29.680Z",
        "projectUpdatedDate": "2026-01-13T11:37:29.680Z",
        "projectStatus": "Active",
        "companyId": "69453c81e603a0a401b6f37b",
        "projectCreatedBy": "69453c81e603a0a401b6f37c",
        "projectUpdatedBy": "69453c81e603a0a401b6f37c",
        "projectLastAccessDate": "2026-01-28T11:43:35.824Z",
        "projectDeleted": false,
        "projectDataLength": 4,
        "projectType": "DOCUMENT",
        "cfDataLength": 1,
        "projectData": [
            {
                "col1": "Part Id",
                "col2": "FileName",
                "col3": "",
                "col4": "Status"
            },
            {
                "col1": "P1_Part_1",
                "col2": "Roe v. Wade, 410 U.S. 113 (1973).docx",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_2",
                "col2": "Roe v. Wade, 410 U.S. 113 (1973).docx",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_3",
                "col2": "Roe v. Wade, 410 U.S. 113 (1973).docx",
                "col3": "",
                "col4": "DATA"
            }
        ],
        "sourceLanguage": "English",
        "targetLanguage": "English"
    },
    {
        "_id": "6966218222d63318094b9aef",
        "projectId": "773378",
        "projectName": "Constitution of the United States",
         "projectDescription": "",
        "projectCategory": "Constitutions",
        "referenceUrl": "https://constitutioncenter.org/media/files/constitution.pdf",
        "projectCreatedDate": "2026-01-13T10:42:09.874Z",
        "projectUpdatedDate": "2026-01-13T10:42:09.874Z",
        "projectStatus": "Active",
        "companyId": "69453c81e603a0a401b6f37b",
        "projectCreatedBy": "69453c81e603a0a401b6f37c",
        "projectUpdatedBy": "69453c81e603a0a401b6f37c",
        "projectLastAccessDate": "2026-01-28T11:19:13.767Z",
        "projectDeleted": false,
        "projectDataLength": 3,
        "projectType": "DOCUMENT",
        "cfDataLength": 1,
        "projectData": [
            {
                "col1": "Part Id",
                "col2": "FileName",
                "col3": "",
                "col4": "Status"
            },
            {
                "col1": "P1_Part_1",
                "col2": "Constitution of the United States.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_2",
                "col2": "Constitution of the United States.pdf",
                "col3": "",
                "col4": "DATA"
            }
        ],
        "sourceLanguage": "English",
        "targetLanguage": "English"
    },
    {
        "_id": "6966269f22d63318094b9c3f",
        "projectId": "297238",
        "projectName": "The Constitution of India 2024",
         "projectDescription": "",
        "projectCategory": "Constitutions",
        "referenceUrl": "http://legislative.gov.in/documents/constitution-of-india/constitution-of-india-AjN2EjMtQWa?pageTitle=Constitution-of-India",
        "projectCreatedDate": "2026-01-13T11:03:58.418Z",
        "projectUpdatedDate": "2026-01-28T12:52:07.228Z",
        "projectStatus": "Active",
        "companyId": "69453c81e603a0a401b6f37b",
        "projectCreatedBy": "69453c81e603a0a401b6f37c",
        "projectUpdatedBy": "69453c81e603a0a401b6f37c",
        "projectLastAccessDate": "2026-01-28T06:29:34.148Z",
        "projectDeleted": false,
        "projectDataLength": 22,
        "projectType": "DOCUMENT",
        "cfDataLength": 1,
        "projectData": [
            {
                "col1": "Part Id",
                "col2": "FileName",
                "col3": "",
                "col4": "Status"
            },
            {
                "col1": "P1_Part_1",
                "col2": "The Constitution of India 2024 (English Version).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_2",
                "col2": "The Constitution of India 2024 (English Version).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_3",
                "col2": "The Constitution of India 2024 (English Version).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_4",
                "col2": "The Constitution of India 2024 (English Version).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_5",
                "col2": "The Constitution of India 2024 (English Version).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_6",
                "col2": "The Constitution of India 2024 (English Version).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_7",
                "col2": "The Constitution of India 2024 (English Version).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_8",
                "col2": "The Constitution of India 2024 (English Version).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_9",
                "col2": "The Constitution of India 2024 (English Version).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_10",
                "col2": "The Constitution of India 2024 (English Version).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_11",
                "col2": "The Constitution of India 2024 (English Version).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_12",
                "col2": "The Constitution of India 2024 (English Version).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_13",
                "col2": "The Constitution of India 2024 (English Version).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_14",
                "col2": "The Constitution of India 2024 (English Version).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_15",
                "col2": "The Constitution of India 2024 (English Version).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_16",
                "col2": "The Constitution of India 2024 (English Version).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_17",
                "col2": "The Constitution of India 2024 (English Version).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_18",
                "col2": "The Constitution of India 2024 (English Version).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_19",
                "col2": "The Constitution of India 2024 (English Version).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_20",
                "col2": "The Constitution of India 2024 (English Version).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_21",
                "col2": "The Constitution of India 2024 (English Version).pdf",
                "col3": "",
                "col4": "DATA"
            }
        ],
        "sourceLanguage": "English",
        "targetLanguage": "English"
    },
    {
        "_id": "6978a9aaf6cb082d800e2d62",
        "projectId": "200484",
        "projectName": "Chhatrapati Shivaji - His Life and Times by  Gajanan Bhaskar Mehendale",
         "projectDescription": "",
        "projectCategory": "history",
        "referenceUrl": "",
        "projectCreatedDate": "2026-01-27T12:03:54.328Z",
        "projectUpdatedDate": "2026-01-28T12:50:12.187Z",
        "projectStatus": "Active",
        "companyId": "69453c81e603a0a401b6f37b",
        "projectCreatedBy": "69453c81e603a0a401b6f37c",
        "projectUpdatedBy": "69453c81e603a0a401b6f37c",
        "projectLastAccessDate": "2026-01-28T05:11:52.366Z",
        "projectDeleted": false,
        "projectDataLength": 5,
        "projectType": "DOCUMENT",
        "cfDataLength": 1,
        "projectData": [
            {
                "col1": "Part Id",
                "col2": "FileName",
                "col3": "",
                "col4": "Status"
            },
            {
                "col1": "P1_Part_1",
                "col2": "Shivaji His Life and Times PDF.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_2",
                "col2": "Shivaji His Life and Times PDF.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_3",
                "col2": "Shivaji His Life and Times PDF.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_4",
                "col2": "Shivaji His Life and Times PDF.pdf",
                "col3": "",
                "col4": "DATA"
            }
        ],
        "sourceLanguage": "English",
        "targetLanguage": "English"
    },
    {
        "_id": "69789ab2110067d978417219",
        "projectId": "241960",
        "projectName": "National Education Policy 2020 (Government of India) ",
         "projectDescription": "",
        "projectCategory": "legal",
        "referenceUrl": "https://www.education.gov.in/sites/upload_files/mhrd/files/NEP_Final_English_0.pdf",
        "projectCreatedDate": "2026-01-27T11:00:02.379Z",
        "projectUpdatedDate": "2026-01-28T12:45:02.264Z",
        "projectStatus": "Active",
        "companyId": "69453c81e603a0a401b6f37b",
        "projectCreatedBy": "69453c81e603a0a401b6f37c",
        "projectUpdatedBy": "69453c81e603a0a401b6f37c",
        "projectLastAccessDate": "2026-01-27T11:04:50.109Z",
        "projectDeleted": false,
        "projectDataLength": 8,
        "projectType": "DOCUMENT",
        "cfDataLength": 1,
        "projectData": [
            {
                "col1": "Part Id",
                "col2": "FileName",
                "col3": "",
                "col4": "Status"
            },
            {
                "col1": "P1_Part_1",
                "col2": "National Education Policy 2020 (India) .pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_2",
                "col2": "National Education Policy 2020 (India) .pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_3",
                "col2": "National Education Policy 2020 (India) .pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_4",
                "col2": "National Education Policy 2020 (India) .pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_5",
                "col2": "National Education Policy 2020 (India) .pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_6",
                "col2": "National Education Policy 2020 (India) .pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_7",
                "col2": "National Education Policy 2020 (India) .pdf",
                "col3": "",
                "col4": "DATA"
            }
        ],
        "sourceLanguage": "English",
        "targetLanguage": "English"
    },
    {
        "_id": "69789461110067d978416f80",
        "projectId": "494622",
        "projectName": "McKinsey - Empowering People to Unlock AI's  Full Potential",
         "projectDescription": "",
        "projectCategory": "Whitepapers",
        "referenceUrl": "https://www.mckinsey.com/capabilities/tech-and-ai/our-insights/superagency-in-the-workplace-empowering-people-to-unlock-ais-full-potential-at-work",
        "projectCreatedDate": "2026-01-27T10:33:04.701Z",
        "projectUpdatedDate": "2026-01-27T11:24:53.060Z",
        "projectStatus": "Active",
        "companyId": "69453c81e603a0a401b6f37b",
        "projectCreatedBy": "69453c81e603a0a401b6f37c",
        "projectUpdatedBy": "69453c81e603a0a401b6f37c",
        "projectLastAccessDate": "2026-01-27T10:39:00.921Z",
        "projectDeleted": false,
        "projectDataLength": 3,
        "projectType": "DOCUMENT",
        "cfDataLength": 1,
        "projectData": [
            {
                "col1": "Part Id",
                "col2": "FileName",
                "col3": "",
                "col4": "Status"
            },
            {
                "col1": "P1_Part_1",
                "col2": "McKinsey - Empowering People- Unlock AIs  Full Potential.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_2",
                "col2": "McKinsey - Empowering People- Unlock AIs  Full Potential.pdf",
                "col3": "",
                "col4": "DATA"
            }
        ],
        "sourceLanguage": "English",
        "targetLanguage": "English"
    },
    {
        "_id": "696625b422d63318094b9bf7",
        "projectId": "672610",
        "projectName": "Startup India / DPIIT policy handbook ",
         "projectDescription": "",
        "projectCategory": "Whitepapers",
        "referenceUrl": "https://www.startupindia.gov.in/content/dam/invest-india/Templates/public/Startup%20India%20Kit_Digital_Jan19.pdf",
        "projectCreatedDate": "2026-01-13T11:00:03.834Z",
        "projectUpdatedDate": "2026-01-28T12:51:27.452Z",
        "projectStatus": "Active",
        "companyId": "69453c81e603a0a401b6f37b",
        "projectCreatedBy": "69453c81e603a0a401b6f37c",
        "projectUpdatedBy": "69453c81e603a0a401b6f37c",
        "projectLastAccessDate": "2026-01-27T10:24:19.897Z",
        "projectDeleted": false,
        "projectDataLength": 2,
        "projectType": "DOCUMENT",
        "cfDataLength": 1,
        "projectData": [
            {
                "col1": "Part Id",
                "col2": "FileName",
                "col3": "",
                "col4": "Status"
            },
            {
                "col1": "P1",
                "col2": "Startup India Kit_v5.pdf",
                "col3": "",
                "col4": "DATA"
            }
        ],
        "sourceLanguage": "English",
        "targetLanguage": "English"
    },
    {
        "_id": "697782ff110067d978415fb9",
        "projectId": "943260",
        "projectName": "Optimal Medical Therapy with or without PCI",
         "projectDescription": "The new england journal of medicine  n engl j med 356;15 www.nejm.org april 12, 2007",
        "projectCategory": "Medical",
        "referenceUrl": "",
        "projectCreatedDate": "2026-01-26T15:06:38.787Z",
        "projectUpdatedDate": "2026-01-26T15:06:38.787Z",
        "projectStatus": "Active",
        "companyId": "69453c81e603a0a401b6f37b",
        "projectCreatedBy": "69453c81e603a0a401b6f37c",
        "projectUpdatedBy": "69453c81e603a0a401b6f37c",
        "projectLastAccessDate": "2026-01-27T10:24:01.586Z",
        "projectDeleted": false,
        "projectDataLength": 3,
        "projectType": "DOCUMENT",
        "cfDataLength": 1,
        "projectData": [
            {
                "col1": "Part Id",
                "col2": "FileName",
                "col3": "",
                "col4": "Status"
            },
            {
                "col1": "P1_Part_1",
                "col2": "Optimal Medical Therapy with or without PCI.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_2",
                "col2": "Optimal Medical Therapy with or without PCI.pdf",
                "col3": "",
                "col4": "DATA"
            }
        ],
        "sourceLanguage": "English",
        "targetLanguage": "English"
    },
    {
        "_id": "6978758d110067d9784166bf",
        "projectId": "208018",
        "projectName": "CDC Guideline for Prescribing Opioids for Chronic Pain — United States, 2016",
         "projectDescription": "",
        "projectCategory": "Medical",
        "referenceUrl": "https://www.cdc.gov/mmwr/volumes/65/rr/rr6501e1.htm#print",
        "projectCreatedDate": "2026-01-27T08:21:32.161Z",
        "projectUpdatedDate": "2026-01-28T12:36:58.891Z",
        "projectStatus": "Active",
        "companyId": "69453c81e603a0a401b6f37b",
        "projectCreatedBy": "69453c81e603a0a401b6f37c",
        "projectUpdatedBy": "69453c81e603a0a401b6f37c",
        "projectLastAccessDate": "2026-01-27T10:23:15.350Z",
        "projectDeleted": false,
        "projectDataLength": 7,
        "projectType": "DOCUMENT",
        "cfDataLength": 1,
        "projectData": [
            {
                "col1": "Part Id",
                "col2": "FileName",
                "col3": "",
                "col4": "Status"
            },
            {
                "col1": "P1_Part_1",
                "col2": "CDC Guideline for Prescribing Opioids for Chronic Pain — United States, 2016 _ MMWR.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_2",
                "col2": "CDC Guideline for Prescribing Opioids for Chronic Pain — United States, 2016 _ MMWR.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_3",
                "col2": "CDC Guideline for Prescribing Opioids for Chronic Pain — United States, 2016 _ MMWR.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_4",
                "col2": "CDC Guideline for Prescribing Opioids for Chronic Pain — United States, 2016 _ MMWR.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_5",
                "col2": "CDC Guideline for Prescribing Opioids for Chronic Pain — United States, 2016 _ MMWR.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_6",
                "col2": "CDC Guideline for Prescribing Opioids for Chronic Pain — United States, 2016 _ MMWR.pdf",
                "col3": "",
                "col4": "DATA"
            }
        ],
        "sourceLanguage": "English",
        "targetLanguage": "English"
    },
    {
        "_id": "69787941110067d97841683a",
        "projectId": "600083",
        "projectName": "Vishaka v. State of Rajasthan (sexual harassment guidelines). ",
         "projectDescription": "",
        "projectCategory": "legal",
        "referenceUrl": "https://jharkhandhighcourt.nic.in/ihcj/orders/wp_criminal_No_666_670_1992.pdf",
        "projectCreatedDate": "2026-01-27T08:37:20.268Z",
        "projectUpdatedDate": "2026-01-28T12:53:15.765Z",
        "projectStatus": "Active",
        "companyId": "69453c81e603a0a401b6f37b",
        "projectCreatedBy": "69453c81e603a0a401b6f37c",
        "projectUpdatedBy": "69453c81e603a0a401b6f37c",
        "projectLastAccessDate": "2026-01-27T10:21:58.542Z",
        "projectDeleted": false,
        "projectDataLength": 2,
        "projectType": "DOCUMENT",
        "cfDataLength": 1,
        "projectData": [
            {
                "col1": "Part Id",
                "col2": "FileName",
                "col3": "",
                "col4": "Status"
            },
            {
                "col1": "P1",
                "col2": "Vishaka & Ors vs State Of Rajasthan & Ors _ 13 August, 1997.pdf",
                "col3": "",
                "col4": "DATA"
            }
        ],
        "sourceLanguage": "English",
        "targetLanguage": "English"
    },
    {
        "_id": "69788c8e110067d978416d39",
        "projectId": "753508",
        "projectName": "Maneka Gandhi v. Union of India ",
         "projectDescription": "",
        "projectCategory": "legal",
        "referenceUrl": "",
        "projectCreatedDate": "2026-01-27T09:59:42.129Z",
        "projectUpdatedDate": "2026-01-27T10:04:46.822Z",
        "projectStatus": "Active",
        "companyId": "69453c81e603a0a401b6f37b",
        "projectCreatedBy": "69453c81e603a0a401b6f37c",
        "projectUpdatedBy": "69453c81e603a0a401b6f37c",
        "projectLastAccessDate": "2026-01-27T10:21:22.301Z",
        "projectDeleted": false,
        "projectDataLength": 12,
        "projectType": "DOCUMENT",
        "cfDataLength": 1,
        "projectData": [
            {
                "col1": "Part Id",
                "col2": "FileName",
                "col3": "",
                "col4": "Status"
            },
            {
                "col1": "P1_Part_1",
                "col2": "Maneka Gandhi v. Union of India.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_2",
                "col2": "Maneka Gandhi v. Union of India.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_3",
                "col2": "Maneka Gandhi v. Union of India.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_4",
                "col2": "Maneka Gandhi v. Union of India.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_5",
                "col2": "Maneka Gandhi v. Union of India.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_6",
                "col2": "Maneka Gandhi v. Union of India.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_7",
                "col2": "Maneka Gandhi v. Union of India.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_8",
                "col2": "Maneka Gandhi v. Union of India.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_9",
                "col2": "Maneka Gandhi v. Union of India.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_10",
                "col2": "Maneka Gandhi v. Union of India.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_11",
                "col2": "Maneka Gandhi v. Union of India.pdf",
                "col3": "",
                "col4": "DATA"
            }
        ],
        "sourceLanguage": "English",
        "targetLanguage": "English"
    },
    {
        "_id": "697882bb110067d978416abd",
        "projectId": "742694",
        "projectName": "Companies Act, 2013 – Director Duties & Governance -  Ch XI, XII and Schedule IV",
         "projectDescription": "",
        "projectCategory": "Constitutions",
        "referenceUrl": "https://www.indiacode.nic.in/bitstream/123456789/2114/5/A2013-18.pdf",
        "projectCreatedDate": "2026-01-27T09:17:46.565Z",
        "projectUpdatedDate": "2026-01-27T09:17:46.565Z",
        "projectStatus": "Active",
        "companyId": "69453c81e603a0a401b6f37b",
        "projectCreatedBy": "69453c81e603a0a401b6f37c",
        "projectUpdatedBy": "69453c81e603a0a401b6f37c",
        "projectLastAccessDate": "2026-01-27T09:19:28.226Z",
        "projectDeleted": false,
        "projectDataLength": 4,
        "projectType": "DOCUMENT",
        "cfDataLength": 1,
        "projectData": [
            {
                "col1": "Part Id",
                "col2": "FileName",
                "col3": "",
                "col4": "Status"
            },
            {
                "col1": "P1_Part_1",
                "col2": "Companies Act, 2013 – Director Duties & Governance -  Ch XI, XII and Schedule IV.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_2",
                "col2": "Companies Act, 2013 – Director Duties & Governance -  Ch XI, XII and Schedule IV.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_3",
                "col2": "Companies Act, 2013 – Director Duties & Governance -  Ch XI, XII and Schedule IV.pdf",
                "col3": "",
                "col4": "DATA"
            }
        ],
        "sourceLanguage": "English",
        "targetLanguage": "English"
    },
    {
        "_id": "69787c5c110067d97841691c",
        "projectId": "475829",
        "projectName": "Right to Information Act 2005 (India). ",
         "projectDescription": "",
        "projectCategory": "Constitutions",
        "referenceUrl": "https://rti.dopt.gov.in/Writereaddata/RTI-Act.pdf",
        "projectCreatedDate": "2026-01-27T08:50:36.298Z",
        "projectUpdatedDate": "2026-01-28T12:46:20.521Z",
        "projectStatus": "Active",
        "companyId": "69453c81e603a0a401b6f37b",
        "projectCreatedBy": "69453c81e603a0a401b6f37c",
        "projectUpdatedBy": "69453c81e603a0a401b6f37c",
        "projectLastAccessDate": "2026-01-27T08:52:15.004Z",
        "projectDeleted": false,
        "projectDataLength": 3,
        "projectType": "DOCUMENT",
        "cfDataLength": 1,
        "projectData": [
            {
                "col1": "Part Id",
                "col2": "FileName",
                "col3": "",
                "col4": "Status"
            },
            {
                "col1": "P1_Part_1",
                "col2": "Right to Information Act 2005.pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_2",
                "col2": "Right to Information Act 2005.pdf",
                "col3": "",
                "col4": "DATA"
            }
        ],
        "sourceLanguage": "English",
        "targetLanguage": "English"
    },
    {
        "_id": "69787669110067d97841673a",
        "projectId": "571829",
        "projectName": "Digital Data Protection Act 2023 (India)",
         "projectDescription": "",
        "projectCategory": "legal",
        "referenceUrl": "https://www.meity.gov.in/static/uploads/2024/06/2bf1f0e9f04e6fb4f8fef35e82c42aa5.pdf",
        "projectCreatedDate": "2026-01-27T08:25:12.469Z",
        "projectUpdatedDate": "2026-01-27T08:25:12.469Z",
        "projectStatus": "Active",
        "companyId": "69453c81e603a0a401b6f37b",
        "projectCreatedBy": "69453c81e603a0a401b6f37c",
        "projectUpdatedBy": "69453c81e603a0a401b6f37c",
        "projectLastAccessDate": "2026-01-27T08:27:23.591Z",
        "projectDeleted": false,
        "projectDataLength": 3,
        "projectType": "DOCUMENT",
        "cfDataLength": 1,
        "projectData": [
            {
                "col1": "Part Id",
                "col2": "FileName",
                "col3": "",
                "col4": "Status"
            },
            {
                "col1": "P1_Part_1",
                "col2": "Digital Data Protection Act 2023 (India).pdf",
                "col3": "",
                "col4": "DATA"
            },
            {
                "col1": "P1_Part_2",
                "col2": "Digital Data Protection Act 2023 (India).pdf",
                "col3": "",
                "col4": "DATA"
            }
        ],
        "sourceLanguage": "English",
        "targetLanguage": "English"
    },
    {
        "_id": "694775ea1ad3a35bdb27b6d5",
        "projectId": "653604",
        "projectName": "John Cockerill India Ltd vs Sanjay Kamalakar Navare - Unstamped NDA ",
        "projectDescription": "John Cockerill India Ltd - NDA Caselaw",
        "projectCreatedDate": "2025-12-21T04:22:01.195Z",
        "projectUpdatedDate": "2026-01-28T12:10:28.213Z",
        "projectStatus": "Active",
        "companyId": "69453c81e603a0a401b6f37b",
        "projectCreatedBy": "69453c81e603a0a401b6f37c",
        "projectUpdatedBy": "69453c81e603a0a401b6f37c",
        "projectLastAccessDate": "2026-01-27T08:16:17.994Z",
        "projectDeleted": false,
        "projectDataLength": 2,
        "projectType": "DOCUMENT",
        "cfDataLength": 1,
        "projectData": [
            {
                "col1": "Part Id",
                "col2": "FileName",
                "col3": "",
                "col4": "Status"
            },
            {
                "col1": "P1",
                "col2": "John Cockerill India Ltd - NDA Caselaw.pdf",
                "col3": "",
                "col4": "DATA"
            }
        ],
        "sourceLanguage": "English",
        "targetLanguage": "English"
    },
    {
    "_id": "695bb574d58ea40715c6bc38",
    "projectId": "204627",
    "projectName": "Biology-K11B10 Cell Cycle and Cell Division",
    "projectDescription": "",
    "projectCategory": "Educational",
    "projectCreatedDate": "2026-01-05T12:58:27.163Z",
    "projectUpdatedDate": "2026-01-05T12:58:27.163Z",
    "projectStatus": "Active",
    "companyId": "6957bd26d58ea40715c6ae92",
    "projectCreatedBy": "6957bd26d58ea40715c6ae93",
    "projectUpdatedBy": "6957bd26d58ea40715c6ae93",
    "projectLastAccessDate": "2026-01-30T11:15:18.963Z",
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "Content",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "K11B10 Cell Cycle and Cell Division.pdf",
            "col3": "120   B IOLOGY  Are you aware that all organisms, even the largest, start their life from a single cell? You may wonder how a single cell then goes on to form such large organisms. Growth and reproduction are characteristics of cells, indeed of all living organisms. All cells reproduce by dividing into two, with each parental cell giving rise to two daughter cells each time they divide. These newly formed daughter cells can themselves grow and divide, giving rise to a new cell population that is formed by the growth and division of a single parental cell and its progeny. In other words, such cycles of growth and division allow a single cell to form a structure consisting of millions of cells.  10.1   C ELL   C YCLE  Cell division is a very important process in all living organisms. During the division of a cell, DNA replication and cell growth also take place. All these processes, i.e., cell division, DNA replication, and cell growth, hence, have to take place in a coordinated way to ensure correct division and formation of progeny cells containing intact genomes. The sequence of events by which a cell duplicates its genome, synthesises the other constituents of the cell and eventually divides into two daughter cells is termed   cell cycle.   Although cell growth (in terms of cytoplasmic increase) is a continuous process, DNA synthesis occurs only during one specific stage in the cell cycle. The replicated chromosomes (DNA) are then distributed to daughter nuclei by a complex series of events during cell division. These events are themselves under genetic control.  C ELL   C YCLE   AND   C ELL   D IVISION  C HAPTER   10  10.1 Cell Cycle 10.2 M Phase 10.3 Significance of Mitosis 10.4 Meiosis 10.5 Significance of Meiosis Reprint 2025-26C ELL   C YCLE   AND   C ELL   D IVISION   121  10.1.1 Phases of Cell Cycle  A typical eukaryotic cell cycle is illustrated by human cells in culture. These cells divide once in approximately every 24 hours (Figure 10.1). However, this duration of cell cycle can vary from organism to organism and also from cell type to cell type. Yeast for example, can progress through the cell cycle in only about 90 minutes. The cell cycle is divided into two basic phases:  l l l l l   Interphase  l l l l l   M Phase (Mitosis phase)  The M Phase represents the phase when the actual cell division or mitosis occurs and the interphase represents the phase between two successive M phases. It is significant to note that in the 24 hour average duration of cell cycle of a human cell, cell division proper lasts for only about an hour. The interphase lasts more than 95% of the duration of cell cycle. The M Phase starts with the nuclear division, corresponding to the separation of daughter chromosomes   (karyokinesis)   and usually ends with division of cytoplasm   (cytokinesis).   The interphase, though called the resting phase, is the time during which the cell is preparing for division by undergoing both cell growth and DNA replication in an orderly manner. The interphase is divided into three further phases:  l l l l l   G 1   phase (Gap 1)  l l l l l   S phase (Synthesis)  l l l l l   G 2   phase (Gap 2)  G 1   phase corresponds to the interval between mitosis and initiation of DNA replication. During G 1   phase, the cell is metabolically active and continuously grows but does not replicate its DNA. S or   synthesis   phase marks the period during which DNA synthesis or replication takes place. During this time the amount of DNA per cell doubles. If the initial amount of DNA is denoted as 2C then it increases to 4C. However, there is no increase in the chromosome number; if the cell had diploid or 2n number of chromosomes at G 1 , even after S phase the number of chromosomes remains the same, i.e., 2n. In animal cells, during the S phase, DNA replication begins in the nucleus, and the centriole duplicates in the cytoplasm. During the G 2  phase, proteins are synthesised in preparation for mitosis while cell growth continues.  How do plants and animals continue to grow all their lives? Do all cells in a plant divide all the time? Do you think all cells continue to divide in all   plants   and animals?   Can   you tell the name and the location   of   tissues having   cells   that divide all their life in higher   plants?   Do animals have similar m e r i s t e m a t i c tissues?  Figure 10.1   A diagrammatic view of cell cycle indicating formation of two cells from one cell  M Phase Reprint 2025-26122   B IOLOGY  Some cells in the adult animals do not appear to exhibit division (e.g., heart cells) and many other cells divide only occasionally , as needed to replace cells that have been lost because of injury or cell death. These cells that do not divide further exit G 1   phase to enter an inactive stage called  quiescent stage (G 0 )   of the cell cycle. Cells in this stage remain metabolically active but no longer proliferate unless called on to do so depending on the requirement of the organism. In animals, mitotic cell division is only seen in the diploid somatic cells. However, there are few exceptions to this where haploid cells divide by mitosis, for example, male honey bees. Against this, the plants can show mitotic divisions in both haploid and diploid cells. From your recollection of examples of alternation of generations in plants (Chapter 3) identify plant species and stages at which mitosis is seen in haploid cells.  10.2   M P HASE  This is the most dramatic period of the cell cycle, involving a major reorganisation of virtually all components of the cell. Since the number of chromosomes in the parent and progeny cells is the same, it is also called as  equational division .   Though for convenience mitosis has been divided into four stages of nuclear division (karyokinesis), it is very essential to understand that cell division is a progressive process and very clear-cut lines cannot be drawn between various stages. Karyokinesis involves following four stages:  l l l l l  Prophase  l l l l l  Metaphase  l l l l l  Anaphase  l l l l l  Telophase  10.2.1   Prophase  Prophase which is the first stage of karyokinesis of mitosis follows the S and G 2   phases of interphase. In the S and G 2   phases, the new DNA molecules formed are not distinct but intertwined. Prophase is marked by the initiation of condensation of chromosomal material. The chromosomal material becomes untangled during the process of chromatin condensation (Figure 10.2 a). The centrosome, which had undergone duplication during S phase of interphase, now begins to move towards opposite poles of the cell. The completion of prophase can thus be marked by the following characteristic events:  l l l l l   Chromosomal material condenses to form compact mitotic chromosomes. Chromosomes are seen to be composed of two chromatids attached together at the centromere.  l l l l l   Centrosome which had undergone duplication during interphase, begins to move towards opposite poles of the cell. Each centrosome radiates out microtubules called asters. The two asters together with spindle fibres forms mitotic apparatus.  You   have   studied mitosis in onion root tip cells.   It has 16 chromosomes   in each cell.   Can you tell   how   many chromosomes   will the cell have at G 1  phase, after S phase, and after M phase? Also, what will be the DNA content of the cells at G 1 , after S and   at   G 2 ,   if   the content   after   M phase is 2C? Reprint 2025-26C ELL   C YCLE   AND   C ELL   D IVISION   123  Cells at the end of prophase, when viewed under the microscope, do not show golgi complexes, endoplasmic reticulum, nucleolus and the nuclear envelope.  10.2.2   Metaphase  The complete disintegration of the nuclear envelope marks the start of the second phase of mitosis, hence the chromosomes are spread through the cytoplasm of the cell. By this stage, condensation of chromosomes is completed and they can be observed clearly under the microscope. This then, is the stage at which morphology of chromosomes is most easily studied. At this stage, metaphase chromosome is made up of two sister chromatids, which are held together by the centromere (Figure 10.2 b). Small disc-shaped structures at the surface of the centromeres are called kinetochores. These structures serve as the sites of attachment of spindle fibres to the chromosomes that are moved into position at the centre of the cell. Hence, the metaphase is characterised by all the chromosomes coming to lie at the equator with one chromatid of each chromosome connected by its kinetochore to spindle fibres from one pole and its sister chromatid connected by its kinetochore to spindle fibres from the opposite pole (Figure 10.2 b). The plane of alignment of the chromosomes at metaphase is referred to as the   metaphase plate.   The key features of metaphase are:  l l l l l   Spindle   fibres   attach   to   kinetochores   of chromosomes.  l l l l l   Chromosomes are moved to spindle equator and get aligned along metaphase plate through spindle fibres to both poles.  10.2.3   Anaphase  At the onset of anaphase, each chromosome arranged at the metaphase plate is split simultaneously and the two daughter chromatids, now referred to as daughter chromosomes of the future daughter nuclei, begin their migration towards the two opposite poles. As each chromosome moves away from the equatorial plate, the centromere of each chromosome remains directed towards the pole and hence at the leading edge, with the arms of the chromosome trailing behind (Figure 10.2 c). Thus, anaphase stage is characterised by   Figure 10.2 a and b   : A diagrammatic view of stages in mitosis Reprint 2025-26124   B IOLOGY  the following key events:  l l l l l   Centromeres split and chromatids separate.  l l l l l   Chromatids move to opposite poles.  10.2.4   Telophase  At the beginning of the final stage of karyokinesis, i.e., telophase, the chromosomes that have reached their respective poles decondense and lose their individuality. The individual chromosomes can no longer be seen and each set of chromatin material tends to collect at each of the two poles (Figure 10.2 d). This is the stage which shows the following key events:  l l l l l   Chromosomes cluster at opposite spindle poles and their identity is lost as discrete elements.  l l l l l   Nuclear envelope develops around the chromosome clusters at each pole forming two daughter nuclei.  l l l l l   Nucleolus, golgi complex and ER reform.  10.2.5   Cytokinesis  Mitosis accomplishes not only the segregation of duplicated chromosomes into daughter nuclei (karyokinesis), but the cell itself is divided into two daughter cells by the separation of cytoplasm called cytokinesis at the end of which cell division gets completed (Figure 10.2 e). In an animal cell, this is achieved by the appearance of a furrow in the plasma membrane. The furrow gradually deepens and ultimately joins in the centre dividing the cell cytoplasm into two. Plant cells however, are enclosed by a relatively inextensible cell wall, thererfore they undergo cytokinesis by a different mechanism. In plant cells, wall formation starts in the centre of the cell and grows outward to meet the existing lateral walls. The formation of the new cell wall begins with the formation of a simple precursor, called the   cell-plate   that represents the middle lamella between the walls of two adjacent cells. At the time of cytoplasmic division, organelles like mitochondria and plastids get distributed between the two daughter cells. In some organisms karyokinesis is not followed by cytokinesis as a result of which multinucleate condition arises leading to the formation of syncytium (e.g., liquid endosperm in coconut).  Figure 10.2 c to e   : A diagrammatic view of stages in Mitosis Reprint 2025-26C ELL   C YCLE   AND   C ELL   D IVISION   125  10.3   Significance of Mitosis  Mitosis or the equational division is usually restricted to the diploid cells only. However, in some lower plants and in some social insects haploid cells also divide by mitosis. It is very essential to understand the significance of this division in the life of an organism. Are you aware of some examples where you have studied about haploid and diploid insects? Mitosis usually results in the production of diploid daughter cells with identical genetic complement. The growth of multicellular organisms is due to mitosis. Cell growth results in disturbing the ratio between the nucleus and the cytoplasm. It therefore becomes essential for the cell to divide to restore the nucleo-cytoplasmic ratio. A very significant contribution of mitosis is cell repair. The cells of the upper layer of the epidermis, cells of the lining of the gut, and blood cells are being constantly replaced. Mitotic divisions in the meristematic tissues – the apical and the lateral cambium, result in a continuous growth of plants throughout their life.  10.4   M EIOSIS  The production of offspring by sexual reproduction includes the fusion of two gametes, each with a complete haploid set of chromosomes. Gametes are formed from specialised diploid cells. This specialised kind of cell division that reduces the chromosome number by half results in the production of haploid daughter cells. This kind of division is called  meiosis.   Meiosis ensures the production of haploid phase in the life cycle of sexually reproducing organisms whereas fertilisation restores the diploid phase. We come across meiosis during gametogenesis in plants and animals. This leads to the formation of haploid gametes. The key features of meiosis are as follows:  l l l l l   Meiosis involves two sequential cycles of nuclear and cell division called  meiosis I   and   meiosis II   but only a single cycle of DNA replication.  l l l l l   Meiosis I is initiated after the parental chromosomes have replicated to produce identical sister chromatids at the S phase.  l l l l l   Meiosis   involves   pairing   of   homologous   chromosomes   and recombination between non-sister chromatids of homologous chromosomes.  l l l l l   Four haploid cells are formed at the end of meiosis II. Meiotic events can be grouped under the following phases:  Meiosis I   Meiosis II  Prophase I   Prophase II Metaphase I   Metaphase II Anaphase I   Anaphase II Telophase I   Telophase II Reprint 2025-26126 126 126 126 126   B IOLOGY  10.4.1 10.4.1 10.4.1 10.4.1 10.4.1 Meiosis I Meiosis I Meiosis I Meiosis I Meiosis I  Prophase I: Prophase I: Prophase I: Prophase I: Prophase I: Prophase of the first meiotic division is typically longer and more complex when compared to prophase of mitosis. It has been further subdivided into the following five phases based on chromosomal behaviour, i.e., Leptotene, Zygotene, Pachytene, Diplotene and Diakinesis. During leptotene leptotene leptotene leptotene leptotene stage the chromosomes become gradually visible under the light microscope. The compaction of chromosomes continues throughout leptotene. This is followed by the second stage of prophase I called zygotene zygotene zygotene zygotene zygotene. During this stage, chromosomes start pairing together and this process of association is called synapsis. Such paired chromosomes   are   called   homologous   chromosomes.   Electron micrographs of this stage, indicate that chromosome synapsis is accompanied   by   the   formation   of   complex   structure   called synaptonemal complex. synaptonemal complex. synaptonemal complex. synaptonemal complex. synaptonemal complex. The complex formed by a pair of synapsed homologous chromosomes is called a bivalent bivalent bivalent bivalent bivalent or a tetrad. However, these are more clearly visible at the next stage. The first two stages of prophase I are relatively short-lived compared to the next stage that is pachytene pachytene pachytene pachytene pachytene. During this stage, the four chromatids of each bivalent chromosomes becomes distinct and clearly appear as tetrads. This stage is characterised by the appearance of recombination nodules, the sites at which crossing over occurs between non-sister chromatids of the homologous chromosomes. Crossing over is the exchange of genetic material between two homologous chromosomes. Crossing over is also an enzyme-mediated process and the enzyme involved is called recombinase. Crossing over leads to recombination of genetic material on the two chromosomes. Recombination between homologous chromosomes is completed by the end of pachytene, leaving the chromosomes linked at the sites of crossing over. The beginning of diplotene diplotene diplotene diplotene diplotene is recognised by the dissolution of the synaptonemal   complex   and   the   tendency   of   the   recombined homologous chromosomes of the bivalents to separate from each other except at the sites of crossovers. These X-shaped structures, are called chiasmata. chiasmata. chiasmata. chiasmata. chiasmata. In oocytes of some vertebrates, diplotene can last for months or years. The final stage of meiotic prophase I is diakinesis. diakinesis. diakinesis. diakinesis. diakinesis. This is marked by terminalisation of chiasmata. During this phase the chromosomes are fully condensed and the meiotic spindle is assembled to prepare the homologous chromosomes for separation. By the end of diakinesis, the nucleolus disappears and the nuclear envelope also breaks down. Diakinesis represents transition to metaphase. Metaphase I: Metaphase I: Metaphase I: Metaphase I: Metaphase I: The bivalent chromosomes align on the equatorial plate (Figure 10.3). The microtubules from the opposite poles of the spindle attach to the kinetochore of homologous chromosomes. Reprint 2025-26C ELL   C YCLE   AND   C ELL   D IVISION   127  Anaphase I:   The homologous chromosomes separate, while sister chromatids remain associated at their centromeres (Figure 10.3).  Telophase   I:   The nuclear membrane and nucleolus reappear, cytokinesis follows and this is called as dyad of cells (Figure 10.3). Although in many cases the chromosomes do undergo some dispersion, they do not reach the extremely extended state of the interphase nucleus. The stage between the two meiotic divisions is called interkinesis and is generally short lived. There is no replication of DNA during interkinesis. Interkinesis is followed by prophase II, a much simpler prophase than prophase I.  10.4.2   Meiosis II  Prophase II:   Meiosis II is initiated immediately after cytokinesis, usually before the chromosomes have fully elongated. In contrast to meiosis I, meiosis II resembles a normal mitosis. The nuclear membrane disappears by the end of prophase II (Figure 10.4). The chromosomes again become compact.  Metaphase II : At this stage, the chromosomes align at the equator and the microtubules from opposite poles of the spindle get attached to the kinetochores (Figure 10.4) of sister chromatids.  Anaphase II:   It begins with the simultaneous splitting of the centromere of each chromosome (which was holding the sister chromatids together), allowing them to move toward opposite poles of the cell (Figure 10.4) by shortening of microtubules attached to kinetochores.  Figure 10.3   Stages of Meiosis I Reprint 2025-26128   B IOLOGY  Telophase II:   Meiosis ends with telophase II, in which the two groups of chromosomes once again get enclosed by a nuclear envelope; cytokinesis follows resulting in the formation of tetrad of cells i.e., four haploid daughter cells (Figure 10.4).  10.5   S IGNIFICANCE   OF   M EIOSIS  Meiosis is the mechanism by which conservation of specific chromosome number of each species is achieved across generations in sexually reproducing organisms, even though the process, per se, paradoxically, results in reduction of chromosome number by half. It also increases the genetic variability in the population of organisms from one generation to the next. Variations are very important for the process of evolution.  Figure 10.4   Stages of Meiosis II Reprint 2025-26",
            "col4": "DATA"
        }
    ],
    "projectDeleted": false,
    "codeFrameColumn": [
        {
            "col1": "col1",
            "col2": "col3",
            "researchObjective": "",
            "filterColumn": [],
            "title": "Biology-K11B10 Cell Cycle and Cell Division",
            "netSetting": true,
            "subNetSetting": true,
            "categorySetting": true,
            "sentimentSetting": true,
            "thinkingBudgetSetting": 2000,
            "questionType": "OPENEND",
            "otherBrand": false,
            "existingCodeFrame": [],
            "isList": false,
            "codeFrameWithDataList": [],
            "codeFrameList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 5394,
                        "CandidatesTokenCount": 2394,
                        "TotalTokenCount": 9333,
                        "thoughtsTokenCount": 1545,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Net": "Cell Cycle Fundamentals",
                                "Subnets": [
                                    {
                                        "Subnet": "Definition & Overview",
                                        "Categories": [
                                            {
                                                "Category": "Definition of Cell Cycle",
                                                "Summary": "The sequence of events where a cell duplicates its genome, synthesizes cell constituents, and divides into two daughter cells."
                                            },
                                            {
                                                "Category": "Importance of Cell Division",
                                                "Summary": "Cell division is crucial for growth, reproduction, and the coordinated formation of progeny cells with intact genomes."
                                            },
                                            {
                                                "Category": "Organismal Growth",
                                                "Summary": "How a single cell leads to the formation of structures consisting of millions of cells through cycles of growth and division."
                                            }
                                        ],
                                        "Summary": "Explains what the cell cycle is and its general importance in living organisms."
                                    },
                                    {
                                        "Subnet": "Main Phases of Cell Cycle",
                                        "Categories": [
                                            {
                                                "Category": "Interphase",
                                                "Summary": "The preparatory phase between two successive M phases, where the cell grows and replicates DNA."
                                            },
                                            {
                                                "Category": "M Phase",
                                                "Summary": "The phase where actual cell division or mitosis occurs, including nuclear and cytoplasmic division."
                                            },
                                            {
                                                "Category": "Duration of Cell Cycle",
                                                "Summary": "The variable length of the cell cycle, e.g., human cells (24 hours), yeast (90 minutes)."
                                            }
                                        ],
                                        "Summary": "Outlines the two primary stages of the cell cycle: interphase and the division phase."
                                    },
                                    {
                                        "Subnet": "Interphase Sub-stages",
                                        "Categories": [
                                            {
                                                "Category": "G1 Phase (Gap 1)",
                                                "Summary": "Interval between mitosis and DNA replication, characterized by metabolic activity and continuous cell growth without DNA replication."
                                            },
                                            {
                                                "Category": "S Phase (Synthesis)",
                                                "Summary": "Period of DNA synthesis and replication, where DNA content doubles (2C to 4C) but chromosome number remains 2n; centrioles duplicate in animal cells."
                                            },
                                            {
                                                "Category": "G2 Phase (Gap 2)",
                                                "Summary": "Proteins are synthesized in preparation for mitosis while cell growth continues."
                                            },
                                            {
                                                "Category": "G0 Phase (Quiescent Stage)",
                                                "Summary": "An inactive stage where cells exit G1, remain metabolically active but do not proliferate unless signaled."
                                            }
                                        ],
                                        "Summary": "Details the specific events and characteristics of each phase within the interphase."
                                    }
                                ],
                                "Summary": "This Net covers the basic definition, purpose, main phases, and sub-stages of the cell cycle, providing a foundational understanding of cell proliferation."
                            },
                            {
                                "Net": "Mitotic Cell Division",
                                "Subnets": [
                                    {
                                        "Subnet": "Mitosis Overview & Karyokinesis",
                                        "Categories": [
                                            {
                                                "Category": "Mitosis Definition",
                                                "Summary": "Also known as equational division, where parent and progeny cells have the same number of chromosomes."
                                            },
                                            {
                                                "Category": "Karyokinesis Stages",
                                                "Summary": "The four stages of nuclear division: Prophase, Metaphase, Anaphase, and Telophase."
                                            },
                                            {
                                                "Category": "Progressive Nature",
                                                "Summary": "Cell division is a continuous process without clear-cut lines between stages."
                                            }
                                        ],
                                        "Summary": "Introduces mitosis as equational division and outlines its nuclear division stages."
                                    },
                                    {
                                        "Subnet": "Early Mitotic Stages",
                                        "Categories": [
                                            {
                                                "Category": "Prophase",
                                                "Summary": "First stage of karyokinesis, marked by chromosomal material condensation, centrosome movement to poles, and formation of mitotic apparatus (asters and spindle fibers)."
                                            },
                                            {
                                                "Category": "Metaphase",
                                                "Summary": "Complete disintegration of the nuclear envelope, chromosomes condense fully, align at the metaphase plate, and spindle fibers attach to kinetochores."
                                            }
                                        ],
                                        "Summary": "Focuses on the initial phases of mitosis, detailing chromosome and cellular structural changes."
                                    },
                                    {
                                        "Subnet": "Late Mitotic Stages",
                                        "Categories": [
                                            {
                                                "Category": "Anaphase",
                                                "Summary": "Centromeres split, sister chromatids separate (now called daughter chromosomes), and migrate to opposite poles."
                                            },
                                            {
                                                "Category": "Telophase",
                                                "Summary": "Chromosomes decondense at poles, nuclear envelope reforms around chromosome clusters, and nucleolus, Golgi complex, and ER reappear."
                                            }
                                        ],
                                        "Summary": "Describes the final stages of nuclear division, including chromosome separation and nuclear reformation."
                                    },
                                    {
                                        "Subnet": "Cytoplasmic Division (Cytokinesis)",
                                        "Categories": [
                                            {
                                                "Category": "Cytoplasmic Division",
                                                "Summary": "The process where the cell's cytoplasm separates, completing cell division after karyokinesis."
                                            },
                                            {
                                                "Category": "Animal Cell Cytokinesis",
                                                "Summary": "Achieved by the formation and deepening of a furrow in the plasma membrane."
                                            },
                                            {
                                                "Category": "Plant Cell Cytokinesis",
                                                "Summary": "Involves the formation of a cell-plate in the center that grows outward to form a new cell wall."
                                            },
                                            {
                                                "Category": "Syncytium Formation",
                                                "Summary": "A condition where karyokinesis occurs without cytokinesis, leading to multinucleate cells (e.g., coconut endosperm)."
                                            }
                                        ],
                                        "Summary": "Explains the process of cytoplasmic division in both animal and plant cells, and exceptions."
                                    }
                                ],
                                "Summary": "This Net provides a comprehensive breakdown of mitosis, detailing each stage of nuclear division (karyokinesis) and subsequent cytoplasmic division (cytokinesis) and their unique characteristics."
                            },
                            {
                                "Net": "Meiotic Cell Division",
                                "Subnets": [
                                    {
                                        "Subnet": "Meiosis Overview & Features",
                                        "Categories": [
                                            {
                                                "Category": "Meiosis Definition",
                                                "Summary": "Specialized cell division that reduces chromosome number by half, producing haploid daughter cells for sexual reproduction."
                                            },
                                            {
                                                "Category": "Gamete Formation",
                                                "Summary": "Process of forming haploid gametes during gametogenesis in plants and animals."
                                            },
                                            {
                                                "Category": "Key Features of Meiosis",
                                                "Summary": "Involves two sequential cycles (Meiosis I & II) with a single DNA replication, homologous chromosome pairing, recombination, and formation of four haploid cells."
                                            }
                                        ],
                                        "Summary": "Introduces meiosis as reductional division and highlights its essential characteristics."
                                    },
                                    {
                                        "Subnet": "Meiosis I Stages",
                                        "Categories": [
                                            {
                                                "Category": "Prophase I",
                                                "Summary": "Longer and more complex, subdivided into Leptotene, Zygotene (synapsis, synaptonemal complex, bivalents), Pachytene (tetrads, recombination nodules, crossing over), Diplotene (chiasmata, dissolution of synaptonemal complex), and Diakinesis (terminalization of chiasmata, spindle assembly)."
                                            },
                                            {
                                                "Category": "Metaphase I",
                                                "Summary": "Bivalent chromosomes align on the equatorial plate, with microtubules from opposite poles attaching to kinetochores of homologous chromosomes."
                                            },
                                            {
                                                "Category": "Anaphase I",
                                                "Summary": "Homologous chromosomes separate and move to opposite poles, while sister chromatids remain attached at their centromeres."
                                            },
                                            {
                                                "Category": "Telophase I",
                                                "Summary": "Nuclear membrane and nucleolus reappear, followed by cytokinesis, forming a dyad of cells; followed by a short interkinesis without DNA replication."
                                            }
                                        ],
                                        "Summary": "Details the specific events occurring during the first meiotic division, including homologous chromosome pairing and separation."
                                    },
                                    {
                                        "Subnet": "Meiosis II Stages",
                                        "Categories": [
                                            {
                                                "Category": "Prophase II",
                                                "Summary": "Initiated after cytokinesis (or interkinesis), resembling a normal mitosis; nuclear membrane disappears and chromosomes become compact."
                                            },
                                            {
                                                "Category": "Metaphase II",
                                                "Summary": "Chromosomes align at the equator, and microtubules from opposite poles attach to kinetochores of sister chromatids."
                                            },
                                            {
                                                "Category": "Anaphase II",
                                                "Summary": "Centromeres split, and sister chromatids separate and move to opposite poles."
                                            },
                                            {
                                                "Category": "Telophase II",
                                                "Summary": "Nuclear envelopes reform around chromosome groups, cytokinesis follows, resulting in a tetrad of four haploid daughter cells."
                                            }
                                        ],
                                        "Summary": "Describes the events of the second meiotic division, which is similar to mitosis but involves haploid cells."
                                    }
                                ],
                                "Summary": "This Net systematically explains meiosis, covering its two main divisions, the distinct stages within each, and the critical genetic processes involved like crossing over."
                            },
                            {
                                "Net": "Significance of Cell Division",
                                "Subnets": [
                                    {
                                        "Subnet": "Significance of Mitosis",
                                        "Categories": [
                                            {
                                                "Category": "Growth of Multicellular Organisms",
                                                "Summary": "Mitosis drives the growth of multicellular organisms by increasing cell numbers."
                                            },
                                            {
                                                "Category": "Cell Repair and Replacement",
                                                "Summary": "Essential for replacing lost cells due to injury, cell death, or regular turnover (e.g., epidermis, gut lining, blood cells)."
                                            },
                                            {
                                                "Category": "Maintenance of Nucleo-Cytoplasmic Ratio",
                                                "Summary": "Cell division restores the optimal ratio between the nucleus and cytoplasm, disturbed by cell growth."
                                            }
                                        ],
                                        "Summary": "Highlights the crucial roles of mitosis in growth, repair, and maintaining cellular balance."
                                    },
                                    {
                                        "Subnet": "Significance of Meiosis",
                                        "Categories": [
                                            {
                                                "Category": "Conservation of Chromosome Number",
                                                "Summary": "Ensures the specific chromosome number of a species is maintained across generations by reducing it by half in gametes, which is restored upon fertilization."
                                            },
                                            {
                                                "Category": "Genetic Variability",
                                                "Summary": "Increases genetic diversity in populations from one generation to the next, which is vital for evolution."
                                            }
                                        ],
                                        "Summary": "Explains the importance of meiosis in genetic continuity and promoting diversity."
                                    }
                                ],
                                "Summary": "This Net discusses the biological importance and contributions of both mitosis and meiosis to organismal life, including growth, repair, reproduction, and evolution."
                            },
                            {
                                "Net": "996 Non specific NET",
                                "Subnets": [
                                    {
                                        "Subnet": "Irrelevant Responses",
                                        "Categories": [
                                            {
                                                "Category": "Uncodable Content",
                                                "Summary": "Responses that are irrelevant to the survey question and cannot be categorized into existing themes, requiring further review or upcoding."
                                            }
                                        ],
                                        "Summary": "Captures responses that are irrelevant or cannot be easily coded into the defined thematic structure."
                                    }
                                ],
                                "Summary": "This Net is for responses that are irrelevant, off-topic, or difficult to categorize within the established codeframe and may require further review."
                            },
                            {
                                "Net": "997 Don’t Know Can’t Say",
                                "Subnets": [
                                    {
                                        "Subnet": "Non-committal/Vague",
                                        "Categories": [
                                            {
                                                "Category": "Ambiguous/Uncertain",
                                                "Summary": "Responses that are non-committal, vague, or express uncertainty, such as 'Don't Know' or 'Can't Say'."
                                            }
                                        ],
                                        "Summary": "Captures responses indicating uncertainty or lack of a clear opinion."
                                    }
                                ],
                                "Summary": "This Net is for responses that are non-committal, ambiguous, or express a lack of knowledge or inability to answer."
                            },
                            {
                                "Net": "998 Not Applicable",
                                "Subnets": [
                                    {
                                        "Subnet": "Not Relevant to Question",
                                        "Categories": [
                                            {
                                                "Category": "Out of Scope",
                                                "Summary": "Responses that are not applicable to the survey question or the overall project context, despite being coherent."
                                            }
                                        ],
                                        "Summary": "Captures responses that are not relevant to the specific question asked."
                                    }
                                ],
                                "Summary": "This Net is for responses that are coherent but are not applicable or relevant to the specific question at hand."
                            },
                            {
                                "Net": "999 None-Nothing",
                                "Subnets": [
                                    {
                                        "Subnet": "Gibberish/Empty",
                                        "Categories": [
                                            {
                                                "Category": "Meaningless Content",
                                                "Summary": "Responses consisting of gibberish, non-language characters, or expressions that convey no meaningful information, including empty responses."
                                            }
                                        ],
                                        "Summary": "Captures responses that are meaningless, indecipherable, or empty."
                                    }
                                ],
                                "Summary": "This Net is for responses that are incomprehensible, nonsensical, contain gibberish, or are entirely empty."
                            }
                        ],
                        "Summary": "The provided biological text comprehensively details the 'Cell Cycle and Cell Division'. It explains the fundamental processes by which cells grow and reproduce, starting from a single cell forming large organisms. The text defines the cell cycle as a coordinated sequence of events including DNA replication, cell growth, and division, ensuring intact genomes in progeny cells. It elaborates on the two main phases: Interphase (G1, S, G2, and G0 for preparation and growth) and M Phase (actual cell division). The document further distinguishes between Mitosis (equational division, essential for growth and repair in somatic cells) and Meiosis (reductional division, vital for sexual reproduction, gamete formation, and increasing genetic variability). Each division type is meticulously broken down into its respective stages (Prophase, Metaphase, Anaphase, Telophase) and associated events, including karyokinesis and cytokinesis, along with their biological significance."
                    }
                }
            ],
            "onlyCodeFrame": true,
            "errorList": [],
            "codeFrameGenerateType": "CODE_FRAME_ONLY",
            "requestType": "CODEFRAME",
            "requestMode": "DEVELOPMENT",
            "sourceLanguage": "English",
            "targetLanguage": "English",
            "response": {
                "status": "OK",
                "data": {
                    "status": "OK",
                    "PromptTokenCount": 5394,
                    "CandidatesTokenCount": 2394,
                    "TotalTokenCount": 9333,
                    "thoughtsTokenCount": 1545,
                    "finishReason": "STOP",
                    "data": [
                        {
                            "Net": "Cell Cycle Fundamentals",
                            "Subnets": [
                                {
                                    "Subnet": "Definition & Overview",
                                    "Categories": [
                                        {
                                            "Category": "Definition of Cell Cycle",
                                            "Summary": "The sequence of events where a cell duplicates its genome, synthesizes cell constituents, and divides into two daughter cells.",
                                            "code": 3
                                        },
                                        {
                                            "Category": "Importance of Cell Division",
                                            "Summary": "Cell division is crucial for growth, reproduction, and the coordinated formation of progeny cells with intact genomes.",
                                            "code": 4
                                        },
                                        {
                                            "Category": "Organismal Growth",
                                            "Summary": "How a single cell leads to the formation of structures consisting of millions of cells through cycles of growth and division.",
                                            "code": 5
                                        }
                                    ],
                                    "Summary": "Explains what the cell cycle is and its general importance in living organisms.",
                                    "code": 2
                                },
                                {
                                    "Subnet": "Main Phases of Cell Cycle",
                                    "Categories": [
                                        {
                                            "Category": "Interphase",
                                            "Summary": "The preparatory phase between two successive M phases, where the cell grows and replicates DNA.",
                                            "code": 7
                                        },
                                        {
                                            "Category": "M Phase",
                                            "Summary": "The phase where actual cell division or mitosis occurs, including nuclear and cytoplasmic division.",
                                            "code": 8
                                        },
                                        {
                                            "Category": "Duration of Cell Cycle",
                                            "Summary": "The variable length of the cell cycle, e.g., human cells (24 hours), yeast (90 minutes).",
                                            "code": 9
                                        }
                                    ],
                                    "Summary": "Outlines the two primary stages of the cell cycle: interphase and the division phase.",
                                    "code": 6
                                },
                                {
                                    "Subnet": "Interphase Sub-stages",
                                    "Categories": [
                                        {
                                            "Category": "G1 Phase (Gap 1)",
                                            "Summary": "Interval between mitosis and DNA replication, characterized by metabolic activity and continuous cell growth without DNA replication.",
                                            "code": 11
                                        },
                                        {
                                            "Category": "S Phase (Synthesis)",
                                            "Summary": "Period of DNA synthesis and replication, where DNA content doubles (2C to 4C) but chromosome number remains 2n; centrioles duplicate in animal cells.",
                                            "code": 12
                                        },
                                        {
                                            "Category": "G2 Phase (Gap 2)",
                                            "Summary": "Proteins are synthesized in preparation for mitosis while cell growth continues.",
                                            "code": 13
                                        },
                                        {
                                            "Category": "G0 Phase (Quiescent Stage)",
                                            "Summary": "An inactive stage where cells exit G1, remain metabolically active but do not proliferate unless signaled.",
                                            "code": 14
                                        }
                                    ],
                                    "Summary": "Details the specific events and characteristics of each phase within the interphase.",
                                    "code": 10
                                }
                            ],
                            "Summary": "This Net covers the basic definition, purpose, main phases, and sub-stages of the cell cycle, providing a foundational understanding of cell proliferation.",
                            "code": 1
                        },
                        {
                            "Net": "Mitotic Cell Division",
                            "Subnets": [
                                {
                                    "Subnet": "Mitosis Overview & Karyokinesis",
                                    "Categories": [
                                        {
                                            "Category": "Mitosis Definition",
                                            "Summary": "Also known as equational division, where parent and progeny cells have the same number of chromosomes.",
                                            "code": 17
                                        },
                                        {
                                            "Category": "Karyokinesis Stages",
                                            "Summary": "The four stages of nuclear division: Prophase, Metaphase, Anaphase, and Telophase.",
                                            "code": 18
                                        },
                                        {
                                            "Category": "Progressive Nature",
                                            "Summary": "Cell division is a continuous process without clear-cut lines between stages.",
                                            "code": 19
                                        }
                                    ],
                                    "Summary": "Introduces mitosis as equational division and outlines its nuclear division stages.",
                                    "code": 16
                                },
                                {
                                    "Subnet": "Early Mitotic Stages",
                                    "Categories": [
                                        {
                                            "Category": "Prophase",
                                            "Summary": "First stage of karyokinesis, marked by chromosomal material condensation, centrosome movement to poles, and formation of mitotic apparatus (asters and spindle fibers).",
                                            "code": 21
                                        },
                                        {
                                            "Category": "Metaphase",
                                            "Summary": "Complete disintegration of the nuclear envelope, chromosomes condense fully, align at the metaphase plate, and spindle fibers attach to kinetochores.",
                                            "code": 22
                                        }
                                    ],
                                    "Summary": "Focuses on the initial phases of mitosis, detailing chromosome and cellular structural changes.",
                                    "code": 20
                                },
                                {
                                    "Subnet": "Late Mitotic Stages",
                                    "Categories": [
                                        {
                                            "Category": "Anaphase",
                                            "Summary": "Centromeres split, sister chromatids separate (now called daughter chromosomes), and migrate to opposite poles.",
                                            "code": 24
                                        },
                                        {
                                            "Category": "Telophase",
                                            "Summary": "Chromosomes decondense at poles, nuclear envelope reforms around chromosome clusters, and nucleolus, Golgi complex, and ER reappear.",
                                            "code": 25
                                        }
                                    ],
                                    "Summary": "Describes the final stages of nuclear division, including chromosome separation and nuclear reformation.",
                                    "code": 23
                                },
                                {
                                    "Subnet": "Cytoplasmic Division (Cytokinesis)",
                                    "Categories": [
                                        {
                                            "Category": "Cytoplasmic Division",
                                            "Summary": "The process where the cell's cytoplasm separates, completing cell division after karyokinesis.",
                                            "code": 27
                                        },
                                        {
                                            "Category": "Animal Cell Cytokinesis",
                                            "Summary": "Achieved by the formation and deepening of a furrow in the plasma membrane.",
                                            "code": 28
                                        },
                                        {
                                            "Category": "Plant Cell Cytokinesis",
                                            "Summary": "Involves the formation of a cell-plate in the center that grows outward to form a new cell wall.",
                                            "code": 29
                                        },
                                        {
                                            "Category": "Syncytium Formation",
                                            "Summary": "A condition where karyokinesis occurs without cytokinesis, leading to multinucleate cells (e.g., coconut endosperm).",
                                            "code": 30
                                        }
                                    ],
                                    "Summary": "Explains the process of cytoplasmic division in both animal and plant cells, and exceptions.",
                                    "code": 26
                                }
                            ],
                            "Summary": "This Net provides a comprehensive breakdown of mitosis, detailing each stage of nuclear division (karyokinesis) and subsequent cytoplasmic division (cytokinesis) and their unique characteristics.",
                            "code": 15
                        },
                        {
                            "Net": "Meiotic Cell Division",
                            "Subnets": [
                                {
                                    "Subnet": "Meiosis Overview & Features",
                                    "Categories": [
                                        {
                                            "Category": "Meiosis Definition",
                                            "Summary": "Specialized cell division that reduces chromosome number by half, producing haploid daughter cells for sexual reproduction.",
                                            "code": 33
                                        },
                                        {
                                            "Category": "Gamete Formation",
                                            "Summary": "Process of forming haploid gametes during gametogenesis in plants and animals.",
                                            "code": 34
                                        },
                                        {
                                            "Category": "Key Features of Meiosis",
                                            "Summary": "Involves two sequential cycles (Meiosis I & II) with a single DNA replication, homologous chromosome pairing, recombination, and formation of four haploid cells.",
                                            "code": 35
                                        }
                                    ],
                                    "Summary": "Introduces meiosis as reductional division and highlights its essential characteristics.",
                                    "code": 32
                                },
                                {
                                    "Subnet": "Meiosis I Stages",
                                    "Categories": [
                                        {
                                            "Category": "Prophase I",
                                            "Summary": "Longer and more complex, subdivided into Leptotene, Zygotene (synapsis, synaptonemal complex, bivalents), Pachytene (tetrads, recombination nodules, crossing over), Diplotene (chiasmata, dissolution of synaptonemal complex), and Diakinesis (terminalization of chiasmata, spindle assembly).",
                                            "code": 37
                                        },
                                        {
                                            "Category": "Metaphase I",
                                            "Summary": "Bivalent chromosomes align on the equatorial plate, with microtubules from opposite poles attaching to kinetochores of homologous chromosomes.",
                                            "code": 38
                                        },
                                        {
                                            "Category": "Anaphase I",
                                            "Summary": "Homologous chromosomes separate and move to opposite poles, while sister chromatids remain attached at their centromeres.",
                                            "code": 39
                                        },
                                        {
                                            "Category": "Telophase I",
                                            "Summary": "Nuclear membrane and nucleolus reappear, followed by cytokinesis, forming a dyad of cells; followed by a short interkinesis without DNA replication.",
                                            "code": 40
                                        }
                                    ],
                                    "Summary": "Details the specific events occurring during the first meiotic division, including homologous chromosome pairing and separation.",
                                    "code": 36
                                },
                                {
                                    "Subnet": "Meiosis II Stages",
                                    "Categories": [
                                        {
                                            "Category": "Prophase II",
                                            "Summary": "Initiated after cytokinesis (or interkinesis), resembling a normal mitosis; nuclear membrane disappears and chromosomes become compact.",
                                            "code": 42
                                        },
                                        {
                                            "Category": "Metaphase II",
                                            "Summary": "Chromosomes align at the equator, and microtubules from opposite poles attach to kinetochores of sister chromatids.",
                                            "code": 43
                                        },
                                        {
                                            "Category": "Anaphase II",
                                            "Summary": "Centromeres split, and sister chromatids separate and move to opposite poles.",
                                            "code": 44
                                        },
                                        {
                                            "Category": "Telophase II",
                                            "Summary": "Nuclear envelopes reform around chromosome groups, cytokinesis follows, resulting in a tetrad of four haploid daughter cells.",
                                            "code": 45
                                        }
                                    ],
                                    "Summary": "Describes the events of the second meiotic division, which is similar to mitosis but involves haploid cells.",
                                    "code": 41
                                }
                            ],
                            "Summary": "This Net systematically explains meiosis, covering its two main divisions, the distinct stages within each, and the critical genetic processes involved like crossing over.",
                            "code": 31
                        },
                        {
                            "Net": "Significance of Cell Division",
                            "Subnets": [
                                {
                                    "Subnet": "Significance of Mitosis",
                                    "Categories": [
                                        {
                                            "Category": "Growth of Multicellular Organisms",
                                            "Summary": "Mitosis drives the growth of multicellular organisms by increasing cell numbers.",
                                            "code": 48
                                        },
                                        {
                                            "Category": "Cell Repair and Replacement",
                                            "Summary": "Essential for replacing lost cells due to injury, cell death, or regular turnover (e.g., epidermis, gut lining, blood cells).",
                                            "code": 49
                                        },
                                        {
                                            "Category": "Maintenance of Nucleo-Cytoplasmic Ratio",
                                            "Summary": "Cell division restores the optimal ratio between the nucleus and cytoplasm, disturbed by cell growth.",
                                            "code": 50
                                        }
                                    ],
                                    "Summary": "Highlights the crucial roles of mitosis in growth, repair, and maintaining cellular balance.",
                                    "code": 47
                                },
                                {
                                    "Subnet": "Significance of Meiosis",
                                    "Categories": [
                                        {
                                            "Category": "Conservation of Chromosome Number",
                                            "Summary": "Ensures the specific chromosome number of a species is maintained across generations by reducing it by half in gametes, which is restored upon fertilization.",
                                            "code": 52
                                        },
                                        {
                                            "Category": "Genetic Variability",
                                            "Summary": "Increases genetic diversity in populations from one generation to the next, which is vital for evolution.",
                                            "code": 53
                                        }
                                    ],
                                    "Summary": "Explains the importance of meiosis in genetic continuity and promoting diversity.",
                                    "code": 51
                                }
                            ],
                            "Summary": "This Net discusses the biological importance and contributions of both mitosis and meiosis to organismal life, including growth, repair, reproduction, and evolution.",
                            "code": 46
                        }
                    ],
                    "Summary": "The provided biological text comprehensively details the 'Cell Cycle and Cell Division'. It explains the fundamental processes by which cells grow and reproduce, starting from a single cell forming large organisms. The text defines the cell cycle as a coordinated sequence of events including DNA replication, cell growth, and division, ensuring intact genomes in progeny cells. It elaborates on the two main phases: Interphase (G1, S, G2, and G0 for preparation and growth) and M Phase (actual cell division). The document further distinguishes between Mitosis (equational division, essential for growth and repair in somatic cells) and Meiosis (reductional division, vital for sexual reproduction, gamete formation, and increasing genetic variability). Each division type is meticulously broken down into its respective stages (Prophase, Metaphase, Anaphase, Telophase) and associated events, including karyokinesis and cytokinesis, along with their biological significance."
                }
            },
            "summaryList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 5030,
                        "CandidatesTokenCount": 2800,
                        "TotalTokenCount": 9257,
                        "thoughtsTokenCount": 1427,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Summary": "<h2>Comprehensive Synthesis Report: Cell Cycle and Cell Division</h2><h3>Introduction to Cell Life and Division</h3><p>All organisms, from the smallest to the largest, originate from a single cell. This foundational cell undergoes cycles of growth and reproduction, enabling the formation of complex structures composed of millions of cells. Cell division is a fundamental process in all living organisms, intricately linked with DNA replication and cell growth. These processes must be highly coordinated to ensure accurate division and the formation of progeny cells with intact genomes. [Source Data]</p><h3>The Cell Cycle: An Overview</h3><p>The **cell cycle** is defined as the sequence of events by which a cell duplicates its genome, synthesizes its other cellular constituents, and ultimately divides into two daughter cells. While cytoplasmic growth is continuous, DNA synthesis occurs only during a specific stage. The replicated chromosomes are then precisely distributed to daughter nuclei through a complex, genetically controlled series of events. [Source Data]</p><h4>Phases of the Cell Cycle</h4><p>A typical eukaryotic cell cycle, as observed in human cells, lasts approximately 24 hours, though this duration can vary significantly across organisms and cell types (e.g., yeast completes its cycle in about 90 minutes). The cell cycle is broadly divided into two main phases:</p><ul><li>**Interphase**: The preparatory phase between two successive M phases, lasting over 95% of the cell cycle duration. During this time, the cell prepares for division through growth and DNA replication in an orderly manner.</li><li>**M Phase (Mitosis phase)**: The actual phase of cell division or mitosis, lasting about an hour in human cells. It begins with nuclear division (karyokinesis) and usually concludes with cytoplasmic division (cytokinesis). [Source Data]</li></ul><p>Interphase itself is further subdivided into three stages:</p><h5>G<sub>1</sub> Phase (Gap 1)</h5><p>This phase is the interval between mitosis and the initiation of DNA replication. During G<sub>1</sub>, the cell remains metabolically active and continuously grows in size but does not replicate its DNA. [Source Data]</p><h5>S Phase (Synthesis)</h5><p>The S phase is characterized by DNA synthesis or replication. During this period, the amount of DNA per cell doubles (e.g., from 2C to 4C). Importantly, the number of chromosomes remains constant (e.g., 2n at G<sub>1</sub> remains 2n after S phase). In animal cells, DNA replication occurs in the nucleus, and the centriole duplicates in the cytoplasm. [Source Data]</p><h5>G<sub>2</sub> Phase (Gap 2)</h5><p>During the G<sub>2</sub> phase, proteins essential for mitosis are synthesized, and cell growth continues in preparation for the upcoming M phase. [Source Data]</p><h4>Quiescent Stage (G<sub>0</sub>)</h4><p>Some adult animal cells, such as heart cells, do not divide. Other cells divide only occasionally, for instance, to replace lost cells due to injury or natural cell death. These cells exit the G<sub>1</sub> phase and enter an inactive **quiescent stage (G<sub>0</sub>)**. Cells in G<sub>0</sub> remain metabolically active but do not proliferate unless specifically signaled based on organismal requirements. Mitotic division in animals is generally observed in diploid somatic cells, with exceptions like male honey bees where haploid cells divide by mitosis. Plants, however, can exhibit mitotic divisions in both haploid and diploid cells. [Source Data]</p><h3>M Phase (Mitosis): Equational Division</h3><p>The M phase is a highly dynamic period involving significant cellular reorganization. It is referred to as **equational division** because the number of chromosomes in the parent cell is conserved in the progeny cells. Mitosis is conceptually divided into four stages of nuclear division (karyokinesis), which represent a continuous, progressive process without distinct demarcations. [Source Data]</p><h4>Karyokinesis: Nuclear Division</h4><p>Karyokinesis encompasses the following four sequential stages:</p><h5>Prophase</h5><p>Prophase follows the S and G<sub>2</sub> phases. It is marked by the initiation of chromatin condensation, causing the intertwined DNA molecules to become compact mitotic chromosomes. Each chromosome at this stage is composed of two sister chromatids joined at the centromere. Concurrently, the centrosome, duplicated during interphase, begins migrating towards opposite poles of the cell, radiating microtubules known as asters. These asters, along with spindle fibers, form the mitotic apparatus. By the end of prophase, organelles like Golgi complexes, endoplasmic reticulum, nucleolus, and the nuclear envelope disappear. [Source Data]</p><h5>Metaphase</h5><p>The complete disintegration of the nuclear envelope signals the start of metaphase, allowing chromosomes to spread throughout the cytoplasm. Chromosomal condensation is complete, making their morphology most discernible. Each metaphase chromosome consists of two sister chromatids held together by a centromere. Small, disc-shaped structures called **kinetochores** on the centromere serve as attachment sites for spindle fibers. Chromosomes align at the cell's center, forming the **metaphase plate**, with kinetochores of sister chromatids attaching to spindle fibers from opposite poles. [Source Data]</p><h5>Anaphase</h5><p>Anaphase commences with the simultaneous splitting of the centromeres, separating the two sister chromatids. These now-independent structures, referred to as daughter chromosomes, migrate towards opposite poles. As they move, the centromere leads the way, with the chromosome arms trailing behind. [Source Data]</p><h5>Telophase</h5><p>Telophase is the final stage of karyokinesis. The chromosomes, having reached their respective poles, decondense and lose their distinct individuality, forming chromatin material clusters at each pole. Key events include the reformation of the nuclear envelope around these chromosome clusters, resulting in two daughter nuclei, and the reappearance of the nucleolus, Golgi complex, and ER. [Source Data]</p><h4>Cytokinesis: Cytoplasmic Division</h4><p>Following karyokinesis, **cytokinesis** divides the cytoplasm, completing cell division. In animal cells, a furrow appears in the plasma membrane, gradually deepening until it divides the cell into two. Plant cells, encased by a rigid cell wall, form a **cell-plate** in the center, which grows outwards to become the new cell wall (middle lamella). During cytoplasmic division, organelles like mitochondria and plastids are distributed between the daughter cells. In some organisms, karyokinesis may not be followed by cytokinesis, leading to multinucleate conditions (syncytium), as seen in the liquid endosperm of coconut. [Source Data]</p><h4>Significance of Mitosis</h4><p>Mitosis primarily occurs in diploid cells, though some lower plants and social insects exhibit it in haploid cells. Its critical roles include:</p><ul><li>**Growth**: The growth of multicellular organisms is fundamentally driven by mitosis.</li><li>**Maintenance of Nucleo-cytoplasmic Ratio**: Cell growth can disrupt the optimal ratio between the nucleus and cytoplasm; division restores this balance.</li><li>**Cell Repair**: Mitosis is crucial for replacing cells lost due to injury or normal wear and tear, such as cells of the epidermis, gut lining, and blood cells.</li><li>**Continuous Plant Growth**: In plants, mitotic divisions in meristematic tissues (apical and lateral cambium) enable continuous growth throughout their lifespan. [Source Data]</li></ul><h3>Meiosis: Reductional Division</h3><p>Sexual reproduction involves the fusion of two haploid gametes, each contributing a complete set of chromosomes. These gametes are derived from specialized diploid cells through a unique cell division called **meiosis**. Meiosis reduces the chromosome number by half, producing haploid daughter cells. This reduction is essential for maintaining the species-specific chromosome number across generations in sexually reproducing organisms, as fertilization then restores the diploid phase. Meiosis occurs during gametogenesis in both plants and animals. [Source Data]</p><h4>Key Features of Meiosis</h4><ul><li>Meiosis involves two sequential cycles of nuclear and cell division, **meiosis I** and **meiosis II**, but only a single round of DNA replication.</li><li>Meiosis I is initiated after parental chromosomes have replicated during the S phase, producing identical sister chromatids.</li><li>It includes the pairing of homologous chromosomes and **crossing over** (recombination) between non-sister chromatids, leading to genetic variability.</li><li>At the end of meiosis II, four haploid cells are formed. [Source Data]</li></ul><p>Meiotic events are grouped under the following phases:</p><h5>Meiosis I</h5><h6>Prophase I</h6><p>Prophase I is exceptionally long and complex compared to mitotic prophase, subdivided into five stages based on chromosomal behavior:</p><ul><li>**Leptotene**: Chromosomes become gradually visible and continue to compact.</li><li>**Zygotene**: Homologous chromosomes begin to pair (**synapsis**), forming **synaptonemal complexes**. These paired homologous chromosomes are called **bivalents** or **tetrads**.</li><li>**Pachytene**: The four chromatids of each bivalent become distinct, clearly appearing as tetrads. This stage is characterized by the formation of **recombination nodules**, where **crossing over** (exchange of genetic material between non-sister chromatids) occurs, mediated by the enzyme recombinase.</li><li>**Diplotene**: The synaptonemal complex dissolves, and recombined homologous chromosomes tend to separate, remaining linked only at the sites of crossovers, forming X-shaped structures called **chiasmata**. Diplotene can be very prolonged in some vertebrate oocytes.</li><li>**Diakinesis**: Marked by the terminalization of chiasmata. Chromosomes are fully condensed, and the meiotic spindle assembles. The nucleolus disappears, and the nuclear envelope breaks down, transitioning into metaphase I. [Source Data]</li></ul><h6>Metaphase I</h6><p>Bivalent chromosomes align on the equatorial plate. Microtubules from opposite poles attach to the kinetochores of homologous chromosomes. [Source Data]</p><h6>Anaphase I</h6><p>Homologous chromosomes separate and move to opposite poles, while sister chromatids remain attached at their centromeres. [Source Data]</p><h6>Telophase I</h6><p>The nuclear membrane and nucleolus reappear. Cytokinesis follows, resulting in a **dyad of cells**. Chromosomes may undergo some dispersion but generally do not reach the extended state of interphase. [Source Data]</p><h6>Interkinesis</h6><p>A short-lived stage between Meiosis I and Meiosis II, during which no DNA replication occurs. [Source Data]</p><h5>Meiosis II</h5><p>Meiosis II closely resembles a normal mitosis and is initiated immediately after cytokinesis, often before chromosomes fully elongate.</p><h6>Prophase II</h6><p>The nuclear membrane disappears, and chromosomes become compact again. [Source Data]</p><h6>Metaphase II</h6><p>Chromosomes align at the equator, and microtubules from opposite poles attach to the kinetochores of sister chromatids. [Source Data]</p><h6>Anaphase II</h6><p>The centromeres of each chromosome split, allowing sister chromatids (now daughter chromosomes) to separate and move towards opposite poles due to the shortening of microtubules. [Source Data]</p><h6>Telophase II</h6><p>Meiosis ends with Telophase II. Nuclear envelopes reform around the two groups of chromosomes at each pole, followed by cytokinesis, resulting in a **tetrad of cells** – four haploid daughter cells. [Source Data]</p><h4>Significance of Meiosis</h4><p>Meiosis is crucial for:</p><ul><li>**Conservation of Chromosome Number**: It ensures that the specific chromosome number of each species is maintained across generations by halving the chromosome number in gametes, which is then restored upon fertilization.</li><li>**Genetic Variability**: Through crossing over and independent assortment of homologous chromosomes during meiosis I, it significantly increases genetic variability within a population from one generation to the next. This variability is vital for evolution. [Source Data]</li></ul><h3>Conclusions and Implications</h3><p>The processes of cell cycle, mitosis, and meiosis are fundamental biological mechanisms essential for life. The cell cycle ensures regulated growth and division, leading to the formation of new cells. Mitosis, as an equational division, is indispensable for the growth, repair, and maintenance of multicellular organisms, ensuring genetic fidelity in somatic cells. Meiosis, a reductional division, is the cornerstone of sexual reproduction, enabling the formation of genetically diverse haploid gametes, thereby preserving species-specific chromosome numbers across generations and fueling evolutionary processes through genetic recombination. The precise coordination and genetic control of these events underscore their biological importance. [Source Data]</p><h3>Action Items and Recommendations</h3><p>Based on the provided biological text, there are no explicit 'Action Items' or 'Stakeholder responsibilities' articulated, as the content is descriptive and explanatory rather than prescriptive for practical implementation or project management. Similarly, there are no 'Actionable Recommendations' presented within this scientific exposition. The text serves as a comprehensive overview of cell division mechanisms and their biological significance. [Source Data]</p>"
                            }
                        ]
                    }
                }
            ]
        }
    ],
    "fileName": "",
    "projectType": "DOCUMENT",
    "__v": 0,
    "headerCols": {
        "col1": "Part Id",
        "col2": "FileName",
        "col3": "Content"
    }
},
{
    "_id": "695e03dcf98d992fd8248eb6",
    "projectId": "853283",
    "projectName": "K12MATH1 - 03 Matrices",
    "projectDescription": "",
    "projectCategory": "Educational",
    "projectCreatedDate": "2026-01-07T06:57:29.133Z",
    "projectUpdatedDate": "2026-01-07T06:57:29.133Z",
    "projectStatus": "Active",
    "companyId": "6957bd3ad58ea40715c6ae98",
    "projectCreatedBy": "6957bd3ad58ea40715c6ae99",
    "projectUpdatedBy": "6957bd3ad58ea40715c6ae99",
    "projectLastAccessDate": "2026-01-30T11:19:26.417Z",
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "Content",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "K12MATH1 - 03 Matrices.pdf",
            "col3": "34 MATHEMATICS v The essence of Mathematics lies in its freedom. — CANTOR v 3.1 Introduction The knowledge of matrices is necessary in various branches of mathematics. Matrices are one of the most powerful tools in mathematics. This mathematical tool simplifies our work to a great extent when compared with other straight forward methods. The evolution of concept of matrices is the result of an attempt to obtain compact and simple methods of solving system of linear equations. Matrices are not only used as a representation of the coefficients in system of linear equations, but utility of matrices far exceeds that use. Matrix notation and operations are used in electronic spreadsheet programs for personal computer, which in turn is used in different areas of business and science like budgeting, sales projection, cost estimation, analysing the results of an experiment etc. Also, many physical operations such as magnification, rotation and reflection through a plane can be represented mathematically by matrices. Matrices are also used in cryptography. This mathematical tool is not only used in certain branches of sciences, but also in genetics, economics, sociology, modern psychology and industrial management. In this chapter, we shall find it interesting to become acquainted with the fundamentals of matrix and matrix algebra. 3.2 Matrix Suppose we wish to express the information that Radha has 15 notebooks. We may express it as [15] with the understanding that the number inside [ ] is the number of notebooks that Radha has. Now, if we have to express that Radha has 15 notebooks and 6 pens. We may express it as [15 6] with the understanding that first number inside [ ] is the number of notebooks while the other one is the number of pens possessed by Radha. Let us now suppose that we wish to express the information of possession Chapter 3 MATRICES Reprint 2025-26MATRICES 35 of notebooks and pens by Radha and her two friends Fauzia and Simran which is as follows: Radha has 15 notebooks and 6 pens, Fauzia has 10 notebooks and 2 pens, Simran has 13 notebooks and 5 pens. Now this could be arranged in the tabular form as follows: Notebooks Pens Radha 15 6 Fauzia 10 2 Simran 13 5 and this can be expressed as or Radha Fauzia Simran Notebooks 15 10 13 Pens 6 2 5 which can be expressed as: In the first arrangement the entries in the first column represent the number of note books possessed by Radha, Fauzia and Simran, respectively and the entries in the second column represent the number of pens possessed by Radha, Fauzia and Simran, Reprint 2025-2636 MATHEMATICS respectively. Similarly, in the second arrangement, the entries in the first row represent the number of notebooks possessed by Radha, Fauzia and Simran, respectively. The entries in the second row represent the number of pens possessed by Radha, Fauzia and Simran, respectively. An arrangement or display of the above kind is called a matrix . Formally, we define matrix as: Definition 1 A matrix is an ordered rectangular array of numbers or functions. The numbers or functions are called the elements or the entries of the matrix. We denote matrices by capital letters. The following are some examples of matrices: 5 – 2 A 0 5 3 6     =       , 1 2 3 2 B 3.5 –1 2 5 3 5 7 i   + −     =         , 3 1 3 C cos tan sin 2 x x x x x   + =   +   In the above examples, the horizontal lines of elements are said to constitute, rows of the matrix and the vertical lines of elements are said to constitute, columns of the matrix. Thus A has 3 rows and 2 columns, B has 3 rows and 3 columns while C has 2 rows and 3 columns. 3.2.1 Order of a matrix A matrix having m rows and n columns is called a matrix of order m × n or simply m × n matrix (read as an m by n matrix). So referring to the above examples of matrices, we have A as 3 × 2 matrix, B as 3 × 3 matrix and C as 2 × 3 matrix. We observe that A has 3 × 2 = 6 elements, B and C have 9 and 6 elements, respectively. In general, an m × n matrix has the following rectangular array: or A = [ a ij ] m × n , 1 ≤ i ≤ m , 1 ≤ j ≤ n i , j ∈ N Thus the i th row consists of the elements a i 1 , a i 2 , a i 3 ,..., a in , while the j th column consists of the elements a 1 j , a 2 j , a 3 j ,..., a mj , In general a ij , is an element lying in the i th row and j th column. We can also call it as the ( i , j ) th element of A. The number of elements in an m × n matrix will be equal to mn . Reprint 2025-26MATRICES 37 A Note In this chapter 1. We shall follow the notation, namely A = [ a ij ] m × n to indicate that A is a matrix of order m × n . 2. We shall consider only those matrices whose elements are real numbers or functions taking real values. We can also represent any point ( x , y ) in a plane by a matrix (column or row) as x y       (or [ x , y ]). For example point P(0, 1) as a matrix representation may be given as 0 P 1   =     or [0 1]. Observe that in this way we can also express the vertices of a closed rectilinear figure in the form of a matrix. For example, consider a quadrilateral ABCD with vertices A (1, 0), B (3, 2), C (1, 3), D (–1, 2). Now, quadrilateral ABCD in the matrix form, can be represented as 2 4 A B C D 1 3 1 1 X 0 2 3 2 × −   =     or 4 2 A 1 0 B 3 2 Y C 1 3 D 1 2 ×       =     −   Thus, matrices can be used as representation of vertices of geometrical figures in a plane. Now, let us consider some examples. Example 1 Consider the following information regarding the number of men and women workers in three factories I, II and III Men workers Women workers I 30 25 II 25 31 III 27 26 Represent the above information in the form of a 3 × 2 matrix. What does the entry in the third row and second column represent? Reprint 2025-2638 MATHEMATICS Solution The information is represented in the form of a 3 × 2 matrix as follows: 30 25 A 25 31 27 26     =       The entry in the third row and second column represents the number of women workers in factory III. Example 2 If a matrix has 8 elements, what are the possible orders it can have? Solution We know that if a matrix is of order m × n , it has mn elements. Thus, to find all possible orders of a matrix with 8 elements, we will find all ordered pairs of natural numbers, whose product is 8. Thus, all possible ordered pairs are (1, 8), (8, 1), (4, 2), (2, 4) Hence, possible orders are 1 × 8, 8 ×1, 4 × 2, 2 × 4 Example 3 Construct a 3 × 2 matrix whose elements are given by 1 | 3 | 2 ij a i j = − . Solution In general a 3 × 2 matrix is given by 11 12 21 22 31 32 A a a a a a a     =       . Now 1 | 3 | 2 ij a i j = − , i = 1, 2, 3 and j = 1, 2. Therefore 11 1 |1 3 1| 1 2 a = − × = 12 1 5 |1 3 2 | 2 2 a = − × = 21 1 1 | 2 3 1| 2 2 a = − × = 22 1 | 2 3 2 | 2 2 a = − × = 31 1 | 3 3 1| 0 2 a = − × = 32 1 3 | 3 3 2 | 2 2 a = − × = Hence the required matrix is given by 5 1 2 1 A 2 2 3 0 2       =         . Reprint 2025-26MATRICES 39 3.3 Types of Matrices In this section, we shall discuss different types of matrices. (i) Column matrix A matrix is said to be a column matrix if it has only one column. For example, 0 3 A 1 1/ 2         = −       is a column matrix of order 4 × 1. In general, A = [ a ij ] m × 1 is a column matrix of order m × 1. (ii) Row matrix A matrix is said to be a row matrix if it has only one row. For example, 1 4 1 B 5 2 3 2 ×   = −     is a row matrix. In general, B = [ b ij ] 1 × n is a row matrix of order 1 × n . (iii) Square matrix A matrix in which the number of rows are equal to the number of columns, is said to be a square matrix . Thus an m × n matrix is said to be a square matrix if m = n and is known as a square matrix of order ‘ n ’. For example 3 1 0 3 A 3 2 1 2 4 3 1 −       =     −   is a square matrix of order 3. In general, A = [ a ij ] m × m is a square matrix of order m . A Note If A = [ a ij ] is a square matrix of order n , then elements (entries) a 11 , a 22 , ... , a nn are said to constitute the diagonal , of the matrix A. Thus, if 1 3 1 A 2 4 1 3 5 6 −     = −       . Then the elements of the diagonal of A are 1, 4, 6. Reprint 2025-2640 MATHEMATICS (iv) Diagonal matrix A square matrix B = [ b ij ] m × m is said to be a diagonal matrix if all its non diagonal elements are zero, that is a matrix B = [ b ij ] m × m is said to be a diagonal matrix if b ij = 0, when i ≠ j . For example, A = [4], 1 0 B 0 2 −   =     , 1.1 0 0 C 0 2 0 0 0 3 −     =       , are diagonal matrices of order 1, 2, 3, respectively. (v) Scalar matrix A diagonal matrix is said to be a scalar matrix if its diagonal elements are equal, that is, a square matrix B = [ b ij ] n × n is said to be a scalar matrix if b ij = 0, when i ≠ j b ij = k , when i = j , for some constant k . For example A = [3], 1 0 B 0 1 −   =   −   , 3 0 0 C 0 3 0 0 0 3     =       are scalar matrices of order 1, 2 and 3, respectively. (vi) Identity matrix A square matrix in which elements in the diagonal are all 1 and rest are all zero is called an identity matrix . In other words, the square matrix A = [ a ij ] n × n is an identity matrix, if 1 if 0 if ij i j a i j =  =  ≠  . We denote the identity matrix of order n by I n . When order is clear from the context, we simply write it as I. For example [1], 1 0 0 1       , 1 0 0 0 1 0 0 0 1           are identity matrices of order 1, 2 and 3, respectively. Observe that a scalar matrix is an identity matrix when k = 1. But every identity matrix is clearly a scalar matrix. Reprint 2025-26MATRICES 41 (vii) Zero matrix A matrix is said to be zero matrix or null matrix if all its elements are zero. For example, [0], 0 0 0 0       , 0 0 0 0 0 0       , [0, 0] are all zero matrices. We denote zero matrix by O. Its order will be clear from the context. 3.3.1 Equality of matrices Definition 2 Two matrices A = [ a ij ] and B = [ b ij ] are said to be equal if (i) they are of the same order (ii) each element of A is equal to the corresponding element of B, that is a ij = b ij for all i and j . For example, 2 3 2 3 and 0 1 0 1             are equal matrices but 3 2 2 3 and 0 1 0 1             are not equal matrices. Symbolically, if two matrices A and B are equal, we write A = B. If 1.5 0 2 6 3 2 x y z a b c −         =             , then x = – 1.5, y = 0, z = 2, a = 6 , b = 3, c = 2 Example 4 If 3 4 2 7 0 6 3 2 6 1 0 6 3 2 2 3 21 0 2 4 21 0 x z y y a c b b + + − −         − − = − − +         − − + −     Find the values of a , b , c , x , y and z . Solution As the given matrices are equal, therefore, their corresponding elements must be equal. Comparing the corresponding elements, we get x + 3 = 0, z + 4 = 6, 2 y – 7 = 3 y – 2 a – 1 = – 3, 0 = 2 c + 2 b – 3 = 2 b + 4, Simplifying, we get a = – 2, b = – 7, c = – 1, x = – 3, y = –5, z = 2 Example 5 Find the values of a , b , c , and d from the following equation: 2 2 4 3 5 4 3 11 24 a b a b c d c d + − −     =     − +     Reprint 2025-2642 MATHEMATICS Solution By equality of two matrices, equating the corresponding elements, we get 2 a + b = 4 5 c – d = 11 a – 2 b = – 3 4 c + 3 d = 24 Solving these equations, we get a = 1, b = 2, c = 3 and d = 4 EXERCISE 3.1 1. In the matrix 2 5 19 7 5 A 35 2 12 2 17 3 1 5   −     = −     −   , write: (i) The order of the matrix, (ii) The number of elements, (iii) Write the elements a 13 , a 21 , a 33 , a 24 , a 23 . 2. If a matrix has 24 elements, what are the possible orders it can have? What, if it has 13 elements? 3. If a matrix has 18 elements, what are the possible orders it can have? What, if it has 5 elements? 4. Construct a 2 × 2 matrix, A = [ a ij ], whose elements are given by: (i) 2 ( ) 2 ij i j a + = (ii) ij i a j = (iii) 2 ( 2 ) 2 ij i j a + = 5. Construct a 3 × 4 matrix, whose elements are given by: (i) 1 | 3 | 2 ij a i j = − + (ii) 2 ij a i j = − 6. Find the values of x , y and z from the following equations: (i) 4 3 5 1 5 y z x     =         (ii) 2 6 2 5 5 8 x y z xy +     =     +     (iii) 9 5 7 x y z x z y z + +         + =         +     7. Find the value of a , b , c and d from the equation: 2 1 5 2 3 0 13 a b a c a b c d − + −     =     − +     Reprint 2025-26MATRICES 43 8. A = [ a ij ] m × n\\ is a square matrix, if (A) m < n (B) m > n (C) m = n (D) None of these 9. Which of the given values of x and y make the following pair of matrices equal 3 7 5 1 2 3 x y x +     + −   , 0 2 8 4 y −       (A) 1 , 7 3 x y − = = (B) Not possible to find (C) y = 7, 2 3 x − = (D) 1 2 , 3 3 x y − − = = 10. The number of all possible matrices of order 3 × 3 with each entry 0 or 1 is: (A) 27 (B) 18 (C) 81 (D) 512 3.4 Operations on Matrices In this section, we shall introduce certain operations on matrices, namely, addition of matrices, multiplication of a matrix by a scalar, difference and multiplication of matrices. 3.4.1 Addition of matrices Suppose Fatima has two factories at places A and B. Each factory produces sport shoes for boys and girls in three different price categories labelled 1, 2 and 3. The quantities produced by each factory are represented as matrices given below: Suppose Fatima wants to know the total production of sport shoes in each price category. Then the total production In category 1 : for boys (80 + 90), for girls (60 + 50) In category 2 : for boys (75 + 70), for girls (65 + 55) In category 3 : for boys (90 + 75), for girls (85 + 75) This can be represented in the matrix form as 80 90 60 50 75 70 65 55 90 75 85 75 + +     + +     + +   . Reprint 2025-2644 MATHEMATICS This new matrix is the sum of the above two matrices. We observe that the sum of two matrices is a matrix obtained by adding the corresponding elements of the given matrices. Furthermore, the two matrices have to be of the same order. Thus, if 11 12 13 21 22 23 A a a a a a a   =     is a 2 × 3 matrix and 11 12 13 21 22 23 B b b b b b b   =     is another 2×3 matrix. Then, we define 11 11 12 12 13 13 21 21 22 22 23 23 A + B a b a b a b a b a b a b + + +   =   + + +   . In general, if A = [ a ij ] and B = [ b ij ] are two matrices of the same order, say m × n . Then, the sum of the two matrices A and B is defined as a matrix C = [ c ij ] m × n , where c ij = a ij + b ij , for all possible values of i and j . Example 6 Given 3 1 1 A 2 3 0   − =     and 2 5 1 B 1 2 3 2     =   −     , find A + B Since A, B are of the same order 2 × 3. Therefore, addition of A and B is defined and is given by 2 3 1 5 1 1 2 3 1 5 0 A + B 1 1 2 2 3 3 0 0 6 2 2     + + − + +     = =     − + +         A Note 1. We emphasise that if A and B are not of the same order, then A + B is not defined. For example if 2 3 A 1 0   =     , 1 2 3 B , 1 0 1   =     then A + B is not defined. 2. We may observe that addition of matrices is an example of binary operation on the set of matrices of the same order. 3.4.2 Multiplication of a matrix by a scalar Now suppose that Fatima has doubled the production at a factory A in all categories (refer to 3.4.1). Reprint 2025-26MATRICES 45 Previously quantities (in standard units) produced by factory A were Revised quantities produced by factory A are as given below: Boys Girls 2 80 2 60 1 2 2 75 2 65 3 2 90 2 85 × ×     × ×     × ×   This can be represented in the matrix form as 160 120 150 130 180 170           . We observe that the new matrix is obtained by multiplying each element of the previous matrix by 2. In general, we may define multiplication of a matrix by a scalar as follows: if A = [ a ij ] m × n is a matrix and k is a scalar, then k A is another matrix which is obtained by multiplying each element of A by the scalar k . In other words, k A = k [ a ij ] m × n = [ k ( a ij )] m × n , that is, ( i , j ) th element of k A is ka ij for all possible values of i and j . For example, if A = 3 1 1.5 5 7 3 2 0 5     −       , then 3A = 3 1 1.5 9 3 4.5 3 5 7 3 3 5 21 9 2 0 5 6 0 15         − = −             Negative of a matrix The negative of a matrix is denoted by – A. We define – A = (– 1) A. Reprint 2025-2646 MATHEMATICS For example, let A = 3 1 5 x     −   , then – A is given by – A = (– 1) 3 1 3 1 A ( 1) 5 5 x x − −     = − =     − −     Difference of matrices If A = [ a ij ], B = [ b ij ] are two matrices of the same order, say m × n , then difference A – B is defined as a matrix D = [ d ij ], where d ij = a ij – b ij , for all value of i and j . In other words, D = A – B = A + (–1) B, that is sum of the matrix A and the matrix – B. Example 7 If 1 2 3 3 1 3 A and B 2 3 1 1 0 2 −     = =     −     , then find 2A – B. Solution We have 2A – B = 2 1 2 3 2 3 1 3 1 3 1 0 2       − − −       = 2 4 6 3 1 3 4 6 2 1 0 2 − −     +     −     = 2 3 4 1 6 3 1 5 3 4 1 6 0 2 2 5 6 0 − + − −     =     + + −     3.4.3 Properties of matrix addition The addition of matrices satisfy the following properties: (i) Commutative Law If A = [ a ij ], B = [ b ij ] are matrices of the same order, say m × n , then A + B = B + A. Now A + B = [ a ij ] + [ b ij ] = [ a ij + b ij ] = [ b ij + a ij ] (addition of numbers is commutative) = ([ b ij ] + [ a ij ]) = B + A (ii) Associative Law For any three matrices A = [ a ij ], B = [ b ij ], C = [ c ij ] of the same order, say m × n , (A + B) + C = A + (B + C). Now (A + B) + C = ([ a ij ] + [ b ij ]) + [ c ij ] = [ a ij + b ij ] + [ c ij ] = [( a ij + b ij ) + c ij ] = [ a ij + ( b ij + c ij )] (Why?) = [ a ij ] + [( b ij + c ij )] = [ a ij ] + ([ b ij ] + [ c ij ]) = A + (B + C) Reprint 2025-26MATRICES 47 (iii) Existence of additive identity Let A = [ a ij ] be an m × n matrix and O be an m × n zero matrix, then A + O = O + A = A. In other words, O is the additive identity for matrix addition. (iv) The existence of additive inverse Let A = [ a ij ] m × n be any matrix, then we have another matrix as – A = [– a ij ] m × n such that A + (– A) = (– A) + A= O. So – A is the additive inverse of A or negative of A. 3.4.4 Properties of scalar multiplication of a matrix If A = [ a ij ] and B = [ b ij ] be two matrices of the same order, say m × n , and k and l are scalars, then (i) k (A +B) = k A + k B, (ii) ( k + l )A = k A + l A (ii) k (A + B) = k ([ a ij ] + [ b ij ]) = k [ a ij + b ij ] = [ k ( a ij + b ij )] = [( k a ij ) + ( k b ij )] = [ k a ij ] + [ k b ij ] = k [ a ij ] + k [ b ij ] = k A + k B (iii) ( k + l ) A = ( k + l ) [ a ij ] = [( k + l ) a ij ] + [ k a ij ] + [ l a ij ] = k [ a ij ] + l [ a ij ] = k A + l A Example 8 If 8 0 2 2 A 4 2 and B 4 2 3 6 5 1 −         = − =         −     , then find the matrix X, such that 2A + 3X = 5B. Solution We have 2A + 3X = 5B or 2A + 3X – 2A = 5B – 2A or 2A – 2A + 3X = 5B – 2A (Matrix addition is commutative) or O + 3X = 5B – 2A (– 2A is the additive inverse of 2A) or 3X = 5B – 2A (O is the additive identity) or X = 1 3 (5B – 2A) or 2 2 8 0 1 X 5 4 2 2 4 2 3 5 1 3 6   −           = − −             −       = 10 10 16 0 1 20 10 8 4 3 25 5 6 12   − −           + −             − − −       Reprint 2025-2648 MATHEMATICS = 10 16 10 0 1 20 8 10 4 3 25 6 5 12 − − +     − +     − − −   = 6 10 1 12 14 3 31 7 − −         − −   = 10 2 3 14 4 3 31 7 3 3 −   −           − −       Example 9 Find X and Y, if 5 2 X Y 0 9   + =     and 3 6 X Y 0 1   − =   −   . Solution We have ( ) ( ) 5 2 3 6 X Y X Y 0 9 0 1     + + − = +     −     . or (X + X) + (Y – Y) = 8 8 0 8       ⇒ 8 8 2X 0 8   =     or X = 8 8 4 4 1 0 8 0 4 2     =         Also (X + Y) – (X – Y) = 5 2 3 6 0 9 0 1     −     −     or (X – X) + (Y + Y) = 5 3 2 6 0 9 1 − −     +   ⇒ 2 4 2Y 0 10 −   =     or Y = 2 4 1 2 1 0 10 0 5 2 − −     =         Example 10 Find the values of x and y from the following equation: 5 3 4 2 7 3 1 2 x y −     +     −     = 7 6 15 14       Solution We have 5 3 4 2 7 3 1 2 x y −     +     −     = 7 6 15 14       ⇒ 2 10 3 4 7 6 14 2 6 1 2 15 14 x y −       + =       −       Reprint 2025-26MATRICES 49 or 2 3 10 4 14 1 2 6 2 x y + −     + − +   = 7 6 15 14       ⇒ 2 3 6 7 6 15 2 4 15 14 x y +     =     −     or 2 x + 3 = 7 and 2 y – 4 = 14 (Why?) or 2 x = 7 – 3 and 2 y = 18 or x = 4 2 and y = 18 2 i.e. x = 2 and y = 9. Example 11 Two farmers Ramkishan and Gurcharan Singh cultivates only three varieties of rice namely Basmati, Permal and Naura. The sale (in Rupees) of these varieties of rice by both the farmers in the month of September and October are given by the following matrices A and B. (i) Find the combined sales in September and October for each farmer in each variety. (ii) Find the decrease in sales from September to October. (iii) If both farmers receive 2% profit on gross sales, compute the profit for each farmer and for each variety sold in October. Solution (i) Combined sales in September and October for each farmer in each variety is given by Reprint 2025-2650 MATHEMATICS (ii) Change in sales from September to October is given by (iii) 2% of B = 2 B 100 × = 0.02 × B = 0.02 = Thus, in October Ramkishan receives ` 100, ` 200 and ` 120 as profit in the sale of each variety of rice, respectively, and Grucharan Singh receives profit of ` 400, ` 200 and ` 200 in the sale of each variety of rice, respectively. 3.4.5 Multiplication of matrices Suppose Meera and Nadeem are two friends. Meera wants to buy 2 pens and 5 story books, while Nadeem needs 8 pens and 10 story books. They both go to a shop to enquire about the rates which are quoted as follows: Pen – ` 5 each, story book – ` 50 each. How much money does each need to spend? Clearly, Meera needs ` (5 × 2 + 50 × 5) that is ` 260, while Nadeem needs (8 × 5 + 50 × 10) ` , that is ` 540. In terms of matrix representation, we can write the above information as follows: Requirements Prices per piece (in Rupees) Money needed (in Rupees) 2 5 8 10       5 50       5 2 5 50 260 8 5 10 50 540 × + ×     =     × + ×     Suppose that they enquire about the rates from another shop, quoted as follows: pen – ` 4 each, story book – ` 40 each. Now, the money required by Meera and Nadeem to make purchases will be respectively ` (4 × 2 + 40 × 5) = ` 208 and ` (8 × 4 + 10 × 40) = ` 432 Reprint 2025-26MATRICES 51 Again, the above information can be represented as follows: Requirements Prices per piece (in Rupees) Money needed (in Rupees) 2 5 8 10       4 40       4 2 40 5 208 8 4 10 4 0 432 × + ×     =     × + ×     Now, the information in both the cases can be combined and expressed in terms of matrices as follows: Requirements Prices per piece (in Rupees) Money needed (in Rupees) 2 5 8 10       5 4 50 40       5 2 5 50 4 2 40 5 8 5 10 5 0 8 4 10 4 0 × + × × + ×     × + × × + ×   = 260 208 540 432       The above is an example of multiplication of matrices. We observe that, for multiplication of two matrices A and B, the number of columns in A should be equal to the number of rows in B. Furthermore for getting the elements of the product matrix, we take rows of A and columns of B, multiply them element-wise and take the sum. Formally, we define multiplication of matrices as follows: The product of two matrices A and B is defined if the number of columns of A is equal to the number of rows of B. Let A = [ a ij ] be an m × n matrix and B = [ b jk ] be an n × p matrix. Then the product of the matrices A and B is the matrix C of order m × p . To get the ( i , k ) th element c ik of the matrix C, we take the i th row of A and k th column of B, multiply them elementwise and take the sum of all these products. In other words, if A = [ a ij ] m × n , B = [ b jk ] n × p , then the i th row of A is [ a i 1 a i 2 ... a in ] and the k th column of B is 1 2 . . . k k nk b b b               , then c ik = a i 1 b 1 k + a i 2 b 2 k + a i 3 b 3 k + ... + a in b nk = 1 n ij jk j a b = ∑ . The matrix C = [ c ik ] m × p is the product of A and B. For example, if 1 1 2 C 0 3 4 −   =     and 2 7 1 D 1 5 4     = −     −   , then the product CD is defined Reprint 2025-2652 MATHEMATICS and is given by 2 7 1 1 2 CD 1 1 0 3 4 5 4   −     = −         −   . This is a 2 × 2 matrix in which each entry is the sum of the products across some row of C with the corresponding entries down some column of D. These four computations are Thus 13 2 CD 17 13 −   =   −   Example 12 Find AB, if 6 9 2 6 0 A and B 2 3 7 9 8     = =         . Solution The matrix A has 2 columns which is equal to the number of rows of B. Hence AB is defined. Now 6(2) 9(7) 6(6) 9(9) 6(0) 9(8) AB 2(2) 3(7) 2(6) 3(9) 2(0) 3(8) + + +   =   + + +   = 12 63 36 81 0 72 4 21 12 27 0 24 + + +     + + +   = 75 117 72 25 39 24       Reprint 2025-26MATRICES 53 Remark If AB is defined, then BA need not be defined. In the above example, AB is defined but BA is not defined because B has 3 column while A has only 2 (and not 3) rows. If A, B are, respectively m × n , k × l matrices, then both AB and BA are defined if and only if n = k and l = m . In particular, if both A and B are square matrices of the same order, then both AB and BA are defined. Non-commutativity of multiplication of matrices Now, we shall see by an example that even if AB and BA are both defined, it is not necessary that AB = BA. Example 13 If 2 3 1 2 3 A and B 4 5 4 2 5 2 1   −     = =     −       , then find AB, BA. Show that AB ≠ BA. Solution Since A is a 2 × 3 matrix and B is 3 × 2 matrix. Hence AB and BA are both defined and are matrices of order 2 × 2 and 3 × 3, respectively. Note that 2 3 1 2 3 AB 4 5 4 2 5 2 1   −     =     −       = 2 8 6",
            "col4": "DATA"
        },
        {
            "col1": "P1_Part_2",
            "col2": "K12MATH1 - 03 Matrices.pdf",
            "col3": "3 10 3 0 4 8 8 10 12 10 5 10 3 − + − + −     =     − + + − + +     and 2 3 2 12 4 6 6 15 1 2 3 BA 4 5 4 20 8 10 12 25 4 2 5 2 1 2 4 4 2 6 5 − − + +     −       = = − − + +       −       − − + +     10 2 21 16 2 37 2 2 11 −     = −     − −   Clearly AB ≠ BA In the above example both AB and BA are of different order and so AB ≠ BA. But one may think that perhaps AB and BA could be the same if they were of the same order. But it is not so, here we give an example to show that even if AB and BA are of same order they may not be same. Example 14 If 1 0 A 0 1   =   −   and 0 1 B 1 0   =     , then 0 1 AB 1 0   =   −   . and 0 1 BA 1 0 −   =     . Clearly AB ≠ BA. Thus matrix multiplication is not commutative. Reprint 2025-2654 MATHEMATICS A Note This does not mean that AB ≠ BA for every pair of matrices A, B for which AB and BA, are defined. For instance, If 1 0 3 0 A , B 0 2 0 4     = =         , then AB = BA = 3 0 0 8       Observe that multiplication of diagonal matrices of same order will be commutative. Zero matrix as the product of two non zero matrices We know that, for real numbers a , b if ab = 0, then either a = 0 or b = 0. This need not be true for matrices, we will observe this through an example. Example 15 Find AB, if 0 1 A 0 2 −   =     and 3 5 B 0 0   =     . Solution We have 0 1 3 5 0 0 AB 0 2 0 0 0 0 −       = =             . Thus, if the product of two matrices is a zero matrix, it is not necessary that one of the matrices is a zero matrix. 3.4.6 Properties of multiplication of matrices The multiplication of matrices possesses the following properties, which we state without proof. 1. The associative law For any three matrices A, B and C. We have (AB) C = A (BC), whenever both sides of the equality are defined. 2. The distributive law For three matrices A, B and C. (i) A (B+C) = AB + AC (ii) (A+B) C = AC + BC, whenever both sides of equality are defined. 3. The existence of multiplicative identity For every square matrix A, there exist an identity matrix of same order such that IA = AI = A. Now, we shall verify these properties by examples. Example 16 If 1 1 1 1 3 1 2 3 4 A 2 0 3 , B 0 2 and C 2 0 2 1 3 1 2 1 4 −     −       = = =       −       − −     , find A(BC), (AB)C and show that (AB)C = A(BC). Reprint 2025-26MATRICES 55 Solution We have 1 1 1 1 3 1 0 1 3 2 4 2 1 AB 2 0 3 0 2 2 0 3 6 0 12 1 18 3 1 2 1 4 3 0 2 9 2 8 1 15 − + + + −                 = = + − + + = −                 − − + − − +         (AB) (C) 2 2 4 0 6 2 8 1 2 1 1 2 3 4 1 18 1 36 2 0 3 36 4 18 2 0 2 1 1 15 1 30 2 0 3 30 4 15 + + − − +     −       = − = − + − + − − +       −       + + − − +     = 4 4 4 7 35 2 39 22 31 2 27 11 −     − −     −   Now BC = 1 6 2 0 3 6 4 3 1 3 1 2 3 4 0 2 0 4 0 0 0 4 0 2 2 0 2 1 1 4 1 8 2 0 3 8 4 4 + + − − +     −       = + + − +       −       − − + − + − − +     = 7 2 3 1 4 0 4 2 7 2 11 8 − −     −     − −   Therefore A(BC) = 7 2 3 1 1 1 1 2 0 3 4 0 4 2 3 1 2 7 2 11 8 − − −         −         − − −     = 7 4 7 2 0 2 3 4 11 1 2 8 14 0 21 4 0 6 6 0 33 2 0 24 21 4 14 6 0 4 9 4 22 3 2 16 + − + + − − + − + −     + + + − − + − − + +     − + + − − + − − − +   = 4 4 4 7 35 2 39 22 31 2 27 11 −     − −     −   . Clearly, (AB) C = A (BC) Reprint 2025-2656 MATHEMATICS Example 17 If 0 6 7 0 1 1 2 A 6 0 8 , B 1 0 2 , C 2 7 8 0 1 2 0 3             = − = = −             −       Calculate AC, BC and (A + B)C. Also, verify that (A + B)C = AC + BC Solution Now, 0 7 8 A + B 5 0 10 8 6 0     = −     −   So (A + B) C = 0 7 8 2 0 14 24 10 5 0 10 2 10 0 30 20 8 6 0 3 16 12 0 28 − +                 − − = − + + =                 − + +         Further AC = 0 6 7 2 0 12 21 9 6 0 8 2 12 0 24 12 7 8 0 3 14 16 0 30 − +                 − − = − + + =                 − + +         and BC = 0 1 1 2 0 2 3 1 1 0 2 2 2 0 6 8 1 2 0 3 2 4 0 2 − +                 − = + + =                 − + −         So AC + BC = 9 1 10 12 8 20 30 2 28             + =             −       Clearly, (A + B) C = AC + BC Example 18 If 1 2 3 A 3 2 1 4 2 1     = −       , then show that A 3 – 23A – 40 I = O Solution We have 2 1 2 3 1 2 3 19 4 8 A A.A 3 2 1 3 2 1 1 12 8 4 2 1 4 2 1 14 6 15             = = − − =                   Reprint 2025-26MATRICES 57 So A 3 = A A 2 = 1 2 3 19 4 8 63 46 69 3 2 1 1 12 8 69 6 23 4 2 1 14 6 15 92 46 63             − = −                   Now A 3 – 23A – 40I = 63 46 69 1 2 3 1 0 0 69 6 23 – 23 3 2 1 – 40 0 1 0 92 46 63 4 2 1 0 0 1             − −                   = 63 46 69 23 46 69 40 0 0 69 6 23 69 46 23 0 40 0 92 46 63 92 46 23 0 0 40 − − − −             − + − − + −             − − − −       = 63 23 40 46 46 0 69 69 0 69 69 0 6 46 40 23 23 0 92 92 0 46 46 0 63 23 40 − − − + − +     − + − + − − +     − + − + − −   = 0 0 0 0 0 0 O 0 0 0     =       Example 19 In a legislative assembly election, a political group hired a public relations firm to promote its candidate in three ways: telephone, house calls, and letters. The cost per contact (in paise) is given in matrix A as A = 40 Telephone 100 Housecall 50 Letter Cost per contact           The number of contacts of each type made in two cities X and Y is given by Telephone Housecall Letter 1000 500 5000 X B Y 3000 1000 10,000 →   =   →   . Find the total amount spent by the group in the two cities X and Y. Reprint 2025-2658 MATHEMATICS Solution We have BA = 40,000 50,000 250, 000 X Y 120,000 + 100,000 +500,000 + + →     →   = 340, 000 X Y 720,000 →     →   So the total amount spent by the group in the two cities is 340,000 paise and 720,000 paise, i.e., ` 3400 and ` 7200, respectively. EXERCISE 3.2 1. Let 2 4 1 3 2 5 A , B , C 3 2 2 5 3 4 −       = = =       −       Find each of the following: (i) A + B (ii) A – B (iii) 3A – C (iv) AB (v) BA 2. Compute the following: (i) a b a b b a b a     +     −     (ii) 2 2 2 2 2 2 2 2 2 2 2 2 a b b c ab bc ac ab a c a b   + +   +     − − + +       (iii) 1 4 6 12 7 6 8 5 16 8 0 5 2 8 5 3 2 4 − −         +             (iv) 2 2 2 2 2 2 2 2 cos sin sin cos sin cos cos sin x x x x x x x x     +             3. Compute the indicated products. (i) a b a b b a b a −         −     (ii) 1 2 3           [2 3 4] (iii) 1 2 1 2 3 2 3 2 3 1 −             (iv) 2 3 4 1 3 5 3 4 5 0 2 4 4 5 6 3 0 5 −                     (v) 2 1 1 0 1 3 2 1 2 1 1 1           −     −   (vi) 2 3 3 1 3 1 0 1 0 2 3 1 −   −         −       Reprint 2025-26MATRICES 59 4. If 1 2 3 3 1 2 4 1 2 A 5 0 2 , B 4 2 5 and C 0 3 2 1 1 1 2 0 3 1 2 3 − −             = = =             − −       , then compute (A+B) and (B – C). Also, verify that A + (B – C) = (A + B) – C. 5. If 2 5 2 3 1 1 3 3 5 5 1 2 4 1 2 4 A and B 3 3 3 5 5 5 7 2 7 6 2 2 3 3 5 5 5                 = =                     , then compute 3A – 5B. 6. Simplify cos sin sin cos cos + sin sin cos cos sin θ θ θ − θ     θ θ     − θ θ θ θ     7. Find X and Y, if (i) 7 0 3 0 X + Y and X – Y 2 5 0 3     = =         (ii) 2 3 2 2 2X + 3Y and 3X 2Y 4 0 1 5 −     = + =     −     8. Find X, if Y = 3 2 1 4       and 2X + Y = 1 0 3 2     −   9. Find x and y , if 1 3 0 5 6 2 0 1 2 1 8 y x       + =             10. Solve the equation for x , y , z and t , if 1 1 3 5 2 3 3 0 2 4 6 x z y t −       + =             11. If 2 1 10 3 1 5 x y −       + =             , find the values of x and y . 12. Given 6 4 3 1 2 3 x y x x y z w w z w +       = +       − +       , find the values of x , y , z and w . Reprint 2025-2660 MATHEMATICS 13. If cos sin 0 F ( ) sin cos 0 0 0 1 x x x x x −     =       , show that F( x ) F( y ) = F( x + y ). 14. Show that (i) 5 1 2 1 2 1 5 1 6 7 3 4 3 4 6 7 − −         ≠                 (ii) 1 2 3 1 1 0 1 1 0 1 2 3 0 1 0 0 1 1 0 1 1 0 1 0 1 1 0 2 3 4 2 3 4 1 1 0 − −                 − ≠ −                         15. Find A 2 – 5A + 6I, if 2 0 1 A 2 1 3 1 1 0     =     −   16. If 1 0 2 A 0 2 1 2 0 3     =       , prove that A 3 – 6A 2 + 7A + 2I = 0 17. If 3 2 1 0 A and I= 4 2 0 1 −     =     −     , find k so that A 2 = k A – 2I 18. If 0 tan 2 A tan 0 2 α   −   =   α       and I is the identity matrix of order 2, show that I + A = (I – A) cos sin sin cos α − α     α α   19. A trust fund has ` 30,000 that must be invested in two different types of bonds. The first bond pays 5% interest per year, and the second bond pays 7% interest per year. Using matrix multiplication, determine how to divide ` 30,000 among the two types of bonds. If the trust fund must obtain an annual total interest of: (a) ` 1800 (b) ` 2000 Reprint 2025-26MATRICES 61 20. The bookshop of a particular school has 10 dozen chemistry books, 8 dozen physics books, 10 dozen economics books. Their selling prices are ` 80, ` 60 and ` 40 each respectively. Find the total amount the bookshop will receive from selling all the books using matrix algebra. Assume X, Y, Z, W and P are matrices of order 2 × n , 3 × k , 2 × p , n × 3 and p × k , respectively. Choose the correct answer in Exercises 21 and 22. 21. The restriction on n , k and p so that PY + WY will be defined are: (A) k = 3, p = n (B) k is arbitrary, p = 2 (C) p is arbitrary, k = 3 (D) k = 2, p = 3 22. If n = p , then the order of the matrix 7X – 5Z is: (A) p × 2 (B) 2 × n (C) n × 3 (D) p × n 3.5. Transpose of a Matrix In this section, we shall learn about transpose of a matrix and special types of matrices such as symmetric and skew symmetric matrices. Definition 3 If A = [ a ij ] be an m × n matrix, then the matrix obtained by interchanging the rows and columns of A is called the transpose of A. Transpose of the matrix A is denoted by A ′ or (A T ). In other words, if A = [ a ij ] m × n , then A ′ = [ a ji ] n × m . For example, if 2 3 3 2 3 5 3 3 0 A 3 1 , then A 1 5 1 0 1 5 5 × ×         = ′ =   −       −       3.5.1 Properties of transpose of the matrices We now state the following properties of transpose of matrices without proof. These may be verified by taking suitable examples. For any matrices A and B of suitable orders, we have (i) (A ′ ) ′ = A, (ii) ( k A) ′ = k A ′ (where k is any constant) (iii) (A + B) ′ = A ′ + B ′ (iv) (A B) ′ = B ′ A ′ Example 20 If 2 1 2 3 3 2 A and B 1 2 4 4 2 0   −   = =         , verify that (i) (A ′ ) ′ = A, (ii) (A + B) ′ = A ′ + B ′ , (iii) ( k B) ′ = k B ′ , where k is any constant. Reprint 2025-2662 MATHEMATICS Solution (i) We have A = ( ) 3 4 3 3 2 3 3 2 A 3 2 A A 4 2 0 4 2 0 2 0         ′ ′ ′ ⇒ = ⇒ = =               Thus (A ′ ) ′ = A (ii) We have A = 3 3 2 , 4 2 0       B = 2 1 2 5 3 1 4 A B 1 2 4 5 4 4   −   − ⇒ + =         Therefore (A + B) ′ = 5 5 3 1 4 4 4     −       Now A ′ = 3 4 2 1 3 2 , B 1 2 , 2 0 2 4         ′ = −             So A ′ + B ′ = 5 5 3 1 4 4 4     −       Thus (A + B) ′ = A ′ + B ′ (iii) We have k B = k 2 1 2 2 2 1 2 4 2 4 k k k k k k − −     =         Then ( k B) ′ = 2 2 1 2 1 2 B 2 4 2 4 k k k k k k k k         ′ − = − =             Thus ( k B) ′ = k B ′ Reprint 2025-26MATRICES 63 Example 21 If [ ] 2 A 4 , B 1 3 6 5 −     = = −       , verify that (AB) ′ = B ′ A ′ . Solution We have A = [ ] 2 4 , B 1 3 6 5 −     = −       then AB = [ ] 2 4 1 3 6 5 −     −       = 2 6 12 4 12 24 5 15 30 − −     −     −   Now A ′ = [–2 4 5] , 1 B 3 6     ′ =     −   B ′ A ′ = [ ] 1 2 4 5 3 2 4 5 6 12 15 (AB) 6 12 24 30 −         ′ − = − =         − − −     Clearly (AB) ′ = B ′ A ′ 3.6 Symmetric and Skew Symmetric Matrices Definition 4 A square matrix A = [ a ij ] is said to be symmetric if A ′ = A, that is, [ a ij ] = [ a ji ] for all possible values of i and j . For example 3 2 3 A 2 1.5 1 3 1 1     = − −     −   is a symmetric matrix as A ′ = A Definition 5 A square matrix A = [ a ij ] is said to be skew symmetric matrix if A ′ = – A, that is a ji = – a ij for all possible values of i and j . Now, if we put i = j , we have a ii = – a ii . Therefore 2 a ii = 0 or a ii = 0 for all i ’s. This means that all the diagonal elements of a skew symmetric matrix are zero. Reprint 2025-2664 MATHEMATICS For example, the matrix 0 B 0 0 e f e g f g     = −     − −   is a skew symmetric matrix as B ′ = –B Now, we are going to prove some results of symmetric and skew-symmetric matrices. Theorem 1 For any square matrix A with real number entries, A + A ′ is a symmetric matrix and A – A ′ is a skew symmetric matrix. Proof Let B = A + A ′ , then B ′ = (A + A ′ ) ′ = A ′ + (A ′ ) ′ (as (A + B) ′ = A ′ + B ′ ) = A ′ + A (as (A ′ ) ′ = A) = A + A ′ (as A + B = B + A) = B Therefore B = A + A ′ is a symmetric matrix Now let C = A – A ′ C ′ = (A – A ′ ) ′ = A ′ – (A ′ ) ′ (Why?) = A ′ – A (Why?) = – (A – A ′ ) = – C Therefore C = A – A ′ is a skew symmetric matrix. Theorem 2 Any square matrix can be expressed as the sum of a symmetric and a skew symmetric matrix. Proof Let A be a square matrix, then we can write 1 1 A (A A ) (A A ) 2 2 ′ ′ = + + − From the Theorem 1, we know that (A + A ′ ) is a symmetric matrix and (A – A ′ ) is a skew symmetric matrix. Since for any matrix A, ( k A) ′ = k A ′ , it follows that 1 (A A ) 2 ′ + Reprint 2025-26MATRICES 65 is symmetric matrix and 1 (A A ) 2 ′ − is skew symmetric matrix. Thus, any square matrix can be expressed as the sum of a symmetric and a skew symmetric matrix. Example 22 Express the matrix 2 2 4 B 1 3 4 1 2 3 − −     = −     − −   as the sum of a symmetric and a skew symmetric matrix. Solution Here B ′ = 2 1 1 2 3 2 4 4 3 −     − −     − −   Let P = 4 3 3 1 1 (B + B ) 3 6 2 2 2 3 2 6 − −     ′ = −     − −   = 3 3 2 2 2 3 3 1 2 3 1 3 2 − −       −       −   −     , Now P ′ = 3 3 2 2 2 3 3 1 2 3 1 3 2 − −       −       −   −     = P Thus P = 1 (B + B ) 2 ′ is a symmetric matrix. Also, let Q = 1 5 0 2 2 0 1 5 1 1 1 (B – B ) 1 0 6 0 3 2 2 2 5 6 0 5 3 0 2 − −     − −         ′ = =       −       −     Reprint 2025-2666 MATHEMATICS Then Q ′ = 1 5 0 2 3 1 0 3 Q 2 5 3 0 2       −   − = −     −       Thus Q = 1 (B – B ) 2 ′ is a skew symmetric matrix. Now 3 3 1 5 2 0 2 2 2 2 2 2 4 3 1 P + Q 3 1 0 3 1 3 4 B 2 2 1 2 3 3 5 1 3 3 0 2 2 − − − −         − −       −       = + = − =         − −       −     − −         Thus, B is represented as the sum of a symmetric and a skew symmetric matrix. EXERCISE 3.3 1. Find the transpose of each of the following matrices: (i) 5 1 2 1           −   (ii) 1 1 2 3 −       (iii) 1 5 6 3 5 6 2 3 1 −         −   2. If 1 2 3 4 1 5 A 5 7 9 and B 1 2 0 2 1 1 1 3 1 − − −         = =         −     , then verify that (i) (A + B) ′ = A ′ + B ′ , (ii) (A – B) ′ = A ′ – B ′ 3. If 3 4 1 2 1 A 1 2 and B 1 2 3 0 1   −     ′ = − =           , then verify that (i) (A + B) ′ = A ′ + B ′ (ii) (A – B) ′ = A ′ – B ′ Reprint 2025-26MATRICES 67 4. If 2 3 1 0 A and B 1 2 1 2 − −     ′ = =         , then find (A + 2B) ′ 5. For the matrices A and B, verify that (AB) ′ = B ′ A ′ , where (i) [ ] 1 A 4 , B 1 2 1 3     = − = −       (ii) [ ] 0 A 1 , B 1 5 7 2     = =       6. If (i) cos sin A sin cos α α   =   − α α   , then verify that A ′ A = I (ii) If sin cos A cos sin α α   =   − α α   , then verify that A ′ A = I 7. (i) Show that the matrix 1 1 5 A 1 2 1 5 1 3 −     = −       is a symmetric matrix. (ii) Show that the matrix 0 1 1 A 1 0 1 1 1 0 −     = −     −   is a skew symmetric matrix. 8. For the matrix 1 5 A 6 7   =     , verify that (i) (A + A ′ ) is a symmetric matrix (ii) (A – A ′ ) is a skew symmetric matrix 9. Find ( ) 1 A A 2 ′ + and ( ) 1 A A 2 ′ − , when 0 A 0 0 a b a c b c     = −     − −   10. Express the following matrices as the sum of a symmetric and a skew symmetric matrix: Reprint 2025-2668 MATHEMATICS (i) 3 5 1 1     −   (ii) 6 2 2 2 3 1 2 1 3 −     − −     −   (iii) 3 3 1 2 2 1 4 5 2 −     − −     − −   (iv) 1 5 1 2     −   Choose the correct answer in the Exercises 11 and 12. 11. If A, B are symmetric matrices of same order, then AB – BA is a (A) Skew symmetric matrix (B) Symmetric matrix (C) Zero matrix (D) Identity matrix 12. If cos sin A , sin cos α − α   =   α α   and A + A ′ = I, then the value of α is (A) 6 π (B) 3 π (C) π (D) 3 2 π 3.7 Invertible Matrices Definition 6 If A is a square matrix of order m , and if there exists another square matrix B of the same order m , such that AB = BA = I, then B is called the inverse matrix of A and it is denoted by A – 1 . In that case A is said to be invertible. For example, let A = 2 3 1 2       and B = 2 3 1 2 −     −   be two matrices. Now AB = 2 3 2 3 1 2 1 2 −         −     = 4 3 6 6 1 0 I 2 2 3 4 0 1 − − +     = =     − − +     Also BA = 1 0 I 0 1   =     . Thus B is the inverse of A, in other words B = A – 1 and A is inverse of B, i.e., A = B –1 Reprint 2025-26MATRICES 69 A Note 1. A rectangular matrix does not possess inverse matrix, since for products BA and AB to be defined and to be equal, it is necessary that matrices A and B should be square matrices of the same order. 2. If B is the inverse of A, then A is also the inverse of B. Theorem 3 (Uniqueness of inverse) Inverse of a square matrix, if it exists, is unique. Proof Let A = [ a ij ] be a square matrix of order m . If possible, let B and C be two inverses of A. We shall show that B = C. Since B is the inverse of A AB = BA = I ... (1) Since C is also the inverse of A AC = CA = I ... (2) Thus B = BI = B (AC) = (BA) C = IC = C Theorem 4 If A and B are invertible matrices of the same order, then (AB) –1 = B –1 A –1 . Proof From the definition of inverse of a matrix, we have (AB) (AB) –1 = 1 or A –1 (AB) (AB) –1 = A –1 I (Pre multiplying both sides by A –1 ) or (A –1 A) B (AB) –1 = A –1 (Since A –1 I = A –1 ) or IB (AB) –1 = A –1 or B (AB) –1 = A –1 or B –1 B (AB) –1 = B –1 A –1 or I (AB) –1 = B –1 A –1 Hence (AB) –1 = B –1 A –1 1. Matrices A and B will be inverse of each other only if (A) AB = BA (B) AB = BA = 0 (C) AB = 0, BA = I (D) AB = BA = I EXERCISE 3.4 Reprint 2025-2670 MATHEMATICS Miscellaneous Examples Example 23 If cos sin A sin cos θ θ   =   − θ θ   , then prove that cos sin A sin cos n n n n n θ θ   =   − θ θ   , n ∈ N . Solution We shall prove the result by using principle of mathematical induction. We have P( n ) : If cos sin A sin cos θ θ   =   − θ θ   , then cos sin A sin cos n n n n n θ θ   =   − θ θ   , n ∈ N P(1) : cos sin A sin cos θ θ   =   − θ θ   , so 1 cos sin A sin cos θ θ   =   − θ θ   Therefore, the result is true for n = 1. Let the result be true for n = k . So P( k ) : cos sin A sin cos θ θ   =   − θ θ   , then cos sin A sin cos k k k k k θ θ   =   − θ θ   Now, we prove that the result holds for n = k +1 Now A k + 1 = cos sin cos sin A A sin cos sin cos k k k k k θ θ θ θ     ⋅ =     − θ θ − θ θ     = cos cos – sin sin cos sin sin cos sin cos cos sin sin sin cos cos k k k k k k k k θ θ θ θ θ θ + θ θ     − θ θ + θ θ − θ θ + θ θ   = cos ( ) sin ( ) cos( 1) sin ( 1) sin ( ) cos ( ) sin ( 1) cos ( 1) k k k k k k k k θ + θ θ + θ + θ + θ     =     − θ + θ θ + θ − + θ + θ     Therefore, the result is true for n = k + 1. Thus by principle of mathematical induction, we have cos sin A sin cos n n n n n θ θ   =   − θ θ   , holds for all natural numbers. Example 24 If A and B are symmetric matrices of the same order, then show that AB is symmetric if and only if A and B commute, that is AB = BA. Solution Since A and B are both symmetric matrices, therefore A ′ = A and B ′ = B. Reprint 2025-26MATRICES 71 Let AB be symmetric, then (AB) ′ = AB But (AB) ′ = B ′ A ′ = BA (Why?) Therefore BA = AB Conversely, if AB = BA, then we shall show that AB is symmetric. Now (AB) ′ = B ′ A ′ = B A (as A and B are symmetric) = AB Hence AB is symmetric. Example 25 Let 2 1 5",
            "col4": "DATA"
        },
        {
            "col1": "P1_Part_3",
            "col2": "K12MATH1 - 03 Matrices.pdf",
            "col3": "2 2 5 A , B , C 3 4 7 4 3 8 −       = = =             . Find a matrix D such that CD – AB = O. Solution Since A, B, C are all square matrices of order 2, and CD – AB is well defined, D must be a square matrix of order 2. Let D = a b c d       . Then CD – AB = 0 gives 2 5 2 1 5 2 3 8 3 4 7 4 a b c d −         −                 = O or 2 5 2 5 3 0 3 8 3 8 43 22 a c b d a c b d + +     −     + +     = 0 0 0 0       or 2 5 3 2 5 3 8 43 3 8 22 a c b d a c b d + − +     + − + −   = 0 0 0 0       By equality of matrices, we get 2 a + 5 c – 3 = 0 ... (1) 3 a + 8 c – 43 = 0 ... (2) 2 b + 5 d = 0 ... (3) and 3 b + 8 d – 22 = 0 ... (4) Solving (1) and (2), we get a = –191, c = 77. Solving (3) and (4), we get b = – 110, d = 44. Reprint 2025-2672 MATHEMATICS Therefore D = 191 110 77 44 a b c d − −     =         Reprint 2025-26",
            "col4": "DATA"
        }
    ],
    "projectDeleted": false,
    "codeFrameColumn": [
        {
            "col1": "col1",
            "col2": "col3",
            "researchObjective": "",
            "filterColumn": [],
            "title": "K12MATH1 - 03 Matrices",
            "netSetting": true,
            "subNetSetting": true,
            "categorySetting": true,
            "sentimentSetting": true,
            "thinkingBudgetSetting": 2000,
            "questionType": "OPENEND",
            "otherBrand": false,
            "existingCodeFrame": [],
            "isList": false,
            "codeFrameWithDataList": [],
            "codeFrameList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 23830,
                        "CandidatesTokenCount": 2040,
                        "TotalTokenCount": 27064,
                        "thoughtsTokenCount": 1194,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Net": "Introduction & Basic Definitions",
                                "Subnets": [
                                    {
                                        "Subnet": "Concept & Utility",
                                        "Categories": [
                                            {
                                                "Category": "Definition of Matrix",
                                                "Summary": "Formal definition of a matrix as an ordered rectangular array of numbers or functions, with elements/entries."
                                            },
                                            {
                                                "Category": "Importance & Applications",
                                                "Summary": "Explains the role of matrices as a powerful mathematical tool, simplifying work and used in various fields like business, science, cryptography, genetics, and economics."
                                            }
                                        ],
                                        "Summary": "Explains what matrices are and why they are useful in mathematics and other fields."
                                    },
                                    {
                                        "Subnet": "Matrix Structure",
                                        "Categories": [
                                            {
                                                "Category": "Elements & Entries",
                                                "Summary": "Refers to the individual numbers or functions contained within a matrix."
                                            },
                                            {
                                                "Category": "Rows & Columns",
                                                "Summary": "Defines horizontal lines as rows and vertical lines as columns within a matrix."
                                            },
                                            {
                                                "Category": "Order of a Matrix",
                                                "Summary": "Describes the dimension of a matrix as m x n, indicating the number of rows (m) and columns (n)."
                                            }
                                        ],
                                        "Summary": "Details the components and arrangement within a matrix."
                                    }
                                ],
                                "Summary": "Covers the foundational understanding of matrices, their representation, and structural components."
                            },
                            {
                                "Net": "Classification of Matrices",
                                "Subnets": [
                                    {
                                        "Subnet": "Structural Types",
                                        "Categories": [
                                            {
                                                "Category": "Column & Row Matrix",
                                                "Summary": "Matrices with only one column or only one row, respectively."
                                            },
                                            {
                                                "Category": "Square Matrix",
                                                "Summary": "A matrix where the number of rows equals the number of columns (m=n)."
                                            }
                                        ],
                                        "Summary": "Categorizes matrices based on their dimensions."
                                    },
                                    {
                                        "Subnet": "Special Element Types",
                                        "Categories": [
                                            {
                                                "Category": "Diagonal Matrix",
                                                "Summary": "A square matrix where all non-diagonal elements are zero."
                                            },
                                            {
                                                "Category": "Scalar Matrix",
                                                "Summary": "A diagonal matrix where all diagonal elements are equal."
                                            },
                                            {
                                                "Category": "Identity Matrix",
                                                "Summary": "A square matrix with all diagonal elements as 1 and all other elements as 0."
                                            },
                                            {
                                                "Category": "Zero Matrix / Null Matrix",
                                                "Summary": "A matrix where all its elements are zero."
                                            }
                                        ],
                                        "Summary": "Classifies matrices based on the specific values or patterns of their elements."
                                    }
                                ],
                                "Summary": "Categorizes matrices based on their specific structural characteristics."
                            },
                            {
                                "Net": "Matrix Operations",
                                "Subnets": [
                                    {
                                        "Subnet": "Basic Arithmetic Operations",
                                        "Categories": [
                                            {
                                                "Category": "Addition of Matrices",
                                                "Summary": "Adding two matrices of the same order by summing corresponding elements."
                                            },
                                            {
                                                "Category": "Scalar Multiplication",
                                                "Summary": "Multiplying each element of a matrix by a scalar (constant)."
                                            },
                                            {
                                                "Category": "Difference of Matrices",
                                                "Summary": "Subtracting two matrices of the same order by finding the difference of corresponding elements (A - B = A + (-1)B)."
                                            }
                                        ],
                                        "Summary": "Covers fundamental arithmetic operations that can be performed on matrices."
                                    },
                                    {
                                        "Subnet": "Matrix Multiplication",
                                        "Categories": [
                                            {
                                                "Category": "Definition & Process",
                                                "Summary": "Defines matrix multiplication and the condition (columns of A = rows of B), detailing the row-by-column element-wise product summation."
                                            },
                                            {
                                                "Category": "Non-Commutativity",
                                                "Summary": "Illustrates that matrix multiplication is generally not commutative (AB ≠ BA), even when both products are defined."
                                            },
                                            {
                                                "Category": "Zero Product",
                                                "Summary": "Shows that the product of two non-zero matrices can be a zero matrix."
                                            }
                                        ],
                                        "Summary": "Explains the specific rules and nuances of multiplying two matrices."
                                    }
                                ],
                                "Summary": "Describes the various mathematical operations that can be performed with matrices."
                            },
                            {
                                "Net": "Properties of Matrix Operations",
                                "Subnets": [
                                    {
                                        "Subnet": "Addition Properties",
                                        "Categories": [
                                            {
                                                "Category": "Commutative Law",
                                                "Summary": "A + B = B + A for matrices of the same order."
                                            },
                                            {
                                                "Category": "Associative Law",
                                                "Summary": "(A + B) + C = A + (B + C) for matrices of the same order."
                                            },
                                            {
                                                "Category": "Additive Identity & Inverse",
                                                "Summary": "Existence of a zero matrix (O) such that A + O = A, and an additive inverse (-A) such that A + (-A) = O."
                                            }
                                        ],
                                        "Summary": "Details the rules and laws governing matrix addition."
                                    },
                                    {
                                        "Subnet": "Multiplication Properties",
                                        "Categories": [
                                            {
                                                "Category": "Associative Law of Multiplication",
                                                "Summary": "(AB)C = A(BC) whenever both sides are defined."
                                            },
                                            {
                                                "Category": "Distributive Laws",
                                                "Summary": "A(B+C) = AB + AC and (A+B)C = AC + BC whenever both sides are defined."
                                            },
                                            {
                                                "Category": "Multiplicative Identity",
                                                "Summary": "Existence of an identity matrix (I) such that AI = IA = A for square matrices."
                                            }
                                        ],
                                        "Summary": "Describes the rules and laws governing matrix multiplication."
                                    },
                                    {
                                        "Subnet": "Scalar Multiplication Properties",
                                        "Categories": [
                                            {
                                                "Category": "Distributivity over Matrix Addition",
                                                "Summary": "k(A + B) = kA + kB for scalar k and matrices A, B."
                                            },
                                            {
                                                "Category": "Distributivity over Scalar Addition",
                                                "Summary": "(k + l)A = kA + lA for scalars k, l and matrix A."
                                            }
                                        ],
                                        "Summary": "Outlines how scalar multiplication interacts with matrix addition."
                                    }
                                ],
                                "Summary": "Focuses on the algebraic rules and laws that govern matrix operations."
                            },
                            {
                                "Net": "Advanced Matrix Concepts",
                                "Subnets": [
                                    {
                                        "Subnet": "Transpose of a Matrix",
                                        "Categories": [
                                            {
                                                "Category": "Definition of Transpose",
                                                "Summary": "The matrix obtained by interchanging the rows and columns of an original matrix (A')."
                                            },
                                            {
                                                "Category": "Properties of Transpose",
                                                "Summary": "Rules like (A')' = A, (kA)' = kA', (A+B)' = A' + B', and (AB)' = B'A'."
                                            }
                                        ],
                                        "Summary": "Covers the definition and properties of interchanging rows and columns of a matrix."
                                    },
                                    {
                                        "Subnet": "Symmetric & Skew Symmetric Matrices",
                                        "Categories": [
                                            {
                                                "Category": "Definition of Symmetric Matrix",
                                                "Summary": "A square matrix A is symmetric if A' = A."
                                            },
                                            {
                                                "Category": "Definition of Skew Symmetric Matrix",
                                                "Summary": "A square matrix A is skew symmetric if A' = -A (diagonal elements are zero)."
                                            },
                                            {
                                                "Category": "Decomposition Theorem",
                                                "Summary": "Any square matrix can be expressed as the sum of a symmetric and a skew symmetric matrix (A = 1/2(A+A') + 1/2(A-A'))."
                                            }
                                        ],
                                        "Summary": "Explores specific types of square matrices based on their relationship with their transpose."
                                    },
                                    {
                                        "Subnet": "Invertible Matrices",
                                        "Categories": [
                                            {
                                                "Category": "Definition of Inverse Matrix",
                                                "Summary": "For a square matrix A, its inverse B (A⁻¹) exists if AB = BA = I."
                                            },
                                            {
                                                "Category": "Uniqueness of Inverse",
                                                "Summary": "Proves that if an inverse of a square matrix exists, it is unique."
                                            },
                                            {
                                                "Category": "Inverse of a Product",
                                                "Summary": "Theorem stating that (AB)⁻¹ = B⁻¹A⁻¹ for invertible matrices A and B."
                                            }
                                        ],
                                        "Summary": "Discusses matrices that have an inverse and related theorems."
                                    }
                                ],
                                "Summary": "Covers more complex matrix properties and types beyond basic operations."
                            },
                            {
                                "Net": "Problem Solving & Applications",
                                "Subnets": [
                                    {
                                        "Subnet": "Worked Examples",
                                        "Categories": [
                                            {
                                                "Category": "Matrix Construction",
                                                "Summary": "Examples demonstrating how to construct a matrix given its order and element definition."
                                            },
                                            {
                                                "Category": "Solving Matrix Equations",
                                                "Summary": "Examples showing how to find unknown variables by equating matrices or solving matrix equations (e.g., 2A + 3X = 5B)."
                                            },
                                            {
                                                "Category": "Practical Word Problems",
                                                "Summary": "Examples applying matrix operations to solve real-world scenarios like sales analysis or cost calculation."
                                            }
                                        ],
                                        "Summary": "Provides step-by-step solutions to various matrix problems."
                                    },
                                    {
                                        "Subnet": "Exercises",
                                        "Categories": [
                                            {
                                                "Category": "Basic Operations Exercises",
                                                "Summary": "Practice problems on matrix addition, subtraction, scalar multiplication, and multiplication."
                                            },
                                            {
                                                "Category": "Matrix Property Verification",
                                                "Summary": "Exercises to verify properties like transpose rules, symmetric/skew-symmetric conditions, and matrix equation solutions."
                                            },
                                            {
                                                "Category": "Conceptual Questions",
                                                "Summary": "Multiple-choice questions testing understanding of matrix definitions and properties."
                                            }
                                        ],
                                        "Summary": "Offers practice problems for self-assessment and reinforcement of concepts."
                                    }
                                ],
                                "Summary": "Showcases the practical use of matrices through examples and practice problems."
                            },
                            {
                                "Net": "Uncodable Responses",
                                "Subnets": [
                                    {
                                        "Subnet": "Non specific NET",
                                        "Categories": [
                                            {
                                                "Category": "Irrelevant / Upcode",
                                                "Summary": "Responses that are irrelevant to the survey question or project and cannot be easily coded into existing themes, requiring further review."
                                            }
                                        ],
                                        "Summary": "Responses that are irrelevant but do not fit other 'uncodable' criteria."
                                    },
                                    {
                                        "Subnet": "Don’t Know Can’t Say",
                                        "Categories": [
                                            {
                                                "Category": "Non-Committal / Vague",
                                                "Summary": "Responses that express uncertainty, indecision, or lack of specific information."
                                            }
                                        ],
                                        "Summary": "Responses that are non-committal or vague."
                                    },
                                    {
                                        "Subnet": "Not Applicable",
                                        "Categories": [
                                            {
                                                "Category": "Not Relevant / Out of Scope",
                                                "Summary": "Responses that are clearly irrelevant or not applicable to the question asked or the survey's context."
                                            }
                                        ],
                                        "Summary": "Responses that are irrelevant or not applicable to the question."
                                    },
                                    {
                                        "Subnet": "None-Nothing",
                                        "Categories": [
                                            {
                                                "Category": "Gibberish / Empty",
                                                "Summary": "Responses consisting of non-language characters, meaningless text, or explicit statements of 'none' or 'nothing'."
                                            }
                                        ],
                                        "Summary": "Responses that are plain gibberish, meaningless, or explicitly state nothing."
                                    }
                                ],
                                "Summary": "Categories for responses that cannot be coded into the main thematic structure."
                            }
                        ],
                        "Summary": "This document is a comprehensive chapter from a mathematics textbook focusing on Matrices. It introduces the fundamental definition and utility of matrices, explaining concepts like elements, rows, columns, and order. The text then delves into various types of matrices, including column, row, square, diagonal, scalar, identity, and zero matrices. A significant portion covers operations on matrices such as addition, scalar multiplication, subtraction, and multiplication, detailing their properties like commutativity, associativity, and the existence of identities and inverses. Further, it explores advanced concepts like the transpose of a matrix, symmetric and skew-symmetric matrices, and invertible matrices, complete with theorems and proofs. The chapter includes numerous examples and exercises to reinforce understanding and demonstrate practical applications."
                    }
                }
            ],
            "onlyCodeFrame": true,
            "errorList": [],
            "codeFrameGenerateType": "CODE_FRAME_ONLY",
            "requestType": "CODEFRAME",
            "requestMode": "DEVELOPMENT",
            "sourceLanguage": "English",
            "targetLanguage": "English",
            "response": {
                "status": "OK",
                "data": {
                    "status": "OK",
                    "PromptTokenCount": 23830,
                    "CandidatesTokenCount": 2040,
                    "TotalTokenCount": 27064,
                    "thoughtsTokenCount": 1194,
                    "finishReason": "STOP",
                    "data": [
                        {
                            "Net": "Introduction & Basic Definitions",
                            "Subnets": [
                                {
                                    "Subnet": "Concept & Utility",
                                    "Categories": [
                                        {
                                            "Category": "Definition of Matrix",
                                            "Summary": "Formal definition of a matrix as an ordered rectangular array of numbers or functions, with elements/entries.",
                                            "code": 3
                                        },
                                        {
                                            "Category": "Importance & Applications",
                                            "Summary": "Explains the role of matrices as a powerful mathematical tool, simplifying work and used in various fields like business, science, cryptography, genetics, and economics.",
                                            "code": 4
                                        }
                                    ],
                                    "Summary": "Explains what matrices are and why they are useful in mathematics and other fields.",
                                    "code": 2
                                },
                                {
                                    "Subnet": "Matrix Structure",
                                    "Categories": [
                                        {
                                            "Category": "Elements & Entries",
                                            "Summary": "Refers to the individual numbers or functions contained within a matrix.",
                                            "code": 6
                                        },
                                        {
                                            "Category": "Rows & Columns",
                                            "Summary": "Defines horizontal lines as rows and vertical lines as columns within a matrix.",
                                            "code": 7
                                        },
                                        {
                                            "Category": "Order of a Matrix",
                                            "Summary": "Describes the dimension of a matrix as m x n, indicating the number of rows (m) and columns (n).",
                                            "code": 8
                                        }
                                    ],
                                    "Summary": "Details the components and arrangement within a matrix.",
                                    "code": 5
                                }
                            ],
                            "Summary": "Covers the foundational understanding of matrices, their representation, and structural components.",
                            "code": 1
                        },
                        {
                            "Net": "Classification of Matrices",
                            "Subnets": [
                                {
                                    "Subnet": "Structural Types",
                                    "Categories": [
                                        {
                                            "Category": "Column & Row Matrix",
                                            "Summary": "Matrices with only one column or only one row, respectively.",
                                            "code": 11
                                        },
                                        {
                                            "Category": "Square Matrix",
                                            "Summary": "A matrix where the number of rows equals the number of columns (m=n).",
                                            "code": 12
                                        }
                                    ],
                                    "Summary": "Categorizes matrices based on their dimensions.",
                                    "code": 10
                                },
                                {
                                    "Subnet": "Special Element Types",
                                    "Categories": [
                                        {
                                            "Category": "Diagonal Matrix",
                                            "Summary": "A square matrix where all non-diagonal elements are zero.",
                                            "code": 14
                                        },
                                        {
                                            "Category": "Scalar Matrix",
                                            "Summary": "A diagonal matrix where all diagonal elements are equal.",
                                            "code": 15
                                        },
                                        {
                                            "Category": "Identity Matrix",
                                            "Summary": "A square matrix with all diagonal elements as 1 and all other elements as 0.",
                                            "code": 16
                                        },
                                        {
                                            "Category": "Zero Matrix / Null Matrix",
                                            "Summary": "A matrix where all its elements are zero.",
                                            "code": 17
                                        }
                                    ],
                                    "Summary": "Classifies matrices based on the specific values or patterns of their elements.",
                                    "code": 13
                                }
                            ],
                            "Summary": "Categorizes matrices based on their specific structural characteristics.",
                            "code": 9
                        },
                        {
                            "Net": "Matrix Operations",
                            "Subnets": [
                                {
                                    "Subnet": "Basic Arithmetic Operations",
                                    "Categories": [
                                        {
                                            "Category": "Addition of Matrices",
                                            "Summary": "Adding two matrices of the same order by summing corresponding elements.",
                                            "code": 20
                                        },
                                        {
                                            "Category": "Scalar Multiplication",
                                            "Summary": "Multiplying each element of a matrix by a scalar (constant).",
                                            "code": 21
                                        },
                                        {
                                            "Category": "Difference of Matrices",
                                            "Summary": "Subtracting two matrices of the same order by finding the difference of corresponding elements (A - B = A + (-1)B).",
                                            "code": 22
                                        }
                                    ],
                                    "Summary": "Covers fundamental arithmetic operations that can be performed on matrices.",
                                    "code": 19
                                },
                                {
                                    "Subnet": "Matrix Multiplication",
                                    "Categories": [
                                        {
                                            "Category": "Definition & Process",
                                            "Summary": "Defines matrix multiplication and the condition (columns of A = rows of B), detailing the row-by-column element-wise product summation.",
                                            "code": 24
                                        },
                                        {
                                            "Category": "Non-Commutativity",
                                            "Summary": "Illustrates that matrix multiplication is generally not commutative (AB ≠ BA), even when both products are defined.",
                                            "code": 25
                                        },
                                        {
                                            "Category": "Zero Product",
                                            "Summary": "Shows that the product of two non-zero matrices can be a zero matrix.",
                                            "code": 26
                                        }
                                    ],
                                    "Summary": "Explains the specific rules and nuances of multiplying two matrices.",
                                    "code": 23
                                }
                            ],
                            "Summary": "Describes the various mathematical operations that can be performed with matrices.",
                            "code": 18
                        },
                        {
                            "Net": "Properties of Matrix Operations",
                            "Subnets": [
                                {
                                    "Subnet": "Addition Properties",
                                    "Categories": [
                                        {
                                            "Category": "Commutative Law",
                                            "Summary": "A + B = B + A for matrices of the same order.",
                                            "code": 29
                                        },
                                        {
                                            "Category": "Associative Law",
                                            "Summary": "(A + B) + C = A + (B + C) for matrices of the same order.",
                                            "code": 30
                                        },
                                        {
                                            "Category": "Additive Identity & Inverse",
                                            "Summary": "Existence of a zero matrix (O) such that A + O = A, and an additive inverse (-A) such that A + (-A) = O.",
                                            "code": 31
                                        }
                                    ],
                                    "Summary": "Details the rules and laws governing matrix addition.",
                                    "code": 28
                                },
                                {
                                    "Subnet": "Multiplication Properties",
                                    "Categories": [
                                        {
                                            "Category": "Associative Law of Multiplication",
                                            "Summary": "(AB)C = A(BC) whenever both sides are defined.",
                                            "code": 33
                                        },
                                        {
                                            "Category": "Distributive Laws",
                                            "Summary": "A(B+C) = AB + AC and (A+B)C = AC + BC whenever both sides are defined.",
                                            "code": 34
                                        },
                                        {
                                            "Category": "Multiplicative Identity",
                                            "Summary": "Existence of an identity matrix (I) such that AI = IA = A for square matrices.",
                                            "code": 35
                                        }
                                    ],
                                    "Summary": "Describes the rules and laws governing matrix multiplication.",
                                    "code": 32
                                },
                                {
                                    "Subnet": "Scalar Multiplication Properties",
                                    "Categories": [
                                        {
                                            "Category": "Distributivity over Matrix Addition",
                                            "Summary": "k(A + B) = kA + kB for scalar k and matrices A, B.",
                                            "code": 37
                                        },
                                        {
                                            "Category": "Distributivity over Scalar Addition",
                                            "Summary": "(k + l)A = kA + lA for scalars k, l and matrix A.",
                                            "code": 38
                                        }
                                    ],
                                    "Summary": "Outlines how scalar multiplication interacts with matrix addition.",
                                    "code": 36
                                }
                            ],
                            "Summary": "Focuses on the algebraic rules and laws that govern matrix operations.",
                            "code": 27
                        },
                        {
                            "Net": "Advanced Matrix Concepts",
                            "Subnets": [
                                {
                                    "Subnet": "Transpose of a Matrix",
                                    "Categories": [
                                        {
                                            "Category": "Definition of Transpose",
                                            "Summary": "The matrix obtained by interchanging the rows and columns of an original matrix (A').",
                                            "code": 41
                                        },
                                        {
                                            "Category": "Properties of Transpose",
                                            "Summary": "Rules like (A')' = A, (kA)' = kA', (A+B)' = A' + B', and (AB)' = B'A'.",
                                            "code": 42
                                        }
                                    ],
                                    "Summary": "Covers the definition and properties of interchanging rows and columns of a matrix.",
                                    "code": 40
                                },
                                {
                                    "Subnet": "Symmetric & Skew Symmetric Matrices",
                                    "Categories": [
                                        {
                                            "Category": "Definition of Symmetric Matrix",
                                            "Summary": "A square matrix A is symmetric if A' = A.",
                                            "code": 44
                                        },
                                        {
                                            "Category": "Definition of Skew Symmetric Matrix",
                                            "Summary": "A square matrix A is skew symmetric if A' = -A (diagonal elements are zero).",
                                            "code": 45
                                        },
                                        {
                                            "Category": "Decomposition Theorem",
                                            "Summary": "Any square matrix can be expressed as the sum of a symmetric and a skew symmetric matrix (A = 1/2(A+A') + 1/2(A-A')).",
                                            "code": 46
                                        }
                                    ],
                                    "Summary": "Explores specific types of square matrices based on their relationship with their transpose.",
                                    "code": 43
                                },
                                {
                                    "Subnet": "Invertible Matrices",
                                    "Categories": [
                                        {
                                            "Category": "Definition of Inverse Matrix",
                                            "Summary": "For a square matrix A, its inverse B (A⁻¹) exists if AB = BA = I.",
                                            "code": 48
                                        },
                                        {
                                            "Category": "Uniqueness of Inverse",
                                            "Summary": "Proves that if an inverse of a square matrix exists, it is unique.",
                                            "code": 49
                                        },
                                        {
                                            "Category": "Inverse of a Product",
                                            "Summary": "Theorem stating that (AB)⁻¹ = B⁻¹A⁻¹ for invertible matrices A and B.",
                                            "code": 50
                                        }
                                    ],
                                    "Summary": "Discusses matrices that have an inverse and related theorems.",
                                    "code": 47
                                }
                            ],
                            "Summary": "Covers more complex matrix properties and types beyond basic operations.",
                            "code": 39
                        },
                        {
                            "Net": "Problem Solving & Applications",
                            "Subnets": [
                                {
                                    "Subnet": "Worked Examples",
                                    "Categories": [
                                        {
                                            "Category": "Matrix Construction",
                                            "Summary": "Examples demonstrating how to construct a matrix given its order and element definition.",
                                            "code": 53
                                        },
                                        {
                                            "Category": "Solving Matrix Equations",
                                            "Summary": "Examples showing how to find unknown variables by equating matrices or solving matrix equations (e.g., 2A + 3X = 5B).",
                                            "code": 54
                                        },
                                        {
                                            "Category": "Practical Word Problems",
                                            "Summary": "Examples applying matrix operations to solve real-world scenarios like sales analysis or cost calculation.",
                                            "code": 55
                                        }
                                    ],
                                    "Summary": "Provides step-by-step solutions to various matrix problems.",
                                    "code": 52
                                },
                                {
                                    "Subnet": "Exercises",
                                    "Categories": [
                                        {
                                            "Category": "Basic Operations Exercises",
                                            "Summary": "Practice problems on matrix addition, subtraction, scalar multiplication, and multiplication.",
                                            "code": 57
                                        },
                                        {
                                            "Category": "Matrix Property Verification",
                                            "Summary": "Exercises to verify properties like transpose rules, symmetric/skew-symmetric conditions, and matrix equation solutions.",
                                            "code": 58
                                        },
                                        {
                                            "Category": "Conceptual Questions",
                                            "Summary": "Multiple-choice questions testing understanding of matrix definitions and properties.",
                                            "code": 59
                                        }
                                    ],
                                    "Summary": "Offers practice problems for self-assessment and reinforcement of concepts.",
                                    "code": 56
                                }
                            ],
                            "Summary": "Showcases the practical use of matrices through examples and practice problems.",
                            "code": 51
                        },
                        {
                            "Net": "Uncodable Responses",
                            "Subnets": [
                                {
                                    "Subnet": "Non specific NET",
                                    "Categories": [
                                        {
                                            "Category": "Irrelevant / Upcode",
                                            "Summary": "Responses that are irrelevant to the survey question or project and cannot be easily coded into existing themes, requiring further review.",
                                            "code": 62
                                        }
                                    ],
                                    "Summary": "Responses that are irrelevant but do not fit other 'uncodable' criteria.",
                                    "code": 61
                                },
                                {
                                    "Subnet": "Don’t Know Can’t Say",
                                    "Categories": [
                                        {
                                            "Category": "Non-Committal / Vague",
                                            "Summary": "Responses that express uncertainty, indecision, or lack of specific information.",
                                            "code": 64
                                        }
                                    ],
                                    "Summary": "Responses that are non-committal or vague.",
                                    "code": 63
                                },
                                {
                                    "Subnet": "Not Applicable",
                                    "Categories": [
                                        {
                                            "Category": "Not Relevant / Out of Scope",
                                            "Summary": "Responses that are clearly irrelevant or not applicable to the question asked or the survey's context.",
                                            "code": 66
                                        }
                                    ],
                                    "Summary": "Responses that are irrelevant or not applicable to the question.",
                                    "code": 65
                                },
                                {
                                    "Subnet": "None-Nothing",
                                    "Categories": [
                                        {
                                            "Category": "Gibberish / Empty",
                                            "Summary": "Responses consisting of non-language characters, meaningless text, or explicit statements of 'none' or 'nothing'.",
                                            "code": 68
                                        }
                                    ],
                                    "Summary": "Responses that are plain gibberish, meaningless, or explicitly state nothing.",
                                    "code": 67
                                }
                            ],
                            "Summary": "Categories for responses that cannot be coded into the main thematic structure.",
                            "code": 60
                        }
                    ],
                    "Summary": "This document is a comprehensive chapter from a mathematics textbook focusing on Matrices. It introduces the fundamental definition and utility of matrices, explaining concepts like elements, rows, columns, and order. The text then delves into various types of matrices, including column, row, square, diagonal, scalar, identity, and zero matrices. A significant portion covers operations on matrices such as addition, scalar multiplication, subtraction, and multiplication, detailing their properties like commutativity, associativity, and the existence of identities and inverses. Further, it explores advanced concepts like the transpose of a matrix, symmetric and skew-symmetric matrices, and invertible matrices, complete with theorems and proofs. The chapter includes numerous examples and exercises to reinforce understanding and demonstrate practical applications."
                }
            },
            "summaryList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 23466,
                        "CandidatesTokenCount": 2460,
                        "TotalTokenCount": 27923,
                        "thoughtsTokenCount": 1997,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Summary": "<h2>Comprehensive Synthesis Report: Fundamentals of Matrices</h2><h3>1. Introduction to Matrices</h3><p>Mathematics, in its essence, thrives on freedom and the development of powerful tools to simplify complex problems. Matrices represent one such indispensable tool, simplifying work extensively compared to straightforward methods, particularly in solving systems of linear equations. Beyond their foundational use in representing coefficients, matrices find widespread application in electronic spreadsheet programs for business (budgeting, sales projection, cost estimation), scientific experiments, physical operations (magnification, rotation, reflection), cryptography, genetics, economics, sociology, modern psychology, and industrial management.</p><h3>2. Definition and Structure of a Matrix</h3><p>A matrix is formally defined as an ordered rectangular array of numbers or functions, which are referred to as its elements or entries. Matrices are typically denoted by capital letters. The elements are arranged in horizontal lines called 'rows' and vertical lines called 'columns'.</p><h4>2.1. Order of a Matrix</h4><p>The 'order' of a matrix specifies its dimensions. An <i>m</i> × <i>n</i> matrix has <i>m</i> rows and <i>n</i> columns. The total number of elements in an <i>m</i> × <i>n</i> matrix is <i>mn</i>. For instance, a 3 × 2 matrix contains 6 elements. The element in the <i>i</i>-th row and <i>j</i>-th column is denoted as <i>a<sub>ij</sub></i>.</p><p>Matrices can also represent points in a plane or vertices of geometrical figures, providing a compact mathematical representation (e.g., a point (x, y) can be represented as [x; y] or [x y]).</p><h3>3. Types of Matrices</h3><p>Matrices are classified into several types based on their structure:</p><ul><li><b>Column Matrix:</b> A matrix with only one column (order <i>m</i> × 1).</li><li><b>Row Matrix:</b> A matrix with only one row (order 1 × <i>n</i>).</li><li><b>Square Matrix:</b> A matrix where the number of rows equals the number of columns (<i>m</i> = <i>n</i>). The elements <i>a<sub>11</sub></i>, <i>a<sub>22</sub></i>, ..., <i>a<sub>nn</sub></i> constitute the 'diagonal' of the matrix.</li><li><b>Diagonal Matrix:</b> A square matrix where all non-diagonal elements are zero (<i>b<sub>ij</sub></i> = 0 when <i>i</i> ≠ <i>j</i>).</li><li><b>Scalar Matrix:</b> A diagonal matrix where all diagonal elements are equal to a constant <i>k</i> (<i>b<sub>ij</sub></i> = 0 when <i>i</i> ≠ <i>j</i>, and <i>b<sub>ij</sub></i> = <i>k</i> when <i>i</i> = <i>j</i>).</li><li><b>Identity Matrix:</b> A square matrix where diagonal elements are all 1 and all other elements are zero. It is denoted by <i>I<sub>n</sub></i> for order <i>n</i>. An identity matrix is a special case of a scalar matrix where <i>k</i> = 1.</li><li><b>Zero Matrix (Null Matrix):</b> A matrix where all its elements are zero. It is denoted by <i>O</i>.</li></ul><h3>4. Equality of Matrices</h3><p>Two matrices A = [<i>a<sub>ij</sub></i>] and B = [<i>b<sub>ij</sub></i>] are considered equal if and only if:</p><ol><li>They are of the same order.</li><li>Each element of A is equal to the corresponding element of B (<i>a<sub>ij</sub></i> = <i>b<sub>ij</sub></i> for all <i>i</i> and <i>j</i>).</li></ol><h3>5. Operations on Matrices</h3><h4>5.1. Addition of Matrices</h4><p>The sum of two matrices A and B, both of the same order <i>m</i> × <i>n</i>, is a new matrix C = [<i>c<sub>ij</sub></i>] of the same order, where <i>c<sub>ij</sub></i> = <i>a<sub>ij</sub></i> + <i>b<sub>ij</sub></i>. Addition is only defined for matrices of identical order.</p><h5>Properties of Matrix Addition:</h5><ul><li><b>Commutative Law:</b> A + B = B + A</li><li><b>Associative Law:</b> (A + B) + C = A + (B + C)</li><li><b>Existence of Additive Identity:</b> A + O = O + A = A (where O is the zero matrix of the same order).</li><li><b>Existence of Additive Inverse:</b> A + (– A) = (– A) + A = O (where – A is the negative of A).</li></ul><h4>5.2. Scalar Multiplication of a Matrix</h4><p>Multiplying a matrix A = [<i>a<sub>ij</sub></i>] by a scalar <i>k</i> results in a new matrix <i>k</i>A, where each element of A is multiplied by <i>k</i>. That is, <i>k</i>A = [<i>k</i>(<i>a<sub>ij</sub></i>)].</p><h5>Properties of Scalar Multiplication:</h5><ul><li><i>k</i>(A + B) = <i>k</i>A + <i>k</i>B</li><li>(<i>k</i> + <i>l</i>)A = <i>k</i>A + <i>l</i>A</li></ul><h4>5.3. Negative and Difference of Matrices</h4><ul><li><b>Negative of a Matrix:</b> The negative of matrix A is – A = (– 1)A.</li><li><b>Difference of Matrices:</b> For two matrices A and B of the same order, A – B is defined as A + (– 1)B, where elements are subtracted component-wise: <i>d<sub>ij</sub></i> = <i>a<sub>ij</sub></i> – <i>b<sub>ij</sub></i>.</li></ul><h4>5.4. Multiplication of Matrices</h4><p>The product of two matrices A and B, denoted AB, is defined only if the number of columns in A is equal to the number of rows in B. If A is an <i>m</i> × <i>n</i> matrix and B is an <i>n</i> × <i>p</i> matrix, their product C = AB will be an <i>m</i> × <i>p</i> matrix. The element <i>c<sub>ik</sub></i> in the <i>i</i>-th row and <i>k</i>-th column of C is calculated by taking the dot product of the <i>i</i>-th row of A and the <i>k</i>-th column of B (i.e., <i>c<sub>ik</sub></i> = Σ <i>a<sub>ij</sub>b<sub>jk</sub></i> from <i>j</i>=1 to <i>n</i>).</p><h5>Key Findings on Matrix Multiplication:</h5><ul><li><b>Non-commutativity:</b> In general, AB ≠ BA, even when both products are defined and of the same order.</li><li><b>Zero Product:</b> It is possible for the product of two non-zero matrices to be a zero matrix.</li></ul><h5>Properties of Matrix Multiplication:</h5><ul><li><b>Associative Law:</b> (AB)C = A(BC), whenever both sides are defined.</li><li><b>Distributive Law:</b><ul><li>A(B+C) = AB + AC</li><li>(A+B)C = AC + BC</li></ul>whenever both sides are defined.</li><li><b>Multiplicative Identity:</b> For every square matrix A, there exists an identity matrix I of the same order such that IA = AI = A.</li></ul><h3>6. Transpose of a Matrix</h3><p>The transpose of an <i>m</i> × <i>n</i> matrix A = [<i>a<sub>ij</sub></i>], denoted A' or A<sup>T</sup>, is an <i>n</i> × <i>m</i> matrix obtained by interchanging the rows and columns of A. Thus, A' = [<i>a<sub>ji</sub></i>].</p><h5>Properties of Transpose:</h5><ul><li>(A')' = A</li><li>(<i>k</i>A)' = <i>k</i>A' (where <i>k</i> is any scalar)</li><li>(A + B)' = A' + B'</li><li>(AB)' = B'A'</li></ul><h3>7. Symmetric and Skew Symmetric Matrices</h3><p>These are special types of square matrices:</p><ul><li><b>Symmetric Matrix:</b> A square matrix A is symmetric if A' = A (i.e., <i>a<sub>ij</sub></i> = <i>a<sub>ji</sub></i>).</li><li><b>Skew Symmetric Matrix:</b> A square matrix A is skew symmetric if A' = – A (i.e., <i>a<sub>ji</sub></i> = – <i>a<sub>ij</sub></i>). A key property is that all diagonal elements of a skew symmetric matrix must be zero (<i>a<sub>ii</sub></i> = 0).</li></ul><h5>Theorems on Symmetric and Skew Symmetric Matrices:</h5><ul><li><b>Theorem 1:</b> For any square matrix A with real number entries, A + A' is a symmetric matrix, and A – A' is a skew symmetric matrix.</li><li><b>Theorem 2:</b> Any square matrix can be expressed uniquely as the sum of a symmetric and a skew symmetric matrix: A = ½(A + A') + ½(A – A').</li></ul><h3>8. Invertible Matrices</h3><p>A square matrix A of order <i>m</i> is called invertible if there exists another square matrix B of the same order <i>m</i> such that AB = BA = I (the identity matrix). In this case, B is the inverse of A (denoted A<sup>-1</sup>), and A is said to be invertible.</p><h5>Key Findings on Invertible Matrices:</h5><ul><li><b>Rectangular Matrices:</b> Rectangular matrices do not possess inverses.</li><li><b>Uniqueness of Inverse:</b> The inverse of a square matrix, if it exists, is unique.</li><li><b>Inverse of a Product:</b> If A and B are invertible matrices of the same order, then (AB)<sup>-1</sup> = B<sup>-1</sup>A<sup>-1</sup>.</li></ul><h3>9. Conclusions</h3><p>The study of matrices provides a robust framework for representing and manipulating data across various fields. Understanding their types, equality conditions, and operations (addition, scalar multiplication, and matrix multiplication) is fundamental. Properties such as commutativity/non-commutativity, associativity, and distributivity define how matrices interact. The concepts of transpose, symmetric, skew-symmetric, and invertible matrices introduce further specialized structures and capabilities crucial for advanced applications.</p><h3>10. Actionable Recommendations</h3><ul><li><b>For Data Representation:</b> Utilize matrix notation for compact and efficient representation of tabular data, such as sales figures, inventory, or experimental results.</li><li><b>For Solving Linear Systems:</b> Employ matrix algebra as a simplified and powerful method for solving systems of linear equations, a common task in engineering, economics, and science.</li><li><b>For Geometric Transformations:</b> Apply matrices to mathematically represent and perform transformations like rotation, scaling, and reflection in computer graphics and physics.</li><li><b>For Cryptography and Security:</b> Leverage matrix operations for encoding and decoding information, enhancing data security.</li><li><b>For Financial Analysis:</b> Integrate matrix operations in financial models for budgeting, sales forecasting, and cost analysis, especially when dealing with multiple variables and scenarios.</li><li><b>For Research and Development:</b> Researchers should apply the theorems related to symmetric and skew-symmetric matrices, and the properties of invertible matrices, for advanced mathematical modeling and problem-solving.</li></ul>"
                            }
                        ]
                    }
                }
            ]
        }
    ],
    "fileName": "",
    "projectType": "DOCUMENT",
    "__v": 0,
    "headerCols": {
        "col1": "Part Id",
        "col2": "FileName",
        "col3": "Content"
    }
},
{
    "_id": "695e1424da2133fe04a5f6c9",
    "projectId": "777203",
    "projectName": "K12PSYH - 07 Social Influence and Group Processes",
    "projectDescription": "",
    "projectCategory":"Educational",
    "projectCreatedDate": "2026-01-07T08:06:57.719Z",
    "projectUpdatedDate": "2026-01-07T08:06:57.719Z",
    "projectStatus": "Active",
    "companyId": "6957bd3ad58ea40715c6ae98",
    "projectCreatedBy": "6957bd3ad58ea40715c6ae99",
    "projectUpdatedBy": "6957bd3ad58ea40715c6ae99",
    "projectLastAccessDate": "2026-01-30T11:30:13.459Z",
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "Content",
            "col4": "Status"
        },
        {
            "col1": "P1",
            "col2": "K12PSYH - 07 Social Influence and Group Processes.pdf",
            "col3": "Psychology 120  SOCIAL INFLUENCE AND GROUP SOCIAL INFLUENCE AND GROUP SOCIAL INFLUENCE AND GROUP SOCIAL INFLUENCE AND GROUP SOCIAL INFLUENCE AND GROUP PROCESSES PROCESSES PROCESSES PROCESSES PROCESSES  After reading this chapter, you would be able to:  understand the nature and types of groups and know how they are formed, and  examine the influence of group on individual behaviour.  Introduction Nature and Formation of Groups  Groupthink   (Box 7.1)  Type of Groups  The Minimal Group Paradigm Experiments  (Box 7.2)  Influence of Group on Individual Behaviour  Social Loafing Group Polarisation  CONTENTS  Key Terms Summary Review Questions Project Ideas Weblinks Pedagogical Hints Reprint 2025-26Chapter 7 •   Social Influence and Group Processes   121  N ATURE   AND   F ORMATION   OF   G ROUPS  What is a Group?  The preceding introduction illustrates the importance of groups in our lives. One question that comes to mind is: “How are groups (e.g., your family, class, and the group with which you play) different from other collections of people?” For example, people who have assembled to watch a cricket match or your school function are at one place, but are not interdependent on each other. They do not have defined roles, status and expectations from each other. In the case of your family, class, and the group with which you play, you will   realise   that   there   is   mutual interdependence, each member has roles, there are status differentials, and there are expectations from each other. Thus, your family, class and playgroup are examples of groups and are different from other collections of people.  A   group   may   be   defined   as   an organised   system   of   two   or   more individuals,   who   are   interacting   and interdependent, who have common motives, have a set of role relationships among its members, and have norms that regulate the behaviour of its members.  Groups   have   the   following   salient characteristics : •   A social unit consisting of two or more individuals who perceive themselves as   belonging   to   the   group.   This characteristic of the group helps in distinguishing one group from the other and gives the group its unique identity. •   A collection of individuals who have common motives and goals. Groups function either working towards a given   goal,   or   away   from   certain threats facing the group. •   A collection of individuals who are interdependent, i.e. what one is doing may have consequences for others.  Think about your day-to-day life and the various social interactions you have. In the morning, before going to school, you interact with your family members; in school, you discuss topics and issues with your teachers and classmates; and after school you phone up, visit or play with your friends.   In each of these instances, you are part of a group which not only provides you the needed support and comfort but also facilitates your growth and development as an individual. Have you ever been away to a place where you were without your family, school, and friends?   How did you feel?   Did you feel there was something vital missing in your life? Our lives are influenced by the nature of group membership we have.   It is, therefore, important to be part of groups which would influence us positively and help us in becoming good citizens. In this chapter, we shall try to understand what groups are and how they influence our behaviour. At this point, it is also important to acknowledge that not only do others influence us, but we, as individuals, are also capable of changing others and the society.  Introduction Reprint 2025-26Psychology 122  Suppose one of the fielders in a cricket team drops an important catch during a match — this will have consequence for the entire team. •   Individuals who are trying to satisfy a need through their joint association also influence each other. •   A gathering of individuals who interact with one another either directly or indirectly. •   A   collection   of   individuals   whose interactions are structured by a set of roles and norms. This means that the group members perform the same functions every time the group meets and the group members adhere to group norms. Norms tell us how we ought to behave in the group and specify the behaviours expected from group members. Groups can be differentiated from other collections of people. For example, a   crowd   is also a collection of people who may be present at a place/situation by chance. Suppose you are going on the road and an accident takes place. Soon a large number of people tend to collect. This is an example of a crowd. There is neither   any   structure   nor   feeling   of belongingness in a crowd. Behaviour of people in crowds is irrational and there is no interdependence among members.  Teams   are special kinds of groups. Members of teams often have comple- mentary skills and are committed to a common goal or purpose. Members are mutually accountable for their activities. In   teams,   there   is   a   positive   synergy attained through the coordinated efforts of the members. The main differences between groups and teams are: •   In groups, performance is dependent on   contributions   of   individual members.   In teams, both individual contributions and teamwork matter. •   In groups, the leader or whoever is heading the group holds responsibility for   the   work.   However   in   teams, although there is a leader, members hold themselves responsible.  Fig.7.1 : Look at these Two Pictures Picture A shows a football team — a group in which members interact with one another, have roles and goals. Picture B depicts an audience watching the football match — a mere collection of people who by some coincidence (may be their interest in football) happened to be in the same place at the same time.  Picture A   Picture B Reprint 2025-26Chapter 7 •   Social Influence and Group Processes   123  An   audience   is also a collection of people who have assembled for a special purpose, may be to watch a cricket match or a movie. Audiences are generally passive but sometimes they go into a frenzy and become mobs. In mobs, there is a definite sense of purpose. There is polarisation in attention, and actions of persons are in a common   direction.   Mob   behaviour   is characterised by homogeneity of thought and behaviour as well as impulsivity.  Why Do People Join Groups?  All of you are members of your family, class and groups with which you interact or play. Similarly, other people are also members of a number of groups at any given   time.   Dif ferent   groups   satisfy different needs, and therefore, we are simultaneously   members   of   dif ferent groups. This sometimes creates pressures for us because there may be competing demands and expectations. Most often we are   able   to   handle   these   competing demands and expectations. People join groups because these groups satisfy a range of needs. In general, people join groups for the following reasons : •   Security   : When we are alone,   we feel insecure.   Groups   reduce   this insecurity. Being with people gives a sense of comfort, and protection. As a result, people feel stronger, and are less vulnerable to threats. •   Status   : When we are members of a group that is perceived to be important by   others,   we   feel   recognised   and experience a sense of power. Suppose your   school   wins   in   an   inter - institutional debate competition, you feel proud and think that you are better than others. •   Self-esteem   : Groups provide feelings of self-worth and establish a positive social identity. Being a member of prestigious groups enhances one’s self-concept. •   Satisfaction of one’s psychological and   social   needs   : Groups satisfy one’s social and psychological needs such as sense of belongingness, giving and   receiving   attention,   love,   and power through a group. •   Goal achievement   : Groups help in achieving such goals which cannot be attained individually. There is power in the majority. •   Provide knowledge and information   : Group membership provides knowledge and information and thus broadens our view. As individuals, we may not have all the required information. Groups supplement   this   infor mation   and knowledge.  Group Formation  In this section, we will see how groups are formed. Basic to group formation is some contact and some for m of interaction between   people.   This   interaction   is facilitated by the following conditions: •   Proximity   : Just think about your group of friends. Would you have been friends if you were not living in the same colony, or going to the same school, or may be playing in the same playground? Probably your answer would be ‘No’. Repeated interactions with the same set of individuals give us a chance to know them, and their interests   and   attitudes.   Common interests, attitudes, and background are important determinants of your liking for your group members. •   Similarity   : Being exposed to someone over a period of time makes us assess our similarities and paves the way for formation of groups. Why do we like people who are similar? Psychologists have given several explanations for this. One explanation is that people prefer consistency and like relationships that are consistent. When two people are Reprint 2025-26Psychology 124  similar, there is consistency and they start liking each other. For example, you like playing football and another person in your class also loves playing football; there is a matching of your interests. There are higher chances that you   may   become   friends.   Another explanation given by psychologists is that when we meet similar people, they reinforce and validate our opinions and values, we feel we are right and thus we start liking them. Suppose you are of the opinion that too much watching of television is not good, because it shows too much violence. You meet someone who   also   has   similar   views.   This validates your opinion, and you start liking the person who was instrumental in validating your opinion. •   Common motives and goals   :   When people have common motives or goals, they get together and form a group which   may   facilitate   their   goal attainment. Suppose you want to teach children in a slum area who are unable to go to school. You cannot do this alone   because   you   have   your   own studies and homework. You, therefore, form a group of like-minded friends and start teaching these children. So you have been able to achieve what you could not have done alone.  Stages of Group Formation  Remember that, like everything else in life, groups develop. You do not become a group member the moment you come together.   Groups   usually   go   through different stages of   formation ,   conflict ,  stabilisation ,   performance,   and   dismissal . Tuckman suggested that groups pass through five developmental sequences. These are: forming, storming, norming, performing and adjourning. •   When group members first meet, there is a great deal of uncertainty about the group, the goal, and how it is to be achieved. People try to know each other and assess whether they will fit in. There   is   excitement   as   well   as apprehensions. This stage is called the  forming stage . •   Often, after this stage, there is a stage of intragroup conflict which is referred to as   storming . In this stage, there is conflict among members about how the target of the group is to be achieved, who is to control the group and its resources, and who is to perform what task. When this stage is complete, some sort of hierarchy of leadership in the group develops and a clear vision as to how to achieve the group goal. •   The   storming   stage   is   followed   by another stage known as   norming . Group members by this time develop norms related to group behaviour. This leads to development of a positive group identity. •   The fourth stage is   performing . By this time, the structure of the group has evolved   and   is   accepted   by   group members. The group moves towards achieving the group goal. For some groups, this may be the last stage of group development. •   However, for some groups, for example, in the case of an organising committee  Activity 7.1  Identifying Stages of Group Formation  Select 10 members from your class randomly and form a committee to plan an open house. See how they go ahead. Give them full autonomy to do all the planning. Other members of the class observe them as they function. Do you see any of these stages emerging? Which were those? What was the order of stages? Which stages were skipped? Discuss in the class. Reprint 2025-26Chapter 7 •   Social Influence and Group Processes   125  for a school function, there may be another stage known as   adjourning stage . In this stage, once the function is over, the group may be disbanded. However, it must be stated that all groups do not always proceed from one stage to the next in such a systematic manner. Sometimes several stages go on simultaneously, while in other instances groups may go back and forth through the various stages or they may just skip some of the stages. During the process of group formation, groups also develop a structure. We should remember that   group structure   develops as   members   interact.   Over   time   this interaction   shows   r egularities   in distribution of task to be per formed, responsibilities assigned to members, and the prestige or relative status of members. Four   important   elements   of   group structure are : •   Roles   are socially defined expectations that individuals in a given situation are expected to fulfil. Roles refer to the typical behaviour that depicts a person in a given social context. You have the role of a son or a daughter and with this role, there are certain   role expectations , i.e. including the behaviour expected of someone in a particular role. As a daughter or a son, you are expected to respect elders, listen to them, and be responsible towards your studies. •   Nor ms   are   expected   standards   of behaviour and beliefs established, agreed upon, and enforced by group members. They may be considered as a group’s ‘unspoken rules’. In your family, there are norms that guide the behaviour of family members. These  Box 7.1   Groupthink  Generally teamwork in groups leads to beneficial results. However, Irving Janis has suggested that cohesion can interfere with effective leadership and can lead to disastrous decisions. Janis discovered a process known as “groupthink” in which a group allows its concerns for unanimity. They, in fact, “override the motivation to realistically appraise courses of action”. It results in the tendency of decision makers to make irrational and uncritical decisions. Groupthink is characterised by the appearance of consensus or unanimous agreement within a group. Each member believes that all members agree upon a particular decision or a policy. No one expresses dissenting opinion because each person believes it would undermine the cohesion of the group and s/he would be unpopular. Studies have shown that such a group has an exaggerated sense of its own power to control events, and tends to ignore or minimise cues from the real world that suggest danger to its plan. In order to preserve the group’s internal harmony and collective well-being, it becomes increasingly out-of-touch with reality. Groupthink is likely to occur in socially homogenous, cohesive groups that are isolated from outsiders, that have no tradition of considering alternatives, and that face a decision with high costs or failures. Examples of several group decisions at the international level can be cited as illustrations of groupthink phenomenon. These decisions turned out to be major fiascos. The Vietnam War is an example. From 1964 to 1967, President Lyndon Johnson and his advisors in the U.S. escalated the Vietnam War thinking that this would bring North Vietnam to the peace table. The escalation decisions were made despite warnings. The grossly miscalculated move resulted in the loss of 56,000 American and more than one million Vietnamese lives and created huge budget deficits. Some ways to counteract or prevent groupthink are: (i) encouraging and rewarding critical thinking and even disagreement among group members, (ii) encouraging groups to   present alternative courses of action, (iii) inviting outside experts to evaluate the group’s decisions, and (iv) encouraging members to seek feedback from trusted others. Reprint 2025-26Psychology 126  norms   represent   shared   ways   of viewing the world. •   Status   refers to the relative social position given to group members by others. This relative position or status may be either   ascribed   (given may be because of one’s seniority) or   achieved  (the   person   has   achieved   status because of expertise or hard work).   By being members of the group, we enjoy the status associated with that group. All   of   us,   therefor e,   strive   to   be members of such groups which are high in status or are viewed favourably by   others.   Even   within   a   group, dif ferent   members   have   dif ferent prestige and status. For example, the captain of a cricket team has a higher status compared to the other members, although all are equally important for the team’s success. •   Cohesiveness   refers to togetherness, binding, or mutual attraction among group members. As the group becomes more cohesive, group members start to think, feel and act as a social unit, and less like isolated individuals. Members of   a   highly   cohesive   group have a greater desire to remain in the group in comparison to those who belong to low cohesive groups. Cohesiveness refers to the team spirit or ‘we feeling’ or a sense of belongingness to the group. It is difficult to leave a cohesive group or to gain membership of a group which is highly cohesive. Extreme cohesiveness however, may sometimes not be in a group’s interest. Psychologists have identified   the   phenomenon   of  groupthink   (see Box 7.1) which is a consequence of extreme cohesiveness.  T YPE   OF   G ROUPS  Groups differ in many respects; some have a   large   number   of   members   (e.g.,   a country), some are small (e.g., a family), some are short-lived (e.g., a committee), some remain together for many years (e.g., religious   groups),   some   are   highly organised (e.g., army, police, etc.), and others are informally organised (e.g., spectators of a match). People may belong to different types of group. Major types of groups are enumerated below : •   primary and secondary groups •   formal and informal groups •   ingroup and outgroup.  Primary and Secondary Groups  A major difference between primary and secondary groups is that primary groups are pre-existing formations which are usually given to the individual whereas secondary groups are those which the individual joins by choice. Thus, family, caste, and religion are primary groups whereas membership of a political party is an example of a secondary group. In a primary group, there is a face-to-face interaction, members have close physical proximity, and they share warm emotional bonds. Primary groups are central to individual’s functioning and have a very major role in developing values and ideals of the individual during the early stages of development.   In   contrast,   secondary groups are those where relationships among members are more impersonal, indirect, and less frequent. In the primary group, boundaries are less permeable, i.e. members do not have the option to choose its membership as compared to secondary groups where it is easy to leave and join another group.  Formal and Informal Groups  These groups differ in the degree to which the functions of the group are stated explicitly and formally. The functions of a formal group are explicitly stated as in the case of an office organisation. The roles to be performed by group members are stated Reprint 2025-26Chapter 7 •   Social Influence and Group Processes   127  in an explicit manner. The formal and infor mal groups differ on the basis of structure. The formation of formal groups is based on some specific rules or laws and members have definite roles. There are a set of norms which help in establishing order. A university is an example of a formal group. On the other hand, the formation of informal groups is not based on   rules   or   laws   and   there   is   close relationship among members.  Ingroup and Outgroup  Just as individuals compare themselves with others in terms of similarities and differences with respect to what they have and what others have, individuals also compare the group they belong to with groups of which they are not a member. The term ‘ingroup’ refers to one’s own group, and ‘outgroup’ refers to another group. For ingroup members, we use the word ‘ we ’ while for outgroup members, the word ‘ they ’   is used. By using the words they and we, one is categorising people as similar or different. It has been found that persons   in   the   ingroup   are   generally supposed   to   be   similar,   are   viewed favourably, and have desirable traits. Members   of   the   outgroup   are   viewed differently and are often perceived negatively in comparison to the ingroup members. Perceptions of ingroup and outgroup affect our social lives. These differences can be easily understood by studying Tajfel’s experiments given in Box 7.2. Although it is common to make these categorisations, it should be appreciated that these categories are not real and are created by us. In some cultures, plurality is celebrated as has been the case in India. We have a unique composite culture which is reflected not only in the lives we live, but also in our art, architecture, and music.  Box 7.2   The Minimal Group Paradigm Experiments  Tajfel and his colleagues were interested in knowing the minimal conditions for intergroup behaviour. ‘Minimal group paradigm’ was developed to answer this question. British school- boys expressed their preference for paintings by two artists — Vassily Kandinsky and Paul Klee. Children were told that it was an experiment on decision-making. They knew the groups in which they were grouped ( Kandinsky group   and   Klee group ). The identity of other group members was hidden using code numbers. The children then distributed money between recipients only by code number and group membership. Sample distribution matrix : Ingroup member   —   7 8 9 10 11 12 13 14 15 16 17 18 19 Outgroup member —   1 3 5 7 9 11 13 15 17 19 21 23 25 You will agree that these groups were created on a flimsy criterion (i.e. preference for paintings by two artists) which had no past history or future. Yet, results showed that children favoured their own group.  Activity 7.2  Ingroup and Outgroup Distinctions  Think   of   any   interinstitutional competition held in the near past. Ask your friends to write a paragraph about your school and its students, and   about   another   school   and students of that school. Ask the class and   list   the   behaviour   and characteristics of your schoolmates, and students of the other school on the board. Observe the differences and discuss in the class. Do you also see similarities? If yes, discuss them too. Reprint 2025-26Psychology 128  I NFLUENCE   OF   G ROUP   ON   I NDIVIDUAL  B EHAVIOUR  We have seen that groups are powerful as   they   are   able   to   influence   the behaviour of individuals. What is the nature of this influence? What impact does the presence of others have on our perfor mance?   We   will   discuss   two situations : (i) an individual performing an activity alone in the presence of others ( social facilitation ), and (ii) an individual performing an activity along with the others as part of a larger group ( social loafing ).  Social Loafing  Social facilitation research suggests that presence of others leads to arousal and can motivate individuals to enhance their performance if they are already good at solving something. This enhancement occurs   when   a   person’s   efforts   are individually   evaluated.   What   would happen if efforts of an individual in a group are pooled so that you look at the performance of the group as a whole? Do you know what often happens? It has been found that individuals work less hard   in   a   group   than   they   do   when performing   alone.   This   points   to   a phenomenon referred to as   ‘social loafing’ . Social loafing is a reduction in individual effort when working on a collective task, i.e. one in which outputs are pooled with those   of   other   group   members.   An example of such a task is the game of tug-of-war. It is not possible for you to identify how much force each member of the   team   has   been   exerting.   Such situations give opportunities to group members   to   relax   and   become   a   free rider.   This   phenomenon   has   been demonstrated in many experiments by Latane and his associates who asked group of male students to clap or cheer as   loudly   as   possible   as   they  (experimenters)   were   interested   in knowing how much noise people make in social settings. They varied the group size; individuals were either alone, or in groups of two, four and six. The results of the study showed that although the total amount of noise rose up, as size increased, the amount of noise produced by each participant dropped. In other words, each participant put in less effort as the group size increased. Why does social loafing occur? The explanations offered are: •   Group members feel less responsible for the overall task being performed and therefore exert less effort. •   Motivation   of   members   decreases because   they   realise   that   their contributions will not be evaluated on individual basis. •   The performance of the group is not to be compared with other groups. •   There is an improper coordination (or no coordination) among members. •   Belonging to the same group is not important for members. It is only an aggregate of individuals. Social loafing may be reduced by: •   Making   the   efforts   of   each   person identifiable. •   Increasing the pressure to work hard (making group members committed to successful task performance). •   Increasing the apparent importance or value of a task. •   Making   people   feel   that   their individual contribution is important. •   Strengthening   group   cohesiveness which increases the motivation for successful group outcome.  Group Polarisation  We all know that important decisions are taken by groups and not by individuals alone. For example, a decision is to be Reprint 2025-26Chapter 7 •   Social Influence and Group Processes   129  this interaction, your views may become stronger. This firm conviction is because of the following three reasons: •   In the company of like-minded people, you   are   likely   to   hear   newer arguments favouring your viewpoints. This will make you more favourable towards capital punishment. •   When you find others also favouring capital punishment, you feel that this view is validated by the public. This is a sort of   bandwagon effect . •   When you find people having similar views, you are likely to perceive them as ingroup. You start identifying with the group, begin showing conformity, and   as   a   consequence   your   views become strengthened.  Activity 7.3  Assessing Polarisation  Give the class a short, 5-item attitude scale developed by your teacher to assess attitudes towards capital punishment. Based on their responses, divide the class into two groups, i.e. those pro-capital punishment and those anti-capital punishment. Now seat these groups into two different rooms and ask them to discuss a recent case in which death sentence has been given by the court. See how the discussion proceeds in the two groups. After the discussion, re-administer the attitude scale to the group members. Examine if, in both groups, positions have hardened in comparison to their initial position as a result of group discussion.  taken   whether   a   school   has   to   be  established in a village. Such a decision  has to be a group decision. We have also  seen that when groups take decisions,  there is a fear that the phenomenon of  groupthink may sometimes occur (see  Box 7.1). Groups show another tendency  referred to as ‘ group polarisation’ . It has  been found that groups are more likely  to   take   extreme   decisions   than  individuals alone. Suppose there is an  employee who has been caught taking  bribe   or   engaging   in   some   other  unethical act. Her/his colleagues are  asked to decide on what punishment s/ he should be given. They may let her/ him go scot-free or decide to terminate  her/his services instead of imposing a  punishment   which   may   be  commensurate with the unethical act s/ he had engaged in. Whatever the initial  position   in   the   group,   this   position  becomes much stronger as a result of  discussions   in   the   group.   This  strengthening   of   the   group’s   initial  position as a result of group interaction  and discussion is referred to as group  polarisation. This may sometimes have  dangerous repercussions as groups may  take extreme positions, i.e. from very  weak to very strong decisions. Why does group polarisation occur? Let us take an example whether capital  punishment should be there. Suppose  you   favour   capital   punishment   for  heinous crimes, what would happen if  you were interacting with and discussing  this issue with like-minded people? After Reprint 2025-26",
            "col4": "DATA"
        }
    ],
    "projectDeleted": false,
    "codeFrameColumn": [
        {
            "col1": "col1",
            "col2": "col3",
            "researchObjective": "",
            "filterColumn": [],
            "title": "K12PSYH - 07 Social Influence and Group Processes",
            "netSetting": true,
            "subNetSetting": true,
            "categorySetting": true,
            "sentimentSetting": true,
            "thinkingBudgetSetting": 2000,
            "questionType": "OPENEND",
            "otherBrand": false,
            "existingCodeFrame": [],
            "isList": false,
            "codeFrameWithDataList": [],
            "codeFrameList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 7221,
                        "CandidatesTokenCount": 2265,
                        "TotalTokenCount": 10873,
                        "thoughtsTokenCount": 1387,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Net": "Nature and Definition of Groups",
                                "Subnets": [
                                    {
                                        "Subnet": "Group Definition & Distinctions",
                                        "Categories": [
                                            {
                                                "Category": "Definition of a Group",
                                                "Summary": "An organized system of two or more interacting, interdependent individuals with common motives, role relationships, and norms."
                                            },
                                            {
                                                "Category": "Group vs. Crowd",
                                                "Summary": "Differentiating a group from a crowd, which lacks structure, belongingness, and exhibits irrational behavior."
                                            },
                                            {
                                                "Category": "Group vs. Audience",
                                                "Summary": "Distinguishing a group from an audience, which is typically passive but can become a mob with a common purpose."
                                            },
                                            {
                                                "Category": "Group vs. Team",
                                                "Summary": "Comparing groups where performance depends on individual contributions versus teams with complementary skills and mutual accountability."
                                            }
                                        ],
                                        "Summary": "Defines a group and differentiates it from crowds, audiences, and mere collections of people, including teams."
                                    },
                                    {
                                        "Subnet": "Characteristics of Groups",
                                        "Categories": [
                                            {
                                                "Category": "Social Unit & Identity",
                                                "Summary": "Groups are perceived as belonging to a unit with a unique identity."
                                            },
                                            {
                                                "Category": "Common Motives & Goals",
                                                "Summary": "Individuals in groups share common motives and work towards shared goals or against common threats."
                                            },
                                            {
                                                "Category": "Interdependence",
                                                "Summary": "Members are interdependent; actions of one have consequences for others."
                                            }
                                        ],
                                        "Summary": "Key attributes that define a social group."
                                    },
                                    {
                                        "Subnet": "Reasons for Joining Groups",
                                        "Categories": [
                                            {
                                                "Category": "Security & Protection",
                                                "Summary": "Groups reduce insecurity and provide comfort and protection."
                                            },
                                            {
                                                "Category": "Status & Self-Esteem",
                                                "Summary": "Membership in important groups provides recognition, power, and enhances self-worth."
                                            },
                                            {
                                                "Category": "Psychological & Social Needs",
                                                "Summary": "Groups satisfy needs like belongingness, attention, love, and power."
                                            },
                                            {
                                                "Category": "Goal Achievement",
                                                "Summary": "Groups enable individuals to achieve goals not attainable alone."
                                            },
                                            {
                                                "Category": "Knowledge & Information",
                                                "Summary": "Group membership provides knowledge and information, broadening perspectives."
                                            }
                                        ],
                                        "Summary": "Explores the various needs and motivations that drive individuals to join groups."
                                    }
                                ],
                                "Summary": "Covers what constitutes a group, its fundamental characteristics, and distinctions from mere collections of people, as well as reasons for joining."
                            },
                            {
                                "Net": "Group Formation and Development",
                                "Subnets": [
                                    {
                                        "Subnet": "Factors Facilitating Formation",
                                        "Categories": [
                                            {
                                                "Category": "Proximity",
                                                "Summary": "Repeated interactions due to shared physical space or environment."
                                            },
                                            {
                                                "Category": "Similarity",
                                                "Summary": "Shared interests, attitudes, values, and backgrounds that lead to liking and group formation."
                                            },
                                            {
                                                "Category": "Common Motives & Goals",
                                                "Summary": "Shared objectives that prompt individuals to unite to achieve them."
                                            }
                                        ],
                                        "Summary": "Conditions that lead to initial contact and interaction for group formation."
                                    },
                                    {
                                        "Subnet": "Stages of Group Formation (Tuckman's Model)",
                                        "Categories": [
                                            {
                                                "Category": "Forming Stage",
                                                "Summary": "Initial uncertainty, getting to know each other, excitement, and apprehensions."
                                            },
                                            {
                                                "Category": "Storming Stage",
                                                "Summary": "Intragroup conflict over goals, control, and task allocation, leading to leadership hierarchy."
                                            },
                                            {
                                                "Category": "Norming Stage",
                                                "Summary": "Development of norms and a positive group identity."
                                            },
                                            {
                                                "Category": "Performing Stage",
                                                "Summary": "Group structure is accepted, and the group moves towards achieving its goals."
                                            },
                                            {
                                                "Category": "Adjourning Stage",
                                                "Summary": "Group disbandment after the completion of its purpose, common for temporary groups."
                                            }
                                        ],
                                        "Summary": "The sequential phases groups often progress through from initial meeting to dissolution, as per Tuckman's model."
                                    }
                                ],
                                "Summary": "Describes the process by which groups are formed and the typical developmental stages they go through."
                            },
                            {
                                "Net": "Group Structure and Dynamics",
                                "Subnets": [
                                    {
                                        "Subnet": "Elements of Group Structure",
                                        "Categories": [
                                            {
                                                "Category": "Roles",
                                                "Summary": "Socially defined expectations and typical behaviors for individuals in a given situation, including role expectations."
                                            },
                                            {
                                                "Category": "Norms",
                                                "Summary": "Expected standards of behavior and beliefs established, agreed upon, and enforced by group members ('unspoken rules')."
                                            },
                                            {
                                                "Category": "Status",
                                                "Summary": "The relative social position given to group members, which can be ascribed or achieved."
                                            },
                                            {
                                                "Category": "Cohesiveness",
                                                "Summary": "The togetherness, binding, or mutual attraction among group members, leading to a 'we feeling' or team spirit."
                                            }
                                        ],
                                        "Summary": "Defines the key components that give a group its internal organization and functional regularities."
                                    },
                                    {
                                        "Subnet": "Impact of Cohesiveness",
                                        "Categories": [
                                            {
                                                "Category": "Benefits of Cohesion",
                                                "Summary": "Increased desire to remain in the group and stronger team spirit."
                                            },
                                            {
                                                "Category": "Risks of Extreme Cohesion",
                                                "Summary": "Potential for negative outcomes like 'groupthink' when cohesion overrides critical thinking."
                                            }
                                        ],
                                        "Summary": "Examines the effects of group cohesiveness, both positive and negative."
                                    }
                                ],
                                "Summary": "Focuses on the internal organization and functional aspects of groups, including roles, norms, status, and cohesiveness."
                            },
                            {
                                "Net": "Types of Groups",
                                "Subnets": [
                                    {
                                        "Subnet": "Primary vs. Secondary Groups",
                                        "Categories": [
                                            {
                                                "Category": "Primary Groups",
                                                "Summary": "Pre-existing, face-to-face, close physical proximity, warm emotional bonds (e.g., family, caste)."
                                            },
                                            {
                                                "Category": "Secondary Groups",
                                                "Summary": "Joined by choice, more impersonal, indirect relationships, permeable boundaries (e.g., political party)."
                                            }
                                        ],
                                        "Summary": "Distinguishes groups based on their formation, interaction patterns, and emotional bonds."
                                    },
                                    {
                                        "Subnet": "Formal vs. Informal Groups",
                                        "Categories": [
                                            {
                                                "Category": "Formal Groups",
                                                "Summary": "Explicitly stated functions, defined roles, specific rules/laws (e.g., university, office organization)."
                                            },
                                            {
                                                "Category": "Informal Groups",
                                                "Summary": "Not based on formal rules, close relationships among members."
                                            }
                                        ],
                                        "Summary": "Categorizes groups based on the explicitness and formality of their functions and structure."
                                    },
                                    {
                                        "Subnet": "Ingroup vs. Outgroup",
                                        "Categories": [
                                            {
                                                "Category": "Ingroup",
                                                "Summary": "One's own group, viewed favorably, members perceived as similar and desirable ('we')."
                                            },
                                            {
                                                "Category": "Outgroup",
                                                "Summary": "Another group, often perceived negatively and differently compared to the ingroup ('they')."
                                            },
                                            {
                                                "Category": "Minimal Group Paradigm",
                                                "Summary": "Experiments demonstrating that even flimsy criteria can lead to ingroup favoritism."
                                            }
                                        ],
                                        "Summary": "Explores how individuals categorize groups they belong to versus those they don't, and its social implications."
                                    }
                                ],
                                "Summary": "Categorizes groups based on various criteria such as formation, formality, and social identity."
                            },
                            {
                                "Net": "Group Influence on Individual Behavior",
                                "Subnets": [
                                    {
                                        "Subnet": "Social Loafing",
                                        "Categories": [
                                            {
                                                "Category": "Definition of Social Loafing",
                                                "Summary": "Reduction in individual effort when working on a collective task where outputs are pooled."
                                            },
                                            {
                                                "Category": "Reasons for Social Loafing",
                                                "Summary": "Less responsibility, decreased motivation due to lack of individual evaluation, poor coordination, or perceived unimportance of group."
                                            },
                                            {
                                                "Category": "Reducing Social Loafing",
                                                "Summary": "Making individual efforts identifiable, increasing task importance, strengthening cohesiveness, and making contributions seem vital."
                                            }
                                        ],
                                        "Summary": "Examines the phenomenon where individuals exert less effort when working in a group compared to working alone."
                                    },
                                    {
                                        "Subnet": "Group Polarisation",
                                        "Categories": [
                                            {
                                                "Category": "Definition of Group Polarisation",
                                                "Summary": "The strengthening of a group's initial position as a result of group interaction and discussion, leading to more extreme decisions."
                                            },
                                            {
                                                "Category": "Reasons for Polarisation",
                                                "Summary": "Exposure to new arguments, validation by like-minded people (bandwagon effect), and identification with the ingroup."
                                            },
                                            {
                                                "Category": "Consequences of Polarisation",
                                                "Summary": "Groups are more likely to take extreme decisions, ranging from very weak to very strong."
                                            }
                                        ],
                                        "Summary": "Focuses on the tendency of groups to make more extreme decisions than individuals alone, and its underlying causes."
                                    },
                                    {
                                        "Subnet": "Groupthink",
                                        "Categories": [
                                            {
                                                "Category": "Definition of Groupthink",
                                                "Summary": "A phenomenon in cohesive groups where the desire for unanimity overrides realistic appraisal of alternatives, leading to irrational decisions."
                                            },
                                            {
                                                "Category": "Characteristics of Groupthink",
                                                "Summary": "Appearance of consensus, suppression of dissenting opinions, exaggerated sense of power, ignoring real-world cues."
                                            },
                                            {
                                                "Category": "Preventing Groupthink",
                                                "Summary": "Encouraging critical thinking, presenting alternatives, inviting external experts, and seeking feedback from trusted others."
                                            }
                                        ],
                                        "Summary": "Explores the negative consequences of extreme group cohesiveness leading to flawed decision-making."
                                    }
                                ],
                                "Summary": "Discusses how the presence and dynamics of groups influence individual performance and decision-making."
                            },
                            {
                                "Net": "996 Non specific NET",
                                "Subnets": [
                                    {
                                        "Subnet": "Uncodable Responses",
                                        "Categories": [
                                            {
                                                "Category": "Irrelevant or Unrelated",
                                                "Summary": "Responses that are not directly relevant to the survey question or main themes of the project."
                                            },
                                            {
                                                "Category": "Vague or Ambiguous",
                                                "Summary": "Responses that lack clarity or specific detail, making them difficult to assign to existing categories."
                                            }
                                        ],
                                        "Summary": "Responses that are irrelevant, vague, or cannot be easily categorized within the established codeframe and may require further review."
                                    }
                                ],
                                "Summary": "Catch-all for responses that are irrelevant but do not fit other specific non-thematic codes, potentially requiring upcoding."
                            },
                            {
                                "Net": "997 Don’t Know Can’t Say",
                                "Subnets": [
                                    {
                                        "Subnet": "Non-committal Responses",
                                        "Categories": [
                                            {
                                                "Category": "Lack of Knowledge",
                                                "Summary": "Explicit statements indicating the respondent does not know or is unsure."
                                            },
                                            {
                                                "Category": "Inability to Comment",
                                                "Summary": "Responses stating the respondent cannot provide an answer or opinion."
                                            }
                                        ],
                                        "Summary": "Responses indicating uncertainty or an inability to provide a definitive answer."
                                    }
                                ],
                                "Summary": "Catch-all for non-committal and vague responses."
                            },
                            {
                                "Net": "998 Not Applicable",
                                "Subnets": [
                                    {
                                        "Subnet": "Responses Not Applicable",
                                        "Categories": [
                                            {
                                                "Category": "Out of Scope",
                                                "Summary": "Responses that are valid but do not apply to the specific question or context."
                                            },
                                            {
                                                "Category": "Exclusion Criteria",
                                                "Summary": "Responses from individuals who do not meet the applicability criteria for the question."
                                            }
                                        ],
                                        "Summary": "Responses that are irrelevant or not applicable to the specific question asked."
                                    }
                                ],
                                "Summary": "Catch-all for responses that are irrelevant or not applicable to the question at hand."
                            },
                            {
                                "Net": "999 None-Nothing",
                                "Subnets": [
                                    {
                                        "Subnet": "Meaningless Responses",
                                        "Categories": [
                                            {
                                                "Category": "Gibberish/Random Characters",
                                                "Summary": "Responses consisting of non-language characters, random letters, or meaningless strings."
                                            },
                                            {
                                                "Category": "Blank/Empty Response",
                                                "Summary": "Respondents who leave the open-ended field empty or provide no meaningful text."
                                            },
                                            {
                                                "Category": "Explicit None/Nothing",
                                                "Summary": "Responses explicitly stating 'None', 'Nothing', or similar negations without further detail."
                                            }
                                        ],
                                        "Summary": "Responses that are plain gibberish, non-language characters, or contain no meaningful content."
                                    }
                                ],
                                "Summary": "Catch-all for responses that are gibberish, non-language characters, or mean nothing."
                            }
                        ],
                        "Summary": "This document provides a comprehensive overview of social influence and group processes, defining groups, detailing their characteristics, formation stages, and structural elements such as roles, norms, status, and cohesiveness. It also explores various types of groups, including primary/secondary, formal/informal, and ingroup/outgroup distinctions, and discusses key phenomena related to group influence on individual behavior like social loafing, group polarization, and groupthink."
                    }
                }
            ],
            "onlyCodeFrame": true,
            "errorList": [],
            "codeFrameGenerateType": "CODE_FRAME_ONLY",
            "requestType": "CODEFRAME",
            "requestMode": "DEVELOPMENT",
            "sourceLanguage": "English",
            "targetLanguage": "English",
            "response": {
                "status": "OK",
                "data": {
                    "status": "OK",
                    "PromptTokenCount": 7221,
                    "CandidatesTokenCount": 2265,
                    "TotalTokenCount": 10873,
                    "thoughtsTokenCount": 1387,
                    "finishReason": "STOP",
                    "data": [
                        {
                            "Net": "Nature and Definition of Groups",
                            "Subnets": [
                                {
                                    "Subnet": "Group Definition & Distinctions",
                                    "Categories": [
                                        {
                                            "Category": "Definition of a Group",
                                            "Summary": "An organized system of two or more interacting, interdependent individuals with common motives, role relationships, and norms.",
                                            "code": 3
                                        },
                                        {
                                            "Category": "Group vs. Crowd",
                                            "Summary": "Differentiating a group from a crowd, which lacks structure, belongingness, and exhibits irrational behavior.",
                                            "code": 4
                                        },
                                        {
                                            "Category": "Group vs. Audience",
                                            "Summary": "Distinguishing a group from an audience, which is typically passive but can become a mob with a common purpose.",
                                            "code": 5
                                        },
                                        {
                                            "Category": "Group vs. Team",
                                            "Summary": "Comparing groups where performance depends on individual contributions versus teams with complementary skills and mutual accountability.",
                                            "code": 6
                                        }
                                    ],
                                    "Summary": "Defines a group and differentiates it from crowds, audiences, and mere collections of people, including teams.",
                                    "code": 2
                                },
                                {
                                    "Subnet": "Characteristics of Groups",
                                    "Categories": [
                                        {
                                            "Category": "Social Unit & Identity",
                                            "Summary": "Groups are perceived as belonging to a unit with a unique identity.",
                                            "code": 8
                                        },
                                        {
                                            "Category": "Common Motives & Goals",
                                            "Summary": "Individuals in groups share common motives and work towards shared goals or against common threats.",
                                            "code": 9
                                        },
                                        {
                                            "Category": "Interdependence",
                                            "Summary": "Members are interdependent; actions of one have consequences for others.",
                                            "code": 10
                                        }
                                    ],
                                    "Summary": "Key attributes that define a social group.",
                                    "code": 7
                                },
                                {
                                    "Subnet": "Reasons for Joining Groups",
                                    "Categories": [
                                        {
                                            "Category": "Security & Protection",
                                            "Summary": "Groups reduce insecurity and provide comfort and protection.",
                                            "code": 12
                                        },
                                        {
                                            "Category": "Status & Self-Esteem",
                                            "Summary": "Membership in important groups provides recognition, power, and enhances self-worth.",
                                            "code": 13
                                        },
                                        {
                                            "Category": "Psychological & Social Needs",
                                            "Summary": "Groups satisfy needs like belongingness, attention, love, and power.",
                                            "code": 14
                                        },
                                        {
                                            "Category": "Goal Achievement",
                                            "Summary": "Groups enable individuals to achieve goals not attainable alone.",
                                            "code": 15
                                        },
                                        {
                                            "Category": "Knowledge & Information",
                                            "Summary": "Group membership provides knowledge and information, broadening perspectives.",
                                            "code": 16
                                        }
                                    ],
                                    "Summary": "Explores the various needs and motivations that drive individuals to join groups.",
                                    "code": 11
                                }
                            ],
                            "Summary": "Covers what constitutes a group, its fundamental characteristics, and distinctions from mere collections of people, as well as reasons for joining.",
                            "code": 1
                        },
                        {
                            "Net": "Group Formation and Development",
                            "Subnets": [
                                {
                                    "Subnet": "Factors Facilitating Formation",
                                    "Categories": [
                                        {
                                            "Category": "Proximity",
                                            "Summary": "Repeated interactions due to shared physical space or environment.",
                                            "code": 19
                                        },
                                        {
                                            "Category": "Similarity",
                                            "Summary": "Shared interests, attitudes, values, and backgrounds that lead to liking and group formation.",
                                            "code": 20
                                        },
                                        {
                                            "Category": "Common Motives & Goals",
                                            "Summary": "Shared objectives that prompt individuals to unite to achieve them.",
                                            "code": 21
                                        }
                                    ],
                                    "Summary": "Conditions that lead to initial contact and interaction for group formation.",
                                    "code": 18
                                },
                                {
                                    "Subnet": "Stages of Group Formation (Tuckman's Model)",
                                    "Categories": [
                                        {
                                            "Category": "Forming Stage",
                                            "Summary": "Initial uncertainty, getting to know each other, excitement, and apprehensions.",
                                            "code": 23
                                        },
                                        {
                                            "Category": "Storming Stage",
                                            "Summary": "Intragroup conflict over goals, control, and task allocation, leading to leadership hierarchy.",
                                            "code": 24
                                        },
                                        {
                                            "Category": "Norming Stage",
                                            "Summary": "Development of norms and a positive group identity.",
                                            "code": 25
                                        },
                                        {
                                            "Category": "Performing Stage",
                                            "Summary": "Group structure is accepted, and the group moves towards achieving its goals.",
                                            "code": 26
                                        },
                                        {
                                            "Category": "Adjourning Stage",
                                            "Summary": "Group disbandment after the completion of its purpose, common for temporary groups.",
                                            "code": 27
                                        }
                                    ],
                                    "Summary": "The sequential phases groups often progress through from initial meeting to dissolution, as per Tuckman's model.",
                                    "code": 22
                                }
                            ],
                            "Summary": "Describes the process by which groups are formed and the typical developmental stages they go through.",
                            "code": 17
                        },
                        {
                            "Net": "Group Structure and Dynamics",
                            "Subnets": [
                                {
                                    "Subnet": "Elements of Group Structure",
                                    "Categories": [
                                        {
                                            "Category": "Roles",
                                            "Summary": "Socially defined expectations and typical behaviors for individuals in a given situation, including role expectations.",
                                            "code": 30
                                        },
                                        {
                                            "Category": "Norms",
                                            "Summary": "Expected standards of behavior and beliefs established, agreed upon, and enforced by group members ('unspoken rules').",
                                            "code": 31
                                        },
                                        {
                                            "Category": "Status",
                                            "Summary": "The relative social position given to group members, which can be ascribed or achieved.",
                                            "code": 32
                                        },
                                        {
                                            "Category": "Cohesiveness",
                                            "Summary": "The togetherness, binding, or mutual attraction among group members, leading to a 'we feeling' or team spirit.",
                                            "code": 33
                                        }
                                    ],
                                    "Summary": "Defines the key components that give a group its internal organization and functional regularities.",
                                    "code": 29
                                },
                                {
                                    "Subnet": "Impact of Cohesiveness",
                                    "Categories": [
                                        {
                                            "Category": "Benefits of Cohesion",
                                            "Summary": "Increased desire to remain in the group and stronger team spirit.",
                                            "code": 35
                                        },
                                        {
                                            "Category": "Risks of Extreme Cohesion",
                                            "Summary": "Potential for negative outcomes like 'groupthink' when cohesion overrides critical thinking.",
                                            "code": 36
                                        }
                                    ],
                                    "Summary": "Examines the effects of group cohesiveness, both positive and negative.",
                                    "code": 34
                                }
                            ],
                            "Summary": "Focuses on the internal organization and functional aspects of groups, including roles, norms, status, and cohesiveness.",
                            "code": 28
                        },
                        {
                            "Net": "Types of Groups",
                            "Subnets": [
                                {
                                    "Subnet": "Primary vs. Secondary Groups",
                                    "Categories": [
                                        {
                                            "Category": "Primary Groups",
                                            "Summary": "Pre-existing, face-to-face, close physical proximity, warm emotional bonds (e.g., family, caste).",
                                            "code": 39
                                        },
                                        {
                                            "Category": "Secondary Groups",
                                            "Summary": "Joined by choice, more impersonal, indirect relationships, permeable boundaries (e.g., political party).",
                                            "code": 40
                                        }
                                    ],
                                    "Summary": "Distinguishes groups based on their formation, interaction patterns, and emotional bonds.",
                                    "code": 38
                                },
                                {
                                    "Subnet": "Formal vs. Informal Groups",
                                    "Categories": [
                                        {
                                            "Category": "Formal Groups",
                                            "Summary": "Explicitly stated functions, defined roles, specific rules/laws (e.g., university, office organization).",
                                            "code": 42
                                        },
                                        {
                                            "Category": "Informal Groups",
                                            "Summary": "Not based on formal rules, close relationships among members.",
                                            "code": 43
                                        }
                                    ],
                                    "Summary": "Categorizes groups based on the explicitness and formality of their functions and structure.",
                                    "code": 41
                                },
                                {
                                    "Subnet": "Ingroup vs. Outgroup",
                                    "Categories": [
                                        {
                                            "Category": "Ingroup",
                                            "Summary": "One's own group, viewed favorably, members perceived as similar and desirable ('we').",
                                            "code": 45
                                        },
                                        {
                                            "Category": "Outgroup",
                                            "Summary": "Another group, often perceived negatively and differently compared to the ingroup ('they').",
                                            "code": 46
                                        },
                                        {
                                            "Category": "Minimal Group Paradigm",
                                            "Summary": "Experiments demonstrating that even flimsy criteria can lead to ingroup favoritism.",
                                            "code": 47
                                        }
                                    ],
                                    "Summary": "Explores how individuals categorize groups they belong to versus those they don't, and its social implications.",
                                    "code": 44
                                }
                            ],
                            "Summary": "Categorizes groups based on various criteria such as formation, formality, and social identity.",
                            "code": 37
                        },
                        {
                            "Net": "Group Influence on Individual Behavior",
                            "Subnets": [
                                {
                                    "Subnet": "Social Loafing",
                                    "Categories": [
                                        {
                                            "Category": "Definition of Social Loafing",
                                            "Summary": "Reduction in individual effort when working on a collective task where outputs are pooled.",
                                            "code": 50
                                        },
                                        {
                                            "Category": "Reasons for Social Loafing",
                                            "Summary": "Less responsibility, decreased motivation due to lack of individual evaluation, poor coordination, or perceived unimportance of group.",
                                            "code": 51
                                        },
                                        {
                                            "Category": "Reducing Social Loafing",
                                            "Summary": "Making individual efforts identifiable, increasing task importance, strengthening cohesiveness, and making contributions seem vital.",
                                            "code": 52
                                        }
                                    ],
                                    "Summary": "Examines the phenomenon where individuals exert less effort when working in a group compared to working alone.",
                                    "code": 49
                                },
                                {
                                    "Subnet": "Group Polarisation",
                                    "Categories": [
                                        {
                                            "Category": "Definition of Group Polarisation",
                                            "Summary": "The strengthening of a group's initial position as a result of group interaction and discussion, leading to more extreme decisions.",
                                            "code": 54
                                        },
                                        {
                                            "Category": "Reasons for Polarisation",
                                            "Summary": "Exposure to new arguments, validation by like-minded people (bandwagon effect), and identification with the ingroup.",
                                            "code": 55
                                        },
                                        {
                                            "Category": "Consequences of Polarisation",
                                            "Summary": "Groups are more likely to take extreme decisions, ranging from very weak to very strong.",
                                            "code": 56
                                        }
                                    ],
                                    "Summary": "Focuses on the tendency of groups to make more extreme decisions than individuals alone, and its underlying causes.",
                                    "code": 53
                                },
                                {
                                    "Subnet": "Groupthink",
                                    "Categories": [
                                        {
                                            "Category": "Definition of Groupthink",
                                            "Summary": "A phenomenon in cohesive groups where the desire for unanimity overrides realistic appraisal of alternatives, leading to irrational decisions.",
                                            "code": 58
                                        },
                                        {
                                            "Category": "Characteristics of Groupthink",
                                            "Summary": "Appearance of consensus, suppression of dissenting opinions, exaggerated sense of power, ignoring real-world cues.",
                                            "code": 59
                                        },
                                        {
                                            "Category": "Preventing Groupthink",
                                            "Summary": "Encouraging critical thinking, presenting alternatives, inviting external experts, and seeking feedback from trusted others.",
                                            "code": 60
                                        }
                                    ],
                                    "Summary": "Explores the negative consequences of extreme group cohesiveness leading to flawed decision-making.",
                                    "code": 57
                                }
                            ],
                            "Summary": "Discusses how the presence and dynamics of groups influence individual performance and decision-making.",
                            "code": 48
                        }
                    ],
                    "Summary": "This document provides a comprehensive overview of social influence and group processes, defining groups, detailing their characteristics, formation stages, and structural elements such as roles, norms, status, and cohesiveness. It also explores various types of groups, including primary/secondary, formal/informal, and ingroup/outgroup distinctions, and discusses key phenomena related to group influence on individual behavior like social loafing, group polarization, and groupthink."
                }
            },
            "summaryList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 6857,
                        "CandidatesTokenCount": 2568,
                        "TotalTokenCount": 10660,
                        "thoughtsTokenCount": 1235,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Summary": "<h2>Comprehensive Synthesis Report: Social Influence and Group Processes</h2><p>This report synthesizes key concepts from the provided text on Social Influence and Group Processes, aiming for comprehensive detail to support auditing and implementation. It covers the nature, formation, types, and influence of groups on individual behavior, including specific phenomena like Groupthink, Social Loafing, and Group Polarization.</p><h3>1. Nature and Formation of Groups</h3><p>The chapter begins by highlighting the pervasive influence of groups in daily life (Data Source: Introduction). It distinguishes groups from mere collections of people based on several critical characteristics:</p><ul><li><b>Definition of a Group:</b> An organized system of two or more interacting and interdependent individuals with common motives, established role relationships, and regulating norms (Data Source: 'What is a Group?' section).</li><li><b>Salient Characteristics:</b><ul><li>Social unit perceiving belonging, creating unique identity.</li><li>Common motives and goals, working towards objectives or away from threats.</li><li>Interdependence, where one member's actions have consequences for others.</li><li>Mutual influence among individuals striving to satisfy needs through association.</li><li>Direct or indirect interaction.</li><li>Interactions structured by defined roles and norms, ensuring consistent behavior.</li></ul></li><li><b>Distinction from Other Collections:</b><ul><li><b>Crowd:</b> A collection of people by chance, lacking structure, belonging, and interdependence; behavior is often irrational.</li><li><b>Teams:</b> Special groups with complementary skills, commitment to a common goal, mutual accountability, and positive synergy. Unlike general groups where performance relies on individual contributions, teams value both individual efforts and teamwork. Leaders in groups hold primary responsibility, whereas in teams, members share responsibility.</li><li><b>Audience:</b> A collection of people assembled for a specific purpose, generally passive, but can become a 'mob' characterized by purpose, polarized attention, homogeneous thought/behavior, and impulsivity.</li></ul></li></ul><h3>2. Why People Join Groups</h3><p>Individuals join groups to satisfy a diverse range of needs, often being members of multiple groups simultaneously, which can create competing demands. The primary reasons for group affiliation include (Data Source: 'Why Do People Join Groups?' section):</p><ul><li><b>Security:</b> Groups reduce feelings of insecurity, providing comfort, protection, and a sense of strength.</li><li><b>Status:</b> Membership in important groups provides recognition and a sense of power.</li><li><b>Self-esteem:</b> Groups foster feelings of self-worth and a positive social identity, especially when associated with prestigious groups.</li><li><b>Satisfaction of Psychological and Social Needs:</b> Groups fulfill needs such as belongingness, attention, love, and power.</li><li><b>Goal Achievement:</b> Groups enable the attainment of goals that are difficult or impossible to achieve individually.</li><li><b>Knowledge and Information:</b> Group membership broadens perspectives and supplements individual knowledge.</li></ul><h3>3. Group Formation and Structure</h3><p>Group formation is rooted in contact and interaction, facilitated by specific conditions (Data Source: 'Group Formation' section):</p><ul><li><b>Proximity:</b> Repeated interactions due to physical closeness (e.g., living in the same colony, attending the same school) allow individuals to get to know each other, discovering common interests, attitudes, and backgrounds.</li><li><b>Similarity:</b> Shared interests, attitudes, and values lead to attraction. Psychologists suggest this is due to a preference for consistency and the validation of one's own opinions and values by similar others.</li><li><b>Common Motives and Goals:</b> Shared objectives bring people together to facilitate goal attainment that might be challenging for an individual.</li></ul><h4>Stages of Group Formation (Tuckman's Model)</h4><p>Groups typically undergo a developmental sequence, though not always systematically (Data Source: 'Stages of Group Formation' section):</p><ul><li><b>Forming Stage:</b> Initial uncertainty about the group, its goals, and member roles. Members assess fit; marked by excitement and apprehension.</li><li><b>Storming Stage:</b> Characterized by intragroup conflict over target achievement, control, and task allocation. Resolution leads to a leadership hierarchy and clear goal vision.</li><li><b>Norming Stage:</b> Development of group norms related to behavior, fostering a positive group identity.</li><li><b>Performing Stage:</b> The group's structure is accepted, and efforts are directed towards achieving the group goal. This may be the final stage for some groups.</li><li><b>Adjourning Stage:</b> For temporary groups (e.g., committees), this stage involves disbandment after the task is completed.</li></ul><p>It's important to note that groups may skip stages, move back and forth, or experience stages simultaneously (Data Source: 'Activity 7.1').</p><h4>Elements of Group Structure</h4><p>As members interact, group structure evolves, defining task distribution, responsibilities, and status (Data Source: 'Group structure' section):</p><ul><li><b>Roles:</b> Socially defined expectations for individuals in a given situation, encompassing 'role expectations' for typical behaviors.</li><li><b>Norms:</b> Expected standards of behavior and beliefs, agreed upon and enforced by members, often serving as 'unspoken rules.'</li><li><b>Status:</b> The relative social position accorded to group members, which can be 'ascribed' (e.g., seniority) or 'achieved' (e.g., expertise). Individuals often seek membership in high-status groups and differentiate status even within a group.</li><li><b>Cohesiveness:</b> Refers to the 'togetherness,' 'we feeling,' or mutual attraction among members. Highly cohesive groups exhibit stronger desires to remain united, but extreme cohesiveness can be detrimental, leading to 'groupthink.'</li></ul><h4>Groupthink (Box 7.1)</h4><p>Groupthink, proposed by Irving Janis, occurs when cohesion impedes effective decision-making, leading to irrational and uncritical decisions due to an overriding concern for unanimity. Characteristics include (Data Source: Box 7.1):</p><ul><li>Appearance of consensus, where members believe everyone agrees and suppress dissenting opinions to maintain group harmony.</li><li>Exaggerated sense of the group's power, leading to disregard for external warnings or dangers.</li><li>Likely in socially homogenous, cohesive, isolated groups lacking a tradition of considering alternatives, especially under high-stakes decisions.</li></ul><p><b>Example:</b> The U.S. escalation of the Vietnam War from 1964-1967 is cited as a prime illustration. To counteract groupthink, the text recommends:</p><ul><li>Encouraging critical thinking and disagreement.</li><li>Promoting the presentation of alternative courses of action.</li><li>Inviting outside experts to evaluate decisions.</li><li>Encouraging members to seek feedback from trusted others.</li></ul><h3>4. Type of Groups</h3><p>Groups vary significantly in size, duration, and organization. Major classifications include (Data Source: 'Type of Groups' section):</p><ul><li><b>Primary and Secondary Groups:</b><ul><li><b>Primary Groups:</b> Pre-existing formations (e.g., family, caste, religion) characterized by face-to-face interaction, close physical proximity, and warm emotional bonds. They are crucial for individual development and possess less permeable boundaries.</li><li><b>Secondary Groups:</b> Groups individuals join by choice (e.g., political party), featuring more impersonal, indirect, and less frequent relationships. They have more permeable boundaries, allowing easier entry and exit.</li></ul></li><li><b>Formal and Informal Groups:</b><ul><li><b>Formal Groups:</b> Functions are explicitly stated, roles are defined, and formation is based on specific rules or laws to establish order (e.g., a university).</li><li><b>Informal Groups:</b> Formation is not rule-based, characterized by close relationships among members.</li></ul></li><li><b>Ingroup and Outgroup:</b><ul><li><b>Ingroup:</b> One's own group, referred to as 'we,' whose members are perceived as similar, viewed favorably, and possessing desirable traits.</li><li><b>Outgroup:</b> Another group, referred to as 'they,' whose members are viewed differently and often negatively compared to the ingroup.</li></ul><p>Tajfel's 'Minimal Group Paradigm' experiments (Box 7.2) demonstrated that even arbitrary group divisions lead to ingroup favoritism, highlighting the artificiality of these categories (Data Source: Box 7.2).</p></li></ul><h3>5. Influence of Group on Individual Behaviour</h3><p>Groups significantly influence individual behavior, affecting performance in the presence of others or as part of a collective (Data Source: 'Influence of Group on Individual Behaviour' section).</p><h4>Social Loafing</h4><p>While social facilitation suggests arousal from others can enhance performance, social loafing describes the reduction in individual effort when working on a collective task where individual outputs are pooled (Data Source: 'Social Loafing' section).</p><ul><li><b>Phenomenon:</b> Individuals work less hard in a group than when alone, as demonstrated by Latane's clapping experiments where individual noise output decreased with increasing group size.</li><li><b>Reasons for Social Loafing:</b><ul><li>Reduced feeling of responsibility for the overall task.</li><li>Decreased motivation due to lack of individual evaluation.</li><li>Absence of comparison with other groups.</li><li>Improper or absent coordination among members.</li><li>Low importance of group belonging for members (seen as an aggregate, not a cohesive unit).</li></ul></li><li><b>Ways to Reduce Social Loafing:</b><ul><li>Making each person's efforts identifiable.</li><li>Increasing pressure and commitment to successful task performance.</li><li>Enhancing the perceived importance or value of the task.</li><li>Making individuals feel their contribution is significant.</li><li>Strengthening group cohesiveness to boost motivation for group success.</li></ul></li></ul><h4>Group Polarization</h4><p>Groups tend to make more extreme decisions than individuals alone, a phenomenon known as group polarization. This involves the strengthening of the group's initial position as a result of interaction and discussion (Data Source: 'Group Polarisation' section).</p><ul><li><b>Phenomenon:</b> Whatever the initial stance, group discussion pushes members towards a more extreme version of that stance. For example, a group initially slightly favoring capital punishment will, after discussion, favor it more strongly.</li><li><b>Reasons for Group Polarization:</b><ul><li><b>Exposure to Newer Arguments:</b> Individuals encounter additional arguments favoring their existing viewpoint from like-minded peers.</li><li><b>Validation by Public/Bandwagon Effect:</b> Finding others with similar views validates one's own position, creating a sense of public endorsement.</li><li><b>Ingroup Identification and Conformity:</b> Identifying with the ingroup and conforming to its perceived stance strengthens individual views.</li></ul></li></ul><h3>6. Conclusions</h3><p>This chapter comprehensively details how groups are fundamental to human existence, providing security, status, and facilitating goal achievement. Understanding the dynamics of group formation, structure, and the potential pitfalls like Groupthink, alongside individual behavioral changes such as Social Loafing and Group Polarization, is crucial. These processes underscore the complex interplay between individual and collective behavior, highlighting that groups can both foster growth and lead to extreme or suboptimal decisions (Data Source: Overall chapter narrative).</p><h3>7. Actionable Recommendations</h3><ul><li><b>For Team Leaders and Managers:</b> To mitigate Groupthink, implement strategies such as actively encouraging critical thinking, soliciting diverse perspectives, engaging external experts for review, and fostering an environment where dissent is valued (Data Source: Box 7.1).</li><li><b>For Educators and Project Organizers:</b> To combat Social Loafing, design tasks that ensure individual contributions are identifiable and valued, clearly articulate the task's importance, and foster strong group cohesiveness (Data Source: 'Social Loafing' section).</li><li><b>For Decision-Making Bodies:</b> Be aware of Group Polarization. Implement structured decision-making processes that encourage exploration of opposing viewpoints before discussions to prevent extreme shifts in opinion (Data Source: 'Group Polarisation' section).</li><li><b>For Individuals:</b> Actively seek membership in groups that promote positive influence and personal development, and cultivate an awareness of group dynamics to make informed choices within collective settings (Data Source: Introduction and general theme of positive influence).</li></ul>"
                            }
                        ]
                    }
                }
            ]
        }
    ],
    "fileName": "",
    "projectType": "DOCUMENT",
    "__v": 0,
    "headerCols": {
        "col1": "Part Id",
        "col2": "FileName",
        "col3": "Content"
    }
},
{
    "_id": "695e1335da2133fe04a5f693",
    "projectId": "269276",
    "projectName": "K12PSYH - 04 Psychological Disorders",
    "projectDescription": "",
    "projectCategory": "Educational",
    "projectCreatedDate": "2026-01-07T08:02:58.361Z",
    "projectUpdatedDate": "2026-01-07T08:02:58.361Z",
    "projectStatus": "Active",
    "companyId": "6957bd3ad58ea40715c6ae98",
    "projectCreatedBy": "6957bd3ad58ea40715c6ae99",
    "projectUpdatedBy": "6957bd3ad58ea40715c6ae99",
    "projectLastAccessDate": "2026-01-30T11:31:16.171Z",
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "Content",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "K12PSYH - 04 Psychological Disorders.pdf",
            "col3": "Chapter 4 • Psychological Disorders 69 PSY PSY PSY PSY PSYCHOLOGICAL DISORDERS CHOLOGICAL DISORDERS CHOLOGICAL DISORDERS CHOLOGICAL DISORDERS CHOLOGICAL DISORDERS Introduction Concepts of Abnormality and Psychological Disorders Classification of Psychological Disorders Factors Underlying Abnormal Behaviour Major Psychological Disorders Anxiety Disorders Obsessive-Compulsive and Related Disorders Trauma- and Stressor-Related Disorders Somatic Symptom and Related Disorders Dissociative Disorders Salient Features of Somatic Symptom and Related Disorders and Dissociative Disorders (Box 4.1) Depressive Disorders Bipolar and Related Disorders Schizophrenia Spectrum and Other Psychotic Disorders Neurodevelopmental Disorders Disruptive, Impulse-Control and Conduct Disorders Feeding and Eating Disorders Substance - Related and Addictive Disorders Effects of Alcohol : Some Facts (Box 4.2) Commonly Abused Substances (Box 4.3) CONTENTS Key Terms Summary Review Questions Project Ideas Weblinks Pedagogical Hints After reading this chapter, you would be able to: understand the basic issues in abnormal behaviour and the criteria used to identify such behaviours, appreciate the factors which cause abnormal behaviour, explain the different models of abnormal behaviour, and describe the major psychological disorders. Reprint 2025-26Psychology 70 C ONCEPTS OF A BNORMALITY AND P SYCHOLOGICAL D ISORDERS Although many definitions of abnormality have been used over the years, none has won universal acceptance. Still, most definitions have certain common features, often called the ‘four Ds’: deviance , distress , dysfunction and danger . That is, psychological disorders are deviant (different, extreme, unusual, even bizarre), distressing (unpleasant and upsetting to the person and to others), dysfunctional (interfering with the person’s ability to carry out daily activities in a constructive You must have come across people who are unhappy, troubled and dissatisfied. Their minds and hearts are filled with sorrow, unrest and tension and they feel that they are unable to move ahead in their lives; they feel life is a painful, uphill struggle, sometimes not worth living. Famous analytical psychologist Carl Jung has quite remarkably said, “How can I be substantial without casting a shadow? I must have a dark side, too, if I am to be whole and by becoming conscious of my shadow, I remember once more that I am a human being like any other”. At times, some of you may have felt nervous before an important examination, tense and concerned about your future career or anxious when someone close to you was unwell. All of us face major problems at some point of our lives. However, some people have an extreme reaction to the problems and stresses of life. In this chapter, we will try to understand what goes wrong when people develop psychological problems, what are the causes and factors which lead to abnormal behaviour, and what are the various signs and symptoms associated with different types of psychological disorders? The study of psychological disorders has intrigued and mystified all cultures for more than 2,500 years. Psychological disorders or mental disorders (as they are commonly referred to), like anything unusual may make us uncomfortable and even a little frightened. Unhappiness, discomfort, anxiety, and unrealised potential are seen all over the world. These failures in living are mainly due to failures in adaptation to life challenges. As you must have studied in the previous chapters, adaptation refers to the person’s ability to modify her/his behaviour in response to changing environmental requirements. When the behaviour cannot be modified according to the needs of the situation, it is said to be maladaptive. Abnormal Psychology is the area within psychology that is focused on maladaptive behaviour – its causes, consequences, and treatment. Introduction way), and possibly dangerous (to the person or to others). This definition is a useful starting point from which we can explore psychological abnormality. Since the word ‘abnormal’ literally means “away from the normal”, it implies deviation from some clearly defined norms or standards. In psychology, we have no ‘ideal model’ or even ‘normal model’ of human behaviour to use as a base for comparison. Various approaches have been used in distinguishing between normal and abnormal behaviours. From these approaches, there emerge two basic and conflicting views: Reprint 2025-26Chapter 4 • Psychological Disorders 71 The first approach views abnormal behaviour as a deviation from social norms . Many psychologists have stated that ‘abnormal’ is simply a label that is given to a behaviour which is deviant from social expectations. Abnormal behaviour, thoughts and emotions are those that differ markedly from a society’s ideas of proper functioning. Each society has norms , which are stated or unstated rules for proper conduct. Behaviours, thoughts and emotions that break societal norms are called abnormal. A society’s norms grow from its particular culture — its history, values, institutions, habits, skills, technology, and arts. Thus, a society whose culture values competition and assertiveness may accept aggressive behaviour, whereas one that emphasises cooperation and family values (such as in India) may consider aggressive behaviour as unacceptable or even abnor mal. A society’s values may change over time, causing its views of what is psychologically abnor mal to change as well. Serious questions have been raised about this definition. It is based on the assumption that socially accepted behaviour is not abnormal, and that normality is nothing more than conformity to social norms. The second approach views abnormal behaviour as maladaptive . Many psychologists believe that the best criterion for determining the normality of behaviour is not whether society accepts it but whether it fosters the well-being of the individual and eventually of the group to which s/he belongs. Well-being is not simply maintenance and survival but also includes growth and fulfilment, i.e. the actualisation of potential, which you must have studied in Maslow’s need hierarchy theory. Accor ding to this criterion, conforming behaviour can be seen as abnormal if it is maladaptive, i.e. if it interferes with optimal functioning and growth. For example, a student in the class prefers to remain silent even when s/he has questions in her/his mind. Describing behaviour as maladaptive implies that a problem exists; it also suggests that vulnerability in the individual, inability to cope, or exceptional stress in the environment have led to problems in life. If you talk to people around, you will see that they have vague ideas about psychological disorders that ar e characterised by superstition, ignorance and fear. Again it is commonly believed that psychological disorder is something to be ashamed of. The stigma attached to mental illness means that people are hesitant to consult a doctor or psychologist because they are ashamed of their problems. Actually, psychological disorder which indicates a failure in adaptation should be viewed as any other illness. Activity 4.1 Talk to three people: one of your friends, a friend of your parents, and your neighbour. Ask them if they have seen someone who is mentally ill or who has mental problems. Try to understand why they find this behaviour abnormal, what are the signs and symptoms shown by this person, what caused this behaviour and can this person be helped. Share the information you elicited in class and see if there are some common features, which make us label others as ‘abnormal’. Historical Background To understand psychological disorders, we would require a brief historical account of how these disorders have been viewed over the ages. When we study the history of abnormal psychology, we find that certain theories have occurred over and over again. One ancient theory that is still encountered today holds that abnormal behaviour can be explained by the Reprint 2025-26Psychology 72 operation of supernatural and magical forces such as evil spirits ( bhoot-pret ), or the devil ( shaitan ). Exorcism , i.e. removing the evil that resides in the individual through countermagic and prayer, is still commonly used. In many societies, the shaman , or medicine man ( ojha ) is a person who is believed to have contact with supernatural forces and is the medium through which spirits communicate with human beings. Through the shaman , an afflicted person can learn which spirits are responsible for her/his problems and what needs to be done to appease them. A recurring theme in the history of abnormal psychology is the belief that individuals behave strangely because their bodies and their brains are not working properly. This is the biological or organic approach . In the modern era, there is evidence that body and brain processes have been linked to many types of maladaptive behaviour. For certain types of disorders, correcting these defective biological processes results in improved functioning. Another approach is the psychological approach . According to this point of view, psychological problems are caused by inadequacies in the way an individual thinks, feels, or perceives the world. All three of these perspectives — supernatural, biological or organic, and psychological — have recurred throughout the history of Western civilisation. In the ancient Western world, it was philosopher- physicians of ancient Greece such as Hippocrates, Socrates, and in particular Plato who developed the organismic approach and viewed disturbed behaviour as arising out of conflicts between emotion and reason. Galen elaborated on the role of the four humours in personal character and temperament. According to him, the material world was made up of four elements, viz. earth, air, fire, and water which combined to form four essential body fluids, viz. blood, black bile, yellow bile, and phlegm. Each of these fluids was seen to be responsible for a different temperament. Imbalances among the humours were believed to cause various disorders. This is similar to the Indian notion of the three doshas of vata, pitta and kapha which were mentioned in the Atharva Veda and Ayurvedic texts. You have already read about it in Chapter 2. In the Middle Ages , demonology and superstition gained renewed importance in the explanation of abnormal behaviour. Demonology related to a belief that people with mental problems were evil and there are numerous instances of ‘witch-hunts’ during this period. During the early Middle Ages, the Christian spirit of charity prevailed and St. Augustine wrote extensively about feelings, mental anguish and conflict. This laid the groundwork for modern psychodynamic theories of abnormal behaviour. The Renaissance Period was marked by increased humanism and curiosity about behaviour. Johann Weyer emphasised psychological conflict and disturbed interpersonal relationships as causes of psychological disorders. He also insisted that ‘witches’ were mentally disturbed and required medical, not theological, treatment. The seventeenth and eighteenth centuries were known as the Age of Reason and Enlightenment , as the scientific method replaced faith and dogma as ways of understanding abnormal behaviour. The growth of a scientific attitude towards psychological disorders in the eighteenth century contributed to the Reform Movement and to increased compassion for people who suffered from these disorders. Reforms of asylums were initiated in both Europe and America. One aspect of the reform movement was the new inclination for deinstitutionalisation which placed Reprint 2025-26Chapter 4 • Psychological Disorders 73 emphasis on providing community care for recovered mentally ill individuals. In recent years, there has been a convergence of these approaches, which has resulted in an interactional , or bio- psycho-social approach . Fr om this perspective, all three factors, i.e. biological, psychological and social play important roles in influencing the expression and outcome of psychological disorders. C LASSIFICATION OF P SYCHOLOGICAL D ISORDERS In order to understand psychological disorders, we need to begin by classifying them. A classification of such disorders consists of a list of categories of specific psychological disor ders grouped into various classes on the basis of some shared characteristics. Classifications are useful because they enable users like psychologists, psychiatrists and social workers to communicate with each other about the disor der and help in understanding the causes of psychological disorders and the processes involved in their development and maintenance. The American Psychiatric Association (APA) has published an official manual describing and classifying various kinds of psychological disorders. The current version of it, the Diagnostic and Statistical Manual of Mental Disorders , 5 th Edition (DSM-5) , presents discrete clinical criteria which indicate the presence or absence of disorders. The classification scheme officially used in India and elsewhere is the tenth revision of the International Classification of Diseases (ICD-10) , which is known as the ICD-10 Classification of Behavioural and Mental Disorders . It was prepared by the World Health Organisation (WHO). For each disorder, a description of the main clinical features or symptoms, and of other associated features including diagnostic guidelines is provided in this scheme. Activity 4.2 Certain behaviours like eating sand would be considered abnormal. But not if it was done after being stranded on a beach in a plane crash. Listed below are ‘abnor mal’ behaviours followed by situations where the behaviours might be considered normal. (i) talking to yourself - you are praying. (ii) standing in the middle of the street waving your arms wildly - you are a traffic policeman. Think about it and list similar examples. F ACTORS U NDERLYING A BNORMAL B EHAVIOUR In order to understand something as complex as abnor mal behaviour, psychologists use different approaches. Each approach in use today emphasises a different aspect of human behaviour, and explains and treats abnormality in line with that aspect. These approaches also emphasise the role of different factors such as biological, psychological and interpersonal, and socio-cultural factors. We will examine some of the approaches which are currently being used to explain abnormal behaviour. Biological factors influence all aspects of our behaviour. A wide range of biological factors such as faulty genes, endocrine imbalances, malnutrition, injuries and other conditions may interfere with normal development and functioning of the human body. These factors may be potential causes of abnormal behaviour. We have already come across the biological model. Accor ding to this model, abnor mal Reprint 2025-26Psychology 74 behaviour has a biochemical or physiological basis. Biological researchers have found that psychological disorders are often related to problems in the transmission of messages from one neuron to another. You have studied in Class XI, that a tiny space called synapse separates one neuron from the next, and the message must move across that space. When an electrical impulse reaches a neuron’s ending, the nerve ending is stimulated to release a chemical, called a neuro-transmitter . Studies indicate that abnormal activity by certain neuro-transmitters can lead to specific psychological disorders. Anxiety disorders have been linked to low activity of the neurotransmitter gamma aminobutyric acid (GABA), schizophrenia to excess activity of dopamine, and depression to low activity of serotonin . Genetic factors have been linked to bipolar and related disorders, schizophrenia, intellectual disability and other psychological disorders. Researchers have not, however, been able to identify the specific genes that are the culprits. It appears that in most cases, no single gene is responsible for a particular behaviour or a psychological disorder. Infact, many genes combine to help bring about our various behaviours and emotional reactions, both functional and dysfunctional. Although there is sound evidence to believe that genetic/ biochemical factors are involved in mental disorders as diverse as schizophrenia, depression, anxiety, etc. but biology alone cannot account for most mental disorders. There are several psychological models which provide a psychological explanation of mental disorders. These models maintain that psychological and interpersonal factors have a significant role to play in abnormal behaviour. These factors include maternal deprivation (separation from the mother, or lack of warmth and stimulation during early years of life), faulty parent-child relationships (rejection, overprotection, over-permissiveness, faulty discipline, etc.), maladaptive family structures (inadequate or disturbed family), and severe stress. The psychological models include the psychodynamic, behavioural, cognitive, and humanistic-existential models. The psychodynamic model is the oldest and most famous of the modern psychological models. You have already read about this model in Chapter 2 on Self and Personality. Psychodynamic theorists believe that behaviour, whether normal or abnormal, is determined by psychological forces within the person of which s/he is not consciously aware. These internal forces are considered dynamic, i.e. they interact with one another and their interaction gives shape to behaviour, thoughts and emotions. Abnormal symptoms are viewed as the result of conflicts between these forces. This model was first formulated by Freud who believed that three central forces shape personality — instinctual needs, drives and impulses (id) , rational thinking (ego) , and moral standards (superego) . Freud stated that abnormal behaviour is a symbolic expression of unconscious mental conflicts that can be generally traced to early childhood or infancy. Another model that emphasises the role of psychological factors is the behavioural model . This model states that both normal and abnormal behaviours are learned and psychological disorders are the result of learning maladaptive ways of behaving. The model concentrates on behaviours that are learned through conditioning and proposes that what has been learned can be unlearned. Learning can take place by classical conditioning (temporal association in which two events repeatedly occur close together in time), operant conditioning (behaviour is followed by a reward), and Reprint 2025-26Chapter 4 • Psychological Disorders 75 social learning (learning by imitating others’ behaviour). These three types of conditioning account for behaviour, whether adaptive or maladaptive. Psychological factors are also emphasised by the cognitive model . This model states that abnormal functioning can result from cognitive problems. People may hold assumptions and attitudes about themselves that are irrational and inaccurate. People may also repeatedly think in illogical ways and make overgeneralisations, that is, they may draw broad, negative conclusions on the basis of a single insignificant event. Another psychological model is the humanistic-existential model which focuses on broader aspects of human existence. Humanists believe that human beings are born with a natural tendency to be friendly, cooperative and constructive, and are driven to self-actualise, i.e. to fulfil this potential for goodness and growth. Existentialists believe that from birth we have total freedom to give meaning to our existence or to avoid that responsibility. Those who shirk from this responsibility would live empty, inauthentic, and dysfunctional lives. In addition to the biological and psychosocial factors, socio-cultural factors such as war and violence, group prejudice and discrimination, economic and employment problems, and rapid social change, put stress on most of us and can also lead to psychological problems in some individuals. According to the socio-cultural model , abnormal behaviour is best understood in light of the social and cultural forces that influence an individual. As behaviour is shaped by societal forces, factors such as family structure and communication, social networks, societal conditions, and societal labels and roles become more important. It has been found that certain family systems are likely to produce abnormal functioning in individual members. Some families have an enmeshed structure in which the members are overinvolved in each other’s activities, thoughts, and feelings. Children from this kind of family may have difficulty in becoming independent in life. The broader social networks in which people operate include their social and professional relationships. Studies have shown that people who are isolated and lack social support, i.e. strong and fulfilling interpersonal relationships in their lives are likely to become more depressed and remain depressed longer than those who have good friendships. Socio-cultural theorists also believe that abnormal functioning is influenced by the societal labels and roles assigned to troubled people. When people break the norms of their society, they are called deviant and ‘mentally ill’. Such labels tend to stick so that the person may be viewed as ‘crazy’ and encouraged to act sick. The person gradually learns to accept and play the sick role, and functions in a disturbed manner. In addition to these models, one of the most widely accepted explanations of abnormal behaviour has been provided by the diathesis-stress model . This model states that psychological disorders develop when a diathesis (biological predisposition to the disorder) is set off by a stressful situation. This model has three components. The first is the diathesis or the presence of some biological aberration which may be inherited. The second component is that the diathesis may carry a vulnerability to develop a psychological disorder. This means that the person is ‘at risk’ or ‘predisposed’ to develop the disorder. The third component is the presence of pathogenic stressors, i.e. factors/ stressors that may lead to psychopathology. If such “at risk” persons Reprint 2025-26Psychology 76 are exposed to these stressors, their predisposition may actually evolve into a disorder. This model has been applied to several disorders including anxiety, depression, and schizophrenia. M AJOR P SYCHOLOGICAL D ISORDERS Anxiety Disorders One day while driving home, Deb felt his heart beating rapidly, he started sweating profusely, and even felt short of breath. He was so scared that he stopped the car and stepped out. In the next few months, these attacks increased and now he was hesitant to drive for fear of being caught in traffic during an attack. Deb started feeling that he had gone crazy and would die. Soon he remained indoors and refused to move out of the house. We experience anxiety when we are waiting to take an examination, or to visit a dentist, or even to give a solo performance. This is normal and expected and even motivates us to do our task well. On the other hand, high levels of anxiety that are distressing and interfere with effective functioning indicate the presence of an anxiety disorder — the most common category of psychological disorders. Everyone has worries and fears. The term anxiety is usually defined as a diffuse, vague, very unpleasant feeling of fear and apprehension. The anxious individual also shows combinations of the following symptoms: rapid heart rate, shortness of breath, diarrhoea, loss of appetite, fainting, dizziness, sweating, sleeplessness, frequent urination and tremors. There are many types of anxiety disorders (see Table 4.1). They include generalised anxiety disorder , which consists of prolonged, vague, unexplained and intense fears that are not attached to any particular object. The symptoms include worry and apprehensive feelings about the future; hypervigilance, which involves constantly scanning the environment for dangers. It is marked by motor tension, as a result of which the person is unable to relax, is restless, and visibly shaky and tense. Another type of anxiety disorder is panic disorder , which consists of recurrent anxiety attacks in which the person experiences intense terror. A panic attack denotes an abrupt surge of intense anxiety rising to a peak when thoughts of a particular stimuli are present. Such thoughts occur in an unpredictable manner. The clinical features include shortness of breath, dizziness, trembling, palpitations, choking, nausea, chest pain or discomfort, fear of going crazy, losing control or dying. You might have met or heard of someone who was afraid to travel in a lift or climb to the tenth floor of a building, or refused to enter a room if s/he saw a lizard. You may have also felt it yourself or seen a friend unable to speak a word of a well-memorised and rehearsed speech before an audience. These kinds of fears are termed as phobias . People who have phobias have irrational fears related to specific objects, people, or situations. Phobias often develop gradually or begin with a generalised anxiety disorder. Phobias can be grouped into three main types, i.e. specific phobias, social phobias, and agoraphobia. Specific phobias are the most commonly occurring type of phobia. This group includes irrational fears such as intense fear of a certain type of animal, or of being in an enclosed space. Intense and incapacitating fear and embarrassment when dealing with others characterises social anxiety disorder (social phobia) . Agoraphobia is the term used when people develop a fear of entering Reprint 2025-26Chapter 4 • Psychological Disorders 77 unfamiliar situations. Many people with agoraphobia are afraid of leaving their home. So their ability to carry out normal life activities is severely limited. Separation anxiety disorder (SAD) is another type of anxiety disorder. Individuals with separation anxiety disorder are fearful and anxious about separation from attachment figures to an extent that is developmentally not appropriate. Children with SAD may have difficulty being in a room by themselves, going to school alone, are fearful of entering new situations, and cling to and shadow their parents’ every move. To avoid separation, children with SAD may fuss, scream, throw severe tantrums, or make suicidal gestures. Obsessive-Compulsive and Related Disorders Have you ever noticed someone washing their hands everytime they touch something, or washing even things like coins, or stepping only within the patterns on the floor or road while walking? People affected by obsessive- compulsive disorder are unable to control their preoccupation with specific ideas or are unable to prevent themselves from repeatedly carrying out a particular act or series of acts that affect their ability to carry out normal activities. Obsessive behaviour is the inability to stop thinking about a particular idea or topic. The person involved, often finds these thoughts to be unpleasant and shameful. Compulsive behaviour is the need to perform certain behaviours over and over again. Many compulsions deal with counting, ordering, checking, touching and washing. Other disorders in this category include hoarding Recall how you felt before your Class X Board examination. How did you feel when the examinations were drawing near (one month before the examinations; one week before the examinations; on the day of the examination, and when you were entering the examination hall)? Also try to recollect what you felt when you were awaiting your results. Write down your experiences in terms of bodily symptoms (e.g. ‘butterflies in the stomach’, clammy hands, excessive perspiration, etc.) as well as mental experiences (e.g. tension, worry, pressure, etc.). Compare your symptoms with those of your classmates and classify them as Mild, Moderate, or Severe. Activity 4.3 1. Generalised Anxiety Disorder : prolonged, vague, unexplained and intense fears that have no object, accompanied by hypervigilance and motor tension. 2. Panic Disorder : frequent anxiety attacks characterised by feelings of intense terror and dread; unpredictable ‘panic attacks’ along with physiological symptoms like breathlessness, palpitations, trembling, dizziness, and a sense of loosing control or even dying. 3. Specific Phobia : irrational fears related to specific objects, interactions with others, and unfamiliar situations. 4. Separation Anxiety Disorder : extreme distress when expecting or going through separation from home or other significant people to whom the individual is immensely attached to. 5. Other disorders included under this category are Selective Mutism, Substance/Medication- Induced Anxiety Disorder, Anxiety Disorder Due to Another Medical condition, etc. Table 4.1 : Major Anxiety Disorders and their Symptoms Reprint 2025-26Psychology 78 disorder, trichotillomania (hair -pulling disorder), excoriation (skin-picking) disorder etc. Trauma- and Stressor-Related Disorders Very often people who have been caught in a natural disaster (such as tsunami) or have been victims of bomb blasts by terrorists, or been in a serious accident or in a war-related situation, experience post-traumatic stress disorder (PTSD). PTSD symptoms vary widely but may include recurrent dreams, flashbacks, impaired concentration, and emotional numbing. Adjustment Disorders and Acute Stress Disorder are also included under this category. Somatic Symptom and Related Disorders These are conditions in which there are physical symptoms in the absence of a physical disease. In these disorders, the individual has psychological difficulties and complains of physical symptoms, for which there is no biological cause. These include conversion disorders, somatic symptom disorder, and illness anxiety disorder among others . Somatic symptom disorder involves a person having persistent body-related symptoms which may or may not be related to any serious medical condition. People with this disorder tend to be overly preoccupied with their symptoms and they continually worry about their health and make frequent visits to doctors. As a result, they experience significant distress and disturbances in their daily life. Illness anxiety disorder involves persistent preoccupation about developing a serious illness and constantly worrying about this possibility. This is accompanied by anxiety about one’s health. Individuals with illness anxiety disorder are overly concerned about undiagnosed disease, negative diagnostic results, do not respond to assurance by doctors, and are easily alarmed about illness such as on hearing about someone else's ill-health or some such news. In general, both somatic symptom disorder and illness anxiety disorder are concerned with medical illnesses. But, the difference lies in the way this concern is expressed. In the case of somatic symptom disorder, this expression is in terms of physical complaints while in case of illness anxiety disorder, as the name suggests, it is the anxiety which is the main concern. The symptoms of conversion disorders are the reported loss of part or all of some basic body functions. Paralysis, blindness, deafness and difficulty in walking are generally among the symptoms reported. These symptoms often occur after a stressful experience and may be quite sudden. Dissociative Disorders Dissociation can be viewed as severance of the connections between ideas and emotions. Dissociation involves feelings of unreality, estrangement, depersonalisation, and sometimes a loss or shift of identity. Sudden temporary alterations of consciousness that blot out painful experiences are a defining characteristic of dissociative disorders . Conditions included in this are Dissociative Amnesia, Dissociative Identity Disorder , and Depersonalisation/Derealisation Disorder . Salient features of somatic symptom and related disorders and dissociative disorders are given in Box 4.1. Dissociative amnesia is characterised by extensive but selective memory loss that has no known organic cause (e.g., head injury). Some people cannot remember Reprint 2025-26Chapter 4 • Psychological Disorders 79 anything about their past. Others can no longer recall specific events, people, places, or objects, while their memory for other events remains intact. A part of dissociative amnesia is dissociative fugue. Essential feature of this could be an unexpected travel away from home and workplace, the assumption of a new identity, and the inability to recall the previous identity. The fugue usually ends when the person suddenly ‘wakes up’ with no memory of the events that occurred during the fugue. This disorder is often associated with an overwhelming stress. Dissociative identity disorder , often referred to as multiple personality , is the most dramatic of the dissociative disorders. It is often associated with traumatic experiences in childhood. In this disorder, the person assumes alternate personalities that may or may not be aware of each other. Depersonalisation/Derealisation disorder involves a dreamlike state in which the person has a sense of being separated both from self and from reality. In depersonalisation, there is a change of self-perception, and the person’s sense of reality is temporarily lost or changed. Depressive Disorders One of the most widely prevalent and recognised of all mental disorders is depression . Depression covers a variety of negative moods and behavioural changes. Depression can refer to a symptom or a disorder . In day-to-day life, we often use the term depression to refer to normal feelings after a significant loss, such as the break-up of a relationship, or the failure to attain a significant goal. Major depressive disorder is defined as a period of depressed mood and/or loss of interest or pleasure in most activities, together with other symptoms which may include change in body weight, constant sleep problems, tiredness, inability to think clearly, agitation, greatly slowed behaviour, and thoughts of death and suicide. Other symptoms include excessive guilt or feelings of worthlessness. Factors Predisposing towards Depression : Genetic make-up, or heredity is an important risk factor for major depression and other depressive disorders. Age is also a risk factor. For instance, women are particularly at risk during young adulthood, while for men Box 4.1 Salient Features of Somatic Symptom and Related Disorders and Dissociative Disorders Dissociative Disorders Dissociative amnesia : The person is unable to recall important, personal information often related to a stressful and traumatic report. The extent of forgetting is beyond normal. Depersonalisation/Derealisation Disorder : The person experiences a change in the person's sense of reality and perception of self. Dissociative identity (multiple personality) Disorder : The person exhibits two or more separate and contrasting personalities, generally associated with a history of abuse. Somatic Symptom and Related Disorders Somatic Symptom Disorder : The person experiences body-related symptoms in the absence of any medical condition (or even if medical condition is present, it is not as serious as the symptoms presented). Illness Anxiety Disorder : The person experiences worry about the possibility of developing a serious medical condition. Conversion : The person suffers from a loss or impairment of motor or sensory function (e.g., paralysis, blindness, etc.) that has no physical cause but may be a response to stress and psychological problems. Reprint 2025-26Psychology 80 the risk is highest in early middle age. Similarly gender also plays a great role in this differential risk addition. For example, women in comparison to men are more likely to report a depressive disorder. Other risk factors ar e experiencing negative life events and lack of social support. Bipolar and Related Disorders Bipolar I disorder involves both mania and depression, which are alternately present and sometimes interrupted by periods of normal mood. Manic episodes rarely appear by themselves; they usually alternate with depression. Bipolar mood disorders were earlier referred to as manic-depressive disorders. Some examples of types of bipolar and related disorders include Bipolar I Disorder, Bipolar II disorder and Cyclothymic Disorder. Every suicide is a misfortune. Suicide takes place throughout the lifespan. Suicide is a result of complex interface of biological, genetic, psychological, sociological, cultural and environmental factors. Some other risk factors are having mental disorders (especially depression and alcohol use disorders), going through natural disasters, experiencing violence, abuse or loss and isolation at any stage of life. Previous suicidal attempt is the strongest risk factor. Often, suicidal behavior indicates difficulties in problem-solving, stress management, and emotional expression. Suicidal thoughts lead to suicidal action only when acting on these thoughts seems to be the only way out of a person ’ s difficulties. These thoughts are heightened under acute emotional and other distress. The ramifications of suicide on social circle and communities tend to be devastating and long-lasting. The stigma surrounding suicide continues despite recent advances in research in this field. Due to this, many people who are contemplating or even attempting suicide do not seek help thus, preventing timely help from reaching them. Therefore improving identification, referral, and management of behaviour are crucial for preventing suicide. Therefore we need to identify vulnerability; comprehend the circumstances leading to such behaviour and accordingly plan interventions. Suicides are preventable. There is a need for comprehensive multi-sectoral approach where the government, media and civil society all play important role as stakeholders. Some measures suggested by WHO include: • limiting access to the means of suicide; • reporting of suicide by media in a responsible way; • bringing in alcohol-related policies; • early identification, treatment and care of people at risk; • training health workers in assessing and managing for suicide; • care for people who attempted suicide and providing community support. Identifying students in distress : Any unexpected or striking change affecting the adolescent ’ s performance, attendance You may have got some bad news in the family (for example, death of a close relative) or watched your favourite character dying in a film or got less marks than you hoped for or lost your pet. This may have made you sad and depressed and hopeless about the future. Try and recall such incidents in your life. List the situations that led to this reaction. Compare your list and reactions with those of others in class. Activity 4.4 Reprint 2025-26Chapter 4 • Psychological Disorders 81 or behaviour should be taken seriously, such as: • lack of interest in common activities • declining grades • decreasing effort • misbehavior in the classroom • mysterious or repeated absence • smoking or drinking, or drug misuse Strengthening students’ self-esteem : Having a positive self-esteem is important in face of distress and helps in coping adequately. In order to foster positive self- esteem in children the following approaches can be useful: • accentuating positive life experiences to develop positive identity. This increases confidence in self. • providing opportunities for development of physical, social and vocational skills. • establishing a trustful communication. • goals for the students should be specific, measurable, achievable, relevant, to be completed within a relevant time frame. Schizophrenia Spectrum and Other Psychotic Disorders Schizophrenia is the descriptive term for a group of psychotic disorders in which personal, social and occupational functioning deteriorate as a result of disturbed thought processes, strange perceptions, unusual emotional states, and motor abnormalities. It is a debilitating disorder. The social and psychological costs of schizophrenia are tremendous, both to patients as well as to their families and society. Symptoms of Schizophrenia The symptoms of schizophrenia can be grouped into three categories, viz. positive symptoms (i.e. excesses of thought, emotion, and behaviour), negative symptoms (i.e. deficits of thought, emotion, and behaviour), and psychomotor symptoms . Positive symptoms are ‘pathological excesses’ or ‘bizarre additions’ to a person’s behaviour. Delusions, disorganised thinking and speech, heightened perception and hallucinations, and inappropriate affect are the ones most often found in schizophrenia. Many people with schizophrenia develop delusions . A delusion is a false belief that is firmly held on inadequate grounds. It is not affected by rational argument, and has no basis in reality. Delusions of persecution are the most common in schizophrenia. People with this delusion believe that they are being plotted against, spied on, slandered, Activity 4.5 Can you list some characters in films you have seen or books you have read who suffered from any of the disorders we have studied here like depression or schizophrenia showing some of these delusions? Can you identify which kind of delusion each of these is? 1. A person who believes that s/he is going to be the next President of India. 2. One who believes that the intelligence agencies/police are conspiring to trap her/him in a spy scandal. 3. One who believes that s/he is the incarnation of God and can make things happen. 4. One who believes that the tsunami occurred to prevent her/him from enjoying her/his holidays. 5. One who believes that her/his actions are controlled by the satellite through a chip implanted in her/his brain by some extraterrestrial beings. Reprint 2025-26Psychology 82 threatened, attacked or deliberately victimised. People with schizophrenia may also experience delusions of reference in which they attach special and personal meaning to the actions of others or to objects and events. In delusions of grandeur , people believe themselves to be specially empowered persons and in delusions of control , they believe that their feelings, thoughts and actions are controlled by others. People with schizophrenia may not be able to think logically and may speak in peculiar ways. These formal thought disorders can make communication extremely difficult. These include rapidly shifting from one topic to another so that the normal structure of thinking is muddled and becomes illogical ( loosening of associations, derailment ), inventing new words or phrases ( neologisms ), and persistent and inappropriate repetition of the same thoughts ( perseveration ). People with schizophrenia may have hallucinations , i.e. perceptions that occur in the absence of external stimuli. Auditory hallucinations are most common in schizophrenia. Patients hear sounds or voices that speak words, phrases and sentences directly to the patient ( second-person hallucination ) or talk to one another referring to the patient as s/he ( third-person hallucination ). Hallucinations can also involve the other senses. These include tactile hallucinations (i.e. forms of tingling, burning), somatic hallucinations (i.e. something happening inside the body such as a snake crawling inside one’s stomach), visual hallucinations (i.e. vague perceptions of colour or distinct visions of people or objects), gustatory hallucinations (i.e. food or drink taste strange), and olfactory hallucinations (i.e. smell of poison or smoke). People with schizophrenia also show inappropriate affect , i.e. emotions that are unsuited to the situation. Negative symptoms are ‘pathological deficits’ and include poverty of speech, blunted and flat affect, loss of volition, and social withdrawal. People with schizophrenia show alogia or poverty of speech, i.e. a reduction in speech and speech content. Many people with schizophrenia show less anger, sadness, joy, and other feelings than most people do. Thus they have blunted affect . Some show no emotions at all, a condition known as flat affect . Also patients with schizophrenia experience avolition , or apathy and an inability to start or complete a course of action. People with this disorder may withdraw socially and become totally focused on their own ideas and fantasies. People with schizophrenia also show psychomotor symptoms . They move less spontaneously or make odd grimaces and gestures. These symptoms may take extreme forms known as catatonia . People in a catatonic stupor remain motionless and silent for long stretches of time. Some show catatonic rigidity , i.e. maintaining a rigid, upright posture for hours. Others exhibit catatonic posturing , i.e. assuming awkward, bizarre positions for long periods of time. Neurodevelopmental Disorders A common feature of the neurodevelopmental disorders is that they manifest in the early stage of development. Often the symptoms appear before the child enters school or during the early stage of schooling. These disorders result in hampering personal, social, academic and occupational functioning. These get characterised as deficits or excesses in a particular Reprint 2025-26Chapter 4 • Psychological Disorders 83 behaviour or delays in achieving a particular age-appropriate behaviour. We will now discuss several disorders like Attention-Deficit/Hyperactivity Disorder (ADHD) , Autism Spectrum Disorder, Intellectual Disability, and Specific Lear ning Disorder . These disorders, if not attended, can lead to more serious and chronic disorders as the child moves into adulthood. The two main features of ADHD are inattention and hyperactivity- impulsivity . Children who are inattentive find it difficult to sustain mental effort during work or play. They have a hard time keeping their minds on any one thing or in following instructions. Common complaints are that the child does not listen, cannot concentrate, does not follow instructions, is disorganised, easily distracted, forgetful, does not finish assignments, and is quick to lose interest in boring activities. Children who are impulsive seem unable to control their immediate reactions or to think before they act. They find it difficult to wait or take turns, have difficulty resisting immediate temptations or delaying gratification. Minor mishaps such as knocking things over are common whereas more serious accidents and injuries can also occur. Hyperactivity also takes many forms. Children with ADHD are in constant motion. Sitting still through a lesson is impossible for them. The child may fidget, squirm, climb and run around the room aimlessly. Parents and teachers describe them as ‘driven by a motor’, always on the go, and talk incessantly. Autism Spectrum Disorder is characterised by widespread impairments in social interaction and communication skills, and stereotyped patterns of behaviours, interests and activities. Children with autism spectrum disorder have marked difficulties in social interaction and communication across different contexts, a restricted range of interests, and strong desire for routine. About 70 per cent of children with autism spectrum disorder have intellectual disabilities. Children with autism spectrum disorder experience profound difficulties in relating to other people. They are unable to initiate social behaviour and seem unresponsive to other people’s feelings. They are unable to share experiences or emotions with others. They also show serious abnormalities in communication and language that persist over time. Many of them never develop speech and those who do, have repetitive and deviant speech patterns. Such children often show narrow patterns of interests and repetitive behaviours such as lining up objects or stereotyped body",
            "col4": "DATA"
        },
        {
            "col1": "P1_Part_2",
            "col2": "K12PSYH - 04 Psychological Disorders.pdf",
            "col3": "movements such as rocking. These motor movements may be self-stimulatory such as hand flapping or self-injurious such as banging their head against the wall. Due to the nature of these difficulties in terms of verbal and non-verbal communication, individuals with autism spectrum disorder tend to experience difficulties in starting, maintaining and even understanding relationships. You have already read about variations in intelligence in Chapter 1. Intellectual disability refers to below average intellectual functioning (with an IQ of approximately 70 or below), and deficits or impairments in adaptive behaviour (i.e. in the areas of communication, self-care, home living, social/interpersonal skills, functional academic skills, work, etc.) which are manifested before the age of 18 years. Table 4.2 describes characteristics of the intellectually disabled persons. Reprint 2025-26Psychology 84 In case of specific learning disorder , the individual experiences difficulty in perceiving or processing information efficiently and accurately. These get manifested during early school years and the individual encounters problems in basic skills in reading, writing and/or mathematics. The affected child tends to perform below average for her/his age. However, individuals may be able to reach acceptable performance levels with additional inputs and efforts. Specific learning disorder is likely to impair functioning and performance in activities/ occupations dependent on the related skills. Disruptive, Impulse-Control and Conduct Disorders The disorders included under this category are Oppositional Defiant Disorder, Conduct Disorder and others. Children with Oppositional Defiant Disorder (ODD) display age-inappropriate amounts of stubbornness, are irritable, defiant, disobedient, and behave in a hostile manner. Individuals with ODD do not see themselves as angry, oppositional, or defiant and often justify their behaviour as reaction to circumstances/demands. Thus, the symptoms of the disorder become entangled with the problematic interactions with others. The terms conduct disorder and antisocial behaviour refer to age-inappropriate actions and attitudes that violate family expectations, societal norms, and the personal or property rights of others. The behaviours typical of conduct disorder include aggressive actions that cause or threaten harm to people or animals, non- aggressive conduct that causes property damage, major deceitfulness or theft, and serious rule violations. Children show many different types of aggressive behaviour, such as verbal aggression (i.e. name-calling, swearing), physical aggression (i.e. hitting, fighting), hostile aggression (i.e. directed at inflicting injury to others), and proactive aggression (i.e. dominating and bullying others without provocation). Feeding and Eating Disorders Another group of disorders which are of special interest to young people are eating disorders . These include anorexia nervosa , bulimia nervosa, and binge eating . In anorexia nervosa , the individual has a distorted body image that leads her/ him to see herself/himself as overweight. Often refusing to eat, exercising compulsively and developing unusual habits such as refusing to eat in front of others, the person with anorexia may lose large amounts of weight and even starve herself/himself to death. In bulimia nervosa , the individual may eat excessive amounts of food, then purge her/his body of food by using medicines such as laxatives or diuretics or by vomiting. The person often feels disgusted and ashamed when s/he binges and is relieved of tension and negative emotions after purging. In binge eating , there are frequent episodes of out-of-control eating. The individual tends to eat at a higher speed than normal and continues eating till s/he feels uncomfortably full. In fact, large amount of food may be eaten even when the individual is not feeling hungry. Substance-Related and Addictive Disorders Addictive behaviour, whether it involves excessive intake of high calorie food resulting in extreme obesity or involving the abuse of substances such as alcohol or cocaine, is one of the most severe problems being faced by society today. Disorders relating to maladaptive behaviours resulting from regular and Reprint 2025-26Chapter 4 • Psychological Disorders 85 consistent use of the substance involved are included under substance related and addictive disorders . These disorders include problems associated with the use and abuse of alcohol, cocaine, tobacco and opiods among others, which alter the way people think, feel and behave. While there are many disorders listed under this category, few frequently used substances are discussed below: Alcohol People who abuse alcohol drink large amounts regularly and rely on it to help them face difficult situations. Eventually the drinking interferes with their social behaviour and ability to think and work. Their bodies then build up a tolerance for alcohol and they need to drink even greater amounts to feel its effects. They also experience withdrawal responses when they stop drinking. Alcoholism destroys millions of families, social relationships and careers. Intoxicated drivers are responsible for many road accidents. It also has serious effects on the children of persons with this disorder. These children have higher rates of psychological problems, particularly anxiety, depression, phobias and substance-related disorders. Excessive drinking can seriously damage physical health. Some of the ill- effects of alcohol on health and psychological functioning are presented in Box 4.2. Table 4.2 : Characteristics of Individuals with Different Levels of Intellectual Disability Area of Mild Moderate Severe Functioning (IQ range = 55 to (IQ range = 35–40 (IQ range = 20–25 to approximately 70) to approximately approximately 35–40) 50–55) and Profound (IQ = below 20–25) Self-help Skills Feeds and dresses Has difficulties and No skills to partial self and cares for requires training but skills, but some can own toilet needs can learn adequate care for personal needs self-help skills on limited basis Speech and Receptive and Receptive and Receptive language Communication expressive language expressive language is limited; is adequate; is adequate; expressive language understands has speech problems is poor communication Academics Optimal learning Very few academic No academic skills environment; third skills; first or second to sixth grade grade is maximal Social Skills Has friends; can Capable of making Not capable of having learn to adjust friends but has real friends; no social quickly difficulty in many interactions social situations Vocational Can hold a job; Sheltered work Generally no Adjustment competitive to semi- environment; usually employment; usually competitive; primarily needs consistent needs constant care unskilled work supervision Adult Living Usually marries, Usually does not No marriage or has children; needs marry or have children; always help during stress children; dependent dependent on others Reprint 2025-26Psychology 86 Heroin Heroin intake significantly interferes with social and occupational functioning. Most abusers further develop a dependence on heroin, revolving their lives around the substance, building up a tolerance for it, and experiencing a withdrawal reaction when they stop taking it. The most direct danger of heroin abuse is an overdose, which slows down the respiratory centres in the brain, almost paralysing breathing, and in many cases causing death. Cocaine Regular use of cocaine may lead to a pattern of abuse in which the person may be intoxicated throughout the day and function poorly in social relationships and at work. It may also cause problems in short-term memory and attention. Dependence may develop, so that cocaine dominates the person’s life, more of the drug is needed to get the desired effects, and stopping it results in feelings of depression, fatigue, sleep problems, irritability and anxiety. Cocaine poses serious dangers. It has dangerous effects on psychological functioning and physical well-being. Some of the commonly abused substances are given in Box 4.3. Box 4.2 Effects of Alcohol : Some Facts • All alcohol beverages contain ethyl alcohol. • This chemical is absorbed into the blood and carried into the central nervous system (brain and spinal cord) where it depresses or slows down functioning. • Ethyl alcohol depresses those areas in the brain that control judgment and inhibition; people become more talkative and friendly, and they feel more confident and happy. • As alcohol is absorbed, it affects other areas of the brain. For example, drinkers are unable to make sound judgments, speech becomes less careful and less clear, and memory falters; many people become emotional, loud and aggressive. • Motor difficulties increase. For example, people become unsteady when they walk and clumsy in performing simple activities; vision becomes blurred and they have trouble in hearing; they have difficulty in driving or in solving simple problems. Box 4.3 Commonly Abused Substances (Following the DSM-5 Classification) • Alcohol • Stimulants : dextroamphetamines, metaamphetamines, cocaine • Caffeine : coffee, tea, caffeinated soda, analgesics, chocolate, cocoa • Cannabis : marijuana or ‘ bhang ’ • Hallucinogens : LSD, mescaline • Inhalants : gasoline, glue, paint thinners, spray paints, typewriter correction fluid, sprays • Tobacco : cigarettes, bidi • Opioid : morphine, heroin, cough syrup, painkillers (analgesics, anaesthetics) • Sedatives, Hypnotics or Anxiolytics : sleeping pills, anti-anxiety medication Reprint 2025-26",
            "col4": "DATA"
        }
    ],
    "projectDeleted": false,
    "codeFrameColumn": [
        {
            "col1": "col1",
            "col2": "col3",
            "researchObjective": "",
            "filterColumn": [],
            "title": "K12PSYH - 04 Psychological Disorders",
            "netSetting": true,
            "subNetSetting": true,
            "categorySetting": true,
            "sentimentSetting": true,
            "thinkingBudgetSetting": 2000,
            "questionType": "OPENEND",
            "otherBrand": false,
            "existingCodeFrame": [],
            "isList": false,
            "codeFrameWithDataList": [],
            "codeFrameList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 11639,
                        "CandidatesTokenCount": 3537,
                        "TotalTokenCount": 16443,
                        "thoughtsTokenCount": 1267,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Net": "Understanding Abnormality",
                                "Subnets": [
                                    {
                                        "Subnet": "Defining Abnormality",
                                        "Categories": [
                                            {
                                                "Category": "The Four Ds",
                                                "Summary": "Psychological disorders characterized by deviance, distress, dysfunction, and danger."
                                            },
                                            {
                                                "Category": "Deviation from Social Norms",
                                                "Summary": "Behavior, thoughts, or emotions that differ markedly from a society’s ideas of proper functioning."
                                            },
                                            {
                                                "Category": "Maladaptive Behavior",
                                                "Summary": "Behavior that interferes with optimal functioning, growth, and adaptation to life challenges."
                                            }
                                        ],
                                        "Summary": "Explores the key criteria and conceptual approaches used to define and identify abnormal behavior."
                                    },
                                    {
                                        "Subnet": "Historical Perspectives",
                                        "Categories": [
                                            {
                                                "Category": "Supernatural/Magical Forces",
                                                "Summary": "Belief that abnormal behavior is caused by evil spirits, devil, or magical forces, often treated by exorcism."
                                            },
                                            {
                                                "Category": "Biological/Organic Approach",
                                                "Summary": "View that abnormal behavior arises from body and brain processes, including imbalances of humours (Galen) or physical ailments."
                                            },
                                            {
                                                "Category": "Psychological Approach",
                                                "Summary": "Problems caused by inadequacies in individual thought, feeling, or perception, leading to modern psychodynamic theories."
                                            }
                                        ],
                                        "Summary": "Traces the historical evolution of explanations for abnormal behavior, from ancient supernatural beliefs to the emergence of biological and psychological viewpoints."
                                    },
                                    {
                                        "Subnet": "Classification Systems",
                                        "Categories": [
                                            {
                                                "Category": "Diagnostic and Statistical Manual (DSM-5)",
                                                "Summary": "Official manual by American Psychiatric Association describing and classifying psychological disorders with discrete clinical criteria."
                                            },
                                            {
                                                "Category": "International Classification of Diseases (ICD-10)",
                                                "Summary": "Classification scheme by World Health Organization, used in India, providing descriptions of clinical features and diagnostic guidelines."
                                            }
                                        ],
                                        "Summary": "Details the formal systems used globally to classify and categorize various psychological disorders for communication and understanding."
                                    }
                                ],
                                "Summary": "Encompasses the core concepts, definitions, historical context, and classification frameworks used to understand psychological disorders."
                            },
                            {
                                "Net": "Factors & Causes of Abnormal Behavior",
                                "Subnets": [
                                    {
                                        "Subnet": "Biological Factors",
                                        "Categories": [
                                            {
                                                "Category": "Genetic Predisposition",
                                                "Summary": "Faulty genes linked to disorders like bipolar, schizophrenia, and intellectual disability."
                                            },
                                            {
                                                "Category": "Neurochemical Imbalances",
                                                "Summary": "Abnormal activity of neurotransmitters (e.g., GABA, dopamine, serotonin) linked to specific disorders like anxiety, schizophrenia, and depression."
                                            },
                                            {
                                                "Category": "Physiological Conditions",
                                                "Summary": "Endocrine imbalances, malnutrition, injuries, and other biological conditions interfering with normal body functioning."
                                            }
                                        ],
                                        "Summary": "Examines the biological underpinnings of abnormal behavior, including genetic, neurochemical, and physiological influences."
                                    },
                                    {
                                        "Subnet": "Psychological Models",
                                        "Categories": [
                                            {
                                                "Category": "Psychodynamic Model",
                                                "Summary": "Behavior determined by unconscious psychological forces and conflicts, often rooted in early childhood (Freud's id, ego, superego)."
                                            },
                                            {
                                                "Category": "Behavioral Model",
                                                "Summary": "Abnormal behaviors are learned through conditioning (classical, operant, social learning) and can be unlearned."
                                            },
                                            {
                                                "Category": "Cognitive Model",
                                                "Summary": "Abnormal functioning results from irrational/inaccurate assumptions, illogical thinking, and overgeneralizations about oneself and the world."
                                            },
                                            {
                                                "Category": "Humanistic-Existential Model",
                                                "Summary": "Focuses on self-actualization, inherent goodness, and freedom to give meaning to existence; dysfunction arises from shirking responsibility."
                                            }
                                        ],
                                        "Summary": "Discusses various psychological theories explaining abnormal behavior, focusing on internal mental processes, learning, and personal meaning."
                                    },
                                    {
                                        "Subnet": "Socio-Cultural Factors & Models",
                                        "Categories": [
                                            {
                                                "Category": "Environmental Stressors",
                                                "Summary": "War, violence, discrimination, economic problems, and rapid social change contributing to psychological problems."
                                            },
                                            {
                                                "Category": "Family & Social Networks",
                                                "Summary": "Maladaptive family structures (e.g., enmeshed families) and lack of social support leading to abnormal functioning or prolonged depression."
                                            },
                                            {
                                                "Category": "Societal Labels & Roles",
                                                "Summary": "How societal norms and labels (e.g., 'mentally ill') can influence individuals to adopt and play a sick role."
                                            }
                                        ],
                                        "Summary": "Explores how societal, cultural, and interpersonal forces, including family dynamics and social labels, influence abnormal behavior."
                                    },
                                    {
                                        "Subnet": "Diathesis-Stress Model",
                                        "Categories": [
                                            {
                                                "Category": "Biological Predisposition (Diathesis)",
                                                "Summary": "Inherited biological aberration or vulnerability making an individual 'at risk' for a psychological disorder."
                                            },
                                            {
                                                "Category": "Pathogenic Stressors",
                                                "Summary": "Environmental factors or stressful situations that trigger the development of a disorder in a predisposed individual."
                                            }
                                        ],
                                        "Summary": "Describes an interactional model where psychological disorders develop from a biological predisposition triggered by stressful life events."
                                    }
                                ],
                                "Summary": "Details the multifactorial causes of abnormal behavior, including biological, psychological, socio-cultural influences, and the integrative diathesis-stress model."
                            },
                            {
                                "Net": "Anxiety & Related Disorders",
                                "Subnets": [
                                    {
                                        "Subnet": "Core Anxiety Disorders",
                                        "Categories": [
                                            {
                                                "Category": "Generalised Anxiety Disorder (GAD)",
                                                "Summary": "Prolonged, vague, unexplained, and intense fears not attached to a specific object, accompanied by worry, hypervigilance, and motor tension."
                                            },
                                            {
                                                "Category": "Panic Disorder",
                                                "Summary": "Recurrent, unpredictable anxiety attacks with intense terror, physiological symptoms (e.g., shortness of breath, palpitations), and fear of losing control or dying."
                                            },
                                            {
                                                "Category": "Phobias",
                                                "Summary": "Irrational fears related to specific objects, people, or situations, including specific phobias, social anxiety disorder, and agoraphobia."
                                            },
                                            {
                                                "Category": "Separation Anxiety Disorder (SAD)",
                                                "Summary": "Extreme distress and anxiety about separation from attachment figures, developmentally inappropriate, often manifesting in children."
                                            }
                                        ],
                                        "Summary": "Covers disorders characterized by excessive and persistent fear, worry, and apprehension, along with physical and psychological symptoms."
                                    },
                                    {
                                        "Subnet": "Obsessive-Compulsive & Related Disorders",
                                        "Categories": [
                                            {
                                                "Category": "Obsessive-Compulsive Disorder (OCD)",
                                                "Summary": "Inability to control preoccupation with specific ideas (obsessions) and/or repeated performance of specific acts (compulsions), causing distress and functional impairment."
                                            },
                                            {
                                                "Category": "Hoarding Disorder",
                                                "Summary": "Difficulty discarding possessions regardless of their actual value, leading to clutter that interferes with living areas."
                                            }
                                        ],
                                        "Summary": "Addresses disorders characterized by recurrent, intrusive thoughts (obsessions) and/or repetitive behaviors (compulsions)."
                                    },
                                    {
                                        "Subnet": "Trauma- & Stressor-Related Disorders",
                                        "Categories": [
                                            {
                                                "Category": "Post-Traumatic Stress Disorder (PTSD)",
                                                "Summary": "Symptoms like recurrent dreams, flashbacks, impaired concentration, and emotional numbing following exposure to a traumatic event."
                                            },
                                            {
                                                "Category": "Adjustment Disorders",
                                                "Summary": "Emotional or behavioral symptoms developing in response to an identifiable stressor, not meeting criteria for another mental disorder."
                                            }
                                        ],
                                        "Summary": "Includes conditions that develop in response to exposure to a traumatic or stressful event."
                                    }
                                ],
                                "Summary": "Categorizes a range of disorders marked by excessive fear, anxiety, or maladaptive responses to obsessive thoughts, compulsions, or traumatic experiences."
                            },
                            {
                                "Net": "Mood & Psychotic Disorders",
                                "Subnets": [
                                    {
                                        "Subnet": "Depressive Disorders",
                                        "Categories": [
                                            {
                                                "Category": "Major Depressive Disorder",
                                                "Summary": "A period of depressed mood, loss of interest/pleasure, changes in weight/sleep, fatigue, impaired concentration, feelings of worthlessness/guilt, and thoughts of death/suicide."
                                            },
                                            {
                                                "Category": "Risk Factors for Depression",
                                                "Summary": "Genetic make-up, age (young adulthood for women, early middle age for men), gender (women higher risk), negative life events, and lack of social support."
                                            }
                                        ],
                                        "Summary": "Focuses on disorders characterized by a pervasive low mood, loss of interest, and other significant behavioral and cognitive symptoms."
                                    },
                                    {
                                        "Subnet": "Bipolar & Related Disorders",
                                        "Categories": [
                                            {
                                                "Category": "Bipolar I Disorder",
                                                "Summary": "Involves alternating episodes of mania (elevated, expansive, or irritable mood) and depression, sometimes with periods of normal mood."
                                            },
                                            {
                                                "Category": "Manic Episodes",
                                                "Summary": "Distinct periods of abnormally and persistently elevated, expansive, or irritable mood and increased goal-directed activity or energy."
                                            }
                                        ],
                                        "Summary": "Describes disorders characterized by significant mood swings, encompassing both depressive and manic episodes."
                                    },
                                    {
                                        "Subnet": "Schizophrenia Spectrum & Other Psychotic Disorders",
                                        "Categories": [
                                            {
                                                "Category": "Positive Symptoms",
                                                "Summary": "Pathological excesses including delusions (persecution, reference, grandeur, control), disorganized thinking/speech (loosening of associations, neologisms), heightened perceptions, and hallucinations (auditory, tactile, somatic, visual, gustatory, olfactory), and inappropriate affect."
                                            },
                                            {
                                                "Category": "Negative Symptoms",
                                                "Summary": "Pathological deficits including poverty of speech (alogia), blunted/flat affect (reduced or no emotions), loss of volition (avolition/apathy), and social withdrawal."
                                            },
                                            {
                                                "Category": "Psychomotor Symptoms",
                                                "Summary": "Odd grimaces, gestures, or abnormal movements, sometimes extreme forms like catatonia (stupor, rigidity, posturing)."
                                            }
                                        ],
                                        "Summary": "Covers severe disorders characterized by a significant deterioration in personal, social, and occupational functioning due to disturbed thought processes, perceptions, emotional states, and motor abnormalities."
                                    },
                                    {
                                        "Subnet": "Suicide Prevention",
                                        "Categories": [
                                            {
                                                "Category": "Risk Factors for Suicidal Behavior",
                                                "Summary": "Mental disorders (depression, alcohol use), natural disasters, violence, abuse, loss, isolation, and previous attempts."
                                            },
                                            {
                                                "Category": "Prevention Strategies",
                                                "Summary": "Limiting access to means, responsible media reporting, alcohol policies, early identification/treatment, health worker training, and community support."
                                            },
                                            {
                                                "Category": "Identifying Student Distress",
                                                "Summary": "Changes in performance, attendance, behavior (lack of interest, declining grades, misconduct, absence, substance misuse)."
                                            }
                                        ],
                                        "Summary": "Addresses the complex issue of suicide, including risk factors, societal impact, and strategies for prevention and intervention."
                                    }
                                ],
                                "Summary": "Covers a spectrum of disorders affecting mood and thought processes, including depressive disorders, bipolar conditions, and severe psychotic disorders like schizophrenia, alongside considerations for suicide prevention."
                            },
                            {
                                "Net": "Developmental & Behavioral Disorders",
                                "Subnets": [
                                    {
                                        "Subnet": "Neurodevelopmental Disorders",
                                        "Categories": [
                                            {
                                                "Category": "Attention-Deficit/Hyperactivity Disorder (ADHD)",
                                                "Summary": "Characterized by persistent patterns of inattention (difficulty sustaining effort, disorganization) and/or hyperactivity-impulsivity (constant motion, difficulty waiting, uncontrolled reactions)."
                                            },
                                            {
                                                "Category": "Autism Spectrum Disorder",
                                                "Summary": "Widespread impairments in social interaction and communication skills, stereotyped behaviors, restricted interests, and a strong desire for routine."
                                            },
                                            {
                                                "Category": "Intellectual Disability",
                                                "Summary": "Below-average intellectual functioning (IQ ≤ 70) and deficits in adaptive behavior (communication, self-care, social skills) manifested before age 18."
                                            },
                                            {
                                                "Category": "Specific Learning Disorder",
                                                "Summary": "Difficulty in perceiving or processing information efficiently and accurately, impacting basic skills in reading, writing, and/or mathematics."
                                            }
                                        ],
                                        "Summary": "Encompasses disorders that manifest early in development, causing deficits or excesses in specific behaviors and impacting personal, social, academic, and occupational functioning."
                                    },
                                    {
                                        "Subnet": "Disruptive, Impulse-Control & Conduct Disorders",
                                        "Categories": [
                                            {
                                                "Category": "Oppositional Defiant Disorder (ODD)",
                                                "Summary": "Age-inappropriate patterns of stubbornness, irritability, defiance, disobedience, and hostile behavior, often justifying actions as reactions to circumstances."
                                            },
                                            {
                                                "Category": "Conduct Disorder",
                                                "Summary": "Age-inappropriate actions and attitudes violating family expectations, societal norms, or the rights of others, including aggressive actions, property damage, deceitfulness, and rule violations."
                                            }
                                        ],
                                        "Summary": "Addresses disorders characterized by problems in self-control of emotions and behaviors, often violating the rights of others or major societal norms."
                                    },
                                    {
                                        "Subnet": "Feeding & Eating Disorders",
                                        "Categories": [
                                            {
                                                "Category": "Anorexia Nervosa",
                                                "Summary": "Distorted body image leading to self-starvation, compulsive exercise, and significant weight loss due to an intense fear of gaining weight."
                                            },
                                            {
                                                "Category": "Bulimia Nervosa",
                                                "Summary": "Recurrent episodes of binge eating followed by compensatory behaviors like purging (vomiting, laxatives) due to a preoccupation with body shape and weight."
                                            },
                                            {
                                                "Category": "Binge Eating Disorder",
                                                "Summary": "Frequent episodes of eating unusually large amounts of food in a short period, often feeling out of control, eating rapidly, and to the point of discomfort, without compensatory behaviors."
                                            }
                                        ],
                                        "Summary": "Covers disorders characterized by persistent disturbance of eating or eating-related behavior that results in altered consumption or absorption of food and significantly impairs physical health or psychosocial functioning."
                                    }
                                ],
                                "Summary": "A category of disorders typically manifesting in childhood or adolescence, affecting development, behavior control, and eating patterns."
                            },
                            {
                                "Net": "Substance-Related & Addictive Disorders",
                                "Subnets": [
                                    {
                                        "Subnet": "Alcohol-Related Disorders",
                                        "Categories": [
                                            {
                                                "Category": "Alcohol Abuse & Dependence",
                                                "Summary": "Regular intake of large amounts of alcohol, leading to tolerance, withdrawal responses, and interference with social behavior, thinking, and work."
                                            },
                                            {
                                                "Category": "Effects of Alcohol",
                                                "Summary": "Depression of CNS, impaired judgment, unclear speech, memory faltering, emotional/aggressive behavior, increased motor difficulties, blurred vision, and difficulty in problem-solving/driving."
                                            },
                                            {
                                                "Category": "Societal & Family Impact",
                                                "Summary": "Destruction of families, social relationships, careers, contribution to road accidents, and higher rates of psychological problems in children of abusers."
                                            }
                                        ],
                                        "Summary": "Details the maladaptive behavioral patterns and health consequences associated with the regular and consistent use of alcohol."
                                    },
                                    {
                                        "Subnet": "Other Commonly Abused Substances",
                                        "Categories": [
                                            {
                                                "Category": "Heroin Abuse & Overdose",
                                                "Summary": "Interference with social/occupational functioning, dependence, tolerance, withdrawal, and risk of overdose causing respiratory paralysis and death."
                                            },
                                            {
                                                "Category": "Cocaine Abuse & Effects",
                                                "Summary": "Intoxication, poor social/work functioning, short-term memory/attention problems, dependence, withdrawal (depression, fatigue, irritability, anxiety), and dangerous psychological/physical effects."
                                            },
                                            {
                                                "Category": "Other Stimulants & Hallucinogens",
                                                "Summary": "Includes dextroamphetamines, metaamphetamines, caffeine, cannabis, LSD, mescaline, inhalants, tobacco, opioids, sedatives, hypnotics, and anxiolytics."
                                            }
                                        ],
                                        "Summary": "Covers the maladaptive behaviors, dependence, and harmful effects resulting from the use and abuse of various other psychoactive substances."
                                    }
                                ],
                                "Summary": "Encompasses disorders stemming from the maladaptive use of various substances, leading to abuse, dependence, and significant psychological and physical harm."
                            },
                            {
                                "Net": "996 Non specific NET",
                                "Subnets": [
                                    {
                                        "Subnet": "Non specific Subnet",
                                        "Categories": [
                                            {
                                                "Category": "Non specific Category",
                                                "Summary": "Responses that are irrelevant, cannot be coded into specific themes, and may require further review."
                                            }
                                        ],
                                        "Summary": "Contains responses that are irrelevant to the established themes but do not fit other catch-all categories."
                                    }
                                ],
                                "Summary": "A broad category for responses that are irrelevant or don't fit into the primary thematic nets, potentially needing upcoding."
                            },
                            {
                                "Net": "997 Don't Know Can't Say",
                                "Subnets": [
                                    {
                                        "Subnet": "Non-committal/Vague",
                                        "Categories": [
                                            {
                                                "Category": "Uncertain/Indecisive",
                                                "Summary": "Responses indicating a lack of knowledge, inability to decide, or general vagueness about the topic."
                                            }
                                        ],
                                        "Summary": "Contains responses that are non-committal or vague in nature."
                                    }
                                ],
                                "Summary": "For responses that express uncertainty, indecision, or a lack of definitive opinion."
                            },
                            {
                                "Net": "998 Not Applicable",
                                "Subnets": [
                                    {
                                        "Subnet": "Irrelevant/Out of Scope",
                                        "Categories": [
                                            {
                                                "Category": "Not Relevant to Question",
                                                "Summary": "Responses that are irrelevant to the specific survey question or the overall project context."
                                            }
                                        ],
                                        "Summary": "Contains responses that are not applicable or relevant to the question asked."
                                    }
                                ],
                                "Summary": "For responses that are deemed irrelevant or outside the scope of the current inquiry."
                            },
                            {
                                "Net": "999 None-Nothing",
                                "Subnets": [
                                    {
                                        "Subnet": "Gibberish/Meaningless",
                                        "Categories": [
                                            {
                                                "Category": "Unintelligible Text",
                                                "Summary": "Responses consisting of plain gibberish, non-language characters, or text that conveys no discernible meaning."
                                            }
                                        ],
                                        "Summary": "Contains responses that are uninterpretable or completely lack coherent meaning."
                                    }
                                ],
                                "Summary": "For responses that are uninterpretable, gibberish, or convey absolutely no information."
                            }
                        ],
                        "Summary": "The provided text, 'Psychological Disorders', offers a comprehensive overview of abnormal behavior. It defines abnormality through criteria like deviance, distress, dysfunction, and danger, and traces historical perspectives from supernatural beliefs to modern bio-psycho-social approaches. The text details various theoretical models explaining abnormal behavior, including biological, psychodynamic, behavioral, cognitive, humanistic-existential, socio-cultural, and the diathesis-stress model. It then systematically describes major psychological disorders categorized by diagnostic manuals (DSM-5, ICD-10), covering anxiety disorders (GAD, panic, phobias, SAD, OCD, PTSD), mood disorders (depressive, bipolar), psychotic disorders (schizophrenia with positive, negative, and psychomotor symptoms), neurodevelopmental disorders (ADHD, autism, intellectual disability, specific learning disorder), disruptive/impulse-control disorders (ODD, conduct disorder), feeding/eating disorders (anorexia, bulimia, binge eating), and substance-related and addictive disorders (alcohol, heroin, cocaine abuse). The text also highlights risk factors, symptoms, and societal impact for many of these conditions, emphasizing the importance of identification, prevention, and support."
                    }
                }
            ],
            "onlyCodeFrame": true,
            "errorList": [],
            "codeFrameGenerateType": "CODE_FRAME_ONLY",
            "requestType": "CODEFRAME",
            "requestMode": "DEVELOPMENT",
            "sourceLanguage": "English",
            "targetLanguage": "English",
            "response": {
                "status": "OK",
                "data": {
                    "status": "OK",
                    "PromptTokenCount": 11639,
                    "CandidatesTokenCount": 3537,
                    "TotalTokenCount": 16443,
                    "thoughtsTokenCount": 1267,
                    "finishReason": "STOP",
                    "data": [
                        {
                            "Net": "Understanding Abnormality",
                            "Subnets": [
                                {
                                    "Subnet": "Defining Abnormality",
                                    "Categories": [
                                        {
                                            "Category": "The Four Ds",
                                            "Summary": "Psychological disorders characterized by deviance, distress, dysfunction, and danger.",
                                            "code": 3
                                        },
                                        {
                                            "Category": "Deviation from Social Norms",
                                            "Summary": "Behavior, thoughts, or emotions that differ markedly from a society’s ideas of proper functioning.",
                                            "code": 4
                                        },
                                        {
                                            "Category": "Maladaptive Behavior",
                                            "Summary": "Behavior that interferes with optimal functioning, growth, and adaptation to life challenges.",
                                            "code": 5
                                        }
                                    ],
                                    "Summary": "Explores the key criteria and conceptual approaches used to define and identify abnormal behavior.",
                                    "code": 2
                                },
                                {
                                    "Subnet": "Historical Perspectives",
                                    "Categories": [
                                        {
                                            "Category": "Supernatural/Magical Forces",
                                            "Summary": "Belief that abnormal behavior is caused by evil spirits, devil, or magical forces, often treated by exorcism.",
                                            "code": 7
                                        },
                                        {
                                            "Category": "Biological/Organic Approach",
                                            "Summary": "View that abnormal behavior arises from body and brain processes, including imbalances of humours (Galen) or physical ailments.",
                                            "code": 8
                                        },
                                        {
                                            "Category": "Psychological Approach",
                                            "Summary": "Problems caused by inadequacies in individual thought, feeling, or perception, leading to modern psychodynamic theories.",
                                            "code": 9
                                        }
                                    ],
                                    "Summary": "Traces the historical evolution of explanations for abnormal behavior, from ancient supernatural beliefs to the emergence of biological and psychological viewpoints.",
                                    "code": 6
                                },
                                {
                                    "Subnet": "Classification Systems",
                                    "Categories": [
                                        {
                                            "Category": "Diagnostic and Statistical Manual (DSM-5)",
                                            "Summary": "Official manual by American Psychiatric Association describing and classifying psychological disorders with discrete clinical criteria.",
                                            "code": 11
                                        },
                                        {
                                            "Category": "International Classification of Diseases (ICD-10)",
                                            "Summary": "Classification scheme by World Health Organization, used in India, providing descriptions of clinical features and diagnostic guidelines.",
                                            "code": 12
                                        }
                                    ],
                                    "Summary": "Details the formal systems used globally to classify and categorize various psychological disorders for communication and understanding.",
                                    "code": 10
                                }
                            ],
                            "Summary": "Encompasses the core concepts, definitions, historical context, and classification frameworks used to understand psychological disorders.",
                            "code": 1
                        },
                        {
                            "Net": "Factors & Causes of Abnormal Behavior",
                            "Subnets": [
                                {
                                    "Subnet": "Biological Factors",
                                    "Categories": [
                                        {
                                            "Category": "Genetic Predisposition",
                                            "Summary": "Faulty genes linked to disorders like bipolar, schizophrenia, and intellectual disability.",
                                            "code": 15
                                        },
                                        {
                                            "Category": "Neurochemical Imbalances",
                                            "Summary": "Abnormal activity of neurotransmitters (e.g., GABA, dopamine, serotonin) linked to specific disorders like anxiety, schizophrenia, and depression.",
                                            "code": 16
                                        },
                                        {
                                            "Category": "Physiological Conditions",
                                            "Summary": "Endocrine imbalances, malnutrition, injuries, and other biological conditions interfering with normal body functioning.",
                                            "code": 17
                                        }
                                    ],
                                    "Summary": "Examines the biological underpinnings of abnormal behavior, including genetic, neurochemical, and physiological influences.",
                                    "code": 14
                                },
                                {
                                    "Subnet": "Psychological Models",
                                    "Categories": [
                                        {
                                            "Category": "Psychodynamic Model",
                                            "Summary": "Behavior determined by unconscious psychological forces and conflicts, often rooted in early childhood (Freud's id, ego, superego).",
                                            "code": 19
                                        },
                                        {
                                            "Category": "Behavioral Model",
                                            "Summary": "Abnormal behaviors are learned through conditioning (classical, operant, social learning) and can be unlearned.",
                                            "code": 20
                                        },
                                        {
                                            "Category": "Cognitive Model",
                                            "Summary": "Abnormal functioning results from irrational/inaccurate assumptions, illogical thinking, and overgeneralizations about oneself and the world.",
                                            "code": 21
                                        },
                                        {
                                            "Category": "Humanistic-Existential Model",
                                            "Summary": "Focuses on self-actualization, inherent goodness, and freedom to give meaning to existence; dysfunction arises from shirking responsibility.",
                                            "code": 22
                                        }
                                    ],
                                    "Summary": "Discusses various psychological theories explaining abnormal behavior, focusing on internal mental processes, learning, and personal meaning.",
                                    "code": 18
                                },
                                {
                                    "Subnet": "Socio-Cultural Factors & Models",
                                    "Categories": [
                                        {
                                            "Category": "Environmental Stressors",
                                            "Summary": "War, violence, discrimination, economic problems, and rapid social change contributing to psychological problems.",
                                            "code": 24
                                        },
                                        {
                                            "Category": "Family & Social Networks",
                                            "Summary": "Maladaptive family structures (e.g., enmeshed families) and lack of social support leading to abnormal functioning or prolonged depression.",
                                            "code": 25
                                        },
                                        {
                                            "Category": "Societal Labels & Roles",
                                            "Summary": "How societal norms and labels (e.g., 'mentally ill') can influence individuals to adopt and play a sick role.",
                                            "code": 26
                                        }
                                    ],
                                    "Summary": "Explores how societal, cultural, and interpersonal forces, including family dynamics and social labels, influence abnormal behavior.",
                                    "code": 23
                                },
                                {
                                    "Subnet": "Diathesis-Stress Model",
                                    "Categories": [
                                        {
                                            "Category": "Biological Predisposition (Diathesis)",
                                            "Summary": "Inherited biological aberration or vulnerability making an individual 'at risk' for a psychological disorder.",
                                            "code": 28
                                        },
                                        {
                                            "Category": "Pathogenic Stressors",
                                            "Summary": "Environmental factors or stressful situations that trigger the development of a disorder in a predisposed individual.",
                                            "code": 29
                                        }
                                    ],
                                    "Summary": "Describes an interactional model where psychological disorders develop from a biological predisposition triggered by stressful life events.",
                                    "code": 27
                                }
                            ],
                            "Summary": "Details the multifactorial causes of abnormal behavior, including biological, psychological, socio-cultural influences, and the integrative diathesis-stress model.",
                            "code": 13
                        },
                        {
                            "Net": "Anxiety & Related Disorders",
                            "Subnets": [
                                {
                                    "Subnet": "Core Anxiety Disorders",
                                    "Categories": [
                                        {
                                            "Category": "Generalised Anxiety Disorder (GAD)",
                                            "Summary": "Prolonged, vague, unexplained, and intense fears not attached to a specific object, accompanied by worry, hypervigilance, and motor tension.",
                                            "code": 32
                                        },
                                        {
                                            "Category": "Panic Disorder",
                                            "Summary": "Recurrent, unpredictable anxiety attacks with intense terror, physiological symptoms (e.g., shortness of breath, palpitations), and fear of losing control or dying.",
                                            "code": 33
                                        },
                                        {
                                            "Category": "Phobias",
                                            "Summary": "Irrational fears related to specific objects, people, or situations, including specific phobias, social anxiety disorder, and agoraphobia.",
                                            "code": 34
                                        },
                                        {
                                            "Category": "Separation Anxiety Disorder (SAD)",
                                            "Summary": "Extreme distress and anxiety about separation from attachment figures, developmentally inappropriate, often manifesting in children.",
                                            "code": 35
                                        }
                                    ],
                                    "Summary": "Covers disorders characterized by excessive and persistent fear, worry, and apprehension, along with physical and psychological symptoms.",
                                    "code": 31
                                },
                                {
                                    "Subnet": "Obsessive-Compulsive & Related Disorders",
                                    "Categories": [
                                        {
                                            "Category": "Obsessive-Compulsive Disorder (OCD)",
                                            "Summary": "Inability to control preoccupation with specific ideas (obsessions) and/or repeated performance of specific acts (compulsions), causing distress and functional impairment.",
                                            "code": 37
                                        },
                                        {
                                            "Category": "Hoarding Disorder",
                                            "Summary": "Difficulty discarding possessions regardless of their actual value, leading to clutter that interferes with living areas.",
                                            "code": 38
                                        }
                                    ],
                                    "Summary": "Addresses disorders characterized by recurrent, intrusive thoughts (obsessions) and/or repetitive behaviors (compulsions).",
                                    "code": 36
                                },
                                {
                                    "Subnet": "Trauma- & Stressor-Related Disorders",
                                    "Categories": [
                                        {
                                            "Category": "Post-Traumatic Stress Disorder (PTSD)",
                                            "Summary": "Symptoms like recurrent dreams, flashbacks, impaired concentration, and emotional numbing following exposure to a traumatic event.",
                                            "code": 40
                                        },
                                        {
                                            "Category": "Adjustment Disorders",
                                            "Summary": "Emotional or behavioral symptoms developing in response to an identifiable stressor, not meeting criteria for another mental disorder.",
                                            "code": 41
                                        }
                                    ],
                                    "Summary": "Includes conditions that develop in response to exposure to a traumatic or stressful event.",
                                    "code": 39
                                }
                            ],
                            "Summary": "Categorizes a range of disorders marked by excessive fear, anxiety, or maladaptive responses to obsessive thoughts, compulsions, or traumatic experiences.",
                            "code": 30
                        },
                        {
                            "Net": "Mood & Psychotic Disorders",
                            "Subnets": [
                                {
                                    "Subnet": "Depressive Disorders",
                                    "Categories": [
                                        {
                                            "Category": "Major Depressive Disorder",
                                            "Summary": "A period of depressed mood, loss of interest/pleasure, changes in weight/sleep, fatigue, impaired concentration, feelings of worthlessness/guilt, and thoughts of death/suicide.",
                                            "code": 44
                                        },
                                        {
                                            "Category": "Risk Factors for Depression",
                                            "Summary": "Genetic make-up, age (young adulthood for women, early middle age for men), gender (women higher risk), negative life events, and lack of social support.",
                                            "code": 45
                                        }
                                    ],
                                    "Summary": "Focuses on disorders characterized by a pervasive low mood, loss of interest, and other significant behavioral and cognitive symptoms.",
                                    "code": 43
                                },
                                {
                                    "Subnet": "Bipolar & Related Disorders",
                                    "Categories": [
                                        {
                                            "Category": "Bipolar I Disorder",
                                            "Summary": "Involves alternating episodes of mania (elevated, expansive, or irritable mood) and depression, sometimes with periods of normal mood.",
                                            "code": 47
                                        },
                                        {
                                            "Category": "Manic Episodes",
                                            "Summary": "Distinct periods of abnormally and persistently elevated, expansive, or irritable mood and increased goal-directed activity or energy.",
                                            "code": 48
                                        }
                                    ],
                                    "Summary": "Describes disorders characterized by significant mood swings, encompassing both depressive and manic episodes.",
                                    "code": 46
                                },
                                {
                                    "Subnet": "Schizophrenia Spectrum & Other Psychotic Disorders",
                                    "Categories": [
                                        {
                                            "Category": "Positive Symptoms",
                                            "Summary": "Pathological excesses including delusions (persecution, reference, grandeur, control), disorganized thinking/speech (loosening of associations, neologisms), heightened perceptions, and hallucinations (auditory, tactile, somatic, visual, gustatory, olfactory), and inappropriate affect.",
                                            "code": 50
                                        },
                                        {
                                            "Category": "Negative Symptoms",
                                            "Summary": "Pathological deficits including poverty of speech (alogia), blunted/flat affect (reduced or no emotions), loss of volition (avolition/apathy), and social withdrawal.",
                                            "code": 51
                                        },
                                        {
                                            "Category": "Psychomotor Symptoms",
                                            "Summary": "Odd grimaces, gestures, or abnormal movements, sometimes extreme forms like catatonia (stupor, rigidity, posturing).",
                                            "code": 52
                                        }
                                    ],
                                    "Summary": "Covers severe disorders characterized by a significant deterioration in personal, social, and occupational functioning due to disturbed thought processes, perceptions, emotional states, and motor abnormalities.",
                                    "code": 49
                                },
                                {
                                    "Subnet": "Suicide Prevention",
                                    "Categories": [
                                        {
                                            "Category": "Risk Factors for Suicidal Behavior",
                                            "Summary": "Mental disorders (depression, alcohol use), natural disasters, violence, abuse, loss, isolation, and previous attempts.",
                                            "code": 54
                                        },
                                        {
                                            "Category": "Prevention Strategies",
                                            "Summary": "Limiting access to means, responsible media reporting, alcohol policies, early identification/treatment, health worker training, and community support.",
                                            "code": 55
                                        },
                                        {
                                            "Category": "Identifying Student Distress",
                                            "Summary": "Changes in performance, attendance, behavior (lack of interest, declining grades, misconduct, absence, substance misuse).",
                                            "code": 56
                                        }
                                    ],
                                    "Summary": "Addresses the complex issue of suicide, including risk factors, societal impact, and strategies for prevention and intervention.",
                                    "code": 53
                                }
                            ],
                            "Summary": "Covers a spectrum of disorders affecting mood and thought processes, including depressive disorders, bipolar conditions, and severe psychotic disorders like schizophrenia, alongside considerations for suicide prevention.",
                            "code": 42
                        },
                        {
                            "Net": "Developmental & Behavioral Disorders",
                            "Subnets": [
                                {
                                    "Subnet": "Neurodevelopmental Disorders",
                                    "Categories": [
                                        {
                                            "Category": "Attention-Deficit/Hyperactivity Disorder (ADHD)",
                                            "Summary": "Characterized by persistent patterns of inattention (difficulty sustaining effort, disorganization) and/or hyperactivity-impulsivity (constant motion, difficulty waiting, uncontrolled reactions).",
                                            "code": 59
                                        },
                                        {
                                            "Category": "Autism Spectrum Disorder",
                                            "Summary": "Widespread impairments in social interaction and communication skills, stereotyped behaviors, restricted interests, and a strong desire for routine.",
                                            "code": 60
                                        },
                                        {
                                            "Category": "Intellectual Disability",
                                            "Summary": "Below-average intellectual functioning (IQ ≤ 70) and deficits in adaptive behavior (communication, self-care, social skills) manifested before age 18.",
                                            "code": 61
                                        },
                                        {
                                            "Category": "Specific Learning Disorder",
                                            "Summary": "Difficulty in perceiving or processing information efficiently and accurately, impacting basic skills in reading, writing, and/or mathematics.",
                                            "code": 62
                                        }
                                    ],
                                    "Summary": "Encompasses disorders that manifest early in development, causing deficits or excesses in specific behaviors and impacting personal, social, academic, and occupational functioning.",
                                    "code": 58
                                },
                                {
                                    "Subnet": "Disruptive, Impulse-Control & Conduct Disorders",
                                    "Categories": [
                                        {
                                            "Category": "Oppositional Defiant Disorder (ODD)",
                                            "Summary": "Age-inappropriate patterns of stubbornness, irritability, defiance, disobedience, and hostile behavior, often justifying actions as reactions to circumstances.",
                                            "code": 64
                                        },
                                        {
                                            "Category": "Conduct Disorder",
                                            "Summary": "Age-inappropriate actions and attitudes violating family expectations, societal norms, or the rights of others, including aggressive actions, property damage, deceitfulness, and rule violations.",
                                            "code": 65
                                        }
                                    ],
                                    "Summary": "Addresses disorders characterized by problems in self-control of emotions and behaviors, often violating the rights of others or major societal norms.",
                                    "code": 63
                                },
                                {
                                    "Subnet": "Feeding & Eating Disorders",
                                    "Categories": [
                                        {
                                            "Category": "Anorexia Nervosa",
                                            "Summary": "Distorted body image leading to self-starvation, compulsive exercise, and significant weight loss due to an intense fear of gaining weight.",
                                            "code": 67
                                        },
                                        {
                                            "Category": "Bulimia Nervosa",
                                            "Summary": "Recurrent episodes of binge eating followed by compensatory behaviors like purging (vomiting, laxatives) due to a preoccupation with body shape and weight.",
                                            "code": 68
                                        },
                                        {
                                            "Category": "Binge Eating Disorder",
                                            "Summary": "Frequent episodes of eating unusually large amounts of food in a short period, often feeling out of control, eating rapidly, and to the point of discomfort, without compensatory behaviors.",
                                            "code": 69
                                        }
                                    ],
                                    "Summary": "Covers disorders characterized by persistent disturbance of eating or eating-related behavior that results in altered consumption or absorption of food and significantly impairs physical health or psychosocial functioning.",
                                    "code": 66
                                }
                            ],
                            "Summary": "A category of disorders typically manifesting in childhood or adolescence, affecting development, behavior control, and eating patterns.",
                            "code": 57
                        },
                        {
                            "Net": "Substance-Related & Addictive Disorders",
                            "Subnets": [
                                {
                                    "Subnet": "Alcohol-Related Disorders",
                                    "Categories": [
                                        {
                                            "Category": "Alcohol Abuse & Dependence",
                                            "Summary": "Regular intake of large amounts of alcohol, leading to tolerance, withdrawal responses, and interference with social behavior, thinking, and work.",
                                            "code": 72
                                        },
                                        {
                                            "Category": "Effects of Alcohol",
                                            "Summary": "Depression of CNS, impaired judgment, unclear speech, memory faltering, emotional/aggressive behavior, increased motor difficulties, blurred vision, and difficulty in problem-solving/driving.",
                                            "code": 73
                                        },
                                        {
                                            "Category": "Societal & Family Impact",
                                            "Summary": "Destruction of families, social relationships, careers, contribution to road accidents, and higher rates of psychological problems in children of abusers.",
                                            "code": 74
                                        }
                                    ],
                                    "Summary": "Details the maladaptive behavioral patterns and health consequences associated with the regular and consistent use of alcohol.",
                                    "code": 71
                                },
                                {
                                    "Subnet": "Other Commonly Abused Substances",
                                    "Categories": [
                                        {
                                            "Category": "Heroin Abuse & Overdose",
                                            "Summary": "Interference with social/occupational functioning, dependence, tolerance, withdrawal, and risk of overdose causing respiratory paralysis and death.",
                                            "code": 76
                                        },
                                        {
                                            "Category": "Cocaine Abuse & Effects",
                                            "Summary": "Intoxication, poor social/work functioning, short-term memory/attention problems, dependence, withdrawal (depression, fatigue, irritability, anxiety), and dangerous psychological/physical effects.",
                                            "code": 77
                                        },
                                        {
                                            "Category": "Other Stimulants & Hallucinogens",
                                            "Summary": "Includes dextroamphetamines, metaamphetamines, caffeine, cannabis, LSD, mescaline, inhalants, tobacco, opioids, sedatives, hypnotics, and anxiolytics.",
                                            "code": 78
                                        }
                                    ],
                                    "Summary": "Covers the maladaptive behaviors, dependence, and harmful effects resulting from the use and abuse of various other psychoactive substances.",
                                    "code": 75
                                }
                            ],
                            "Summary": "Encompasses disorders stemming from the maladaptive use of various substances, leading to abuse, dependence, and significant psychological and physical harm.",
                            "code": 70
                        }
                    ],
                    "Summary": "The provided text, 'Psychological Disorders', offers a comprehensive overview of abnormal behavior. It defines abnormality through criteria like deviance, distress, dysfunction, and danger, and traces historical perspectives from supernatural beliefs to modern bio-psycho-social approaches. The text details various theoretical models explaining abnormal behavior, including biological, psychodynamic, behavioral, cognitive, humanistic-existential, socio-cultural, and the diathesis-stress model. It then systematically describes major psychological disorders categorized by diagnostic manuals (DSM-5, ICD-10), covering anxiety disorders (GAD, panic, phobias, SAD, OCD, PTSD), mood disorders (depressive, bipolar), psychotic disorders (schizophrenia with positive, negative, and psychomotor symptoms), neurodevelopmental disorders (ADHD, autism, intellectual disability, specific learning disorder), disruptive/impulse-control disorders (ODD, conduct disorder), feeding/eating disorders (anorexia, bulimia, binge eating), and substance-related and addictive disorders (alcohol, heroin, cocaine abuse). The text also highlights risk factors, symptoms, and societal impact for many of these conditions, emphasizing the importance of identification, prevention, and support."
                }
            },
            "summaryList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 11275,
                        "CandidatesTokenCount": 4900,
                        "TotalTokenCount": 17543,
                        "thoughtsTokenCount": 1368,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Summary": "<h2>Chapter 4: Psychological Disorders - A Comprehensive Synthesis</h2><h3>1. Introduction to Psychological Disorders</h3>This chapter provides a comprehensive overview of psychological disorders, starting with an exploration of the fundamental concepts of abnormality and the various criteria used for identification. It delves into the historical evolution of understanding these conditions, outlines classification systems, and examines the multifaceted factors that contribute to abnormal behavior, before detailing major psychological disorders (e.g., Introduction).<p><b>1.1 Concepts of Abnormality and Psychological Disorders</b><br>Defining abnormality is complex, with no universally accepted definition. However, most definitions incorporate the 'four Ds':</p><ul><li><b>Deviance:</b> Behavior that is different, extreme, unusual, or bizarre (e.g., abnormal means 'away from the normal').</li><li><b>Distress:</b> Unpleasant and upsetting to the individual and others (e.g., people who are unhappy, troubled, and dissatisfied).</li><li><b>Dysfunction:</b> Interfering with a person's ability to carry out daily activities constructively (e.g., maladaptive behavior).</li><li><b>Danger:</b> Potentially harmful to oneself or others (e.g., possible danger to the person or to others).</li></ul><p>Two primary approaches distinguish normal from abnormal behaviors (e.g., various approaches have been used):</p><ul><li><b>Deviation from Social Norms:</b> Views abnormal behavior as differing markedly from societal expectations, which are culturally determined. This approach is critiqued for assuming social conformity equals normality (e.g., societal norms grow from its particular culture).</li><li><b>Maladaptive Behavior:</b> Considers behavior abnormal if it interferes with an individual's well-being, growth, and fulfillment (e.g., interferes with optimal functioning and growth). This suggests vulnerability or an inability to cope with environmental stress (e.g., vulnerability in the individual, inability to cope, or exceptional stress).</li></ul><p>There is a significant stigma attached to mental illness, leading people to hesitate in seeking professional help (e.g., stigma attached to mental illness means that people are hesitant to consult a doctor or psychologist).</p><p><b>1.2 Historical Background</b><br>The understanding of psychological disorders has evolved through various perspectives (e.g., history of abnormal psychology):</p><ul><li><b>Supernatural/Magical:</b> Ancient belief in evil spirits (bhoot-pret), the devil (shaitan), leading to practices like exorcism and reliance on shamans (ojha) (e.g., still commonly used).</li><li><b>Biological/Organic:</b> Posits that abnormal behavior results from improper body and brain functioning. This approach finds modern support in linking biological processes to maladaptive behavior (e.g., correcting these defective biological processes results in improved functioning).</li><li><b>Psychological:</b> Attributes problems to inadequacies in an individual's thinking, feeling, or perception of the world (e.g., caused by inadequacies in the way an individual thinks, feels, or perceives the world).</li></ul><p>Historically, in the ancient Western world, philosopher-physicians like Hippocrates, Socrates, and Plato developed the organismic approach, viewing disturbed behavior as conflicts between emotion and reason. Galen's theory of four humours (blood, black bile, yellow bile, phlegm) causing temperament imbalances paralleled the Indian concept of three doshas (vata, pitta, kapha) (e.g., Galen elaborated on the role of the four humours). The Middle Ages saw a resurgence of demonology and witch-hunts, though figures like St. Augustine laid groundwork for psychodynamic theories. The Renaissance introduced humanism, with Johann Weyer advocating for medical treatment for the 'mentally disturbed' rather than theological. The Age of Reason and Enlightenment (17th-18th centuries) emphasized the scientific method, leading to asylum reforms and the movement towards deinstitutionalization and community care (e.g., Reform Movement). Modern understanding converges into an <b>interactional, or bio-psycho-social approach</b>, recognizing the interplay of biological, psychological, and social factors (e.g., all three factors... play important roles).</p><p><b>1.3 Classification of Psychological Disorders</b><br>Classification systems are essential for communication among professionals (psychologists, psychiatrists, social workers) and for understanding causes and treatment (e.g., enable users... to communicate with each other). The primary classification manuals are:</p><ul><li>The American Psychiatric Association's <b>Diagnostic and Statistical Manual of Mental Disorders, 5th Edition (DSM-5)</b>, which provides discrete clinical criteria (e.g., current version of it).</li><li>The World Health Organisation's <b>International Classification of Diseases (ICD-10) Classification of Behavioural and Mental Disorders</b>, used in India and globally, providing descriptions of symptoms and diagnostic guidelines (e.g., classification scheme officially used in India and elsewhere).</li></ul><h3>2. Factors Underlying Abnormal Behavior</h3><p>Psychologists utilize diverse approaches to explain abnormal behavior, each emphasizing different aspects and factors (e.g., psychologists use different approaches).</p><p><b>2.1 Biological Factors</b><br>These encompass faulty genes, endocrine imbalances, malnutrition, and injuries, which can disrupt normal development and functioning. The biological model posits a biochemical or physiological basis for abnormal behavior, linking psychological disorders to problems in neurotransmitter activity (e.g., abnormal activity by certain neuro-transmitters can lead to specific psychological disorders). For instance, low GABA activity is linked to anxiety, excess dopamine to schizophrenia, and low serotonin to depression. Genetic factors are implicated in disorders like bipolar disorder and schizophrenia, though typically multiple genes are involved rather than a single culprit (e.g., no single gene is responsible for a particular behavior or a psychological disorder).</p><p><b>2.2 Psychological Models</b><br>These models highlight the significant role of psychological and interpersonal factors (e.g., psychological and interpersonal factors have a significant role).</p><ul><li><b>Psychodynamic Model:</b> Formulated by Freud, it suggests behavior stems from unconscious psychological forces and conflicts, often rooted in early childhood (e.g., abnormal behavior is a symbolic expression of unconscious mental conflicts).</li><li><b>Behavioral Model:</b> Proposes that both normal and abnormal behaviors are learned through conditioning (classical, operant, social learning) and thus can be unlearned (e.g., what has been learned can be unlearned).</li><li><b>Cognitive Model:</b> Focuses on how irrational assumptions, inaccurate attitudes, and illogical thinking patterns (e.g., overgeneralisations) contribute to abnormal functioning (e.g., abnormal functioning can result from cognitive problems).</li><li><b>Humanistic-Existential Model:</b> Humanists believe in an innate drive for self-actualization, while existentialists emphasize freedom to create meaning. Dysfunction arises from thwarting this potential or shirking responsibility (e.g., those who shirk from this responsibility would live empty, inauthentic, and dysfunctional lives).</li></ul><p><b>2.3 Socio-Cultural Factors</b><br>External stressors such as war, discrimination, economic issues, and rapid social change can induce psychological problems. The socio-cultural model explains abnormal behavior through the lens of social and cultural forces, including family structure (e.g., enmeshed structure), social networks (e.g., isolated and lack social support), and societal labels/roles, which can perpetuate 'sick' behavior (e.g., the person gradually learns to accept and play the sick role).</p><p><b>2.4 Diathesis-Stress Model</b><br>This widely accepted model proposes that psychological disorders develop from an interaction of a biological predisposition (diathesis) and stressful life situations (e.g., psychological disorders develop when a diathesis... is set off by a stressful situation). It involves:</p><ul><li>A biological aberration (inherited diathesis).</li><li>Vulnerability or predisposition to a disorder.</li><li>Pathogenic stressors that trigger the disorder in 'at-risk' individuals (e.g., presence of pathogenic stressors).</li></ul><h3>3. Major Psychological Disorders</h3><p><b>3.1 Anxiety Disorders</b><br>Characterized by distressing and functionally interfering high levels of anxiety (e.g., most common category of psychological disorders). Symptoms include rapid heart rate, shortness of breath, dizziness, and tension (e.g., symptoms: rapid heart rate, shortness of breath, diarrhoea, loss of appetite, fainting, dizziness, sweating, sleeplessness, frequent urination and tremors).</p><ul><li><b>Generalised Anxiety Disorder:</b> Prolonged, vague, intense, unexplained fears without a specific object, accompanied by hypervigilance and motor tension (e.g., constantly scanning the environment for dangers).</li><li><b>Panic Disorder:</b> Recurrent intense terror attacks, sudden and unpredictable, with physical symptoms like breathlessness, palpitations, and fear of losing control or dying (e.g., abrupt surge of intense anxiety rising to a peak).</li><li><b>Phobias:</b> Irrational fears of specific objects, people, or situations. Includes specific phobias (e.g., intense fear of a certain type of animal), social anxiety disorder (social phobia) (e.g., incapacitating fear and embarrassment when dealing with others), and agoraphobia (fear of unfamiliar situations or leaving home) (e.g., afraid of leaving their home).</li><li><b>Separation Anxiety Disorder (SAD):</b> Developmentally inappropriate fear and anxiety about separation from attachment figures, leading to distress and avoidance behaviors (e.g., cling to and shadow their parents’ every move).</li></ul><p><b>3.2 Obsessive-Compulsive and Related Disorders</b><br>Involve inability to control preoccupation with specific ideas (obsessions) or to prevent repeated actions (compulsions) (e.g., unable to control their preoccupation with specific ideas). Obsessions are unpleasant, shameful thoughts, while compulsions are repetitive behaviors like counting, checking, or washing (e.g., Many compulsions deal with counting, ordering, checking, touching and washing). Other disorders include hoarding, trichotillomania (hair-pulling), and excoriation (skin-picking) (e.g., other disorders in this category include hoarding disorder, trichotillomania (hair -pulling disorder), excoriation (skin-picking) disorder etc.).</p><p><b>3.3 Trauma- and Stressor-Related Disorders</b><br>Occur after exposure to traumatic or stressful events (e.g., natural disaster, bomb blasts). Post-traumatic stress disorder (PTSD) symptoms include recurrent dreams, flashbacks, impaired concentration, and emotional numbing (e.g., PTSD symptoms vary widely). Adjustment Disorders and Acute Stress Disorder also fall into this category.</p><p><b>3.4 Somatic Symptom and Related Disorders</b><br>These involve physical symptoms without a physical disease, where psychological difficulties manifest physically (e.g., physical symptoms in the absence of a physical disease).</p><ul><li><b>Somatic Symptom Disorder:</b> Persistent body-related symptoms, with excessive preoccupation and worry about health, leading to distress and daily life disturbances (e.g., tend to be overly preoccupied with their symptoms).</li><li><b>Illness Anxiety Disorder:</b> Persistent preoccupation about developing a serious illness, accompanied by anxiety, often not reassured by medical results (e.g., constantly worrying about this possibility).</li><li><b>Conversion Disorders:</b> Loss of basic body functions (e.g., paralysis, blindness) often occurring after a stressful experience, with no physical cause (e.g., may be quite sudden).</li></ul><p><b>3.5 Dissociative Disorders</b><br>Characterized by a severance of connections between ideas and emotions, involving feelings of unreality, depersonalization, or loss of identity, often blotting out painful experiences (e.g., sudden temporary alterations of consciousness that blot out painful experiences).</p><ul><li><b>Dissociative Amnesia:</b> Extensive but selective memory loss without organic cause, sometimes involving dissociative fugue (unexpected travel and new identity) (e.g., cannot remember anything about their past).</li><li><b>Dissociative Identity Disorder (Multiple Personality):</b> The presence of two or more distinct alternate personalities, often linked to childhood trauma (e.g., often associated with traumatic experiences in childhood).</li><li><b>Depersonalisation/Derealisation Disorder:</b> A dreamlike state where the person feels separated from self (depersonalization) and reality (derealization) (e.g., person’s sense of reality is temporarily lost or changed).</li></ul><p><b>3.6 Depressive Disorders</b><br>Depression is a widely prevalent mental disorder, manifesting as negative moods and behavioral changes. Major depressive disorder involves a depressed mood and/or loss of interest/pleasure in activities, along with symptoms like weight change, sleep problems, fatigue, cognitive difficulties, agitation, and thoughts of death/suicide (e.g., Major depressive disorder is defined as a period of depressed mood). Predisposing factors include genetic make-up, age (e.g., women particularly at risk during young adulthood), gender (women more likely to report), negative life events, and lack of social support (e.g., experiencing negative life events and lack of social support).</p><p><b>3.7 Bipolar and Related Disorders</b><br>Characterized by alternating episodes of mania (elevated, expansive, or irritable mood) and depression, sometimes with periods of normal mood (e.g., involve both mania and depression). Examples include Bipolar I, Bipolar II, and Cyclothymic Disorder.</p><p><b>Conclusions on Suicide:</b> Suicide is a complex and devastating outcome of biological, genetic, psychological, sociological, cultural, and environmental factors. Mental disorders (especially depression and alcohol use disorders), natural disasters, violence, abuse, loss, and isolation are significant risk factors, with previous attempts being the strongest (e.g., Previous suicidal attempt is the strongest risk factor). Suicidal behavior often reflects difficulties in problem-solving, stress management, and emotional expression (e.g., Suicidal thoughts lead to suicidal action only when acting on these thoughts seems to be the only way out).<br><b>Actionable Recommendations for Suicide Prevention (WHO suggestions):</b></p><ul><li>Limiting access to means of suicide (e.g., limiting access to the means of suicide).</li><li>Responsible media reporting of suicide (e.g., reporting of suicide by media in a responsible way).</li><li>Implementing alcohol-related policies (e.g., bringing in alcohol-related policies).</li><li>Early identification, treatment, and care for at-risk individuals (e.g., early identification, treatment and care of people at risk).</li><li>Training health workers in suicide assessment and management (e.g., training health workers in assessing and managing for suicide).</li><li>Providing care for attempt survivors and community support (e.g., care for people who attempted suicide and providing community support).</li></ul><p><b>Stakeholder Responsibilities:</b> Government, media, and civil society all play crucial roles (e.g., a need for comprehensive multi-sectoral approach where the government, media and civil society all play important role as stakeholders).<br><b>Identifying Students in Distress:</b> Look for unexpected changes in performance, attendance, or behavior (e.g., lack of interest in common activities, declining grades, smoking or drinking, or drug misuse).<br><b>Strengthening Students' Self-Esteem:</b> Foster positive identity, provide skill development opportunities, establish trustful communication, and set specific, measurable, achievable, relevant, time-bound (SMART) goals (e.g., accentuating positive life experiences to develop positive identity).</p><p><b>3.8 Schizophrenia Spectrum and Other Psychotic Disorders</b><br>Schizophrenia is a debilitating psychotic disorder marked by deteriorated personal, social, and occupational functioning due to disturbed thought processes, strange perceptions, unusual emotions, and motor abnormalities (e.g., a group of psychotic disorders). Symptoms are categorized as positive, negative, and psychomotor.</p><ul><li><b>Positive Symptoms (pathological excesses):</b><br><ul><li><b>Delusions:</b> False, firmly held beliefs despite contrary evidence (e.g., not affected by rational argument). Common types include persecution, reference, grandeur, and control (e.g., Delusions of persecution are the most common).</li><li><b>Disorganized Thinking and Speech:</b> Difficulty thinking logically, leading to incoherent speech (e.g., rapidly shifting from one topic to another), neologisms (inventing new words), and perseveration (repeated thoughts).</li><li><b>Heightened Perception and Hallucinations:</b> Perceptions without external stimuli (e.g., perceptions that occur in the absence of external stimuli). Auditory hallucinations (voices) are most common, but others involve tactile, somatic, visual, gustatory, and olfactory senses (e.g., auditory hallucinations are most common).</li><li><b>Inappropriate Affect:</b> Emotions unsuited to the situation (e.g., emotions that are unsuited to the situation).</li></ul></li><li><b>Negative Symptoms (pathological deficits):</b><br><ul><li><b>Alogia (Poverty of Speech):</b> Reduced speech content and quantity (e.g., a reduction in speech and speech content).</li><li><b>Blunted and Flat Affect:</b> Reduced or absent emotional expression (e.g., Some show no emotions at all).</li><li><b>Avolition (Apathy):</b> Inability to initiate or complete actions (e.g., an inability to start or complete a course of action).</li><li><b>Social Withdrawal:</b> Retreat from social interaction, focused on internal fantasies (e.g., withdraw socially and become totally focused on their own ideas and fantasies).</li></ul></li><li><b>Psychomotor Symptoms:</b> Odd grimaces, gestures, or catatonia (motionlessness, rigidity, or bizarre posturing for extended periods) (e.g., remain motionless and silent for long stretches of time).</li></ul><p><b>3.9 Neurodevelopmental Disorders</b><br>Manifest early in development, often before or during early schooling, hampering personal, social, academic, and occupational functioning (e.g., manifest in the early stage of development). If untreated, these can lead to chronic disorders in adulthood (e.g., can lead to more serious and chronic disorders as the child moves into adulthood).</p><ul><li><b>Attention-Deficit/Hyperactivity Disorder (ADHD):</b> Characterized by inattention (difficulty sustaining mental effort, disorganization, distractibility) and hyperactivity-impulsivity (constant motion, fidgeting, inability to wait or delay gratification) (e.g., children with ADHD are in constant motion).</li><li><b>Autism Spectrum Disorder:</b> Widespread impairments in social interaction and communication, restricted interests, and stereotyped behaviors. About 70% also have intellectual disabilities (e.g., marked difficulties in social interaction and communication). They struggle with social initiation, emotional sharing, and may have repetitive speech patterns or self-stimulatory/self-injurious motor movements (e.g., lining up objects or stereotyped body movements such as rocking).</li><li><b>Intellectual Disability:</b> Below-average intellectual functioning (IQ &le; 70) and deficits in adaptive behavior (e.g., communication, self-care, social skills) manifested before age 18 (e.g., deficits or impairments in adaptive behaviour). Categorized into mild, moderate, severe, and profound, impacting self-help, speech, academics, social skills, and vocational adjustment (e.g., Table 4.2: Characteristics of Individuals with Different Levels of Intellectual Disability).</li><li><b>Specific Learning Disorder:</b> Difficulty perceiving or processing information efficiently, affecting basic skills in reading, writing, or mathematics in early school years (e.g., individual encounters problems in basic skills in reading, writing and/or mathematics).</li></ul><p><b>3.10 Disruptive, Impulse-Control and Conduct Disorders</b><br>Involve age-inappropriate actions that violate expectations and rights of others (e.g., age-inappropriate actions and attitudes).</p><ul><li><b>Oppositional Defiant Disorder (ODD):</b> Stubbornness, irritability, defiance, and hostile behavior in age-inappropriate amounts (e.g., display age-inappropriate amounts of stubbornness). Individuals often justify their behavior and don't see themselves as defiant.</li><li><b>Conduct Disorder:</b> Aggressive actions harming people/animals, property damage, deceitfulness/theft, and serious rule violations (e.g., aggressive actions that cause or threaten harm to people or animals). Aggression can be verbal, physical, hostile, or proactive (e.g., dominating and bullying others without provocation).</li></ul><p><b>3.11 Feeding and Eating Disorders</b><br>Of particular interest to young people (e.g., Another group of disorders which are of special interest to young people are eating disorders).</p><ul><li><b>Anorexia Nervosa:</b> Distorted body image leading to self-starvation, compulsive exercise, and unusual eating habits, potentially fatal (e.g., Often refusing to eat, exercising compulsively).</li><li><b>Bulimia Nervosa:</b> Binge eating followed by purging (laxatives, vomiting), accompanied by feelings of disgust and shame (e.g., relieved of tension and negative emotions after purging).</li><li><b>Binge Eating:</b> Frequent episodes of out-of-control eating, consuming large amounts of food quickly, even when not hungry, leading to uncomfortable fullness (e.g., Individual tends to eat at a higher speed than normal).</li></ul><p><b>3.12 Substance-Related and Addictive Disorders</b><br>Maladaptive behaviors resulting from regular, consistent substance use and abuse, altering thinking, feeling, and behavior (e.g., problems associated with the use and abuse of alcohol, cocaine, tobacco and opiods among others).</p><ul><li><b>Alcohol:</b> Abuse involves large, regular intake, interfering with social and occupational functioning, leading to tolerance and withdrawal. Destroys families and careers, causes accidents, and negatively impacts children (e.g., Alcoholism destroys millions of families, social relationships and careers). Physically damages health by depressing the central nervous system, affecting judgment, speech, memory, emotions, and motor skills (e.g., affects other areas of the brain).</li><li><b>Heroin:</b> Significantly interferes with social/occupational functioning, leading to dependence, tolerance, and withdrawal. Overdose can be fatal due to respiratory paralysis (e.g., Most direct danger of heroin abuse is an overdose).</li><li><b>Cocaine:</b> Regular use can lead to intoxication, poor social/work functioning, and problems with short-term memory/attention. Dependence develops, with withdrawal causing depression, fatigue, sleep problems, irritability, and anxiety. Poses serious dangers to psychological and physical well-being (e.g., It may also cause problems in short-term memory and attention).</li></ul><p>Commonly abused substances include alcohol, stimulants (dextroamphetamines, cocaine), caffeine, cannabis, hallucinogens, inhalants, tobacco, opioids, and sedatives/hypnotics/anxiolytics (e.g., Box 4.3: Commonly Abused Substances).</p>"
                            }
                        ]
                    }
                }
            ]
        }
    ],
    "fileName": "",
    "projectType": "DOCUMENT",
    "__v": 0,
    "headerCols": {
        "col1": "Part Id",
        "col2": "FileName",
        "col3": "Content"
    }
},
{
    "_id": "695e0d3dda2133fe04a5f50f",
    "projectId": "377077",
    "projectName": "K12PHY2 - 06 SEMICONDUCTOR ELECTRONICS MATERIALS, DEVICES AND SIMPLE CIRCUITS",
    "projectDescription": "",
    "projectCategory":"Educational",
    "projectCreatedDate": "2026-01-07T07:37:30.409Z",
    "projectUpdatedDate": "2026-01-07T07:37:30.409Z",
    "projectStatus": "Active",
    "companyId": "6957bd3ad58ea40715c6ae98",
    "projectCreatedBy": "6957bd3ad58ea40715c6ae99",
    "projectUpdatedBy": "6957bd3ad58ea40715c6ae99",
    "projectLastAccessDate": "2026-01-30T11:33:44.503Z",
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "Content",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "K12PHY2 - 06 SEMICONDUCTOR ELECTRONICS MATERIALS, DEVICES AND SIMPLE CIRCUITS.pdf",
            "col3": "14.1 I NTRODUCTION Devices in which a controlled flow of electrons can be obtained are the basic building blocks of all the electronic circuits. Before the discovery of transistor in 1948, such devices were mostly vacuum tubes (also called valves) like the vacuum diode which has two electrodes, viz., anode (often called plate) and cathode; triode which has three electrodes – cathode, plate and grid; tetrode and pentode (respectively with 4 and 5 electrodes). In a vacuum tube, the electrons are supplied by a heated cathode and the controlled flow of these electrons in vacuum is obtained by varying the voltage between its different electrodes. Vacuum is required in the inter-electrode space; otherwise the moving electrons may lose their energy on collision with the air molecules in their path. In these devices the electrons can flow only from the cathode to the anode (i.e., only in one direction). Therefore, such devices are generally referred to as valves . These vacuum tube devices are bulky, consume high power, operate generally at high voltages (~100 V) and have limited life and low reliability. The seed of the development of modern solid-state semiconductor electronics goes back to 1930’s when it was realised that some solid- state semiconductors and their junctions offer the possibility of controlling the number and the direction of flow of charge carriers through them. Simple excitations like light, heat or small applied voltage can change the number of mobile charges in a semiconductor. Note that the supply Chapter Fourteen SEMICONDUCTOR ELECTRONICS: MATERIALS, DEVICES AND SIMPLE CIRCUITS Reprint 2025-26Physics 324 and flow of charge carriers in the semiconductor devices are within the solid itself , while in the earlier vacuum tubes/valves, the mobile electrons were obtained from a heated cathode and they were made to flow in an evacuated space or vacuum. No external heating or large evacuated space is required by the semiconductor devices. They are small in size, consume low power, operate at low voltages and have long life and high reliability. Even the Cathode Ray Tubes (CRT) used in television and computer monitors which work on the principle of vacuum tubes are being replaced by Liquid Crystal Display (LCD) monitors with supporting solid state electronics. Much before the full implications of the semiconductor devices was formally understood, a naturally occurring crystal of galena (Lead sulphide, PbS) with a metal point contact attached to it was used as detector of radio waves. In the following sections, we will introduce the basic concepts of semiconductor physics and discuss some semiconductor devices like junction diodes (a 2-electrode device) and bipolar junction transistor (a 3-electrode device). A few circuits illustrating their applications will also be described. 14.2 C LASSIFICATION OF M ETALS , C ONDUCTORS AND S EMICONDUCTORS On the basis of conductivity On the basis of the relative values of electrical conductivity ( s ) or resistivity ( r = 1/ s ), the solids are broadly classified as: (i) Metals: They possess very low resistivity (or high conductivity). r ~ 10 –2 – 10 –8 W m s ~ 10 2 – 10 8 S m –1 (ii) Semiconductors: They have resistivity or conductivity intermediate to metals and insulators. r ~ 10 –5 – 10 6 W m s ~ 10 5 – 10 –6 S m –1 (iii) Insulators: They have high resistivity (or low conductivity). r ~ 10 11 – 10 19 W m s ~ 10 –11 – 10 –19 S m –1 The values of r and s given above are indicative of magnitude and could well go outside the ranges as well. Relative values of the resistivity are not the only criteria for distinguishing metals, insulators and semiconductors from each other. There are some other differences, which will become clear as we go along in this chapter. Our interest in this chapter is in the study of semiconductors which could be: (i) Elemental semiconductors : Si and Ge (ii) Compound semiconductors : Examples are: ∑ Inorganic: CdS, GaAs, CdSe, InP, etc. ∑ Organic: anthracene, doped pthalocyanines, etc. ∑ Organic polymers: polypyrrole, polyaniline, polythiophene, etc. Most of the currently available semiconductor devices are based on elemental semiconductors Si or Ge and compound inorganic semiconductors. However, after 1990, a few semiconductor devices using Reprint 2025-26325 Semiconductor Electronics: Materials, Devices and Simple Circuits organic semiconductors and semiconducting polymers have been developed signalling the birth of a futuristic technology of polymer- electronics and molecular-electronics. In this chapter, we will restrict ourselves to the study of inorganic semiconductors, particularly elemental semiconductors Si and Ge. The general concepts introduced here for discussing the elemental semiconductors, by-and-large, apply to most of the compound semiconductors as well. On the basis of energy bands According to the Bohr atomic model, in an isolated atom the energy of any of its electrons is decided by the orbit in which it revolves. But when the atoms come together to form a solid they are close to each other. So the outer orbits of electrons from neighbouring atoms would come very close or could even overlap. This would make the nature of electron motion in a solid very different from that in an isolated atom. Inside the crystal each electron has a unique position and no two electrons see exactly the same pattern of surrounding charges. Because of this, each electron will have a different energy level . These different energy levels with continuous energy variation form what are called energy bands . The energy band which includes the energy levels of the valence electrons is called the valence band . The energy band above the valence band is called the conduction band . With no external energy, all the valence electrons will reside in the valence band. If the lowest level in the conduction band happens to be lower than the highest level of the valence band, the electrons from the valence band can easily move into the conduction band. Normally the conduction band is empty. But when it overlaps on the valence band electrons can move freely into it. This is the case with metallic conductors. If there is some gap between the conduction band and the valence band, electrons in the valence band all remain bound and no free electrons are available in the conduction band. This makes the material an insulator. But some of the electrons from the valence band may gain external energy to cross the gap between the conduction band and the valence band. Then these electrons will move into the conduction band. At the same time they will create vacant energy levels in the valence band where other valence electrons can move. Thus the process creates the possibility of conduction due to electrons in conduction band as well as due to vacancies in the valence band. Let us consider what happens in the case of Si or Ge crystal containing N atoms. For Si, the outermost orbit is the third orbit ( n = 3), while for Ge it is the fourth orbit ( n = 4). The number of electrons in the outermost orbit is 4 (2 s and 2 p electrons). Hence, the total number of outer electrons in the crystal is 4 N . The maximum possible number of electrons in the outer orbit is 8 (2 s + 6 p electrons). So, for the 4 N valence electrons there are 8 N available energy states. These 8 N discrete energy levels can either form a continuous band or they may be grouped in different bands depending upon the distance between the atoms in the crystal (see box on Band Theory of Solids). At the distance between the atoms in the crystal lattices of Si and Ge, the energy band of these 8 N states is split apart into two which are separated by an energy gap E g (Fig. 14.1). The lower band which is completely occupied by the 4 N valence electrons at temperature of absolute zero is the valence band. The other band consisting of 4 N energy states, called the conduction band , is completely empty at absolute zero. Reprint 2025-26Physics 326 The lowest energy level in the conduction band is shown as E C and highest energy level in the valence band is shown as E V . Above E C and below E V there are a large number of closely spaced energy levels, as shown in Fig. 14.1. The gap between the top of the valence band and bottom of the conduction band is called the energy band gap (Energy gap E g ). It may be large, small, or zero, depending upon the material. These different situations, are depicted in Fig. 14.2 and discussed below: Case I: This refers to a situation, as shown in Fig. 14.2(a). One can have a metal either when the conduction band is partially filled and the balanced band is partially empty or when the conduction and valance bands overlap. When there is overlap electrons from valence band can easily move into the conduction band. This situation makes a large number of electrons available for electrical conduction. When the valence band is partially empty, electrons from its lower level can move to higher level making conduction possible. Therefore, the resistance of such materials is low or the conductivity is high. FIGURE 14.2 Difference between energy bands of (a) metals, (b) insulators and (c) semiconductors. FIGURE 14.1 The energy band positions in a semiconductor at 0 K. The upper band, called the conduction band, consists of infinitely large number of closely spaced energy states. The lower band, called the valence band, consists of closely spaced completely filled energy states. Reprint 2025-26327 Semiconductor Electronics: Materials, Devices and Simple Circuits Case II: In this case, as shown in Fig. 14.2(b), a large band gap E g exists ( E g > 3 eV). There are no electrons in the conduction band, and therefore no electrical conduction is possible. Note that the energy gap is so large that electrons cannot be excited from the valence band to the conduction band by thermal excitation. This is the case of insulators . Case III: This situation is shown in Fig. 14.2(c). Here a finite but small band gap ( E g < 3 eV) exists. Because of the small band gap, at room temperature some electrons from valence band can acquire enough energy to cross the energy gap and enter the conduction band . These electrons (though small in numbers) can move in the conduction band. Hence, the resistance of semiconductors is not as high as that of the insulators. In this section we have made a broad classification of metals, conductors and semiconductors. In the section which follows you will learn the conduction process in semiconductors. 14.3 I NTRINSIC S EMICONDUCTOR We shall take the most common case of Ge and Si whose lattice structure is shown in Fig. 14.3. These structures are called the diamond-like structures. Each atom is surrounded by four nearest neighbours. We know that Si and Ge have four valence electrons. In its crystalline structure, every Si or Ge atom tends to share one of its four valence electrons with each of its four nearest neighbour atoms, and also to take share of one electron from each such neighbour. These shared electron pairs are referred to as forming a covalent bond or simply a valence bond . The two shared electrons can be assumed to shuttle back-and-forth between the associated atoms holding them together strongly. Figure 14.4 schematically shows the 2-dimensional representation of Si or Ge structure shown in Fig. 14.3 which overemphasises the covalent bond. It shows an idealised picture in which no bonds are broken (all bonds are intact). Such a situation arises at low temperatures. As the temperature increases, more thermal energy becomes available to these electrons and some of these electrons may break–away (becoming free electrons contributing to conduction). The thermal energy effectively ionises only a few atoms in the crystalline lattice and creates a vacancy in the bond as shown in Fig. 14.5(a). The neighbourhood, from which the free electron (with charge – q ) has come out leaves a vacancy with an effective charge (+ q ). This vacancy with the effective positive electronic charge is called a hole . The hole behaves as an apparent free particle with effective positive charge. In intrinsic semiconductors, the number of free electrons, n e is equal to the number of holes, n h . That is n e = n h = n i (14.1) where n i is called intrinsic carrier concentration. Semiconductors posses the unique property in which, apart from electrons, the holes also move. Suppose there is a hole at site 1 as shown FIGURE 14.3 Three-dimensional dia- mond-like crystal structure for Carbon, Silicon or Germanium with respective lattice spacing a equal to 3.56, 5.43 and 5.66 Å. Reprint 2025-26Physics 328 in Fig. 14.5(a). The movement of holes can be visualised as shown in Fig. 14.5(b). An electron from the covalent bond at site 2 may jump to the vacant site 1 (hole). Thus, after such a jump, the hole is at site 2 and the site 1 has now an electron. Therefore, apparently, the hole has moved from site 1 to site 2. Note that the electron originally set free [Fig. 14.5(a)] is not involved in this process of hole motion. The free electron moves completely independently as conduction electron and gives rise to an electron current, I e under an applied electric field. Remember that the motion of hole is only a convenient way of describing the actual motion of bound electrons, whenever there is an empty bond anywhere in the crystal. Under the action of an electric field, these holes move towards negative potential giving the hole current, I h . The total current, I is thus the sum of the electron current I e and the hole current I h : I = I e + I h (14.2) It may be noted that apart from the process of generation of conduction electrons and holes, a simultaneous process of recombination occurs in which the electrons recombine with the holes. At equilibrium, the rate of generation is equal to the rate of recombination of charge carriers. The recombination occurs due to an electron colliding with a hole. FIGURE 14.4 Schematic two-dimensional representation of Si or Ge structure showing covalent bonds at low temperature (all bonds intact). +4 symbol indicates inner cores of Si or Ge. FIGURE 14.5 (a) Schematic model of generation of hole at site 1 and conduction electron due to thermal energy at moderate temperatures. (b) Simplified representation of possible thermal motion of a hole. The electron from the lower left hand covalent bond (site 2) goes to the earlier hole site1, leaving a hole at its site indicating an apparent movement of the hole from site 1 to site 2. (a) (b) Reprint 2025-26329 Semiconductor Electronics: Materials, Devices and Simple Circuits E XAMPLE 14.1 An intrinsic semiconductor will behave like an insulator at T = 0 K as shown in Fig. 14.6(a). It is the thermal energy at higher temperatures ( T > 0K), which excites some electrons from the valence band to the conduction band. These thermally excited electrons at T > 0 K, partially occupy the conduction band. Therefore, the energy-band diagram of an intrinsic semiconductor will be as shown in Fig. 14.6(b). Here, some electrons are shown in the conduction band. These have come from the valence band leaving equal number of holes there. Example 14.1 C, Si and Ge have same lattice structure. Why is C insulator while Si and Ge intrinsic semiconductors? Solution The 4 bonding electrons of C, Si or Ge lie, respectively, in the second, third and fourth orbit. Hence, energy required to take out an electron from these atoms (i.e., ionisation energy E g ) will be least for Ge, followed by Si and highest for C. Hence, number of free electrons for conduction in Ge and Si are significant but negligibly small for C. 14.4 E XTRINSIC S EMICONDUCTOR The conductivity of an intrinsic semiconductor depends on its temperature, but at room temperature its conductivity is very low. As such, no important electronic devices can be developed using these semiconductors. Hence there is a necessity of improving their conductivity. This can be done by making use of impurities. When a small amount, say, a few parts per million (ppm), of a suitable impurity is added to the pure semiconductor, the conductivity of the semiconductor is increased manifold. Such materials are known as e xtrinsic semiconductors or impurity semiconductors . The deliberate addition of a desirable impurity is called doping and the impurity atoms are called dopants . Such a material is also called a doped semiconductor . The dopant has to be such that it does not distort the original pure semiconductor lattice. It occupies only a very few of the original semiconductor atom sites in the crystal. A necessary condition to attain this is that the sizes of the dopant and the semiconductor atoms should be nearly the same. There are two types of dopants used in doping the tetravalent Si or Ge: (i) Pentavalent (valency 5); like Arsenic (As), Antimony (Sb), Phosphorous (P), etc. FIGURE 14.6 (a) An intrinsic semiconductor at T = 0 K behaves like insulator. (b) At T > 0 K, four thermally generated electron-hole pairs. The filled circles ( ) represent electrons and empty circles ( ) represent holes. Reprint 2025-26Physics 330 (ii) Trivalent (valency 3); like Indium (In), Boron (B), Aluminium (Al), etc. We shall now discuss how the doping changes the number of charge carriers (and hence the conductivity) of semiconductors. Si or Ge belongs to the fourth group in the Periodic table and, therefore, we choose the dopant element from nearby fifth or third group, expecting and taking care that the size of the dopant atom is nearly the same as that of Si or Ge. Interestingly, the pentavalent and trivalent dopants in Si or Ge give two entirely different types of semiconductors as discussed below. (i) n-type semiconductor Suppose we dope Si or Ge with a pentavalent element as shown in Fig. 14.7. When an atom of +5 valency element occupies the position of an atom in the crystal lattice of Si, four of its electrons bond with the four silicon neighbours while the fifth remains very weakly bound to its parent atom. This is because the four electrons participating in bonding are seen as part of the effective core of the atom by the fifth electron. As a result the ionisation energy required to set this electron free is very small and even at room temperature it will be free to move in the lattice of the semiconductor. For example, the energy required is ~ 0.01 eV for germanium, and 0.05 eV for silicon, to separate this electron from its atom. This is in contrast to the energy required to jump the forbidden band (about 0.72 eV for germanium and about 1.1 eV for silicon) at room temperature in the intrinsic semiconductor. Thus, the pentavalent dopant is donating one extra electron for conduction and hence is known as donor impurity. The number of electrons made available for conduction by dopant atoms depends strongly upon the doping level and is independent of any increase in ambient temperature. On the other hand, the number of free electrons (with an equal number of holes) generated by Si atoms, increases weakly with temperature. In a doped semiconductor the total number of conduction electrons n e is due to the electrons contributed by donors and those generated intrinsically, while the total number of holes n h is only due to the holes from the intrinsic source. But the rate of recombination of holes would increase due to the increase in the number of electrons. As a result, the number of holes would get reduced further. Thus, with proper level of doping the number of conduction electrons can be made much larger than the number of holes. Hence in an extrinsic FIGURE 14.7 (a) Pentavalent donor atom (As, Sb, P, etc.) doped for tetravalent Si or Ge giving n- type semiconductor, and (b) Commonly used schematic representation of n-type material which shows only the fixed cores of the substituent donors with one additional effective positive charge and its associated extra electron. Reprint 2025-26331 Semiconductor Electronics: Materials, Devices and Simple Circuits semiconductor doped with pentavalent impurity, electrons become the majority carriers and holes the minority carriers . These semiconductors are, therefore, known as n -type semiconductors . For n-type semiconductors, we have, n e >> n h (14.3) (ii) p-type semiconductor This is obtained when Si or Ge is doped with a trivalent impurity like Al, B, In, etc. The dopant has one valence electron less than Si or Ge and, therefore, this atom can form covalent bonds with neighbouring three Si atoms but does not have any electron to offer to the fourth Si atom. So the bond between the fourth neighbour and the trivalent atom has a vacancy or hole as shown in Fig. 14.8. Since the neighbouring Si atom in the lattice wants an electron in place of a hole, an electron in the outer orbit of an atom in the neighbourhood may jump to fill this vacancy, leaving a vacancy or hole at its own site. Thus the hole is available for conduction. Note that the trivalent foreign atom becomes effectively negatively charged when it shares fourth electron with neighbouring Si atom. Therefore, the dopant atom of p-type material can be treated as core of one negative charge along with its associated hole as shown in Fig. 14.8(b). It is obvious that one acceptor atom gives one hole . These holes are in addition to the intrinsically generated holes while the source of conduction electrons is only intrinsic generation. Thus, for such a material, the holes are the majority carriers and electrons are minority carriers. Therefore, extrinsic semiconductors doped with trivalent impurity are called p -type semiconductors . For p-type semiconductors, the recombination process will reduce the number ( n i )of intrinsically generated electrons to n e . We have, for p-type semiconductors n h >> n e (14.4) Note that the crystal maintains an overall charge neutrality as the charge of additional charge carriers is just equal and opposite to that of the ionised cores in the lattice. In extrinsic semiconductors, because of the abundance of majority current carriers, the minority carriers produced thermally have more chance of meeting majority carriers and thus getting destroyed. Hence, the dopant, by adding a large number of current carriers of one type, which become the majority carriers, indirectly helps to reduce the intrinsic concentration of minority carriers. The semiconductor’s energy band structure is affected by doping. In the case of extrinsic semiconductors, additional energy states due to donor impurities ( E D ) and acceptor impurities ( E A ) also exist. In the energy band diagram of n-type Si semiconductor, the donor energy level E D is slightly below the bottom E C of the conduction band and electrons from this level move into the conduction band with very small supply of energy. At room temperature, most of the donor atoms get ionised but very few (~10 12 ) atoms of Si get ionised. So the conduction band will have most electrons coming from the donor impurities, as shown in Fig. 14.9(a). Similarly, FIGURE 14.8 (a) Trivalent acceptor atom (In, Al, B etc.) doped in tetravalent Si or Ge lattice giving p-type semicon- ductor. (b) Commonly used schematic representation of p-type material which shows only the fixed core of the substituent acceptor with one effective additional negative charge and its associated hole. Reprint 2025-26Physics 332 E XAMPLE 14.2 for p-type semiconductor, the acceptor energy level E A is slightly above the top E V of the valence band as shown in Fig. 14.9(b). With very small supply of energy an electron from the valence band can jump to the level E A and ionise the acceptor negatively. (Alternately, we can also say that with very small supply of energy the hole from level E A sinks down into the valence band. Electrons rise up and holes fall down when they gain external energy.) At room temperature, most of the acceptor atoms get ionised leaving holes in the valence band. Thus at room temperature the density of holes in the valence band is predominantly due to impurity in the extrinsic semiconductor. The electron and hole concentration in a semiconductor in thermal equilibrium is given by n e n h = n i 2 (14.5) Though the above description is grossly approximate and hypothetical, it helps in understanding the difference between metals, insulators and semiconductors (extrinsic and intrinsic) in a simple manner. The difference in the resistivity of C, Si and Ge depends upon the energy gap between their conduction and valence bands. For C (diamond), Si and Ge, the energy gaps are 5.4 eV, 1.1 eV and 0.7 eV, respectively. Sn also is a group IV element but it is a metal because the energy gap in its case is 0 eV. FIGURE 14.9 Energy bands of (a) n-type semiconductor at T > 0K, (b) p-type semiconductor at T > 0K. Example 14.2 Suppose a pure Si crystal has 5 × 10 28 atoms m –3 . It is doped by 1 ppm concentration of pentavalent As. Calculate the number of electrons and holes. Given that n i =1.5 × 10 16 m –3 . Solution Note that thermally generated electrons ( n i ~10 16 m –3 ) are negligibly small as compared to those produced by doping. Therefore, n e ª ª ª ª ª N D . Since n e n h = n i 2 , The number of holes n h = (2.25 × 10 32 )/(5 ×10 22 ) ~ 4.5 × 10 9 m –3 Reprint 2025-26333 Semiconductor Electronics: Materials, Devices and Simple Circuits 14.5 p-n J UNCTION A p-n junction is the basic building block of many semiconductor devices like diodes, transistor, etc. A clear understanding of the junction behaviour is important to analyse the working of other semiconductor devices. We will now try to understand how a junction is formed and how the junction behaves under the influence of external applied voltage (also called bias ). 14.5.1 p-n junction formation Consider a thin p-type silicon (p-Si) semiconductor wafer. By adding precisely a small quantity of pentavelent impurity, part of the p-Si wafer can be converted into n-Si. There are several processes by which a semiconductor can be formed. The wafer now contains p-region and n-region and a metallurgical junction between p-, and n- region. Two important processes occur during the formation of a p-n junction: diffusion and drift . We know that in an n-type semiconductor, the concentration of electrons (number of electrons per unit volume) is more compared to the concentration of holes. Similarly, in a p-type semiconductor, the concentration of holes is more than the concentration of electrons. During the formation of p-n junction, and due to the concentration gradient across p-, and n- sides, holes diffuse from p-side to n-side (p Æ n) and electrons diffuse from n-side to p-side (n Æ p). This motion of charge carries gives rise to diffusion current across the junction. When an electron diffuses from n Æ p, it leaves behind an ionised donor on n-side. This ionised donor (positive charge) is immobile as it is bonded to the surrounding atoms. As the electrons continue to diffuse from n Æ p, a layer of positive charge (or positive space-charge region) on n-side of the junction is developed. Similarly, when a hole diffuses from p Æ n due to the concentration gradient, it leaves behind an ionised acceptor (negative charge) which is immobile. As the holes continue to diffuse, a layer of negative charge (or negative space-charge region) on the p-side of the junction is developed. This space-charge region on either side of the junction together is known as depletion region as the electrons and holes taking part in the initial movement across the junction depleted the region of its free charges (Fig. 14.10). The thickness of depletion region is of the order of one-tenth of a micrometre. Due to the positive space-charge region on n-side of the junction and negative space charge region on p-side of the junction, an electric field directed from positive charge towards negative charge develops. Due to this field, an electron on p-side of the junction moves to n-side and a hole on n-side of the junction moves to p- side. The motion of charge carriers due to the electric field is called drift. Thus a drift current, which is opposite in direction to the diffusion current (Fig. 14.10) starts. FIGURE 14.10 p-n junction formation process. Formation and working of p-n junction diode http://hyperphysics.phy-astr.gsu.edu/hbase/solids/pnjun.html Reprint 2025-26Physics 334 E XAMPLE 14.3 Initially, diffusion current is large and drift current is small. As the diffusion process continues, the space-charge regions on either side of the junction extend, thus increasing the electric field strength and hence drift current. This process continues until the diffusion current equals the drift current. Thus a p-n junction is formed. In a p-n junction under equilibrium there is no net current. The loss of electrons from the n-region and the gain of electron by the p-region causes a difference of potential across the junction of the two regions. The polarity of this potential is such as to oppose further flow of carriers so that a condition of equilibrium exists. Figure 14.11 shows the p-n junction at equilibrium and the potential across the junction. The n-material has lost electrons, and p material has acquired electrons. The n material is thus positive relative to the p material. Since this potential tends to prevent the movement of electron from the n region into the p region, it is often called a barrier potential . Example 14.3 Can we take one slab of p-type semiconductor and physically join it to another n-type semiconductor to get p-n junction? Solution No! Any slab, howsoever flat, will have roughness much larger than the inter-atomic crystal spacing (~2 to 3 Å) and hence continuous contact at the atomic level will not be possible. The junction will behave as a discontinuity for the flowing charge carriers. 14.6 S EMICONDUCTOR D IODE A semiconductor diode [Fig. 14.12(a)] is basically a p-n junction with metallic contacts provided at the ends for the application of an external voltage. It is a two terminal device. A p-n junction diode is symbolically represented as shown in Fig. 14.12(b). The direction of arrow indicates the conventional direction of current (when the diode is under forward bias). The equilibrium barrier potential can be altered by applying an external voltage V across the diode. The situation of p-n junction diode under equilibrium (without bias) is shown in Fig. 14.11(a) and (b). 14.6.1 p-n junction diode under forward bias When an external voltage V is applied across a semiconductor diode such that p-side is connected to the positive terminal of the battery and n-side to the negative terminal [Fig. 14.13(a)], it is said to be forward biased . The applied voltage mostly drops across the depletion region and the voltage drop across the p-side and n-side of the junction is negligible. (This is because the resistance of the depletion region – a region where there are no charges – is very high compared to the resistance of n-side and p-side.) The direction of the applied voltage ( V ) is opposite to the FIGURE 14.11 (a) Diode under equilibrium (V = 0), (b) Barrier potential under no bias. FIGURE 14.12 (a) Semiconductor diode, (b) Symbol for p-n junction diode. n p Reprint 2025-26335 Semiconductor Electronics: Materials, Devices and Simple Circuits built-in potential V 0 . As a result, the depletion layer width decreases and the barrier height is reduced [Fig. 14.13(b)]. The effective barrier height under forward bias is ( V 0 – V ). If the applied voltage is small, the barrier potential will be reduced only slightly below the equilibrium value, and only a small number of carriers in the material—those that happen to be in the uppermost energy levels—will possess enough energy to cross the junction. So the current will be small. If we increase the applied voltage significantly, the barrier height will be reduced and more number of carriers will have the required energy. Thus the current increases. Due to the applied voltage, electrons from n-side cross the depletion region and reach p-side (where they are minority carries). Similarly, holes from p-side cross the junction and reach the n-side (where they are minority carries). This process under forward bias is known as minority carrier injection. At the junction boundary, on each side, the minority carrier concentration increases significantly compared to the locations far from the junction. Due to this concentration gradient, the injected electrons on p-side diffuse from the junction edge of p-side to the other end of p-side. Likewise, the injected holes on n-side diffuse from the junction edge of n-side to the other end of n-side (Fig. 14.14). This motion of charged carriers on either side gives rise to current. The total diode forward current is sum of hole diffusion current and conventional current due to electron diffusion. The magnitude of this current is usually in mA. 14.6.2 p-n junction diode under reverse bias When an external voltage ( V ) is applied across the diode such that n-side is positive and p-side is negative, it is said to be reverse biased [Fig.14.15(a)]. The applied voltage mostly drops across the depletion region. The direction of applied voltage is same as the direction of barrier potential. As a result, the barrier height increases and the depletion region widens due to the change in the electric field. The effective barrier height under reverse bias is ( V 0 + V ), [Fig. 14.15(b)]. This suppresses the flow of electrons from n Æ p and holes from p Æ n. Thus, diffusion current, decreases enormously compared to the diode under forward bias. The electric field direction of the junction is such that if electrons on p-side or holes on n-side in their random motion come close to the junction, they will be swept to its majority zone. This drift of carriers gives rise to current. The drift current is of the order of a few m A. This is quite low because it is due to the motion of carriers from their minority side to their majority side across the junction. The drift current is also there under forward bias but it is negligible ( m A) when compared with current due to injected carriers which is usually in mA. The diode reverse current is not very much dependent on the applied voltage. Even a small voltage is sufficient to sweep the minority carriers from one side of the junction to the other side of the junction. The current FIGURE 14.13 (a) p-n junction diode under forward bias, (b) Barrier potential (1) without battery, (2) Low battery voltage, and (3) High voltage battery. FIGURE 14.14 Forward bias minority carrier injection. Reprint 2025-26Physics 336 is not limited by the magnitude of the applied voltage but is limited due to the concentration of the minority carrier on either side of the junction. The current under reverse bias is essentially voltage independent upto a critical reverse bias voltage, known as breakdown voltage ( V br ). When V = V br , the diode reverse current increases sharply. Even a slight increase in the bias voltage causes large change in the current. If the reverse current is not limited by an external circuit below the rated value (specified by the manufacturer) the p-n junction will get destroyed. Once it exceeds the rated value, the diode gets destroyed due to overheating. This can happen even for the diode under forward bias, if the forward current exceeds the rated value. The circuit arrangement for studying the V - I characteristics of a diode, (i.e., the variation of current as a function of applied voltage) are shown in Fig. 14.16(a) and (b). The battery is connected to the diode through a potentiometer (or reheostat) so that the applied voltage to the diode can be changed. For different values of voltages, the value of the current is noted. A graph between V and I is obtained as in Fig. 14.16(c). Note that in forward bias measurement, we use a milliammeter since the expected current is large (as explained in the earlier section) while a micrometer is used in reverse bias to measure the current. You can see in Fig. 14.16(c) that in forward FIGURE 14.15 (a) Diode under reverse bias, (b) Barrier potential under reverse bias. FIGURE 14.16 Experimental circuit arrangement for studying V - I characteristics of a p-n junction diode (a) in forward bias, (b) in reverse bias. (c) Typical V - I characteristics of a silicon diode. Reprint 2025-26337 Semiconductor Electronics: Materials, Devices and Simple Circuits E XAMPLE 14.4 bias, the current first increases very slowly, almost negligibly, till the voltage across the diode crosses a certain value. After the characteristic voltage, the diode current increases significantly (exponentially), even for a very small increase in the diode bias voltage. This voltage is called the threshold voltage or cut-in voltage (~0.2V for germanium diode and ~0.7 V for silicon diode). For the diode in reverse bias, the current is very small (~ m A) and almost remains constant with change in bias. It is called reverse saturation current . However, for special cases, at very high reverse bias (break down voltage), the current suddenly increases. This special action of the diode is discussed later in Section 14.8. The general purpose diode are not used beyond the reverse saturation current region. The above discussion shows that the p-n junction diode primerly allows the flow of current only in one direction (forward bias). The forward bias resistance is low as compared to the reverse bias resistance. This property is used for rectification of ac voltages as discussed in the next section. For diodes, we define a quantity called dynamic resistance as the ratio of small change in voltage D V to a small change in current D I: d V r I ∆ = ∆ (14.6) Example 14.4 The V-I characteristic of a silicon diode is shown in the Fig. 14.17. Calculate the resistance of the diode at (a) I D = 15 mA and (b) V D = –10 V. FIGURE 14.17 Solution Considering the diode characteristics as a straight line between I = 10 mA to I = 20 mA passing through the origin, we can calculate the resistance using Ohm’s law. (a) From the curve, at I = 20 mA, V = 0.8 V; I = 10 mA, V = 0.7 V r f b = D V / D I = 0.1V/10 mA = 10 W (b) From the curve at V = –10 V, I = –1 m A, Therefore, r rb = 10 V/1 m A= 1.0 × 10 7 W Reprint 2025-26Physics 338 14.7 A PPLICATION OF J UNCTION D IODE AS A R ECTIFIER From the V-I characteristic of a junction diode we see that it allows current to pass only when it is forward biased. So if an alternating voltage is applied across a diode the current flows only in that part of the cycle when the diode is forward biased. This property is used to rectify alternating voltages and the circuit used for this purpose is called a rectifier . If an alternating voltage is applied across a diode in series with a load, a pulsating voltage will appear across the load only during the half cycles of the ac input during which the diode is forward biased. Such rectifier circuit, as shown in Fig. 14.18, is called a half-wave rectifier . The secondary of a transformer supplies the desired ac voltage across terminals A and B. When the voltage at A is positive, the diode is forward biased and it conducts. When A is negative, the diode is reverse-biased and it does not conduct. The reverse saturation current of a diode is negligible and can be considered equal to zero for practical purposes. (The reverse breakdown voltage of the diode must be sufficiently higher than the peak ac voltage at the secondary of the transformer to protect the diode from reverse breakdown.) Therefore, in the positive half-cycle of ac there is a current through the load resistor R L and we get an output voltage, as shown in Fig. 14.18(b), whereas there is no current in the negative half- cycle. In the next positive half-cycle, again we get the output voltage. Thus, the output voltage, though still varying, is restricted to only one direction and is said to be rectified . Since the rectified output of this circuit is only for half of the input ac wave it is called as half-wave rectifier . The circuit using two diodes, shown in Fig. 14.19(a), gives output rectified voltage corresponding to both the positive as well as negative half of the ac cycle. Hence, it is known as full-wave rectifier . Here the p-side of the two diodes are connected to the ends of the secondary of the transformer. The n-side of the diodes are connected together and the output is taken between this common point of diodes and the midpoint of the secondary of the transformer. So for a full-wave rectifier the secondary of the transformer is provided with a centre tapping and so it is called centre-tap transformer . As can be seen from Fig.14.19(c) the voltage rectified by each diode is only half the total secondary voltage. Each diode rectifies only for half the cycle, but the two do so for alternate cycles. Thus, the output between their common terminals and the centre- tap of the transformer becomes a full-wave rectifier output. (Note that there is another circuit of full wave rectifier which does not need a centre- tap transformer but needs four diodes.) Suppose the input voltage to A FIGURE 14.18 (a) Half-wave rectifier circuit, (b) Input ac voltage and output voltage waveforms from the rectifier circuit. Reprint 2025-26339 Semiconductor Electronics: Materials, Devices and Simple Circuits with respect to the centre tap at any instant is positive. It is clear that, at that instant, voltage at B being out of phase will be negative as shown in Fig.14.19(b). So, diode D 1 gets forward biased and conducts (while D 2 being reverse biased is not conducting). Hence, during this positive half cycle we get an output current (and a output voltage across the load resistor R L ) as shown in Fig.14.19(c). In the course of the ac cycle when the voltage at A becomes negative with respect to centre tap, the voltage at B would be positive. In this part of the cycle diode D 1 would not conduct but diode D 2 would, giving an output current and output voltage (across R L ) during the negative half cycle",
            "col4": "DATA"
        },
        {
            "col1": "P1_Part_2",
            "col2": "K12PHY2 - 06 SEMICONDUCTOR ELECTRONICS MATERIALS, DEVICES AND SIMPLE CIRCUITS.pdf",
            "col3": "of the input ac. Thus, we get output voltage during both the positive as well as the negative half of the cycle. Obviously, this is a more efficient circuit for getting rectified voltage or current than the half- wave rectifier. The rectified voltage is in the form of pulses of the shape of half sinusoids. Though it is unidirectional it does not have a steady value. To get steady dc output from the pulsating voltage normally a capacitor is connected across the output terminals (parallel to the load R L ). One can also use an inductor in series with R L for the same purpose. Since these additional circuits appear to filter out the ac ripple and give a pure dc voltage, so they are called filters. Now we shall discuss the role of capacitor in filtering. When the voltage across the capacitor is rising, it gets charged. If there is no external load, it remains charged to the peak voltage of the rectified output. When there is a load, it gets discharged through the load and the voltage across it begins to fall. In the next half-cycle of rectified output it again gets charged to the peak value (Fig. 14.20). The rate of fall of the voltage across the capacitor depends inversely upon the product of capacitance C and the effective resistance R L used in the circuit and is called the time constant . To make the time constant large value of C should be large. So capacitor input filters use large capacitors. The output voltage obtained by using capacitor input filter is nearer to the peak voltage of the rectified voltage. This type of filter is most widely used in power supplies. FIGURE 14.19 (a) A Full-wave rectifier circuit; (b) Input wave forms given to the diode D 1 at A and to the diode D 2 at B; (c) Output waveform across the load R L connected in the full-wave rectifier circuit. Reprint 2025-26Physics 340 FIGURE 14.20 (a) A full-wave rectifier with capacitor filter, (b) Input and output voltage of rectifier in (a). Reprint 2025-26",
            "col4": "DATA"
        }
    ],
    "projectDeleted": false,
    "codeFrameColumn": [
        {
            "col1": "col1",
            "col2": "col3",
            "researchObjective": "",
            "filterColumn": [],
            "title": "K12PHY2 - 06 SEMICONDUCTOR ELECTRONICS MATERIALS, DEVICES AND SIMPLE CIRCUITS",
            "netSetting": true,
            "subNetSetting": true,
            "categorySetting": true,
            "sentimentSetting": true,
            "thinkingBudgetSetting": 2000,
            "questionType": "OPENEND",
            "otherBrand": false,
            "existingCodeFrame": [],
            "isList": false,
            "codeFrameWithDataList": [],
            "codeFrameList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 10620,
                        "CandidatesTokenCount": 3398,
                        "TotalTokenCount": 15438,
                        "thoughtsTokenCount": 1420,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Net": "Introduction to Semiconductor Devices",
                                "Subnets": [
                                    {
                                        "Subnet": "Evolution and Comparison",
                                        "Categories": [
                                            {
                                                "Category": "Vacuum Tube Characteristics",
                                                "Summary": "Older devices (diodes, triodes, tetrodes, pentodes) that are bulky, consume high power, operate at high voltages, have limited life and low reliability."
                                            },
                                            {
                                                "Category": "Semiconductor Advantages",
                                                "Summary": "Modern solid-state devices that are small in size, consume low power, operate at low voltages, and have long life and high reliability."
                                            },
                                            {
                                                "Category": "Transition from Vacuum Tubes",
                                                "Summary": "Replacement of older technologies like Cathode Ray Tubes (CRTs) by Liquid Crystal Display (LCD) monitors with solid-state electronics."
                                            }
                                        ],
                                        "Summary": "Overview of semiconductor electronics, contrasting it with vacuum tube technology and highlighting the advantages of semiconductors."
                                    },
                                    {
                                        "Subnet": "Basic Principles",
                                        "Categories": [
                                            {
                                                "Category": "Charge Carrier Control",
                                                "Summary": "Ability of solid-state semiconductors and their junctions to control the number and direction of flow of charge carriers."
                                            },
                                            {
                                                "Category": "Excitations for Mobile Charges",
                                                "Summary": "Simple excitations like light, heat, or small applied voltage can change the number of mobile charges in a semiconductor."
                                            },
                                            {
                                                "Category": "Galena Crystal Detector",
                                                "Summary": "Early use of a naturally occurring crystal of galena (Lead sulphide, PbS) with a metal point contact as a detector of radio waves."
                                            }
                                        ],
                                        "Summary": "Fundamental concepts explaining how semiconductors operate by controlling charge carrier flow within the solid."
                                    }
                                ],
                                "Summary": "An overview of semiconductor electronics, comparing it to older vacuum tube technology and introducing its fundamental operating principles."
                            },
                            {
                                "Net": "Material Classification",
                                "Subnets": [
                                    {
                                        "Subnet": "Classification by Conductivity",
                                        "Categories": [
                                            {
                                                "Category": "Metals (High Conductivity)",
                                                "Summary": "Solids with very low resistivity (~10⁻² – 10⁻⁸ Ωm) or high conductivity (~10² – 10⁸ S m⁻¹)."
                                            },
                                            {
                                                "Category": "Semiconductors (Intermediate Conductivity)",
                                                "Summary": "Solids with resistivity or conductivity intermediate to metals and insulators (~10⁻⁵ – 10⁶ Ωm or ~10⁵ – 10⁻⁶ S m⁻¹)."
                                            },
                                            {
                                                "Category": "Insulators (Low Conductivity)",
                                                "Summary": "Solids with high resistivity (~10¹¹ – 10¹⁹ Ωm) or low conductivity (~10⁻¹¹ – 10⁻¹⁹ S m⁻¹)."
                                            }
                                        ],
                                        "Summary": "Categorizing materials based on their electrical conductivity or resistivity values."
                                    },
                                    {
                                        "Subnet": "Classification by Energy Bands",
                                        "Categories": [
                                            {
                                                "Category": "Energy Band Theory",
                                                "Summary": "Concept that in solids, atomic energy levels combine to form continuous energy bands, specifically valence and conduction bands."
                                            },
                                            {
                                                "Category": "Metals (Overlapping Bands)",
                                                "Summary": "Conduction band is partially filled, or it overlaps with the valence band, allowing free movement of electrons for conduction."
                                            },
                                            {
                                                "Category": "Insulators (Large Band Gap)",
                                                "Summary": "Large energy band gap (Eg > 3 eV) between valence and conduction bands, preventing electron excitation and electrical conduction."
                                            },
                                            {
                                                "Category": "Semiconductors (Small Band Gap)",
                                                "Summary": "Finite but small energy band gap (Eg < 3 eV), allowing some electrons to cross to the conduction band at room temperature for limited conduction."
                                            }
                                        ],
                                        "Summary": "Explaining material properties based on the arrangement and energy separation of valence and conduction bands."
                                    },
                                    {
                                        "Subnet": "Types of Semiconductors",
                                        "Categories": [
                                            {
                                                "Category": "Elemental Semiconductors",
                                                "Summary": "Single-element semiconductors like Silicon (Si) and Germanium (Ge)."
                                            },
                                            {
                                                "Category": "Inorganic Compound Semiconductors",
                                                "Summary": "Semiconductors formed from inorganic compounds, such as CdS, GaAs, CdSe, InP."
                                            },
                                            {
                                                "Category": "Organic/Polymer Semiconductors",
                                                "Summary": "Emerging class of semiconductors including organic materials (e.g., anthracene, doped pthalocyanines) and organic polymers (e.g., polypyrrole, polyaniline)."
                                            }
                                        ],
                                        "Summary": "Differentiating between various forms of semiconductor materials based on their chemical composition."
                                    }
                                ],
                                "Summary": "Categorization of materials (metals, semiconductors, insulators) based on their electrical properties and energy band structures, along with the specific types of semiconductor materials."
                            },
                            {
                                "Net": "Intrinsic and Extrinsic Semiconductors",
                                "Subnets": [
                                    {
                                        "Subnet": "Intrinsic Semiconductors",
                                        "Categories": [
                                            {
                                                "Category": "Covalent Bonding",
                                                "Summary": "In Si or Ge crystals, each atom shares four valence electrons with its four nearest neighbors, forming covalent bonds."
                                            },
                                            {
                                                "Category": "Electron-Hole Generation",
                                                "Summary": "At temperatures above 0K, thermal energy excites some electrons from covalent bonds, creating free electrons and vacancies called holes."
                                            },
                                            {
                                                "Category": "Conduction Mechanism",
                                                "Summary": "Total current is the sum of electron current (Ie) from free electrons and hole current (Ih) from the apparent movement of holes."
                                            }
                                        ],
                                        "Summary": "Properties and conduction mechanism of pure semiconductors where electron-hole pairs are generated by thermal energy."
                                    },
                                    {
                                        "Subnet": "Extrinsic (Doped) Semiconductors",
                                        "Categories": [
                                            {
                                                "Category": "Doping Process",
                                                "Summary": "Deliberate addition of a small amount of suitable impurity (dopant) to a pure semiconductor to increase its conductivity."
                                            },
                                            {
                                                "Category": "Dopant Characteristics",
                                                "Summary": "Dopant atoms should not distort the original crystal lattice and should have a size similar to the semiconductor atoms."
                                            },
                                            {
                                                "Category": "Charge Neutrality",
                                                "Summary": "Despite added charge carriers, the overall crystal maintains electrical neutrality, as carrier charges are balanced by ionized dopant cores."
                                            }
                                        ],
                                        "Summary": "The process of adding impurities to semiconductors to enhance their electrical conductivity and create impurity semiconductors."
                                    },
                                    {
                                        "Subnet": "n-type Semiconductors",
                                        "Categories": [
                                            {
                                                "Category": "Pentavalent Dopants",
                                                "Summary": "Impurities from Group 15 (valency 5), like Arsenic (As), Antimony (Sb), Phosphorous (P), which donate an extra electron for conduction."
                                            },
                                            {
                                                "Category": "Donor Impurity",
                                                "Summary": "Pentavalent dopants are called donor impurities because they donate an electron to the conduction band with very small ionization energy."
                                            },
                                            {
                                                "Category": "Majority/Minority Carriers (Electrons)",
                                                "Summary": "In n-type semiconductors, electrons become the majority carriers (ne >> nh) and holes are the minority carriers."
                                            }
                                        ],
                                        "Summary": "Characteristics of semiconductors doped with pentavalent impurities, resulting in excess electrons as majority carriers."
                                    },
                                    {
                                        "Subnet": "p-type Semiconductors",
                                        "Categories": [
                                            {
                                                "Category": "Trivalent Dopants",
                                                "Summary": "Impurities from Group 13 (valency 3), like Indium (In), Boron (B), Aluminium (Al), which create a vacancy or hole."
                                            },
                                            {
                                                "Category": "Acceptor Impurity",
                                                "Summary": "Trivalent dopants are called acceptor impurities because they create a hole, which can accept an electron from the valence band."
                                            },
                                            {
                                                "Category": "Majority/Minority Carriers (Holes)",
                                                "Summary": "In p-type semiconductors, holes become the majority carriers (nh >> ne) and electrons are the minority carriers."
                                            }
                                        ],
                                        "Summary": "Characteristics of semiconductors doped with trivalent impurities, resulting in excess holes as majority carriers."
                                    }
                                ],
                                "Summary": "Detailed explanation of intrinsic semiconductors and how doping with pentavalent or trivalent impurities creates extrinsic n-type and p-type semiconductors, altering their conductivity and carrier concentrations."
                            },
                            {
                                "Net": "p-n Junctions",
                                "Subnets": [
                                    {
                                        "Subnet": "Junction Formation",
                                        "Categories": [
                                            {
                                                "Category": "Diffusion Current",
                                                "Summary": "Charge carriers (holes from p-side, electrons from n-side) move across the junction due to concentration gradient."
                                            },
                                            {
                                                "Category": "Depletion Region",
                                                "Summary": "A region on either side of the junction depleted of free charges, formed by immobile ionized donor and acceptor atoms."
                                            },
                                            {
                                                "Category": "Drift Current",
                                                "Summary": "Motion of charge carriers (electrons from p-side, holes from n-side) due to the electric field created by the space-charge regions, opposite to diffusion current."
                                            }
                                        ],
                                        "Summary": "The process by which a p-n junction is formed, involving diffusion and drift of charge carriers and the creation of a depletion region."
                                    },
                                    {
                                        "Subnet": "Equilibrium Conditions",
                                        "Categories": [
                                            {
                                                "Category": "Barrier Potential (V₀)",
                                                "Summary": "A potential difference developed across the junction due to the separation of charges in the depletion region, opposing further carrier flow."
                                            },
                                            {
                                                "Category": "No Net Current",
                                                "Summary": "At equilibrium, diffusion current equals drift current, resulting in zero net current across the junction."
                                            },
                                            {
                                                "Category": "Immobile Ionized Cores",
                                                "Summary": "Ionized donor atoms (positive) on the n-side and ionized acceptor atoms (negative) on the p-side, fixed within the crystal lattice."
                                            }
                                        ],
                                        "Summary": "Description of a p-n junction at thermal equilibrium without external voltage, detailing the barrier potential and charge balance."
                                    }
                                ],
                                "Summary": "Explanation of how p-n junctions are formed through diffusion and drift currents, leading to a depletion region and a barrier potential at equilibrium."
                            },
                            {
                                "Net": "Semiconductor Diodes",
                                "Subnets": [
                                    {
                                        "Subnet": "Diode Structure and Symbol",
                                        "Categories": [
                                            {
                                                "Category": "p-n Junction with Contacts",
                                                "Summary": "A semiconductor diode is essentially a p-n junction with metallic contacts at the ends for external voltage application."
                                            },
                                            {
                                                "Category": "Two-Terminal Device",
                                                "Summary": "A basic electronic component with two terminals (anode and cathode)."
                                            },
                                            {
                                                "Category": "Diode Symbol",
                                                "Summary": "Represented by an arrow indicating the conventional direction of current flow when forward biased."
                                            }
                                        ],
                                        "Summary": "Physical structure and symbolic representation of a semiconductor diode."
                                    },
                                    {
                                        "Subnet": "Forward Bias",
                                        "Categories": [
                                            {
                                                "Category": "Connection & Voltage Drop",
                                                "Summary": "p-side connected to positive, n-side to negative terminal; applied voltage drops across the depletion region."
                                            },
                                            {
                                                "Category": "Barrier Reduction & Current Flow",
                                                "Summary": "Depletion layer width decreases, barrier height reduces (V₀ - V), allowing significant current (mA) due to minority carrier injection."
                                            },
                                            {
                                                "Category": "Minority Carrier Injection",
                                                "Summary": "Electrons cross to p-side, holes cross to n-side, diffusing away from the junction and contributing to current."
                                            }
                                        ],
                                        "Summary": "Behavior of a p-n junction diode when forward biased, leading to current flow."
                                    },
                                    {
                                        "Subnet": "Reverse Bias",
                                        "Categories": [
                                            {
                                                "Category": "Connection & Voltage Increase",
                                                "Summary": "n-side connected to positive, p-side to negative; applied voltage adds to barrier potential."
                                            },
                                            {
                                                "Category": "Barrier Widening & Low Current",
                                                "Summary": "Depletion layer widens, barrier height increases (V₀ + V), suppressing diffusion current and resulting in very small current (μA) from minority carrier drift."
                                            },
                                            {
                                                "Category": "Breakdown Voltage (Vbr)",
                                                "Summary": "A critical reverse bias voltage where the diode reverse current increases sharply, potentially destroying the junction if current is not limited."
                                            }
                                        ],
                                        "Summary": "Behavior of a p-n junction diode when reverse biased, leading to negligible current flow until breakdown."
                                    },
                                    {
                                        "Subnet": "V-I Characteristics & Resistance",
                                        "Categories": [
                                            {
                                                "Category": "Threshold/Cut-in Voltage",
                                                "Summary": "Characteristic voltage (e.g., ~0.7V for Si) after which forward current increases exponentially with small voltage increase."
                                            },
                                            {
                                                "Category": "Reverse Saturation Current",
                                                "Summary": "Very small, almost constant current (~μA) under reverse bias, largely independent of applied voltage until breakdown."
                                            },
                                            {
                                                "Category": "Dynamic Resistance (r_d)",
                                                "Summary": "Defined as the ratio of a small change in voltage (ΔV) to a small change in current (ΔI) in the operating region."
                                            }
                                        ],
                                        "Summary": "The graphical representation of current versus voltage for a diode, showing its different operating regions and resistance."
                                    }
                                ],
                                "Summary": "A comprehensive look at semiconductor diodes, covering their structure, behavior under forward and reverse bias, and their characteristic V-I curve."
                            },
                            {
                                "Net": "Applications of Junction Diodes",
                                "Subnets": [
                                    {
                                        "Subnet": "Half-Wave Rectifiers",
                                        "Categories": [
                                            {
                                                "Category": "Circuit Configuration",
                                                "Summary": "A single diode connected in series with a load, receiving an AC input from a transformer secondary."
                                            },
                                            {
                                                "Category": "Output Waveform",
                                                "Summary": "Produces a pulsating DC output voltage only during the positive half-cycles of the AC input when the diode is forward biased."
                                            },
                                            {
                                                "Category": "Current Flow Direction",
                                                "Summary": "Allows current to flow only in one direction, effectively rectifying half of the AC waveform."
                                            }
                                        ],
                                        "Summary": "Detailed explanation of half-wave rectification, where only one half-cycle of an AC input is converted to pulsating DC."
                                    },
                                    {
                                        "Subnet": "Full-Wave Rectifiers",
                                        "Categories": [
                                            {
                                                "Category": "Centre-Tap Rectifier Circuit",
                                                "Summary": "Uses two diodes and a center-tapped transformer to rectify both positive and negative half-cycles of the AC input."
                                            },
                                            {
                                                "Category": "Output Waveform",
                                                "Summary": "Provides output voltage during both positive and negative half-cycles of the AC input, resulting in a more efficient rectified output than half-wave."
                                            },
                                            {
                                                "Category": "Bridge Rectifier Alternative",
                                                "Summary": "Mention of another full-wave rectifier circuit that does not require a center-tap transformer but uses four diodes."
                                            }
                                        ],
                                        "Summary": "Detailed explanation of full-wave rectification, converting both halves of an AC input into pulsating DC, typically using a center-tap transformer."
                                    },
                                    {
                                        "Subnet": "Filters for DC Output",
                                        "Categories": [
                                            {
                                                "Category": "Capacitor Filter Function",
                                                "Summary": "A capacitor connected across the output terminals (parallel to the load) charges during voltage rise and discharges through the load during fall, smoothing the pulsating DC."
                                            },
                                            {
                                                "Category": "Inductor Filter Alternative",
                                                "Summary": "An inductor can be used in series with the load for the same purpose of filtering out AC ripples."
                                            },
                                            {
                                                "Category": "Time Constant & Smoothing",
                                                "Summary": "The rate of voltage fall across the capacitor depends inversely on the RC time constant; large capacitance values are used for better smoothing, providing an output nearer to the peak voltage."
                                            }
                                        ],
                                        "Summary": "Components and techniques used to convert the pulsating DC output of rectifiers into a steady DC voltage."
                                    }
                                ],
                                "Summary": "Exploring the practical applications of p-n junction diodes, primarily as rectifiers in power supplies to convert alternating current (AC) to direct current (DC), including half-wave, full-wave rectification, and filtering."
                            },
                            {
                                "Net": "996 Non specific NET",
                                "Subnets": [
                                    {
                                        "Subnet": "Non specific Subnet",
                                        "Categories": [
                                            {
                                                "Category": "Non specific Category",
                                                "Summary": "Responses that are irrelevant, cannot be easily coded into existing themes, or require further upcoding."
                                            }
                                        ],
                                        "Summary": "Captures responses that are irrelevant to the established codeframe themes or require further analysis."
                                    }
                                ],
                                "Summary": "This net is for responses that are irrelevant but could not be categorized into other specific themes easily and may need further upcoding."
                            },
                            {
                                "Net": "997 Don’t Know Can’t Say",
                                "Subnets": [
                                    {
                                        "Subnet": "Don’t Know Can’t Say Subnet",
                                        "Categories": [
                                            {
                                                "Category": "Non-Committal/Vague",
                                                "Summary": "Responses that indicate uncertainty, lack of knowledge, or are generally non-committal or vague."
                                            }
                                        ],
                                        "Summary": "Captures non-committal or vague responses that express uncertainty or lack of information."
                                    }
                                ],
                                "Summary": "This net is for responses that are non-committal and vague, such as 'Don't Know' or 'Can't Say'."
                            },
                            {
                                "Net": "998 Not Applicable",
                                "Subnets": [
                                    {
                                        "Subnet": "Not Applicable Subnet",
                                        "Categories": [
                                            {
                                                "Category": "Irrelevant/Not Applicable",
                                                "Summary": "Responses that are irrelevant or not applicable to the question asked or the project at hand."
                                            }
                                        ],
                                        "Summary": "Captures responses that are irrelevant or not applicable to the specific question or project context."
                                    }
                                ],
                                "Summary": "This net is for responses that are irrelevant and not applicable to the question."
                            },
                            {
                                "Net": "999 None-Nothing",
                                "Subnets": [
                                    {
                                        "Subnet": "None-Nothing Subnet",
                                        "Categories": [
                                            {
                                                "Category": "Gibberish/Meaningless",
                                                "Summary": "Responses consisting of plain gibberish, non-language characters, or expressions that mean nothing."
                                            }
                                        ],
                                        "Summary": "Captures responses that are incomprehensible, contain gibberish, or convey no meaningful information."
                                    }
                                ],
                                "Summary": "This net is for responses that are plain gibberish, non-language characters, or mean nothing."
                            }
                        ],
                        "Summary": "This document provides a comprehensive overview of semiconductor electronics, contrasting them with vacuum tubes and highlighting their advantages. It classifies materials into metals, semiconductors, and insulators based on conductivity and energy band theory. Detailed explanations of intrinsic semiconductors, the generation of electron-hole pairs, and the creation of extrinsic (n-type and p-type) semiconductors through doping are provided. The text further elaborates on the formation and behavior of p-n junctions, including depletion regions and barrier potential, under equilibrium, forward bias, and reverse bias conditions. Finally, it describes the semiconductor diode's V-I characteristics and its application as half-wave and full-wave rectifiers, along with the function of filters to obtain steady DC output."
                    }
                }
            ],
            "onlyCodeFrame": true,
            "errorList": [],
            "codeFrameGenerateType": "CODE_FRAME_ONLY",
            "requestType": "CODEFRAME",
            "requestMode": "DEVELOPMENT",
            "sourceLanguage": "English",
            "targetLanguage": "English",
            "response": {
                "status": "OK",
                "data": {
                    "status": "OK",
                    "PromptTokenCount": 10620,
                    "CandidatesTokenCount": 3398,
                    "TotalTokenCount": 15438,
                    "thoughtsTokenCount": 1420,
                    "finishReason": "STOP",
                    "data": [
                        {
                            "Net": "Introduction to Semiconductor Devices",
                            "Subnets": [
                                {
                                    "Subnet": "Evolution and Comparison",
                                    "Categories": [
                                        {
                                            "Category": "Vacuum Tube Characteristics",
                                            "Summary": "Older devices (diodes, triodes, tetrodes, pentodes) that are bulky, consume high power, operate at high voltages, have limited life and low reliability.",
                                            "code": 3
                                        },
                                        {
                                            "Category": "Semiconductor Advantages",
                                            "Summary": "Modern solid-state devices that are small in size, consume low power, operate at low voltages, and have long life and high reliability.",
                                            "code": 4
                                        },
                                        {
                                            "Category": "Transition from Vacuum Tubes",
                                            "Summary": "Replacement of older technologies like Cathode Ray Tubes (CRTs) by Liquid Crystal Display (LCD) monitors with solid-state electronics.",
                                            "code": 5
                                        }
                                    ],
                                    "Summary": "Overview of semiconductor electronics, contrasting it with vacuum tube technology and highlighting the advantages of semiconductors.",
                                    "code": 2
                                },
                                {
                                    "Subnet": "Basic Principles",
                                    "Categories": [
                                        {
                                            "Category": "Charge Carrier Control",
                                            "Summary": "Ability of solid-state semiconductors and their junctions to control the number and direction of flow of charge carriers.",
                                            "code": 7
                                        },
                                        {
                                            "Category": "Excitations for Mobile Charges",
                                            "Summary": "Simple excitations like light, heat, or small applied voltage can change the number of mobile charges in a semiconductor.",
                                            "code": 8
                                        },
                                        {
                                            "Category": "Galena Crystal Detector",
                                            "Summary": "Early use of a naturally occurring crystal of galena (Lead sulphide, PbS) with a metal point contact as a detector of radio waves.",
                                            "code": 9
                                        }
                                    ],
                                    "Summary": "Fundamental concepts explaining how semiconductors operate by controlling charge carrier flow within the solid.",
                                    "code": 6
                                }
                            ],
                            "Summary": "An overview of semiconductor electronics, comparing it to older vacuum tube technology and introducing its fundamental operating principles.",
                            "code": 1
                        },
                        {
                            "Net": "Material Classification",
                            "Subnets": [
                                {
                                    "Subnet": "Classification by Conductivity",
                                    "Categories": [
                                        {
                                            "Category": "Metals (High Conductivity)",
                                            "Summary": "Solids with very low resistivity (~10⁻² – 10⁻⁸ Ωm) or high conductivity (~10² – 10⁸ S m⁻¹).",
                                            "code": 12
                                        },
                                        {
                                            "Category": "Semiconductors (Intermediate Conductivity)",
                                            "Summary": "Solids with resistivity or conductivity intermediate to metals and insulators (~10⁻⁵ – 10⁶ Ωm or ~10⁵ – 10⁻⁶ S m⁻¹).",
                                            "code": 13
                                        },
                                        {
                                            "Category": "Insulators (Low Conductivity)",
                                            "Summary": "Solids with high resistivity (~10¹¹ – 10¹⁹ Ωm) or low conductivity (~10⁻¹¹ – 10⁻¹⁹ S m⁻¹).",
                                            "code": 14
                                        }
                                    ],
                                    "Summary": "Categorizing materials based on their electrical conductivity or resistivity values.",
                                    "code": 11
                                },
                                {
                                    "Subnet": "Classification by Energy Bands",
                                    "Categories": [
                                        {
                                            "Category": "Energy Band Theory",
                                            "Summary": "Concept that in solids, atomic energy levels combine to form continuous energy bands, specifically valence and conduction bands.",
                                            "code": 16
                                        },
                                        {
                                            "Category": "Metals (Overlapping Bands)",
                                            "Summary": "Conduction band is partially filled, or it overlaps with the valence band, allowing free movement of electrons for conduction.",
                                            "code": 17
                                        },
                                        {
                                            "Category": "Insulators (Large Band Gap)",
                                            "Summary": "Large energy band gap (Eg > 3 eV) between valence and conduction bands, preventing electron excitation and electrical conduction.",
                                            "code": 18
                                        },
                                        {
                                            "Category": "Semiconductors (Small Band Gap)",
                                            "Summary": "Finite but small energy band gap (Eg < 3 eV), allowing some electrons to cross to the conduction band at room temperature for limited conduction.",
                                            "code": 19
                                        }
                                    ],
                                    "Summary": "Explaining material properties based on the arrangement and energy separation of valence and conduction bands.",
                                    "code": 15
                                },
                                {
                                    "Subnet": "Types of Semiconductors",
                                    "Categories": [
                                        {
                                            "Category": "Elemental Semiconductors",
                                            "Summary": "Single-element semiconductors like Silicon (Si) and Germanium (Ge).",
                                            "code": 21
                                        },
                                        {
                                            "Category": "Inorganic Compound Semiconductors",
                                            "Summary": "Semiconductors formed from inorganic compounds, such as CdS, GaAs, CdSe, InP.",
                                            "code": 22
                                        },
                                        {
                                            "Category": "Organic/Polymer Semiconductors",
                                            "Summary": "Emerging class of semiconductors including organic materials (e.g., anthracene, doped pthalocyanines) and organic polymers (e.g., polypyrrole, polyaniline).",
                                            "code": 23
                                        }
                                    ],
                                    "Summary": "Differentiating between various forms of semiconductor materials based on their chemical composition.",
                                    "code": 20
                                }
                            ],
                            "Summary": "Categorization of materials (metals, semiconductors, insulators) based on their electrical properties and energy band structures, along with the specific types of semiconductor materials.",
                            "code": 10
                        },
                        {
                            "Net": "Intrinsic and Extrinsic Semiconductors",
                            "Subnets": [
                                {
                                    "Subnet": "Intrinsic Semiconductors",
                                    "Categories": [
                                        {
                                            "Category": "Covalent Bonding",
                                            "Summary": "In Si or Ge crystals, each atom shares four valence electrons with its four nearest neighbors, forming covalent bonds.",
                                            "code": 26
                                        },
                                        {
                                            "Category": "Electron-Hole Generation",
                                            "Summary": "At temperatures above 0K, thermal energy excites some electrons from covalent bonds, creating free electrons and vacancies called holes.",
                                            "code": 27
                                        },
                                        {
                                            "Category": "Conduction Mechanism",
                                            "Summary": "Total current is the sum of electron current (Ie) from free electrons and hole current (Ih) from the apparent movement of holes.",
                                            "code": 28
                                        }
                                    ],
                                    "Summary": "Properties and conduction mechanism of pure semiconductors where electron-hole pairs are generated by thermal energy.",
                                    "code": 25
                                },
                                {
                                    "Subnet": "Extrinsic (Doped) Semiconductors",
                                    "Categories": [
                                        {
                                            "Category": "Doping Process",
                                            "Summary": "Deliberate addition of a small amount of suitable impurity (dopant) to a pure semiconductor to increase its conductivity.",
                                            "code": 30
                                        },
                                        {
                                            "Category": "Dopant Characteristics",
                                            "Summary": "Dopant atoms should not distort the original crystal lattice and should have a size similar to the semiconductor atoms.",
                                            "code": 31
                                        },
                                        {
                                            "Category": "Charge Neutrality",
                                            "Summary": "Despite added charge carriers, the overall crystal maintains electrical neutrality, as carrier charges are balanced by ionized dopant cores.",
                                            "code": 32
                                        }
                                    ],
                                    "Summary": "The process of adding impurities to semiconductors to enhance their electrical conductivity and create impurity semiconductors.",
                                    "code": 29
                                },
                                {
                                    "Subnet": "n-type Semiconductors",
                                    "Categories": [
                                        {
                                            "Category": "Pentavalent Dopants",
                                            "Summary": "Impurities from Group 15 (valency 5), like Arsenic (As), Antimony (Sb), Phosphorous (P), which donate an extra electron for conduction.",
                                            "code": 34
                                        },
                                        {
                                            "Category": "Donor Impurity",
                                            "Summary": "Pentavalent dopants are called donor impurities because they donate an electron to the conduction band with very small ionization energy.",
                                            "code": 35
                                        },
                                        {
                                            "Category": "Majority/Minority Carriers (Electrons)",
                                            "Summary": "In n-type semiconductors, electrons become the majority carriers (ne >> nh) and holes are the minority carriers.",
                                            "code": 36
                                        }
                                    ],
                                    "Summary": "Characteristics of semiconductors doped with pentavalent impurities, resulting in excess electrons as majority carriers.",
                                    "code": 33
                                },
                                {
                                    "Subnet": "p-type Semiconductors",
                                    "Categories": [
                                        {
                                            "Category": "Trivalent Dopants",
                                            "Summary": "Impurities from Group 13 (valency 3), like Indium (In), Boron (B), Aluminium (Al), which create a vacancy or hole.",
                                            "code": 38
                                        },
                                        {
                                            "Category": "Acceptor Impurity",
                                            "Summary": "Trivalent dopants are called acceptor impurities because they create a hole, which can accept an electron from the valence band.",
                                            "code": 39
                                        },
                                        {
                                            "Category": "Majority/Minority Carriers (Holes)",
                                            "Summary": "In p-type semiconductors, holes become the majority carriers (nh >> ne) and electrons are the minority carriers.",
                                            "code": 40
                                        }
                                    ],
                                    "Summary": "Characteristics of semiconductors doped with trivalent impurities, resulting in excess holes as majority carriers.",
                                    "code": 37
                                }
                            ],
                            "Summary": "Detailed explanation of intrinsic semiconductors and how doping with pentavalent or trivalent impurities creates extrinsic n-type and p-type semiconductors, altering their conductivity and carrier concentrations.",
                            "code": 24
                        },
                        {
                            "Net": "p-n Junctions",
                            "Subnets": [
                                {
                                    "Subnet": "Junction Formation",
                                    "Categories": [
                                        {
                                            "Category": "Diffusion Current",
                                            "Summary": "Charge carriers (holes from p-side, electrons from n-side) move across the junction due to concentration gradient.",
                                            "code": 43
                                        },
                                        {
                                            "Category": "Depletion Region",
                                            "Summary": "A region on either side of the junction depleted of free charges, formed by immobile ionized donor and acceptor atoms.",
                                            "code": 44
                                        },
                                        {
                                            "Category": "Drift Current",
                                            "Summary": "Motion of charge carriers (electrons from p-side, holes from n-side) due to the electric field created by the space-charge regions, opposite to diffusion current.",
                                            "code": 45
                                        }
                                    ],
                                    "Summary": "The process by which a p-n junction is formed, involving diffusion and drift of charge carriers and the creation of a depletion region.",
                                    "code": 42
                                },
                                {
                                    "Subnet": "Equilibrium Conditions",
                                    "Categories": [
                                        {
                                            "Category": "Barrier Potential (V₀)",
                                            "Summary": "A potential difference developed across the junction due to the separation of charges in the depletion region, opposing further carrier flow.",
                                            "code": 47
                                        },
                                        {
                                            "Category": "No Net Current",
                                            "Summary": "At equilibrium, diffusion current equals drift current, resulting in zero net current across the junction.",
                                            "code": 48
                                        },
                                        {
                                            "Category": "Immobile Ionized Cores",
                                            "Summary": "Ionized donor atoms (positive) on the n-side and ionized acceptor atoms (negative) on the p-side, fixed within the crystal lattice.",
                                            "code": 49
                                        }
                                    ],
                                    "Summary": "Description of a p-n junction at thermal equilibrium without external voltage, detailing the barrier potential and charge balance.",
                                    "code": 46
                                }
                            ],
                            "Summary": "Explanation of how p-n junctions are formed through diffusion and drift currents, leading to a depletion region and a barrier potential at equilibrium.",
                            "code": 41
                        },
                        {
                            "Net": "Semiconductor Diodes",
                            "Subnets": [
                                {
                                    "Subnet": "Diode Structure and Symbol",
                                    "Categories": [
                                        {
                                            "Category": "p-n Junction with Contacts",
                                            "Summary": "A semiconductor diode is essentially a p-n junction with metallic contacts at the ends for external voltage application.",
                                            "code": 52
                                        },
                                        {
                                            "Category": "Two-Terminal Device",
                                            "Summary": "A basic electronic component with two terminals (anode and cathode).",
                                            "code": 53
                                        },
                                        {
                                            "Category": "Diode Symbol",
                                            "Summary": "Represented by an arrow indicating the conventional direction of current flow when forward biased.",
                                            "code": 54
                                        }
                                    ],
                                    "Summary": "Physical structure and symbolic representation of a semiconductor diode.",
                                    "code": 51
                                },
                                {
                                    "Subnet": "Forward Bias",
                                    "Categories": [
                                        {
                                            "Category": "Connection & Voltage Drop",
                                            "Summary": "p-side connected to positive, n-side to negative terminal; applied voltage drops across the depletion region.",
                                            "code": 56
                                        },
                                        {
                                            "Category": "Barrier Reduction & Current Flow",
                                            "Summary": "Depletion layer width decreases, barrier height reduces (V₀ - V), allowing significant current (mA) due to minority carrier injection.",
                                            "code": 57
                                        },
                                        {
                                            "Category": "Minority Carrier Injection",
                                            "Summary": "Electrons cross to p-side, holes cross to n-side, diffusing away from the junction and contributing to current.",
                                            "code": 58
                                        }
                                    ],
                                    "Summary": "Behavior of a p-n junction diode when forward biased, leading to current flow.",
                                    "code": 55
                                },
                                {
                                    "Subnet": "Reverse Bias",
                                    "Categories": [
                                        {
                                            "Category": "Connection & Voltage Increase",
                                            "Summary": "n-side connected to positive, p-side to negative; applied voltage adds to barrier potential.",
                                            "code": 60
                                        },
                                        {
                                            "Category": "Barrier Widening & Low Current",
                                            "Summary": "Depletion layer widens, barrier height increases (V₀ + V), suppressing diffusion current and resulting in very small current (μA) from minority carrier drift.",
                                            "code": 61
                                        },
                                        {
                                            "Category": "Breakdown Voltage (Vbr)",
                                            "Summary": "A critical reverse bias voltage where the diode reverse current increases sharply, potentially destroying the junction if current is not limited.",
                                            "code": 62
                                        }
                                    ],
                                    "Summary": "Behavior of a p-n junction diode when reverse biased, leading to negligible current flow until breakdown.",
                                    "code": 59
                                },
                                {
                                    "Subnet": "V-I Characteristics & Resistance",
                                    "Categories": [
                                        {
                                            "Category": "Threshold/Cut-in Voltage",
                                            "Summary": "Characteristic voltage (e.g., ~0.7V for Si) after which forward current increases exponentially with small voltage increase.",
                                            "code": 64
                                        },
                                        {
                                            "Category": "Reverse Saturation Current",
                                            "Summary": "Very small, almost constant current (~μA) under reverse bias, largely independent of applied voltage until breakdown.",
                                            "code": 65
                                        },
                                        {
                                            "Category": "Dynamic Resistance (r_d)",
                                            "Summary": "Defined as the ratio of a small change in voltage (ΔV) to a small change in current (ΔI) in the operating region.",
                                            "code": 66
                                        }
                                    ],
                                    "Summary": "The graphical representation of current versus voltage for a diode, showing its different operating regions and resistance.",
                                    "code": 63
                                }
                            ],
                            "Summary": "A comprehensive look at semiconductor diodes, covering their structure, behavior under forward and reverse bias, and their characteristic V-I curve.",
                            "code": 50
                        },
                        {
                            "Net": "Applications of Junction Diodes",
                            "Subnets": [
                                {
                                    "Subnet": "Half-Wave Rectifiers",
                                    "Categories": [
                                        {
                                            "Category": "Circuit Configuration",
                                            "Summary": "A single diode connected in series with a load, receiving an AC input from a transformer secondary.",
                                            "code": 69
                                        },
                                        {
                                            "Category": "Output Waveform",
                                            "Summary": "Produces a pulsating DC output voltage only during the positive half-cycles of the AC input when the diode is forward biased.",
                                            "code": 70
                                        },
                                        {
                                            "Category": "Current Flow Direction",
                                            "Summary": "Allows current to flow only in one direction, effectively rectifying half of the AC waveform.",
                                            "code": 71
                                        }
                                    ],
                                    "Summary": "Detailed explanation of half-wave rectification, where only one half-cycle of an AC input is converted to pulsating DC.",
                                    "code": 68
                                },
                                {
                                    "Subnet": "Full-Wave Rectifiers",
                                    "Categories": [
                                        {
                                            "Category": "Centre-Tap Rectifier Circuit",
                                            "Summary": "Uses two diodes and a center-tapped transformer to rectify both positive and negative half-cycles of the AC input.",
                                            "code": 73
                                        },
                                        {
                                            "Category": "Output Waveform",
                                            "Summary": "Provides output voltage during both positive and negative half-cycles of the AC input, resulting in a more efficient rectified output than half-wave.",
                                            "code": 74
                                        },
                                        {
                                            "Category": "Bridge Rectifier Alternative",
                                            "Summary": "Mention of another full-wave rectifier circuit that does not require a center-tap transformer but uses four diodes.",
                                            "code": 75
                                        }
                                    ],
                                    "Summary": "Detailed explanation of full-wave rectification, converting both halves of an AC input into pulsating DC, typically using a center-tap transformer.",
                                    "code": 72
                                },
                                {
                                    "Subnet": "Filters for DC Output",
                                    "Categories": [
                                        {
                                            "Category": "Capacitor Filter Function",
                                            "Summary": "A capacitor connected across the output terminals (parallel to the load) charges during voltage rise and discharges through the load during fall, smoothing the pulsating DC.",
                                            "code": 77
                                        },
                                        {
                                            "Category": "Inductor Filter Alternative",
                                            "Summary": "An inductor can be used in series with the load for the same purpose of filtering out AC ripples.",
                                            "code": 78
                                        },
                                        {
                                            "Category": "Time Constant & Smoothing",
                                            "Summary": "The rate of voltage fall across the capacitor depends inversely on the RC time constant; large capacitance values are used for better smoothing, providing an output nearer to the peak voltage.",
                                            "code": 79
                                        }
                                    ],
                                    "Summary": "Components and techniques used to convert the pulsating DC output of rectifiers into a steady DC voltage.",
                                    "code": 76
                                }
                            ],
                            "Summary": "Exploring the practical applications of p-n junction diodes, primarily as rectifiers in power supplies to convert alternating current (AC) to direct current (DC), including half-wave, full-wave rectification, and filtering.",
                            "code": 67
                        }
                    ],
                    "Summary": "This document provides a comprehensive overview of semiconductor electronics, contrasting them with vacuum tubes and highlighting their advantages. It classifies materials into metals, semiconductors, and insulators based on conductivity and energy band theory. Detailed explanations of intrinsic semiconductors, the generation of electron-hole pairs, and the creation of extrinsic (n-type and p-type) semiconductors through doping are provided. The text further elaborates on the formation and behavior of p-n junctions, including depletion regions and barrier potential, under equilibrium, forward bias, and reverse bias conditions. Finally, it describes the semiconductor diode's V-I characteristics and its application as half-wave and full-wave rectifiers, along with the function of filters to obtain steady DC output."
                }
            },
            "summaryList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 10256,
                        "CandidatesTokenCount": 3726,
                        "TotalTokenCount": 15447,
                        "thoughtsTokenCount": 1465,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Summary": "<h2>Comprehensive Synthesis Report: Semiconductor Electronics</h2><h3>1. Introduction to Semiconductor Devices</h3><p>Historically, electronic circuits relied on vacuum tubes (valves) like diodes, triodes, tetrodes, and pentodes. These devices operated by controlling the flow of electrons in a vacuum, supplied by a heated cathode. They were characterized by their bulkiness, high power consumption, high operating voltages (around 100V), limited lifespan, and low reliability. (Data Source)</p><p>The advent of solid-state semiconductor electronics, rooted in discoveries from the 1930s, revolutionized this field. Semiconductors offer the ability to control the number and direction of charge carriers within a solid itself, requiring no external heating or evacuated space. This results in devices that are small, consume low power, operate at low voltages, and boast long life and high reliability. Modern displays, such as LCDs, exemplify this shift, replacing older Cathode Ray Tubes (CRTs). Early applications even include the use of galena crystals as radio wave detectors. This report will delve into the fundamental physics of semiconductors, focusing on devices like junction diodes and bipolar junction transistors, along with their circuit applications. (Data Source)</p><h3>2. Classification of Materials</h3><h4>2.1. Classification Based on Electrical Conductivity</h4><p>Solids are broadly categorized based on their electrical conductivity (σ) or resistivity (ρ = 1/σ):</p><ul><li><b>Metals:</b> Possess very low resistivity (ρ ~ 10⁻² – 10⁻⁸ Ωm) or high conductivity (σ ~ 10² – 10⁸ S m⁻¹).</li><li><b>Semiconductors:</b> Exhibit intermediate resistivity (ρ ~ 10⁻⁵ – 10⁶ Ωm) or conductivity (σ ~ 10⁵ – 10⁻⁶ S m⁻¹) between metals and insulators.</li><li><b>Insulators:</b> Have high resistivity (ρ ~ 10¹¹ – 10¹⁹ Ωm) or low conductivity (σ ~ 10⁻¹¹ – 10⁻¹⁹ S m⁻¹).</li></ul><p>These ranges are indicative, and other criteria, particularly energy band structures, further distinguish them. (Data Source)</p><p>Semiconductors can be:</p><ul><li><b>Elemental:</b> Silicon (Si) and Germanium (Ge).</li><li><b>Compound:</b> Inorganic (e.g., CdS, GaAs, InP) and Organic (e.g., anthracene, polyaniline).</li></ul><p>While most current devices use elemental Si/Ge or inorganic compounds, organic semiconductors are a burgeoning field. This report focuses primarily on inorganic elemental semiconductors. (Data Source)</p><h4>2.2. Classification Based on Energy Bands</h4><p>In solids, atomic energy levels broaden into continuous energy bands due to the close proximity of atoms. Key bands are:</p><ul><li><b>Valence Band:</b> Contains valence electrons.</li><li><b>Conduction Band:</b> Lies above the valence band; normally empty.</li></ul><p>The **energy band gap (E<sub>g</sub>)** is the energy difference between the top of the valence band (E<sub>V</sub>) and the bottom of the conduction band (E<sub>C</sub>). E<sub>g</sub> dictates a material's electrical properties: (Data Source)</p><ul><li><b>Metals (Case I):</b> Either the conduction band is partially filled, or the valence and conduction bands overlap (E<sub>g</sub> = 0). This allows electrons to move freely, resulting in high conductivity and low resistance. (Data Source)</li><li><b>Insulators (Case II):</b> Have a large energy gap (E<sub>g</sub> &gt; 3 eV). Electrons cannot gain enough thermal energy to cross this gap, leaving the conduction band empty and preventing electrical conduction. (Data Source)</li><li><b>Semiconductors (Case III):</b> Possess a finite but small energy gap (E<sub>g</sub> &lt; 3 eV). At room temperature, some valence electrons can acquire enough thermal energy to jump to the conduction band, enabling a degree of conduction. These excited electrons leave behind vacant energy levels in the valence band, known as holes, which also contribute to conduction. (Data Source)</li></ul><p>For Si and Ge, the 8N outer electron states (for N atoms) split into two bands separated by E<sub>g</sub>. The lower valence band is completely occupied at 0K, while the upper conduction band is empty. At T &gt; 0K, thermal energy allows some electrons to transition to the conduction band. (Data Source)</p><h3>3. Intrinsic Semiconductors</h3><p>Intrinsic semiconductors like Si and Ge have a diamond-like crystal structure, where each atom forms covalent bonds with four neighbors. At low temperatures (0K), all bonds are intact, and the material behaves as an insulator. (Data Source)</p><p>At higher temperatures (T &gt; 0K), thermal energy can break some covalent bonds, freeing electrons. This creates a **free electron** (charge -q) in the conduction band and a **hole** (effective charge +q) in the valence bond. In intrinsic semiconductors, the number of free electrons (n<sub>e</sub>) equals the number of holes (n<sub>h</sub>), denoted as n<sub>i</sub> (intrinsic carrier concentration). (Data Source)</p><p>Holes behave as apparent free particles moving towards negative potential. When an electron from an adjacent bond fills a hole, the hole appears to move. Both free electrons (contributing to electron current, I<sub>e</sub>) and holes (contributing to hole current, I<sub>h</sub>) enable current flow. The total current (I) is the sum: I = I<sub>e</sub> + I<sub>h</sub>. (Data Source)</p><p>Simultaneously, electrons and holes recombine. At equilibrium, the rate of generation equals the rate of recombination. Carbon, with an E<sub>g</sub> of 5.4 eV, acts as an insulator, while Si (1.1 eV) and Ge (0.7 eV) are intrinsic semiconductors due to their smaller band gaps allowing thermal excitation. (Data Source)</p><h3>4. Extrinsic Semiconductors (Doping)</h3><p>Intrinsic semiconductors have low conductivity at room temperature, limiting their practical use. **Doping** involves adding a small, suitable impurity (dopant) to a pure semiconductor to significantly increase its conductivity. Dopant atoms must have similar size to the host atoms to avoid lattice distortion. (Data Source)</p><h4>4.1. n-type Semiconductor</h4><p>Doping Si or Ge (tetravalent) with a **pentavalent** impurity (e.g., Arsenic, Antimony, Phosphorus from Group 5) creates an n-type semiconductor. Four valence electrons of the dopant form covalent bonds, while the fifth electron is very weakly bound. This 'extra' electron requires minimal ionization energy (e.g., ~0.01 eV for Ge, ~0.05 eV for Si) to become free, far less than E<sub>g</sub>. (Data Source)</p><p>These dopants are called **donor impurities** as they donate extra electrons. In n-type semiconductors, electrons are the **majority carriers**, and holes (primarily from intrinsic generation) are **minority carriers** (n<sub>e</sub> &gt;&gt; n<sub>h</sub>). The donor energy level (E<sub>D</sub>) lies just below the conduction band (E<sub>C</sub>), allowing easy electron transition. (Data Source)</p><h4>4.2. p-type Semiconductor</h4><p>Doping Si or Ge with a **trivalent** impurity (e.g., Indium, Boron, Aluminium from Group 3) creates a p-type semiconductor. The dopant has one less valence electron than Si/Ge, forming only three covalent bonds and leaving a **vacancy or hole** in the fourth bond. (Data Source)</p><p>These dopants are called **acceptor impurities** as they readily accept an electron from a neighboring bond, effectively creating a mobile hole. In p-type semiconductors, holes are the **majority carriers**, and electrons (from intrinsic generation) are **minority carriers** (n<sub>h</sub> &gt;&gt; n<sub>e</sub>). The acceptor energy level (E<sub>A</sub>) lies just above the valence band (E<sub>V</sub>), allowing electrons from the valence band to easily jump into this level, thereby creating holes in the valence band. (Data Source)</p><p>In both n-type and p-type semiconductors, overall charge neutrality is maintained, as the charge of additional carriers is balanced by the ionized dopant cores. Doping also indirectly reduces the intrinsic concentration of minority carriers due to increased recombination rates. (Data Source)</p><p>A fundamental relationship in semiconductors in thermal equilibrium is n<sub>e</sub>n<sub>h</sub> = n<sub>i</sub><sup>2</sup>. (Data Source)</p><h3>5. p-n Junction</h3><h4>5.1. p-n Junction Formation</h4><p>A p-n junction, a core component of many semiconductor devices, is formed by precisely converting a portion of a p-type wafer into n-type material. This cannot be achieved by merely physically joining two different semiconductor slabs due to atomic-level discontinuity. (Data Source)</p><p>Two key processes occur during formation:</p><ul><li><b>Diffusion:</b> Due to concentration gradients, majority carriers move across the junction: holes from p-side to n-side (p &rarr; n), and electrons from n-side to p-side (n &rarr; p). This creates a **diffusion current**.</li><li><b>Drift:</b> As electrons diffuse from n &rarr; p, they leave behind immobile, ionized donor atoms (positive charge) on the n-side. Similarly, diffusing holes from p &rarr; n leave behind immobile, ionized acceptor atoms (negative charge) on the p-side. These immobile charges form space-charge regions, collectively known as the **depletion region**, which is devoid of free carriers.</li></ul><p>An electric field develops across the depletion region, directed from positive (n-side) to negative (p-side) charges. This field causes minority carriers (electrons on p-side, holes on n-side) to move across the junction, creating a **drift current** in the opposite direction to the diffusion current. (Data Source)</p><p>Initially, diffusion current is dominant, but as the depletion region widens and the electric field strengthens, the drift current increases until it balances the diffusion current. At equilibrium, no net current flows. The potential difference established across the junction, opposing further carrier flow, is called the **barrier potential (V<sub>0</sub>)**. The n-material becomes positive relative to the p-material. (Data Source)</p><h3>6. Semiconductor Diode</h3><p>A semiconductor diode is essentially a p-n junction with metallic contacts, functioning as a two-terminal device. Its symbol features an arrow indicating the conventional direction of current under forward bias. (Data Source)</p><h4>6.1. p-n Junction Diode Under Forward Bias</h4><p>When the p-side is connected to the positive terminal and the n-side to the negative terminal of an external voltage (V), the diode is **forward biased**. This applied voltage primarily drops across the depletion region, opposing the built-in barrier potential (V<sub>0</sub>). Consequently, the depletion layer width decreases, and the effective barrier height is reduced to (V<sub>0</sub> - V). (Data Source)</p><p>With sufficient applied voltage, a large number of majority carriers (electrons from n-side, holes from p-side) gain enough energy to cross the reduced barrier. This phenomenon is called **minority carrier injection**, as carriers become minority carriers on the opposite side of the junction. These injected carriers diffuse away from the junction, leading to a significant **forward current**, typically in the milliampere (mA) range. (Data Source)</p><h4>6.2. p-n Junction Diode Under Reverse Bias</h4><p>When the n-side is positive and the p-side is negative, the diode is **reverse biased**. The applied voltage adds to the barrier potential, increasing the effective barrier height to (V<sub>0</sub> + V) and widening the depletion region. This strongly suppresses the flow of majority carriers across the junction. (Data Source)</p><p>However, the electric field sweeps any minority carriers (electrons on the p-side, holes on the n-side) that come near the junction to their respective majority regions. This creates a very small **drift current**, known as the **reverse saturation current**, typically in the microampere (µA) range. This current is nearly constant and independent of the applied voltage until the **breakdown voltage (V<sub>br</sub>)** is reached. Beyond V<sub>br</sub>, the current sharply increases, potentially destroying the diode if not limited. (Data Source)</p><h4>6.3. V-I Characteristics and Dynamic Resistance</h4><p>The **V-I characteristic curve** illustrates the diode's behavior. In forward bias, current remains very small until a **threshold voltage** or **cut-in voltage** is reached (~0.2V for Germanium, ~0.7V for Silicon), after which current increases exponentially. In reverse bias, current remains very small (reverse saturation current) and nearly constant until V<sub>br</sub>. (Data Source)</p><p>The **dynamic resistance (r<sub>d</sub>)** of a diode is defined as the ratio of a small change in voltage (ΔV) to a small change in current (ΔI): r<sub>d</sub> = ΔV/ΔI. (Data Source)</p><h3>7. Application of Junction Diode as a Rectifier</h3><p>The diode's property of allowing current flow primarily in one direction (forward bias) makes it suitable for **rectification**, converting alternating current (AC) to pulsating direct current (DC). Rectifier circuits are used for this purpose. (Data Source)</p><h4>7.1. Half-wave Rectifier</h4><p>A half-wave rectifier uses a single diode in series with a load resistor (R<sub>L</sub>). During the positive half-cycle of the AC input, the diode is forward biased and conducts, producing an output voltage across R<sub>L</sub>. During the negative half-cycle, the diode is reverse biased and does not conduct (assuming negligible reverse current). The output is a series of positive voltage pulses, representing only half of the input AC wave. (Data Source)</p><h4>7.2. Full-wave Rectifier</h4><p>A **center-tap full-wave rectifier** uses two diodes and a center-tapped transformer. During the positive half-cycle of the input AC, one diode conducts, and during the negative half-cycle, the other diode conducts. This results in output voltage pulses for both positive and negative half-cycles of the input, making it more efficient than a half-wave rectifier. (Data Source)</p><p>To obtain a steady DC output from these pulsating voltages, **filter circuits** are used. A common filter involves connecting a capacitor across the load resistor. The capacitor charges during voltage peaks and discharges through the load when the voltage falls, thereby smoothing out the AC ripple and providing a voltage closer to the peak rectified voltage. The effectiveness of the filter depends on the time constant (RC), requiring large capacitance values for better filtering. (Data Source)</p><h3>8. Conclusions and Recommendations</h3><h4>8.1. Conclusions</h4><p>Semiconductor electronics fundamentally transformed circuit design by replacing bulky, inefficient vacuum tubes with miniature, reliable, and energy-efficient solid-state devices. The unique electrical properties of semiconductors arise from their intermediate conductivity, governed by their small energy band gap. Doping allows precise control over charge carrier concentrations, creating n-type (electron majority) and p-type (hole majority) materials. The p-n junction, formed by diffusion and drift processes establishing a depletion region and barrier potential, is the foundational structure for devices like diodes. Diodes exhibit unidirectional current flow, characterized by distinct forward and reverse bias behaviors, making them essential for rectification in power supply circuits. Full-wave rectification with filtering provides a more stable DC output, critical for modern electronic applications.</p><h4>8.2. Actionable Recommendations</h4><p>Based on this synthesis, for educational and practical applications:</p><ol><li><b>Emphasize Energy Band Theory:</b> Instructors should prioritize thorough teaching of energy band theory and its direct correlation to material classification (metals, insulators, semiconductors) to provide a robust theoretical foundation.</li><li><b>Hands-on Doping Simulation/Experiments:</b> Develop laboratory exercises or simulations demonstrating the impact of different dopant types (pentavalent vs. trivalent) and concentrations on semiconductor conductivity and carrier type to reinforce theoretical concepts.</li><li><b>Practical Diode Characteristics:</b> Implement practical experiments for plotting V-I characteristics of various diodes (e.g., Germanium vs. Silicon) to empirically observe cut-in voltage, reverse saturation, and breakdown, linking theory to real-world device behavior.</li><li><b>Rectifier Circuit Building:</b> Require students to construct and test both half-wave and full-wave rectifier circuits, incorporating filter capacitors, to solidify understanding of AC-to-DC conversion and the role of filtering.</li><li><b>Troubleshooting Focus:</b> Incorporate scenarios involving diode failure modes (e.g., exceeding breakdown voltage, overheating from excessive forward current) to highlight device limitations and safe operating practices for future engineers.</li></ol>"
                            }
                        ]
                    }
                }
            ]
        }
    ],
    "fileName": "",
    "projectType": "DOCUMENT",
    "__v": 0,
    "headerCols": {
        "col1": "Part Id",
        "col2": "FileName",
        "col3": "Content"
    }
},
{
    "_id": "695e09cada2133fe04a5f3ff",
    "projectId": "460913",
    "projectName": "K12PHY1 - 04 MOVING CHARGES AND MAGNETISM",
    "projectDescription": "",
    "projectCategory":"Educational",
    "projectCreatedDate": "2026-01-07T07:22:47.754Z",
    "projectUpdatedDate": "2026-01-07T07:22:47.754Z",
    "projectStatus": "Active",
    "companyId": "6957bd3ad58ea40715c6ae98",
    "projectCreatedBy": "6957bd3ad58ea40715c6ae99",
    "projectUpdatedBy": "6957bd3ad58ea40715c6ae99",
    "projectLastAccessDate": "2026-01-30T11:35:04.970Z",
    "projectData": [
        {
            "col1": "Part Id",
            "col2": "FileName",
            "col3": "Content",
            "col4": "Status"
        },
        {
            "col1": "P1_Part_1",
            "col2": "K12PHY1 - 04 MOVING CHARGES AND MAGNETISM.pdf",
            "col3": "4.1 I NTRODUCTION Both Electricity and Magnetism have been known for more than 2000 years. However, it was only about 200 years ago, in 1820, that it was realised that they were intimately related. During a lecture demonstration in the summer of 1820, Danish physicist Hans Christian Oersted noticed that a current in a straight wire caused a noticeable deflection in a nearby magnetic compass needle. He investigated this phenomenon. He found that the alignment of the needle is tangential to an imaginary circle which has the straight wire as its centre and has its plane perpendicular to the wire. This situation is depicted in Fig.4.1(a). It is noticeable when the current is large and the needle sufficiently close to the wire so that the earth’s magnetic field may be ignored. Reversing the direction of the current reverses the orientation of the needle [Fig. 4.1(b)]. The deflection increases on increasing the current or bringing the needle closer to the wire. Iron filings sprinkled around the wire arrange themselves in concentric circles with the wire as the centre [Fig. 4.1(c)]. Oersted concluded that moving charges or currents produced a magnetic field in the surrounding space . Following this, there was intense experimentation. In 1864, the laws obeyed by electricity and magnetism were unified and formulated by Chapter Four MOVING CHARGES AND MAGNETISM Reprint 2025-26Physics 108 James Maxwell who then realised that light was electromagnetic waves. Radio waves were discovered by Hertz, and produced by J.C.Bose and G. Marconi by the end of the 19 th century. A remarkable scientific and technological progress took place in the 20 th century. This was due to our increased understanding of electromagnetism and the invention of devices for production, amplification, transmission and detection of electromagnetic waves. In this chapter, we will see how magnetic field exerts forces on moving charged particles, like electrons, protons, and current-carrying wires. We shall also learn how currents produce magnetic fields. We shall see how particles can be accelerated to very high energies in a cyclotron. We shall study how currents and voltages are detected by a galvanometer. In this and subsequent Chapter on magnetism, we adopt the following convention: A current or a field (electric or magnetic) emerging out of the plane of the paper is depicted by a dot ( § ). A current or a field going into the plane of the paper is depicted by a cross (  ) * . Figures. 4.1(a) and 4.1(b) correspond to these two situations, respectively. 4.2 M AGNETIC F ORCE 4.2.1 Sources and fields Before we introduce the concept of a magnetic field B , we shall recapitulate what we have learnt in Chapter 1 about the electric field E. We have seen that the interaction between two charges can be considered in two stages. The charge Q, the source of the field, produces an electric field E , where FIGURE 4.1 The magnetic field due to a straight long current-carrying wire. The wire is perpendicular to the plane of the paper. A ring of compass needles surrounds the wire. The orientation of the needles is shown when (a) the current emerges out of the plane of the paper, (b) the current moves into the plane of the paper. (c) The arrangement of iron filings around the wire. The darkened ends of the needle represent north poles. The effect of the earth’s magnetic field is neglected. * A dot appears like the tip of an arrow pointed at you, a cross is like the feathered tail of an arrow moving away from you. Hans Christian Oersted (1777–1851) Danish physicist and chemist, professor at Copenhagen. He observed that a compass needle suffers a deflection when placed near a wire carrying an electric current. This discovery gave the first empirical evidence of a connection between electric and magnetic phenomena. HANS CHRISTIAN OERSTED (1777–1851) Reprint 2025-26109 Moving Charges and Magnetism E = Q ˆ r / (4 p e 0 ) r 2 (4.1) where ˆ r is unit vector along r , and the field E is a vector field. A charge q interacts with this field and experiences a force F given by F = q E = q Q ˆ r / (4 p e 0 ) r 2 (4.2) As pointed out in the Chapter 1, the field E is not just an artefact but has a physical role. It can convey energy and momentum and is not established instantaneously but takes finite time to propagate. The concept of a field was specially stressed by Faraday and was incorporated by Maxwell in his unification of electricity and magnetism. In addition to depending on each point in space, it can also vary with time, i.e., be a function of time. In our discussions in this chapter, we will assume that the fields do not change with time. The field at a particular point can be due to one or more charges. If there are more charges the fields add vectorially. You have already learnt in Chapter 1 that this is called the principle of superposition. Once the field is known, the force on a test charge is given by Eq. (4.2). Just as static charges produce an electric field, the currents or moving charges produce (in addition) a magnetic field, denoted by B (r) , again a vector field. It has several basic properties identical to the electric field. It is defined at each point in space (and can in addition depend on time). Experimentally, it is found to obey the principle of superposition: the magnetic field of several sources is the vector addition of magnetic field of each individual source . 4.2.2 Magnetic Field, Lorentz Force Let us suppose that there is a point charge q (moving with a velocity v and, located at r at a given time t ) in presence of both the electric field E (r) and the magnetic field B (r) . The force on an electric charge q due to both of them can be written as F = q [ E (r ) + v × B ( r )] ∫ F electric + F magnetic (4.3) This force was given first by H.A. Lorentz based on the extensive experiments of Ampere and others. It is called the Lorentz force . You have already studied in detail the force due to the electric field. If we look at the interaction with the magnetic field, we find the following features. (i) It depends on q , v and B (charge of the particle, the velocity and the magnetic field). Force on a negative charge is opposite to that on a positive charge. (ii) The magnetic force q [ v × B ] includes a vector product of velocity and magnetic field. The vector product makes the force due to magnetic HENDRIK ANTOON LORENTZ (1853 – 1928) Hendrik Antoon Lorentz (1853 – 1928) Dutch theoretical physicist, professor at Leiden. He investigated the relationship between electricity, magnetism, and mechanics. In order to explain the observed effect of magnetic fields on emitters of light (Zeeman effect), he postulated the existence of electric charges in the atom, for which he was awarded the Nobel Prize in 1902. He derived a set of transformation equations (known after him, as Lorentz transformation equations) by some tangled mathematical arguments, but he was not aware that these equations hinge on a new concept of space and time. Reprint 2025-26Physics 110 field vanish (become zero) if velocity and magnetic field are parallel or anti-parallel. The force acts in a (sideways) direction perpendicular to both the velocity and the magnetic field. Its direction is given by the screw rule or right hand rule for vector (or cross) product as illustrated in Fig. 4.2. (iii) The magnetic force is zero if charge is not moving (as then | v |= 0). Only a moving charge feels the magnetic force. The expression for the magnetic force helps us to define the unit of the magnetic field, if one takes q , F and v , all to be unity in the force equation F = q [ v × B ] = q v B sin q ˆ n , where q is the angle between v and B [see Fig. 4.2 (a)]. The magnitude of magnetic field B is 1 SI unit, when the force acting on a unit charge (1 C), moving perpendicular to B with a speed 1m/s, is one newton. Dimensionally, we have [ B ] = [ F/qv ] and the unit of B are Newton second / (coulomb metre). This unit is called tesla ( T ) named after Nikola Tesla (1856 – 1943). Tesla is a rather large unit. A smaller unit (non-SI) called gauss (=10 –4 tesla) is also often used. The earth’s magnetic field is about 3.6 × 10 –5 T. 4.2.3 Magnetic force on a current-carrying conductor We can extend the analysis for force due to magnetic field on a single moving charge to a straight rod carrying current. Consider a rod of a uniform cross-sectional area A and length l . We shall assume one kind of mobile carriers as in a conductor (here electrons). Let the number density of these mobile charge carriers in it be n . Then the total number of mobile charge carriers in it is nlA . For a steady current I in this conducting rod, we may assume that each mobile carrier has an average drift velocity v d (see Chapter 3). In the presence of an external magnetic field B , the force on these carriers is: F = ( nlA ) q v d ´ ´ ´ ´ ´ B where q is the value of the charge on a carrier. Now nq v d is the current density j and |( nq v d )| A is the current I (see Chapter 3 for the discussion of current and current density). Thus, F = [( nq v d ) l A ] × B = [ j Al ] ´ ´ ´ ´ ´ B = I l ´ ´ ´ ´ ´ B (4.4) where l is a vector of magnitude l , the length of the rod, and with a direction identical to the current I . Note that the current I is not a vector. In the last step leading to Eq. (4.4), we have transferred the vector sign from j to l . Equation (4.4) holds for a straight rod. In this equation, B is the external magnetic field. It is not the field produced by the current-carrying rod. If the wire has an arbitrary shape we can calculate the Lorentz force on it by considering it as a collection of linear strips d l j and summing j j Id ×   F B l This summation can be converted to an integral in most cases. FIGURE 4.2 The direction of the magnetic force acting on a charged particle. (a) The force on a positively charged particle with velocity v and making an angle q with the magnetic field B is given by the right-hand rule. (b) A moving charged particle q is deflected in an opposite sense to – q in the presence of magnetic field. Reprint 2025-26111 Moving Charges and Magnetism E XAMPLE 4.1 Example 4.1 A straight wire of mass 200 g and length 1.5 m carries a current of 2 A. It is suspended in mid-air by a uniform horizontal magnetic field B ( Fig. 4.3). What is the magnitude of the magnetic field? FIGURE 4.3 Solution From Eq. (4.4), we find that there is an upward force F , of magnitude IlB, . For mid-air suspension, this must be balanced by the force due to gravity: m g = I lB m g B I l = 0.2 9.8 0.65 T 2 1.5 × = = × Note that it would have been sufficient to specify m/ l , the mass per unit length of the wire. The earth’s magnetic field is approximately 4 × 10 –5 T and we have ignored it. Example 4.2 If the magnetic field is parallel to the positive y -axis and the charged particle is moving along the positive x -axis (Fig. 4.4), which way would the Lorentz force be for (a) an electron (negative charge), (b) a proton (positive charge). FIGURE 4.4 Solution The velocity v of particle is along the x -axis, while B , the magnetic field is along the y -axis, so v × B is along the z -axis (screw rule or right-hand thumb rule). So, (a) for electron it will be along – z axis. (b) for a positive charge (proton) the force is along + z axis. E XAMPLE 4.2 Charged particles moving in a magnetic field. Interactive demonstration: http://www.phys.hawaii.edu/~teb/optics/java/partmagn/index.html Reprint 2025-26Physics 112 4.3 M OTION IN A M AGNETIC F IELD We will now consider, in greater detail, the motion of a charge moving in a magnetic field. We have learnt in Mechanics (see Class XI book, Chapter 5) that a force on a particle does work if the force has a component along (or opposed to) the direction of motion of the particle. In the case of motion of a charge in a magnetic field, the magnetic force is perpendicular to the velocity of the particle. So no work is done and no change in the magnitude of the velocity is produced (though the direction of momentum may be changed). [Notice that this is unlike the force due to an electric field, q E , which can have a component parallel (or antiparallel) to motion and thus can transfer energy in addition to momentum.] We shall consider motion of a charged particle in a uniform magnetic field. First consider the case of v perpendicular to B . The perpendicular force, q v × B , acts as a centripetal force and produces a circular motion perpendicular to the magnetic field. The particle will describe a circle if v and B are perpendicular to each other (Fig. 4.5). If velocity has a component along B , this component remains unchanged as the motion along the magnetic field will not be affected by the magnetic field. The motion in a plane perpendicular to B is as before a circular one, thereby producing a helical motion (Fig. 4.6). You have already learnt in earlier classes (See Class XI, Chapter 3) that if r is the radius of the circular path of a particle, then a force of m v 2 / r , acts perpendicular to the path towards the centre of the circle, and is called the centripetal force. If the velocity v is perpendicular to the magnetic field B, the magnetic force is perpendicular to both v and B and acts like a centripetal force. It has a magnitude q v B . Equating the two expressions for centripetal force, m v 2 / r = q v B , which gives r = m v / qB (4.5) for the radius of the circle described by the charged particle. The larger the momentum, the larger is the radius and bigger the circle described. If w is the angular frequency, then v = w r . So, w = 2 p n = q B / m [4.6(a)] which is independent of the velocity or energy . Here n is the frequency of rotation. The independence of n from energy has important application in the design of a cyclotron. The time taken for one revolution is T = 2 p / w ∫ 1/ n . If there is a component of the velocity parallel to the magnetic field (denoted by v | | ), it will make the particle move along the field and the path of the FIGURE 4.5 Circular motion FIGURE 4.6 Helical motion Reprint 2025-26113 Moving Charges and Magnetism E XAMPLE 4.3 Example 4.3 What is the radius of the path of an electron (mass 9 × 10 -31 kg and charge 1.6 × 10 –19 C) moving at a speed of 3 × 10 7 m/s in a magnetic field of 6 × 10 –4 T perpendicular to it? What is its frequency? Calculate its energy in keV. ( 1 eV = 1.6 × 10 –19 J). Solution Using Eq. (4.5) we find r = m v / ( qB ) = 9 ×1 0 –31 kg × 3 × 10 7 m s –1 / ( 1.6 × 10 –19 C × 6 × 10 –4 T ) = 28 × 10 –2 m = 28 cm n = v / (2 p r ) = 17×10 6 s –1 = 17×10 6 Hz =17 MHz. E = (½ ) mv 2 = (½ ) 9 × 10 –31 kg × 9 × 10 14 m 2 /s 2 = 40.5 ×10 –17 J ≈ 4×10 –16 J = 2.5 keV. particle would be a helical one (Fig. 4.6). The distance moved along the magnetic field in one rotation is called pitch p . Using Eq. [4.6 (a)], we have p = v || T = 2 p m v | | / q B [4.6(b)] The radius of the circular component of motion is called the radius of the helix . 4.4 M A G N E T I C F I E L D D U E T O A C U R R E N T E LEMENT , B IOT -S AVART L AW All magnetic fields that we know are due to currents (or moving charges) and due to intrinsic magnetic moments of particles. Here, we shall study the relation between current and the magnetic field it produces. It is given by the Biot-Savart’s law. Fig. 4.7 shows a finite conductor XY carrying current I . Consider an infinitesimal element d l of the conductor. The magnetic field d B due to this element is to be determined at a point P which is at a distance r from it. Let q be the angle between d l and the displacement vector r . According to Biot-Savart’s law, the magnitude of the magnetic field d B is proportional to the current I , the element length |d l | , and inversely proportional to the square of the distance r . Its direction * is perpendicular to the plane containing d l and r . Thus, in vector notation, d I d r B r ∝ × l 3 = × μ 0 3 4 π I d r l r [4.7(a)] where m 0 /4 p is a constant of proportionality. The above expression holds when the medium is vacuum. FIGURE 4.7 Illustration of the Biot-Savart law. The current element I d l produces a field d B at a distance r . The ƒ sign indicates that the field is perpendicular to the plane of this page and directed into it. * The sense of d l × r is also given by the Right Hand Screw rule : Look at the plane containing vectors d l and r . Imagine moving from the first vector towards second vector. If the movement is anticlockwise, the resultant is towards you. If it is clockwise, the resultant is away from you. Reprint 2025-26Physics 114 E XAMPLE 4.4 The magnitude of this field is, μ θ = π 0 2 d sin d 4 I l r B [4.7(b)] where we have used the property of cross-product. Equation [4.7 (a)] constitutes our basic equation for the magnetic field. The proportionality constant in SI units has the exact value, 7 0 10 Tm/A 4 μ − = π [4.7(c)] We call μ 0 the permeability of free space (or vacuum). The Biot-Savart law for the magnetic field has certain similarities, as well as, differences with the Coulomb’s law for the electrostatic field. Some of these are: (i) Both are long range, since both depend inversely on the square of distance from the source to the point of interest. The principle of superposition applies to both fields. [In this connection, note that the magnetic field is linear in the source I d l just as the electrostatic field is linear in its source: the electric charge.] (ii) The electrostatic field is produced by a scalar source, namely, the electric charge. The magnetic field is produced by a vector source I d l . (iii) The electrostatic field is along the displacement vector joining the source and the field point. The magnetic field is perpendicular to the plane containing the displacement vector r and the current element I d l . (iv) There is an angle dependence in the Biot-Savart law which is not present in the electrostatic case. In Fig. 4.7, the magnetic field at any point in the direction of d l (the dashed line) is zero. Along this line, θ = 0, sin θ = 0 and from Eq. [4.7(a)], |d B | = 0. There is an interesting relation between ε 0 , the permittivity of free space; μ 0 , the permeability of free space; and c , the speed of light in vacuum: ( ) 0 0 0 0 4 4 μ ε μ ε = π π ( ) 7 9 1 10 9 10 − = × 8 2 2 1 1 (3 10 ) c = = × We will discuss this connection further in Chapter 8 on the electromagnetic waves. Since the speed of light in vacuum is constant, the product μ 0 ε 0 is fixed in magnitude. Choosing the value of either ε 0 or μ 0 , fixes the value of the other. In SI units, μ 0 is fixed to be equal to 4 π × 10 –7 in magnitude. Example 4.4 An element ˆ ∆ = ∆ x i l is placed at the origin and carries a large current I = 10 A (Fig. 4.8). What is the magnetic field on the y - axis at a distance of 0.5 m. ∆ x = 1 cm. FIGURE 4.8 Reprint 2025-26115 Moving Charges and Magnetism Solution 0 2 d sin |d | 4 I l r μ θ = π B [using Eq. (4.7)] 2 d 10 m l x − = ∆ = , I = 10 A, r = 0.5 m = y , 7 0 T m / 4 10 A μ − π = θ = 90° ; sin θ = 1 7 2 2 10 10 10 d 25 10 − − − × × = × B = 4 × 10 –8 T The direction of the field is in the + z -direction. This is so since, ˆ ˆ d = ∆ × i × j x y r l ( ) ˆ ˆ y x = ∆ i × j ˆ y x = ∆ k We remind you of the following cyclic property of cross-products, ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ; ; × = × = × = i j k j k i k i j Note that the field is small in magnitude. In the next section, we shall use the Biot-Savart law to calculate the magnetic field due to a circular loop. 4.5 M AGNETIC F IELD ON THE A XIS OF A C IRCULAR C URRENT L OOP In this section, we shall evaluate the magnetic field due to a circular coil along its axis. The evaluation entails summing up the effect of infinitesimal current elements ( I d l ) mentioned in the previous section. We assume that the current I is steady and that the evaluation is carried out in free space (i.e., vacuum). Fig. 4.9 depicts a circular loop carrying a steady current I . The loop is placed in the plane with its centre at the origin O and has a radius R . The x - axis is the axis of the loop. We wish to calculate the magnetic field at the point P on this axis. Let x be the distance of P from the centre O of the loop. Consider a conducting element d l of the loop. This is shown in Fig. 4.9. The magnitude d B of the magnetic field due to d l is given by the Biot-Savart law [Eq. 4.7(a)], 0 3 4 = × r I d dB r μ π l (4.8) Now r 2 = x 2 + R 2 . Further, any element of the loop will be perpendicular to the displacement vector from the element to the axial point. For example, the element d l in Fig. 4.9 is in the plane, whereas, the displacement vector r from d l to the axial point P is in the plane. Hence | d l × r |= r d l . Thus, ( ) π 0 2 2 d d 4 I l B x R μ = + (4.9) FIGURE 4.9 Magnetic field on the axis of a current carrying circular loop of radius R. Shown are the magnetic field d B (due to a line element d l ) and its components along and perpendicular to the axis. E XAMPLE 4.4 Reprint 2025-26Physics 116 The direction of d B is shown in Fig. 4.9. It is perpendicular to the plane formed by d l and r . It has an x -component d B x and a component perpendicular to x -axis, d B ⊥ . When the components perpendicular to the x - axis are summed over, they cancel out and we obtain a null result. For example, the d B ⊥ component due to d l is cancelled by the contribution due to the diametrically opposite d l element, shown in Fig. 4.9. Thus, only the x - component survives. The net contribution along x - direction can be obtained by integrating d B x = d B cos θ over the loop. For Fig. 4.9, 2 2 1/ 2 cos ( ) R x R θ = + (4.10) From Eqs. (4.9) and (4.10), ( ) π 0 3/ 2 2 2 d d 4 x I l R B x R μ = + The summation of elements d l over the loop yields 2 π R , the circumference of the loop. Thus, the magnetic field at P due to entire circular loop is ( ) 2 0 3 / 2 2 2 ˆ ˆ 2 x I R B x R μ = = + B i i (4.11) As a special case of the above result, we may obtain the field at the centre of the loop. Here x = 0, and we obtain, 0 0 ˆ 2 I R μ = B i (4.12) The magnetic field lines due to a circular wire form closed loops and are shown in Fig. 4.10. The direction of the magnetic field is given by (another) right-hand thumb rule stated below: Curl the palm of your right hand around the circular wire with the fingers pointing in the direction of the current. The right-hand thumb gives the direction of the magnetic field. FIGURE 4.10 The magnetic field lines for a current loop. The direction of the field is given by the right-hand thumb rule described in the text. The upper side of the loop may be thought of as the north pole and the lower side as the south pole of a magnet. Reprint 2025-26117 Moving Charges and Magnetism E XAMPLE 4.5 Example 4.5 A straight wire carrying a current of 12 A is bent into a semi-circular arc of radius 2.0 cm as shown in Fig. 4.11(a). Consider the magnetic field B at the centre of the arc. (a) What is the magnetic field due to the straight segments? (b) In what way the contribution to B from the semicircle differs from that of a circular loop and in what way does it resemble? (c) Would your answer be different if the wire were bent into a semi-circular arc of the same radius but in the opposite way as shown in Fig. 4.11(b)? FIGURE 4.11 Solution (a) d l and r for each element of the straight segments are parallel. Therefore, d l × r = 0. Straight segments do not contribute to | B| . (b) For all segments of the semicircular arc, d l × r are all parallel to each other (into the plane of the paper). All such contributions add up in magnitude. Hence direction of B for a semicircular arc is given by the right-hand rule and magnitude is half that of a circular loop. Thus B is 1.9 × 10 –4 T normal to the plane of the paper going into it. (c) Same magnitude of B but opposite in direction to that in (b). Example 4.6 Consider a tightly wound 100 turn coil of radius 10 cm, carrying a current of 1 A. What is the magnitude of the magnetic field at the centre of the coil? Solution Since the coil is tightly wound, we may take each circular element to have the same radius R = 10 cm = 0.1 m. The number of turns N = 100. The magnitude of the magnetic field is, –7 2 0 –1 4 10 10 1 2 2 10 NI B R μ π × × × = = × 4 2 10 − = π × 4 6 28 10 T . − = × 4.6 A MPERE ’ S C IRCUITAL L AW There is an alternative and appealing way in which the Biot-Savart law may be expressed. Ampere’s circuital law considers an open surface with a boundary (Fig. 4.12). The surface has current passing through it. We consider the boundary to be made up of a number of small line elements. Consider one such element of length dl. We take the value of the tangential component of the magnetic field, B t , at this element and multiply it by the FIGURE 4.12 E XAMPLE 4.6 Reprint 2025-26Physics 118 length of that element dl [Note: B t dl= B . d l ]. All such products are added together. We consider the limit as the lengths of elements get smaller and their number gets larger. The sum then tends to an integral. Ampere’s law states that this integral is equal to μ 0 times the total current passing through the surface, i.e., “ B. d l = μ 0 I [4.13(a)] where I is the total current through the surface. The integral is taken over the closed loop coinciding with the boundary C of the surface. The relation above involves a sign-convention, given by the right-hand rule. Let the fingers of the right-hand be curled in the sense the boundary is traversed in the loop integral “ B. d l . Then the direction of the thumb gives the sense in which the current I is regarded as positive. For several applications, a much simplified version of Eq. [4.13(a)] proves sufficient. We shall assume that, in such cases, it is possible to choose the loop (called an amperian loop ) such that at each point of the loop, either (i) B is tangential to the loop and is a non-zero constant B, or (ii) B is normal to the loop, or (iii) B vanishes. Now, let L be the length (part) of the loop for which B is tangential . Let I e be the current enclosed by the loop. Then, Eq. (4.13) reduces to, BL = μ 0 I e [4.13(b)] When there is a system with a symmetry such as for a straight infinite current-carrying wire in Fig. 4.13, the Ampere’s law enables an easy evaluation of the magnetic field, much the same way Gauss’ law helps in determination of the electric field. This is exhibited in the Example 4.8 below. The boundary of the loop chosen is a circle and magnetic field is tangential to the circumference of the circle. The law gives, for the left hand side of Eq. [4.13 (b)], B. 2 π r . We find that the magnetic field at a distance r outside the wire is tangential and given by B × 2 π r = μ 0 I , B = μ 0 I / (2 π r ) (4.14) The above result for the infinite wire is interesting from several points of view. ANDRE AMPERE (1775 –1836) Andre Ampere (1775 – 1836) Andre Marie Ampere was a French physicist, mathematician and chemist who founded the science of electrodynamics. Ampere was a child prodigy who mastered advanced mathematics by the age of 12. Ampere grasped the significance of Oersted’s discovery. He carried out a large series of experiments to explore the relationship between current electricity and magnetism. These investigations culminated in 1827 with the publication of the ‘Mathematical Theory of Electrodynamic Pheno- mena Deduced Solely from Experiments’. He hypo- thesised that all magnetic phenomena are due to circulating electric currents. Ampere was humble and absent- minded. He once forgot an invitation to dine with the Emperor Napoleon. He died of pneumonia at the age of 61. His gravestone bears the epitaph: Tandem Feli x (Happy at last). (i) It implies that the field at every point on a circle of radius r , (with the wire along the axis), is same in magnitude. In other words, the magnetic field Reprint 2025-26119 Moving Charges and Magnetism E XAMPLE 4.7 possesses what is called a cylindrical symmetry . The field that normally can depend on three coordinates depends only on one: r . Whenever there is symmetry, the solutions simplify. (ii) The field direction at any point on this circle is tangential to it. Thus, the lines of constant magnitude of magnetic field form concentric circles. Notice now, in Fig. 4.1(c), the iron filings form concentric circles. These lines called magnetic field lines form closed loops. This is unlike the electrostatic field lines which originate from positive charges and end at negative charges. The expression for the magnetic field of a straight wire provides a theoretical justification to Oersted’s experiments. (iii) Another interesting point to note is that even though the wire is infinite, the field due to it at a non-zero distance is not infinite. It tends to blow up only when we come very close to the wire. The field is directly proportional to the current and inversely proportional to the distance from the (infinitely long) current source. (iv) There exists a simple rule to determine the direction of the magnetic field due to a long wire. This rule, called the right-hand rule * , is: Grasp the wire in your right hand with your extended thumb pointing in the direction of the current. Your fingers will curl around in the direction of the magnetic field. Ampere’s circuital law is not new in content from Biot-Savart law. Both relate the magnetic field and the current, and both express the same physical consequences of a steady electrical current. Ampere’s law is to Biot-Savart law, what Gauss’s law is to Coulomb’s law. Both, Ampere’s and Gauss’s law relate a physical quantity on the periphery or boundary (magnetic or electric field) to another physical quantity, namely, the source, in the interior (current or charge). We also note that Ampere’s circuital law holds for steady currents which do not fluctuate with time. The following example will help us understand what is meant by the term enclosed current. Example 4.7 Figure 4.13 shows a long straight wire of a circular cross-section (radius a ) carrying steady current I . The current I is uniformly distributed across this cross-section. Calculate the magnetic field in the region r < a and r > a . FIGURE 4.13 * Note that there are two distinct right-hand rules: One which gives the direction of B on the axis of current-loop and the other which gives direction of B for a straight conducting wire. Fingers and thumb play different roles in the two. Reprint 2025-26Physics 120 E XAMPLE 4.7 Solution (a) Consider the case r > a . The Amperian loop, labelled 2, is a circle concentric with the cross-section. For this loop, L = 2 π r I e = Current enclosed by the loop = I The result is the familiar expression for a long straight wire B (2 π r ) = μ 0 I π 0 2 I B r μ = [4.15(a)] 1 B r ∝ ( r > a ) Now the current enclosed I e is not I , but is less than this value. Since the current distribution is uniform, the current enclosed is, I I r a e =       π π 2 2 2 2 Ir a = Using Ampere’s law, π 2 0 2 (2 ) I r B r a μ = B I a r =       μ 0 2 2 π [4.15(b)] B ∝ r ( r < a ) FIGURE 4.14 Figure (4.14) shows a plot of the magnitude of B with distance r from the centre of the wire. The direction of the field is tangential to the respective circular loop (1 or 2) and given by the right-hand rule described earlier in this section. This example possesses the required symmetry so that Ampere’s law can be applied readily. It should be noted that while Ampere’s circuital law holds for any loop, it may not always facilitate an evaluation of the magnetic field in every case. For example, for the case of the circular loop discussed in Section 4.5, it cannot be applied to extract the simple expression B = μ 0 I/2R [Eq. (4.12)] for the field at the centre of the loop. However, there exists a large number of situations of high symmetry where the law can be conveniently applied. We shall use it in the next section to calculate Reprint 2025-26121 Moving Charges and Magnetism the magnetic field produced by a commonly used and very useful magnetic system: the solenoid . 4.7 T HE S OLENOID We shall discuss a long solenoid. By long solenoid we mean that the solenoid’s length is large compared to its radius. It consists of a long wire wound in the form of a helix where the neighbouring turns are closely spaced. So each turn can be regarded as a circular loop. The net magnetic field is the vector sum of the fields due to all the turns. Enamelled wires are used for winding so that turns are insulated from each other. Figure 4.15 displays the magnetic field lines for a finite solenoid. We show a section of this solenoid in an enlarged manner in Fig. 4.15(a). Figure 4.15(b) shows the entire finite solenoid with its magnetic field. In Fig. 4.15(a), it is clear from the circular loops that the field between two neighbouring turns vanishes. In Fig. 4.15(b), we see that the field at the interior mid-point P is uniform, strong and along the axis of the solenoid. The field at the exterior mid-point Q is weak and moreover is along the axis of the solenoid with no perpendicular or normal component. As the FIGURE 4.15 (a) The magnetic field due to a section of the solenoid which has been stretched out for clarity. Only the exterior semi-circular part is shown. Notice how the circular loops between neighbouring turns tend to cancel. (b) The magnetic field of a finite solenoid. FIGURE 4.16 The magnetic field of a very long solenoid. We consider a rectangular Amperian loop abcd to determine the field. Reprint 2025-26Physics 122 E XAMPLE 4.8 solenoid is made longer it appears like a long cylindrical metal sheet. Figure 4.16 represents this idealised picture. The field outside the solenoid approaches zero. We shall assume that the field outside is zero. The field inside becomes everywhere parallel to the axis. Consider a rectangular Amperian loop abcd. Along cd the field is zero as argued above. Along transverse sections bc and ad, the field component is zero. Thus, these two sections make no contribution. Let the field along ab be B . Thus, the relevant length of the Amperian loop is, L = h . Let n be the number of turns per unit length, then the total number of turns is nh . The enclosed current is, I e = I (n h ) , where I is the current in the solenoid. From Ampere’s circuital law [Eq. 4.13 (b)] BL = μ 0 I e , B h = μ 0 I (n h ) B = μ 0 n I (4.16) The direction of the field is given by the right-hand rule. The solenoid is commonly used to obtain a uniform magnetic field. We shall see in the next chapter that a large field is possible by inserting a soft iron core inside the solenoid. Example 4.8 A solenoid of length 0.5 m has a radius of 1 cm and is made up of 500 turns. It carries a current of 5 A. What is the magnitude of the magnetic field inside the solenoid? Solution The number of turns per unit length is, 500 1000 0.5 n = = turns/m The length l = 0.5 m and radius r = 0.01 m. Thus, l / a = 50 i.e., l >> a . Hence, we can use the long solenoid formula, namely, Eq. (4.20) B = μ 0 n I = 4 π × 10 –7 × 10 3 × 5 = 6.28 × 10 –3 T FIGURE 4.17 Two long straight parallel conductors carrying steady currents I a and I b and separated by a distance d. B a is the magnetic field set up by conductor ‘a’ at conductor ‘b’. 4.8 F ORCE BETWEEN T WO P ARALLEL C URRENTS , THE A MPERE We have learnt that there exists a magnetic field due to a conductor carrying a current which obeys the Biot-Savart law. Further, we have learnt that an external magnetic field will exert a force on a current-carrying conductor. This follows from the Lorentz force formula. Thus, it is logical to expect that two current-carrying conductors placed near each other will exert (magnetic) forces on each other. In the period 1820-25, Ampere studied the nature of this magnetic force and its dependence on the magnitude of the current, on the shape and size of the conductors, as well as, the distances between the conductors. In this section, we shall take the simple example of two parallel current- carrying conductors, which will perhaps help us to appreciate Ampere’s painstaking work. Reprint 2025-26123 Moving Charges and Magnetism Figure 4.17 shows two long parallel conductors a and b separated by a distance d and carrying (parallel) currents I a and I b , respectively. The conductor ‘a’ produces, the same magnetic field B a at all points along the conductor ‘b’. The right-hand rule tells us that the direction of this field is downwards (when the conductors are placed horizontally). Its magnitude",
            "col4": "DATA"
        },
        {
            "col1": "P1_Part_2",
            "col2": "K12PHY1 - 04 MOVING CHARGES AND MAGNETISM.pdf",
            "col3": "is given by Eq. [4.15(a)] or from Ampere’s circuital law, 0 2 a a I B d μ = π The conductor ‘b’ carrying a current I b will experience a sideways force due to the field B a . The direction of this force is towards the conductor ‘a’ (Verify this). We label this force as F ba , the force on a segment L of ‘b’ due to ‘a’. The magnitude of this force is given by Eq. (4.4), = ba b a F I LB μ π = 0 2 a b I I L d (4.17) It is of course possible to compute the force on ‘a’ due to ‘b’. From considerations similar to above we can find the force F ab , on a segment of length L of ‘a’ due to the current in ‘b’. It is equal in magnitude to F ba , and directed towards ‘b’. Thus, F ba = – F ab (4.18) Note that this is consistent with Newton’s third Law. Thus, at least for parallel conductors and steady currents, we have shown that the Biot-Savart law and the Lorentz force yield results in accordance with Newton’s third Law * . We have seen from above that currents flowing in the same direction attract each other. One can show that oppositely directed currents repel each other. Thus, Parallel currents attract, and antiparallel currents repel. This rule is the opposite of what we find in electrostatics. Like (same sign) charges repel each other, but like (parallel) currents attract each other. Let f ba represent the magnitude of the force F ba per unit length. Then, from Eq. (4.17), π 0 2 a b ba I I f d μ = (4.19) The above expression is used to define the ampere (A), which is one of the seven SI base units. * It turns out that when we have time-dependent currents and/or charges in motion, Newton’s third law may not hold for forces between charges and/or conductors. An essential consequence of the Newton’s third law in mechanics is conservation of momentum of an isolated system. This, however, holds even for the case of time-dependent situations with electromagnetic fields, provided the momentum carried by fields is also taken into account. Reprint 2025-26Physics 124 E XAMPLE 4.9 The ampere is the value of that steady current which, when maintained in each of the two very long, straight, parallel conductors of negligible cross-section, and placed one metre apart in vacuum, would produce on each of these conductors a force equal to 2 × 10 –7 newtons per metre of length. This definition of the ampere was adopted in 1946. It is a theoretical definition. In practice, one must eliminate the effect of the earth’s magnetic field and substitute very long wires by multiturn coils of appropriate geometries. An instrument called the current balance is used to measure this mechanical force. The SI unit of charge, namely, the coulomb, can now be defined in terms of the ampere. When a steady current of 1A is set up in a conductor, the quantity of charge that flows through its cross-section in 1s is one coulomb (1C). Example 4.9 The horizontal component of the earth’s magnetic field at a certain place is 3.0 ×10 –5 T and the direction of the field is from the geographic south to the geographic north. A very long straight conductor is carrying a steady current of 1A. What is the force per unit length on it when it is placed on a horizontal table and the direction of the current is (a) east to west; (b) south to north? Solution F = I l × B F = IlB sin θ The force per unit length is f = F / l = I B sin θ (a) When the current is flowing from east to west, θ = 90° Hence, f = I B = 1 × 3 × 10 –5 = 3 × 10 –5 N m –1 This is larger than the value 2×10 –7 Nm –1 quoted in the definition of the ampere. Hence it is important to eliminate the effect of the earth’s magnetic field and other stray fields while standardising the ampere. The direction of the force is downwards. This direction may be obtained by the directional property of cross product of vectors. (b) When the current is flowing from south to north, θ = 0 o f = 0 Hence there is no force on the conductor. 4.9 T ORQUE ON C URRENT L OOP , M AGNETIC D IPOLE 4.9.1 Torque on a rectangular current loop in a uniform magnetic field We now show that a rectangular loop carrying a steady current I and placed in a uniform magnetic field experiences a torque. It does not experience a net force. This behaviour is analogous to that of electric dipole in a uniform electric field (Section 1.11). Reprint 2025-26125 Moving Charges and Magnetism We first consider the simple case when the rectangular loop is placed such that the uniform magnetic field B is in the plane of the loop. This is illustrated in Fig. 4.18(a). The field exerts no force on the two arms AD and BC of the loop. It is perpendicular to the arm AB of the loop and exerts a force F 1 on it which is directed into the plane of the loop. Its magnitude is, F 1 = I b B Similarly, it exerts a force F 2 on the arm CD and F 2 is directed out of the plane of the paper. F 2 = I b B = F 1 Thus, the net force on the loop is zero . There is a torque on the loop due to the pair of forces F 1 and F 2 . Figure 4.18(b) shows a view of the loop from the AD end. It shows that the torque on the loop tends to rotate it anticlockwise. This torque is (in magnitude), 1 2 2 2 a a F F τ = + ( ) 2 2 a a IbB IbB I ab B = + = = I A B (4.20) where A = ab is the area of the rectangle. We next consider the case when the plane of the loop, is not along the magnetic field, but makes an angle with it. We take the angle between the field and the normal to the coil to be angle θ (The previous case corresponds to θ = π /2) . Figure 4.19 illustrates this general case. The forces on the arms BC and DA are equal, opposite, and act along the axis of the coil, which connects the centres of mass of BC and DA. Being collinear along the axis they cancel each other, resulting in no net force or torque. The forces on arms AB and CD are F 1 and F 2 . They too are equal and opposite, with magnitude, F 1 = F 2 = I b B But they are not collinear! This results in a couple as before. The torque is, however, less than the earlier case when plane of loop was along the magnetic field. This is because the perpendicular distance between the forces of the couple has decreased. Figure 4.19(b) is a view of the arrangement from the AD end and it illustrates these two forces constituting a couple. The magnitude of the torque on the loop is, 1 2 sin sin 2 2 a a F F τ θ θ = + = I ab B sin θ = I A B sin θ (4.21) FIGURE 4.18 (a) A rectangular current-carrying coil in uniform magnetic field. The magnetic moment m points downwards. The torque τ τ τ τ τ is along the axis and tends to rotate the coil anticlockwise. (b) The couple acting on the coil. Reprint 2025-26Physics 126 E XAMPLE 4.10 As θ ‡ 0, the perpendicular distance between the forces of the couple also approaches zero. This makes the forces collinear and the net force and torque zero. The torques in Eqs. (4.20) and (4.21) can be expressed as vector product of the magnetic moment of the coil and the magnetic field. We define the magnetic moment of the current loop as, m = I A (4.22) where the direction of the area vector A is given by the right-hand thumb rule and is directed into the plane of the paper in Fig. 4.18. Then as the angle between m and B is θ , Eqs. (4.20) and (4.21) can be expressed by one expression (4.23) This is analogous to the electrostatic case (Electric dipole of dipole moment p e in an electric field E ). τ = p E × e As is clear from Eq. (4.22), the dimensions of the magnetic moment are [A][L 2 ] and its unit is Am 2 . From Eq. (4.23), we see that the torque τ τ τ τ τ vanishes when m is either parallel or antiparallel to the magnetic field B . This indicates a state of equilibrium as there is no torque on the coil (this also applies to any object with a magnetic moment m ). When m and B are parallel the equilibrium is a stable one. Any small rotation of the coil produces a torque which brings it back to its original position. When they are antiparallel, the equilibrium is unstable as any rotation produces a torque which increases with the amount of rotation. The presence of this torque is also the reason why a small magnet or any magnetic dipole aligns itself with the external magnetic field. If the loop has N closely wound turns, the expression for torque, Eq. (4.23), still holds, with m = N I A (4.24) Example 4.10 A 100 turn closely wound circular coil of radius 10 cm carries a current of 3.2 A. (a) What is the field at the centre of the coil? (b) What is the magnetic moment of this coil? The coil is placed in a vertical plane and is free to rotate about a horizontal axis which coincides with its diameter. A uniform magnetic field of 2T in the horizontal direction exists such that initially the axis of the coil is in the direction of the field. The coil rotates through an angle of 90° under the influence of the magnetic field. (c) What are the magnitudes of the torques on the coil in the initial and final position? (d) What is the angular speed acquired by the coil when it has rotated by 90°? The moment of inertia of the coil is 0.1 kg m 2 . FIGURE 4.19 (a) The area vector of the loop ABCD makes an arbitrary angle θ with the magnetic field. (b) Top view of the loop. The forces F 1 and F 2 acting on the arms AB and CD are indicated. Reprint 2025-26127 Moving Charges and Magnetism E XAMPLE 4.11 E XAMPLE 4.10 Solution (a) From Eq. (4.12) 0 2 NI B R μ = Here, N = 100; I = 3.2 A, and R = 0.1 m. Hence, = × × × − − 4 10 10 2 10 5 1 (using π × 3.2 = 10) = 2 × 1 0 –3 T The direction is given by the right-hand thumb rule. (b) The magnetic moment is given by Eq. (4.24), m = N I A = N I π r 2 = 100 × 3.2 × 3.14 × 10 –2 = 10 A m 2 The direction is once again given by the right-hand thumb rule. (c) τ = m × B [from Eq. (4.23)] = m B sin θ Initially, θ = 0. Thus, initial torque τ i = 0. Finally, θ = π /2 (or 90 º ). Thus, final torque τ f = m B = 10 × 2 = 20 N m. (d) From Newton’s second law, I where I is the moment of inertia of the coil. From chain rule, d d d d d d d d = = t t ω ω θ ω ω θ θ Using this, I d sin d = m B ω ω θ θ Integrating from θ = 0 to θ = π /2, Example 4.11 (a) A current-carrying circular loop lies on a smooth horizontal plane. Can a uniform magnetic field be set up in such a manner that the loop turns around itself (i.e., turns about the vertical axis). (b) A current-carrying circular loop is located in a uniform external magnetic field. If the loop is free to turn, what is its orientation of stable equilibrium? Show that in this orientation, the flux of Reprint 2025-26Physics 128 E XAMPLE 4.11 the total field (external field + field produced by the loop) is maximum. (c) A loop of irregular shape carrying current is located in an external magnetic field. If the wire is flexible, why does it change to a circular shape? Solution (a) No, because that would require τ τ τ τ τ to be in the vertical direction. But τ τ τ τ τ = I A × B , and since A of the horizontal loop is in the vertical direction, τ would be in the plane of the loop for any B . (b) Orientation of stable equilibrium is one where the area vector A of the loop is in the direction of external magnetic field. In this orientation, the magnetic field produced by the loop is in the same direction as external field, both normal to the plane of the loop, thus giving rise to maximum flux of the total field. (c) It assumes circular shape with its plane normal to the field to maximise flux, since for a given perimeter, a circle encloses greater area than any other shape. 4.9.2 Circular current loop as a magnetic dipole In this section, we shall consider the elementary magnetic element: the current loop. We shall show that the magnetic field (at large distances) due to current in a circular current loop is very similar in behaviour to the electric field of an electric dipole. In Section 4.5, we have evaluated the magnetic field on the axis of a circular loop, of a radius R , carrying a steady current I . The magnitude of this field is [(Eq. (4.11)], ( ) μ = + 2 0 3 / 2 2 2 2 I R B x R and its direction is along the axis and given by the right-hand thumb rule (Fig. 4.10). Here, x is the distance along the axis from the centre of the loop. For x >> R , we may drop the R 2 term in the denominator. Thus, μ = 2 0 3 2 IR B x Note that the area of the loop A = π R 2 . Thus, μ = π 0 3 2 IA B x As earlier, we define the magnetic moment m to have a magnitude IA, m = I A . Hence, B m ≃ μ 0 3 2 π x π μ = 0 3 2 4 x m [4.25(a)] The expression of Eq. [4.25(a)] is very similar to an expression obtained earlier for the electric field of a dipole. The similarity may be seen if we substitute, μ ε → 0 0 1/ Reprint 2025-26129 Moving Charges and Magnetism e → m p (electrostatic dipole) → B E (electrostatic field) We then obtain, 3 0 2 4 e x ε = π p E which is precisely the field for an electric dipole at a point on its axis. considered in Chapter 1, Section 1.9 [Eq. (1.20)]. It can be shown that the above analogy can be carried further. We had found in Chapter 1 that the electric field on the perpendicular bisector of the dipole is given by [See Eq.(1.21)], E ≃ p e x 4 0 3 π ε where x is the distance from the dipole. If we replace p ‡ m and 0 0 1/ μ ε → in the above expression, we obtain the result for B for a point in the plane of the loop at a distance x from the centre. For x >>R, B m ≃ μ 0 3 4 π x x R ; >> [4.25(b)] The results given by Eqs. [4.25(a)] and [4.25(b)] become exact for a point magnetic dipole. The results obtained above can be shown to apply to any planar loop: a planar current loop is equivalent to a magnetic dipole of dipole moment m = I A , which is the analogue of electric dipole moment p . Note, however, a fundamental difference: an electric dipole is built up of two elementary units — the charges (or electric monopoles). In magnetism, a magnetic dipole (or a current loop) is the most elementary element. The equivalent of electric charges, i.e., magnetic monopoles, are not known to exist. We have shown that a current loop (i) produces a magnetic field (see Fig. 4 . 10) and behaves like a magnetic dipole at large distances, and (ii) is subject to torque like a magnetic needle. This led Ampere to suggest that all magnetism is due to circulating currents. This seems to be partly true and no magnetic monopoles have been seen so far. However, elementary particles such as an electron or a proton also carry an intrinsic magnetic moment, not accounted by circulating currents. 4.10 T HE M OVING C OIL G ALVANOMETER Currents and voltages in circuits have been discussed extensively in Chapters 3. But how do we measure them? How do we claim that current in a circuit is 1.5 A or the voltage drop across a resistor is 1.2 V? Figure 4.20 exhibits a very useful instrument for this purpose: the moving coil galvanometer (MCG). It is a device whose principle can be understood on the basis of our discussion in Section 4.9. The galvanometer consists of a coil, with many turns, free to rotate about a fixed axis (Fig. 4.20), in a uniform radial magnetic field. There is a cylindrical soft iron core which not only makes the field radial but also increases the strength of the magnetic field. When a current flows through the coil, a torque acts on it. This torque is given by Eq. (4.20) to be τ = NI AB Reprint 2025-26Physics 130 where the symbols have their usual meaning. Since the field is radial by design, we have taken sin θ = 1 in the above expression for the torque. The magnetic torque NIAB tends to rotate the coil. A spring S p provides a counter torque k φ that balances the magnetic torque NIAB; resulting in a steady angular deflection φ . In equilibrium k φ = NI AB where k is the torsional constant of the spring; i.e. the restoring torque per unit twist. The deflection φ is indicated on the scale by a pointer attached to the spring. We have φ =       NAB k I (4.26) The quantity in brackets is a constant for a given galvanometer. The galvanometer can be used in a number of ways. It can be used as a detector to check if a current is flowing in the circuit. We have come across this usage in the Wheatstone’s bridge arrangement. In this usage the neutral position of the pointer (when no current is flowing through the galvanometer) is in the middle of the scale and not at the left end as shown in Fig.4.20. Depending on the direction of the current, the pointer’s deflection is either to the right or the left. The galvanometer cannot as such be used as an ammeter to measure the value of the current in a given circuit. This is for two reasons: (i) Galvanometer is a very sensitive device, it gives a full- scale deflection for a current of the order of μ A. (ii) For measuring currents, the galvanometer has to be connected in series, and as it has a large resistance, this will change the value of the current in the circuit. To overcome these difficulties, one attaches a small resistance r s , called shunt resistance , in parallel with the galvanometer coil; so that most of the current passes through the shunt. The resistance of this arrangement is, R G r s / (R G + r s ) ≃ r s if R G >> r s If r s has small value, in relation to the resistance of the rest of the circuit R c , the effect of introducing the measuring instrument is also small and negligible. This arrangement is schematically shown in Fig. 4.21. The scale of this ammeter is calibrated and then graduated to read off the current value with ease. We define the current sensitivity of the galvanometer as the deflection per unit current . From Eq. (4.26) this current sensitivity is, NAB I k φ = (4.27) A convenient way for the manufacturer to increase the sensitivity is to increase the number of turns N . We choose galvanometers having sensitivities of value, required by our experiment. FIGURE 4.20 The moving coil galvanometer. Its elements are described in the text. Depending on the requirement, this device can be used as a current detector or for measuring the value of the current (ammeter) or voltage (voltmeter). FIGURE 4.21 Conversion of a galvanometer (G) to an ammeter by the introduction of a shunt resistance r s of very small value in parallel. Reprint 2025-26131 Moving Charges and Magnetism The galvanometer can also be used as a voltmeter to measure the voltage across a given section of the circuit. For this it must be connected in parallel with that section of the circuit. Further, it must draw a very small current, otherwise the voltage measurement will disturb the original set up by an amount which is very large. Usually we like to keep the disturbance due to the measuring device below one per cent. To ensure this, a large resistance R is connected in series with the galvanometer. This arrangement is schematically depicted in Fig.4.22. Note that the resistance of the voltmeter is now, R G + R ≃ R : large The scale of the voltmeter is calibrated to read off the voltage value with ease. We define the voltage sensitivity as the deflection per unit voltage . From Eq. (4.26), φ V NAB k I V NAB k R =       =       1 (4.28) An interesting point to note is that increasing the current sensitivity may not necessarily increase the voltage sensitivity. Let us take Eq. (4.27) which provides a measure of current sensitivity. If N → 2N , i.e., we double the number of turns, then 2 I I φ φ → Thus, the current sensitivity doubles. However, the resistance of the galvanometer is also likely to double, since it is proportional to the length of the wire. In Eq. (4.28), N → 2N , and R → 2R , thus the voltage sensitivity, V V φ φ → remains unchanged. So in general, the modification needed for conversion of a galvanometer to an ammeter will be different from what is needed for converting it into a voltmeter. Example 4.12 In the circuit (Fig. 4.23) the current is to be measured. What is the value of the current if the ammeter shown (a) is a galvanometer with a resistance R G = 60.00 Ω ; (b) is a galvanometer described in (a) but converted to an ammeter by a shunt resistance r s = 0.02 Ω ; (c) is an ideal ammeter with zero resistance? FIGURE 4.23 FIGURE 4.22 Conversion of a galvanometer (G) to a voltmeter by the introduction of a resistance R of large value in series. E XAMPLE 4.12 Reprint 2025-26Physics 132 E XAMPLE 4.12 Solution (a) Total resistance in the circuit is, 3 63 G R + = Ω . Hence, I = 3/63 = 0.048 A. (b) Resistance of the galvanometer converted to an ammeter is, R r R r G s G s + = × + 60 0 02 60 0 02 Ω Ω Ω . ( . ) ≃ 0.02 Ω Total resistance in the circuit is, 0.02 3 3.02 Ω + Ω = Ω . Hence, I = 3/3.02 = 0.99 A. (c) For the ideal ammeter with zero resistance, I = 3/3 = 1.00 A Reprint 2025-26",
            "col4": "DATA"
        }
    ],
    "projectDeleted": false,
    "codeFrameColumn": [
        {
            "col1": "col1",
            "col2": "col3",
            "researchObjective": "",
            "filterColumn": [],
            "title": "K12PHY1 - 04 MOVING CHARGES AND MAGNETISM",
            "netSetting": true,
            "subNetSetting": true,
            "categorySetting": true,
            "sentimentSetting": true,
            "thinkingBudgetSetting": 2000,
            "questionType": "OPENEND",
            "otherBrand": false,
            "existingCodeFrame": [],
            "isList": false,
            "codeFrameWithDataList": [],
            "codeFrameList": [],
            "onlyCodeFrame": true,
            "errorList": [],
            "codeFrameGenerateType": "CODE_FRAME_ONLY",
            "requestType": "CODEFRAME",
            "requestMode": "DEVELOPMENT",
            "sourceLanguage": "English",
            "targetLanguage": "English",
            "response": {
                "status": "OK",
                "data": {
                    "status": "OK",
                    "PromptTokenCount": 15848,
                    "CandidatesTokenCount": 3370,
                    "TotalTokenCount": 20588,
                    "thoughtsTokenCount": 1370,
                    "finishReason": "STOP",
                    "data": [
                        {
                            "Net": "Introduction to Magnetism & Historical Context",
                            "Subnets": [
                                {
                                    "Subnet": "Historical Discoveries",
                                    "Categories": [
                                        {
                                            "Category": "Oersted's Experiment",
                                            "Summary": "Describes Hans Christian Oersted's observation of a compass needle deflection near a current-carrying wire, demonstrating the magnetic effect of electric currents.",
                                            "code": 3
                                        },
                                        {
                                            "Category": "Unification of EM by Maxwell",
                                            "Summary": "Mentions James Maxwell's work in unifying electricity and magnetism, leading to the understanding of light as electromagnetic waves.",
                                            "code": 4
                                        },
                                        {
                                            "Category": "EM Wave Discoveries",
                                            "Summary": "Notes discoveries of radio waves by Hertz, Bose, and Marconi, highlighting technological progress in electromagnetism.",
                                            "code": 5
                                        }
                                    ],
                                    "Summary": "Focuses on the initial key experiments and realizations about electricity and magnetism.",
                                    "code": 2
                                },
                                {
                                    "Subnet": "Basic Concepts & Conventions",
                                    "Categories": [
                                        {
                                            "Category": "Magnetic Field Production",
                                            "Summary": "States that moving charges or currents produce magnetic fields in the surrounding space.",
                                            "code": 7
                                        },
                                        {
                                            "Category": "Field Representation Conventions",
                                            "Summary": "Explains the conventions for depicting currents or fields emerging from (dot) or going into (cross) the plane of the paper.",
                                            "code": 8
                                        },
                                        {
                                            "Category": "Superposition Principle for B-field",
                                            "Summary": "Notes that magnetic fields obey the principle of superposition, allowing vector addition of fields from multiple sources.",
                                            "code": 9
                                        }
                                    ],
                                    "Summary": "Outlines foundational ideas and representational conventions used throughout the study of magnetism.",
                                    "code": 6
                                }
                            ],
                            "Summary": "This section introduces the historical background of electromagnetism, particularly Oersted's pivotal discovery linking electric currents to magnetic fields, laying the groundwork for further understanding.",
                            "code": 1
                        },
                        {
                            "Net": "Magnetic Force",
                            "Subnets": [
                                {
                                    "Subnet": "Lorentz Force on Moving Charges",
                                    "Categories": [
                                        {
                                            "Category": "Lorentz Force Equation",
                                            "Summary": "Defines the total force (electric and magnetic) on a point charge, F = q[E + (v × B)], introduced by H.A. Lorentz.",
                                            "code": 12
                                        },
                                        {
                                            "Category": "Characteristics of Magnetic Force",
                                            "Summary": "Details properties such as dependence on charge, velocity, and magnetic field, its perpendicular direction to v and B, and its being zero for stationary charges or parallel/anti-parallel v and B.",
                                            "code": 13
                                        },
                                        {
                                            "Category": "Unit of Magnetic Field (Tesla)",
                                            "Summary": "Defines the SI unit of magnetic field, Tesla (T), and its relation to gauss, along with Earth's magnetic field strength.",
                                            "code": 14
                                        }
                                    ],
                                    "Summary": "Describes the force experienced by a point charge moving in combined electric and magnetic fields.",
                                    "code": 11
                                },
                                {
                                    "Subnet": "Magnetic Force on Current-Carrying Conductors",
                                    "Categories": [
                                        {
                                            "Category": "Force on Straight Wire",
                                            "Summary": "Derives the force on a straight rod carrying current in an external magnetic field, F = I (l × B).",
                                            "code": 16
                                        },
                                        {
                                            "Category": "Force on Arbitrary Shape Wire",
                                            "Summary": "Explains how to calculate the Lorentz force on a wire of arbitrary shape by integrating over infinitesimal linear strips.",
                                            "code": 17
                                        },
                                        {
                                            "Category": "Mid-Air Suspension Example",
                                            "Summary": "Illustrates calculating the magnetic field required to suspend a current-carrying wire against gravity.",
                                            "code": 18
                                        }
                                    ],
                                    "Summary": "Extends the concept of magnetic force from single charges to macroscopic current-carrying conductors.",
                                    "code": 15
                                },
                                {
                                    "Subnet": "Force Between Parallel Currents",
                                    "Categories": [
                                        {
                                            "Category": "Direction of Force",
                                            "Summary": "States that parallel currents attract each other, and anti-parallel currents repel each other, opposite to electrostatics.",
                                            "code": 20
                                        },
                                        {
                                            "Category": "Magnitude of Force per Unit Length",
                                            "Summary": "Provides the formula for the force per unit length between two long parallel current-carrying conductors, f/L = μ₀ I_a I_b / (2πd).",
                                            "code": 21
                                        },
                                        {
                                            "Category": "Definition of Ampere",
                                            "Summary": "Explains how the force between parallel conductors is used to define the SI base unit of current, the Ampere.",
                                            "code": 22
                                        }
                                    ],
                                    "Summary": "Discusses the mutual magnetic forces between two parallel current-carrying conductors, which led to the definition of the Ampere.",
                                    "code": 19
                                }
                            ],
                            "Summary": "This section details the fundamental forces exerted by magnetic fields on moving charges and current-carrying conductors, including the Lorentz force equation.",
                            "code": 10
                        },
                        {
                            "Net": "Magnetic Fields from Currents",
                            "Subnets": [
                                {
                                    "Subnet": "Biot-Savart Law",
                                    "Categories": [
                                        {
                                            "Category": "Law Definition & Equation",
                                            "Summary": "Defines the Biot-Savart law for the magnetic field dB due to an infinitesimal current element I dl, d B = (μ₀/4π) * (I dl × r / r³).",
                                            "code": 25
                                        },
                                        {
                                            "Category": "Similarities with Coulomb's Law",
                                            "Summary": "Highlights similarities like inverse square distance dependence and superposition principle.",
                                            "code": 26
                                        },
                                        {
                                            "Category": "Differences from Coulomb's Law",
                                            "Summary": "Points out differences such as vector source (I dl) vs. scalar source (charge), perpendicular field direction, and angle dependence.",
                                            "code": 27
                                        }
                                    ],
                                    "Summary": "Introduces the Biot-Savart law as a fundamental method for calculating magnetic fields produced by current elements.",
                                    "code": 24
                                },
                                {
                                    "Subnet": "Magnetic Field of Circular Current Loop",
                                    "Categories": [
                                        {
                                            "Category": "Field on Axis of Loop",
                                            "Summary": "Derives the magnetic field along the axis of a circular current loop, B_x = (μ₀ I R² / 2(x² + R²)^(3/2)) i-hat.",
                                            "code": 29
                                        },
                                        {
                                            "Category": "Field at Center of Loop",
                                            "Summary": "Specifies the magnetic field at the center of a circular loop (x=0), B = μ₀ I / (2R).",
                                            "code": 30
                                        },
                                        {
                                            "Category": "Right-Hand Thumb Rule for Loops",
                                            "Summary": "Explains the right-hand thumb rule for determining the direction of the magnetic field due to a circular current loop.",
                                            "code": 31
                                        }
                                    ],
                                    "Summary": "Applies the Biot-Savart law to calculate the magnetic field produced by a circular current loop, specifically on its axis and at its center.",
                                    "code": 28
                                },
                                {
                                    "Subnet": "Ampere's Circuital Law",
                                    "Categories": [
                                        {
                                            "Category": "Law Definition & Equation",
                                            "Summary": "States Ampere's circuital law: ∫ B · dl = μ₀ I_enclosed, relating the line integral of the magnetic field around a closed loop to the total current passing through the surface bounded by the loop.",
                                            "code": 33
                                        },
                                        {
                                            "Category": "Applications for Symmetry",
                                            "Summary": "Explains that Ampere's law simplifies magnetic field calculations for systems with high symmetry (e.g., straight wires).",
                                            "code": 34
                                        },
                                        {
                                            "Category": "Right-Hand Rule for Ampere's Law",
                                            "Summary": "Describes the right-hand rule for determining the positive direction of enclosed current based on the loop traversal direction.",
                                            "code": 35
                                        }
                                    ],
                                    "Summary": "Presents Ampere's circuital law as an alternative, often simpler, method for calculating magnetic fields in highly symmetric situations.",
                                    "code": 32
                                },
                                {
                                    "Subnet": "Magnetic Field of Solenoids",
                                    "Categories": [
                                        {
                                            "Category": "Definition & Characteristics",
                                            "Summary": "Describes a solenoid as a long wire wound in a helix, producing a uniform and strong magnetic field inside and a weak field outside.",
                                            "code": 37
                                        },
                                        {
                                            "Category": "Field Inside a Long Solenoid",
                                            "Summary": "Derives the magnetic field inside a long solenoid using Ampere's law, B = μ₀ n I, where n is the number of turns per unit length.",
                                            "code": 38
                                        },
                                        {
                                            "Category": "Uniform Field Application",
                                            "Summary": "Highlights that solenoids are commonly used to obtain uniform magnetic fields.",
                                            "code": 39
                                        }
                                    ],
                                    "Summary": "Explores the magnetic field characteristics and calculation for long solenoids, a common magnetic system.",
                                    "code": 36
                                }
                            ],
                            "Summary": "This net focuses on the fundamental laws and principles governing the generation of magnetic fields by electric currents, and their application to various configurations like wires, loops, and solenoids.",
                            "code": 23
                        },
                        {
                            "Net": "Motion of Charges in Magnetic Fields",
                            "Subnets": [
                                {
                                    "Subnet": "Circular Motion in Perpendicular B-field",
                                    "Categories": [
                                        {
                                            "Category": "Centripetal Force from B-field",
                                            "Summary": "Explains that when velocity is perpendicular to the magnetic field, the magnetic force (qvB) acts as a centripetal force, leading to circular motion.",
                                            "code": 42
                                        },
                                        {
                                            "Category": "Radius of Circular Path",
                                            "Summary": "Derives the radius of the circular path of a charged particle: r = mv / (qB).",
                                            "code": 43
                                        },
                                        {
                                            "Category": "Angular Frequency (Cyclotron Frequency)",
                                            "Summary": "Defines the angular frequency ω = qB/m, noting its independence from velocity or energy, which is important for cyclotron design.",
                                            "code": 44
                                        }
                                    ],
                                    "Summary": "Details the behavior of charged particles moving perpendicular to a uniform magnetic field, resulting in circular trajectories.",
                                    "code": 41
                                },
                                {
                                    "Subnet": "Helical Motion in Angled B-field",
                                    "Categories": [
                                        {
                                            "Category": "Velocity Component Parallel to B",
                                            "Summary": "Explains that the velocity component parallel to the magnetic field remains unaffected.",
                                            "code": 46
                                        },
                                        {
                                            "Category": "Combination of Motions",
                                            "Summary": "Describes how the combination of circular motion (from perpendicular velocity) and linear motion (from parallel velocity) results in a helical path.",
                                            "code": 47
                                        },
                                        {
                                            "Category": "Pitch of Helix",
                                            "Summary": "Defines the pitch of the helical path as the distance moved along the magnetic field in one rotation, p = v_parallel * T.",
                                            "code": 48
                                        }
                                    ],
                                    "Summary": "Describes the more general case of charged particle motion when the velocity has components both parallel and perpendicular to the magnetic field, leading to helical trajectories.",
                                    "code": 45
                                },
                                {
                                    "Subnet": "Work Done by Magnetic Force",
                                    "Categories": [
                                        {
                                            "Category": "No Work Done by Magnetic Force",
                                            "Summary": "States that the magnetic force does no work on the particle because it is always perpendicular to the velocity.",
                                            "code": 50
                                        },
                                        {
                                            "Category": "No Change in Kinetic Energy/Speed",
                                            "Summary": "Concludes that the magnitude of the velocity (speed) and kinetic energy of the particle remain unchanged, though the direction of momentum changes.",
                                            "code": 51
                                        }
                                    ],
                                    "Summary": "Analyzes the work-energy implications of magnetic forces on moving charges.",
                                    "code": 49
                                }
                            ],
                            "Summary": "This net describes the dynamics of charged particles when subjected to magnetic fields, including circular and helical paths and the concept of work done by magnetic forces.",
                            "code": 40
                        },
                        {
                            "Net": "Magnetic Dipoles & Torque",
                            "Subnets": [
                                {
                                    "Subnet": "Torque on a Current Loop",
                                    "Categories": [
                                        {
                                            "Category": "Rectangular Loop in Uniform B-field",
                                            "Summary": "Analyzes the forces and torque on a rectangular current loop placed in a uniform magnetic field.",
                                            "code": 54
                                        },
                                        {
                                            "Category": "Torque Equation",
                                            "Summary": "Derives the torque acting on a current loop as τ = IAB sinθ or vectorially as τ = m × B, where A is the area and θ is the angle between the area vector and B.",
                                            "code": 55
                                        },
                                        {
                                            "Category": "Equilibrium Positions",
                                            "Summary": "Identifies stable (m parallel to B) and unstable (m anti-parallel to B) equilibrium states where the net torque is zero.",
                                            "code": 56
                                        }
                                    ],
                                    "Summary": "Explores the interaction of current-carrying loops with external magnetic fields, leading to the generation of torque.",
                                    "code": 53
                                },
                                {
                                    "Subnet": "Current Loop as a Magnetic Dipole",
                                    "Categories": [
                                        {
                                            "Category": "Definition of Magnetic Moment",
                                            "Summary": "Defines the magnetic moment m of a current loop as m = IA (or NIA for N turns), with its direction given by the right-hand thumb rule.",
                                            "code": 58
                                        },
                                        {
                                            "Category": "Magnetic Field of Dipole at Large Distances",
                                            "Summary": "Shows that at large distances, the magnetic field of a circular current loop resembles that of an electric dipole.",
                                            "code": 59
                                        },
                                        {
                                            "Category": "Analogy with Electric Dipole",
                                            "Summary": "Draws parallels between the magnetic moment (m) and electric dipole moment (p_e), and their respective fields.",
                                            "code": 60
                                        }
                                    ],
                                    "Summary": "Demonstrates how a current-carrying loop behaves as an elementary magnetic dipole, with a defined magnetic moment and characteristic field.",
                                    "code": 57
                                },
                                {
                                    "Subnet": "Absence of Magnetic Monopoles",
                                    "Categories": [
                                        {
                                            "Category": "Fundamental Difference",
                                            "Summary": "Highlights the fundamental difference that magnetic dipoles are the most elementary magnetic elements, unlike electric dipoles which are built from charges.",
                                            "code": 62
                                        },
                                        {
                                            "Category": "No Known Magnetic Monopoles",
                                            "Summary": "States that magnetic monopoles (equivalent to electric charges) are not known to exist.",
                                            "code": 63
                                        }
                                    ],
                                    "Summary": "Discusses the fundamental difference between magnetic and electric dipoles, particularly the observed absence of magnetic monopoles.",
                                    "code": 61
                                }
                            ],
                            "Summary": "This net investigates the behavior of current loops as magnetic dipoles, including the torque they experience in a magnetic field and the concept of magnetic moment.",
                            "code": 52
                        },
                        {
                            "Net": "Measurement & Devices",
                            "Subnets": [
                                {
                                    "Subnet": "Moving Coil Galvanometer (MCG)",
                                    "Categories": [
                                        {
                                            "Category": "Principle & Construction",
                                            "Summary": "Explains the working principle of the MCG based on the torque on a current-carrying coil in a radial magnetic field, with elements like a coil, spring, and soft iron core.",
                                            "code": 66
                                        },
                                        {
                                            "Category": "Deflection Equation",
                                            "Summary": "Derives the angular deflection φ = (NAB/k)I, showing proportionality between deflection and current.",
                                            "code": 67
                                        },
                                        {
                                            "Category": "Current Sensitivity",
                                            "Summary": "Defines current sensitivity as deflection per unit current, φ/I = NAB/k, and factors influencing it.",
                                            "code": 68
                                        }
                                    ],
                                    "Summary": "Details the construction, working principle, and characteristics of the moving coil galvanometer.",
                                    "code": 65
                                },
                                {
                                    "Subnet": "Galvanometer as Ammeter",
                                    "Categories": [
                                        {
                                            "Category": "Conversion Method",
                                            "Summary": "Explains how a galvanometer is converted into an ammeter by connecting a small shunt resistance (r_s) in parallel with the coil.",
                                            "code": 70
                                        },
                                        {
                                            "Category": "Reasons for Shunt Resistance",
                                            "Summary": "Discusses why a shunt is needed: to bypass most current (due to galvanometer sensitivity) and reduce overall resistance for series connection.",
                                            "code": 71
                                        }
                                    ],
                                    "Summary": "Describes the modification of a galvanometer to measure current effectively in a circuit.",
                                    "code": 69
                                },
                                {
                                    "Subnet": "Galvanometer as Voltmeter",
                                    "Categories": [
                                        {
                                            "Category": "Conversion Method",
                                            "Summary": "Explains how a galvanometer is converted into a voltmeter by connecting a large resistance (R) in series with the coil.",
                                            "code": 73
                                        },
                                        {
                                            "Category": "Reasons for Series Resistance",
                                            "Summary": "Discusses why a large series resistance is needed: to draw minimal current and prevent disturbance to the circuit when connected in parallel for voltage measurement.",
                                            "code": 74
                                        },
                                        {
                                            "Category": "Voltage Sensitivity",
                                            "Summary": "Defines voltage sensitivity as deflection per unit voltage, φ/V = NAB/(kR_G+kR), and notes that increasing current sensitivity doesn't necessarily increase voltage sensitivity.",
                                            "code": 75
                                        }
                                    ],
                                    "Summary": "Describes the modification of a galvanometer to measure voltage effectively across a circuit section.",
                                    "code": 72
                                }
                            ],
                            "Summary": "This net covers the working of the moving coil galvanometer and its adaptations for measuring electrical quantities like current (ammeter) and voltage (voltmeter).",
                            "code": 64
                        }
                    ],
                    "Summary": "The provided text is a chapter from a physics textbook focusing on \"Moving Charges and Magnetism\". It introduces the historical discovery of the relationship between electricity and magnetism by Oersted, followed by detailed explanations of fundamental concepts. Key topics include magnetic force on moving charges (Lorentz force) and current-carrying conductors, motion of charged particles in uniform magnetic fields (circular and helical paths), the Biot-Savart law for calculating magnetic fields due to current elements and circular loops, Ampere's circuital law for calculating magnetic fields in symmetric configurations (like long straight wires and solenoids), forces between parallel current-carrying conductors leading to the definition of the Ampere, torque on current loops behaving as magnetic dipoles, and the working principle and applications of the moving coil galvanometer (as a detector, ammeter, and voltmeter). The chapter includes examples, figures, and historical notes on key scientists like Oersted, Lorentz, and Ampere."
                }
            },
            "summaryList": [
                {
                    "status": "OK",
                    "data": {
                        "status": "OK",
                        "PromptTokenCount": 15517,
                        "CandidatesTokenCount": 4136,
                        "TotalTokenCount": 21045,
                        "thoughtsTokenCount": 1392,
                        "finishReason": "STOP",
                        "data": [
                            {
                                "Summary": "<h2>Comprehensive Synthesis Report: Moving Charges and Magnetism</h2><h3>1. Introduction and Historical Context (4.1)</h3><p>The relationship between electricity and magnetism, though both phenomena were known for over two millennia, was only established in 1820 by Danish physicist Hans Christian Oersted. During a lecture demonstration, Oersted observed that an electric current in a straight wire caused a noticeable deflection in a nearby magnetic compass needle. His investigations revealed that the needle's alignment was tangential to an imaginary circle centered on the wire, with its plane perpendicular to the wire. The direction of deflection reversed with the current, and its magnitude increased with current or proximity to the wire. Iron filings sprinkled around the wire formed concentric circles, leading Oersted to conclude that moving charges or currents produce a magnetic field in the surrounding space.</p><p>This discovery spurred intense experimentation, culminating in James Maxwell's unification of the laws of electricity and magnetism in 1864, which also revealed light as an electromagnetic wave. This foundational understanding laid the groundwork for significant technological advancements in the 20th century, including the discovery of radio waves by Hertz and their production by J.C. Bose and G. Marconi. This chapter explores how magnetic fields exert forces on moving charged particles and current-carrying wires, how currents generate magnetic fields, applications like particle acceleration in a cyclotron, and current/voltage detection via a galvanometer.</p><p><strong>Convention Note:</strong> A dot (&#x2022;) depicts a current or field emerging out of the plane, while a cross (&#x2024;) indicates it going into the plane.</p><h3>2. Magnetic Force (4.2)</h3><h4>2.1 Sources and Fields (4.2.1)</h4><p>Similar to the concept of an electric field (E) produced by a static charge (Q), where an electric force F = qE acts on a charge (q), moving charges or currents produce an additional magnetic field (B). Both E and B are vector fields, defined at each point in space and potentially varying with time. They convey energy and momentum and are not established instantaneously. The principle of superposition applies to magnetic fields, meaning the magnetic field from multiple sources is the vector sum of individual source fields.</p><h4>2.2 Lorentz Force (4.2.2)</h4><p>The total force on a point charge (q) moving with velocity (v) in the presence of both an electric field E(r) and a magnetic field B(r) is given by the Lorentz force equation:</p><p>F = q [E(r) + v &#xD7; B(r)] = F<sub>electric</sub> + F<sub>magnetic</sub></p><p>Key features of the magnetic force F<sub>magnetic</sub> = q(v &#xD7; B) include:</p><ul><li>It depends on the charge (q), velocity (v), and magnetic field (B). The force on a negative charge is opposite to that on a positive charge.</li><li>It involves a vector cross product (v &#xD7; B), meaning the magnetic force is always perpendicular to both the velocity and the magnetic field. This implies the magnetic force does no work on the particle, only changing its direction of motion.</li><li>The force is zero if v is parallel or antiparallel to B.</li><li>The force is zero if the charge is not moving (v = 0). Only moving charges experience a magnetic force.</li></ul><p>The direction of the magnetic force is determined by the screw rule or right-hand rule for vector products. The SI unit for the magnetic field is the Tesla (T), named after Nikola Tesla. One Tesla is defined as the magnetic field magnitude where a force of one Newton acts on a unit charge (1 C) moving perpendicular to B with a speed of 1 m/s. A smaller unit, the gauss (1 G = 10<sup>-4</sup> T), is also commonly used (e.g., Earth's magnetic field is ~3.6 &#xD7; 10<sup>-5</sup> T).</p><h4>2.3 Magnetic Force on a Current-Carrying Conductor (4.2.3)</h4><p>The concept of magnetic force extends from a single moving charge to a current-carrying conductor. For a straight rod of length (l) and uniform cross-sectional area (A) carrying a steady current (I), with mobile charge carriers of number density (n) and drift velocity (v<sub>d</sub>), the total magnetic force in an external magnetic field B is:</p><p>F = I (l &#xD7; B)</p><p>where l is a vector of magnitude equal to the rod's length and direction identical to the current. This equation holds for straight conductors. For arbitrarily shaped wires, the total force is found by integrating over infinitesimal linear strips.</p><h3>3. Motion in a Magnetic Field (4.3)</h3><p>Since the magnetic force is always perpendicular to the velocity of a charged particle, it does no work and thus does not change the magnitude of the particle's velocity (its kinetic energy remains constant). It only alters the direction of the velocity (and momentum).</p><ul><li><strong>Circular Motion:</strong> If a charged particle's velocity (v) is perpendicular to a uniform magnetic field (B), the magnetic force (q v B) acts as a centripetal force, causing the particle to move in a circular path. The radius (r) of this circular path is given by:</li></ul><p>r = mv / qB</p><p>The angular frequency (&#x3c9; = qB/m) and frequency (n = qB/2&#x3c0;m) of rotation are independent of the particle's velocity or energy, a crucial principle in cyclotron design.</p><ul><li><strong>Helical Motion:</strong> If the velocity has a component parallel to the magnetic field (v<sub>||</sub>), this component remains unaffected by the magnetic field. The motion perpendicular to B remains circular, combining to form a helical path. The distance moved along the magnetic field during one rotation is called the pitch (p), given by:</li></ul><p>p = v<sub>||</sub> T = 2&#x3c0;m v<sub>||</sub> / qB</p><h3>4. Magnetic Field Due to a Current Element: Biot-Savart Law (4.4)</h3><p>All known magnetic fields originate from currents (moving charges) or intrinsic magnetic moments of particles. The relationship between a current element and the magnetic field it produces is described by the Biot-Savart law. For an infinitesimal current element (I dl) at a distance (r) from a point (P), making an angle (&#x3b8;) with the displacement vector (r), the magnetic field (dB) at P has a magnitude:</p><p>dB = (&#x3bc;<sub>0</sub> / 4&#x3c0;) * (I dl sin&#x3b8; / r<sup>2</sup>)</p><p>and its direction is perpendicular to the plane containing dl and r, given by the right-hand screw rule (dl &#xD7; r). &#x3bc;<sub>0</sub> / 4&#x3c0; is the constant of proportionality, with &#x3bc;<sub>0</sub> being the permeability of free space (4&#x3c0; &#xD7; 10<sup>-7</sup> T m/A).</p><p><strong>Similarities and Differences with Coulomb&#x2019;s Law:</strong></p><ul><li><strong>Similarities:</strong> Both are long-range forces (inverse square dependence on distance). Both obey the principle of superposition. Magnetic field is linear in its source (I dl), like electrostatic field is linear in charge.</li><li><strong>Differences:</strong> Electrostatic fields are produced by scalar sources (electric charge), while magnetic fields are produced by vector sources (I dl). Electrostatic fields are along the displacement vector, while magnetic fields are perpendicular to the plane containing the displacement vector and the current element. Biot-Savart law has an angle dependence (sin&#x3b8;), which is absent in Coulomb&#x2019;s law (magnetic field is zero along the direction of the current element).</li></ul><p>There's a fundamental relation between &#x3bc;<sub>0</sub> (permeability of free space), &#x3b5;<sub>0</sub> (permittivity of free space), and c (speed of light in vacuum): c = 1/&#x221A;(&#x3bc;<sub>0</sub>&#x3b5;<sub>0</sub>).</p><h3>5. Magnetic Field on the Axis of a Circular Current Loop (4.5)</h3><p>Using the Biot-Savart law and integrating over a circular loop of radius R carrying a steady current I, the magnetic field (B) at a point P on its axis at a distance x from the center is:</p><p>B = (&#x3bc;<sub>0</sub> I R<sup>2</sup> / 2) * (1 / (x<sup>2</sup> + R<sup>2</sup>)<sup>3/2</sup>)</p><p>The direction of this field is along the axis, given by the right-hand thumb rule: curling fingers in the direction of current, the thumb points in the direction of the magnetic field. At the center of the loop (x=0), the field simplifies to B = &#x3bc;<sub>0</sub> I / (2R). The magnetic field lines due to a circular wire form closed loops.</p><h3>6. Ampere&#x2019;s Circuital Law (4.6)</h3><p>Ampere&#x2019;s circuital law provides an alternative way to express the relationship between current and magnetic field, especially useful for systems with high symmetry. It states that the line integral of the magnetic field (B) around any closed loop (C) is equal to &#x3bc;<sub>0</sub> times the total current (I<sub>enclosed</sub>) passing through the surface bounded by that loop:</p><p>&#x222E; B &#x2022; dl = &#x3bc;<sub>0</sub> I<sub>enclosed</sub></p><p>A right-hand rule is used for the sign convention: if the fingers of the right hand curl in the direction of the loop integral, the thumb indicates the positive direction of the enclosed current. For highly symmetric cases, an 'Amperian loop' can be chosen such that B is either tangential and constant, normal, or zero along the loop, simplifying the equation to BL = &#x3bc;<sub>0</sub> I<sub>enclosed</sub>.</p><p><strong>Application to an Infinite Straight Wire:</strong> For an infinitely long straight wire carrying current I, Ampere's law readily gives the magnetic field at a distance r:</p><p>B = &#x3bc;<sub>0</sub> I / (2&#x3c0;r)</p><p>Properties of this field include cylindrical symmetry, tangential direction, and field lines forming concentric circles (unlike electrostatic fields which originate and end on charges). The magnitude of the field is directly proportional to current and inversely proportional to distance. Another right-hand rule helps determine the direction: grasping the wire with the thumb pointing in the current direction, fingers curl in the magnetic field direction. Ampere's law is to Biot-Savart law what Gauss's law is to Coulomb's law, relating a boundary quantity to an interior source. It applies to steady currents.</p><h3>7. The Solenoid (4.7)</h3><p>A long solenoid (length significantly greater than radius) consists of a long wire wound in a helix with closely spaced turns. Each turn acts as a circular loop, and the net magnetic field is the vector sum. For an ideal long solenoid, the magnetic field outside approaches zero. Inside, the field is uniform, strong, and parallel to the axis. Applying Ampere&#x2019;s circuital law to a rectangular Amperian loop enclosing turns of the solenoid, the magnetic field inside is found to be:</p><p>B = &#x3bc;<sub>0</sub> n I</p><p>where n is the number of turns per unit length and I is the current. Solenoids are commonly used to produce uniform magnetic fields, which can be further amplified by inserting a soft iron core.</p><h3>8. Force Between Two Parallel Currents, The Ampere (4.8)</h3><p>Two current-carrying conductors exert magnetic forces on each other. This is a direct consequence of one conductor producing a magnetic field (Biot-Savart law) and the other conductor experiencing a force in that field (Lorentz force). For two long, straight parallel conductors separated by distance d and carrying currents I<sub>a</sub> and I<sub>b</sub>:</p><ul><li>The force per unit length (f) between them is: f = &#x3bc;<sub>0</sub> I<sub>a</sub> I<sub>b</sub> / (2&#x3c0;d)</li><li><strong>Parallel currents attract each other.</strong></li><li><strong>Antiparallel currents repel each other.</strong></li></ul><p>This behavior is opposite to electrostatic forces where like charges repel. This force is consistent with Newton's third law for steady currents. The force per unit length is used to define the Ampere (A), one of the seven SI base units:</p><p><strong>Action Item (Definition of Ampere):</strong> One Ampere is defined as that steady current which, when maintained in each of two very long, straight, parallel conductors of negligible cross-section, placed one meter apart in vacuum, would produce on each conductor a force equal to 2 &#xD7; 10<sup>-7</sup> Newtons per meter of length. The SI unit of charge, the Coulomb (C), is then defined as the quantity of charge that flows through a conductor&#x2019;s cross-section in 1 second when a steady current of 1A is established.</p><h3>9. Torque on Current Loop, Magnetic Dipole (4.9)</h3><h4>9.1 Torque on a rectangular current loop in a uniform magnetic field (4.9.1)</h4><p>A rectangular loop carrying a steady current I in a uniform magnetic field experiences no net force but does experience a torque. This is analogous to an electric dipole in a uniform electric field. If the plane of the loop makes an angle &#x3b8; with the magnetic field (or, more precisely, if &#x3b8; is the angle between the magnetic field and the normal to the loop), the magnitude of the torque (&#x3c4;) is:</p><p>&#x3c4; = IAB sin&#x3b8;</p><p>where A is the area of the loop and B is the magnetic field strength. We define the magnetic moment (m) of the current loop as m = IA, where the direction of the area vector A is given by the right-hand thumb rule. The torque can then be expressed as a vector product:</p><p>&#x3c4; = m &#xD7; B</p><p>The torque vanishes when m is parallel (&#x3b8;=0) or antiparallel (&#x3b8;=180&#xB0;) to B. Parallel alignment represents stable equilibrium, while antiparallel is unstable equilibrium. For a coil with N closely wound turns, the magnetic moment is m = NIA. The presence of this torque causes small magnets or magnetic dipoles to align with external magnetic fields.</p><h4>9.2 Circular current loop as a magnetic dipole (4.9.2)</h4><p>At large distances (x >> R), the magnetic field produced by a circular current loop on its axis resembles that of an electric dipole. The magnetic field (B) on the axis is approximately:</p><p>B &#x2248; &#x3bc;<sub>0</sub> / (2&#x3c0;x<sup>3</sup>) * m</p><p>where m = IA is the magnetic moment. This analogy extends to the field on the perpendicular bisector as well. A planar current loop effectively behaves as a magnetic dipole. A key fundamental difference is that while electric dipoles are composed of two elementary charges (monopoles), magnetic monopoles have not been observed; thus, a current loop (magnetic dipole) is considered the most elementary magnetic element. Ampere suggested that all magnetism is due to circulating currents, a concept partly supported, though elementary particles also possess intrinsic magnetic moments not solely due to circulating currents.</p><h3>10. The Moving Coil Galvanometer (4.10)</h3><p>The moving coil galvanometer (MCG) is a sensitive device used for detecting and measuring currents and voltages, based on the principle of torque on a current loop in a magnetic field. It consists of a coil with many turns, free to rotate in a uniform radial magnetic field, often enhanced by a soft iron core. When current (I) flows through the coil, a magnetic torque (&#x3c4; = NIAB) acts on it (since the field is radial, sin&#x3b8; = 1). This torque is balanced by a restoring torque (k&#x3c6;) from a spring, leading to an angular deflection (&#x3c6;):</p><p>&#x3c6; = (NAB / k) I</p><p>where k is the torsional constant of the spring. The deflection &#x3c6; is proportional to the current I.</p><ul><li><strong>As a Current Detector:</strong> Used to check current flow (e.g., in Wheatstone&#x2019;s bridge), with a neutral position in the middle of the scale.</li><li><strong>As an Ammeter:</strong> To measure current, a galvanometer must be modified because it is very sensitive (full-scale deflection for &#x3bc;A currents) and has high resistance (which would alter circuit current if connected in series). To convert it, a small shunt resistance (r<sub>s</sub>) is connected in parallel with the galvanometer coil. Most of the current bypasses the galvanometer through the shunt. The resulting ammeter has very low resistance and is connected in series in the circuit. Current sensitivity is defined as deflection per unit current (&#x3c6;/I = NAB/k).</li><li><strong>As a Voltmeter:</strong> To measure voltage, a galvanometer must be connected in parallel across the circuit section of interest and draw very little current to avoid disturbing the circuit. This is achieved by connecting a large resistance (R) in series with the galvanometer. The resulting voltmeter has very high resistance. Voltage sensitivity is defined as deflection per unit voltage (&#x3c6;/V = NAB / (kR<sub>total</sub>), where R<sub>total</sub> &#x2248; R). Interestingly, increasing current sensitivity (e.g., by doubling N) does not necessarily increase voltage sensitivity, as the galvanometer's resistance also doubles.</li></ul><p><strong>Actionable Recommendations:</strong></p><ul><li>To measure current accurately, connect a galvanometer with a low shunt resistance in series with the circuit.</li><li>To measure voltage accurately, connect a galvanometer with a high series resistance in parallel with the circuit component.</li></ul>"
                            }
                        ]
                    }
                }
            ]
        }
    ],
    "fileName": "",
    "projectType": "DOCUMENT",
    "__v": 0,
    "headerCols": {
        "col1": "Part Id",
        "col2": "FileName",
        "col3": "Content"
    }
}

]