AI for Social Good Using Technology to Create Real-World Impact
20 Feb 2026 10:00h - 11:00h
AI for Social Good Using Technology to Create Real-World Impact
Summary
The panel discussed how AI, when coupled with open digital public infrastructure, can deliver population-scale impact across education, health, and agriculture [1-3][8-11]. James Manyika highlighted AI’s rapid progress and early successes, citing AlphaFold’s protein-structure database that now serves over three million researchers, with India as the fourth-largest user [14-16]. He argued that expanding access requires coordinated digital public infrastructure and open networks-such as India’s UPI, Bhashini, and Google’s Project Vani, which provides free multilingual speech data for over 100 Indic languages [19-24][25]. Nandan Nilekani reinforced that open networks act as general-purpose platforms that let AI agents simplify transactions for users, using UPI as a model and stressing the importance of language localization [77-82][84-95]. Sang-Boo Kim described the World Bank’s AgriConnect as an open-stack, farmer-centric service that can be extended to health and education, illustrating the need for universal standards [102-108]. Kiran Mazumdar-Shaw outlined a “health stack” built on India’s consent-based data sharing, enabling AI-driven risk profiling, insurance integration, and universal healthcare, while also envisioning AI-augmented biology such as virtual cells [119-128][133-140]. Sunil Wadhwani explained that Digital Public Infrastructure supplies data pipelines and distribution channels that make AI models for TB diagnosis, treatment adherence, and rapid reading assessment scalable and low-cost, reaching millions of patients and students [170-179][190-204][208-218]. He noted growing interest from other low- and middle-income countries to adopt these solutions, with the World Bank and India working to replicate models in Africa, Brazil, and the Philippines [229-233][322-324]. The panel agreed that cheap inference is critical for mass diffusion; Nilekani gave the example of plugging improved weather models into AgriConnect to serve ten million farmers [246-254]. Manyika added that India’s infrastructure allowed Google’s Neural GCM to deliver monsoon forecasts to 38 million farmers, demonstrating the power of open networks [264-268]. In a closing lightning round, participants called for rapid scaling: Nilekani for massive diffusion to farmers, Mazumdar-Shaw for a sustainable universal health-care standard, and others for broader global dissemination [301-308][311-313][328-330]. The moderator concluded that AI’s benefits can only be realized through inclusive open networks and invited further participation via Google.org impact challenges [340-344]. Overall, the discussion underscored that coordinated, decentralized digital public infrastructure combined with multilingual AI agents is seen as the essential foundation for achieving population-scale transformation in education, health, agriculture, and beyond [19-21][36-37][41].
Keypoints
Major discussion points
– Open digital public infrastructure (DPI) and interoperable networks are the essential coordination layer that lets AI turn human intent into real-world action at population scale.
James Manyika stresses that “digital public infrastructure and open networks… provide the coordination layer” [19-21]; Nandan Nilekani cites UPI as a model of an open network that enabled massive growth [77-78]; Sunil Wadhwani describes DPI as the “data pipelines” and “distribution channels” that make AI models usable in health and education [168-174].
– AI-driven, multilingual agents can remove complexity for users and accelerate diffusion of technology, especially in agriculture, health and education.
Nandan explains that AI agents on open networks “remove complexity for the user” and enable inclusion for farmers and small producers [80-82]; Manyika cites the Gemini-powered agriculture pilot that gives farmers multilingual AI assistance [30-33]; Sunil details AI-based TB diagnosis from a cough and rapid reading-assessment tools for millions of children [196-204][208-214].
– Language localization and data-rich “digital stacks” are critical for scaling AI services in India and beyond.
Manyika highlights Project Vani’s speech data for 100+ Indic languages [23-26]; Nandan points out that India’s multilingual agents must handle code-mixing and that initiatives like Bhashini and AI-for-Bharat are breaking language barriers [84-95]; Kiran Mazumdar-Shaw notes India’s consent-based health data stack that can be leveraged for risk profiling and insurance [122-128].
– The cost of AI inference must drop dramatically; low-cost, plug-and-play models are needed for massive diffusion.
Nandan warns that “the cost of AI inference has to drop dramatically” for it to work at scale [246-249]; he gives the example of plugging Google’s improved weather model into an open AgriConnect network to reach millions of farmers [250-254]; Manyika adds that the Indian government’s infrastructure allowed the Neural GCM monsoon model to reach ~38 million farmers [264-268].
– India’s open-network models are being packaged as blueprints for global replication, with the World Bank and other partners seeking standardized, scalable approaches.
Sang-Boo Kim describes AgriConnect as a “universal network” that can expand to health and education and that the World Bank is working to replicate the model in other countries [102-108][229-236]; Sunil notes growing interest from governments in the Global South to adopt India-built AI platforms [321-326].
Overall purpose / goal of the discussion
The summit convened global leaders to demonstrate how open, interoperable digital public infrastructure can serve as a “global, interoperable coordination rail” that powers AI-driven solutions for education, healthcare, agriculture and other public services, with the aim of achieving population-scale impact both in India and worldwide [1-4][7][10-12].
Overall tone
The conversation remained upbeat, collaborative and forward-looking. Speakers repeatedly expressed optimism about AI’s transformative potential, celebrated concrete successes (e.g., AlphaFold usage, TB detection, reading-assessment pilots), and used enthusiastic language (“extraordinary,” “powerful,” “holy grail”). Applause and a “lightning-round” at the end reinforced a celebratory, solution-focused atmosphere, with no noticeable shift toward criticism or pessimism throughout the session.
Speakers
– James Manyika
– Role/Title: Senior Vice President at Google, leading research, labs, and technology in society; Co-chair of the UN High-Level Advisory Board on AI.
– Area of Expertise: Artificial Intelligence, AI policy, technology for society. [S8][S9][S10]
– Nandan Nilekani
– Role/Title: Co-founder and Chairman of Infosys; Co-founder of Networks for Humanity; Global leader in Digital Public Infrastructure.
– Area of Expertise: Digital public infrastructure, open networks, fintech, AI diffusion. [S3][S4]
– Sangbu Kim
– Role/Title: Vice President for Digital and AI at the World Bank.
– Area of Expertise: Digital economy, AI-enabled development, open standards, agriculture, health, education. [S2][S1]
– Kiran Mazumdar-Shaw
– Role/Title: Chairperson, Biocon Group; Biotech entrepreneur, healthcare visionary, philanthropist.
– Area of Expertise: Biotechnology, healthcare innovation, AI in health, universal health care. [S14]
– Sunil Wadhwani
– Role/Title: Co-founder of the Wadhwani Institute for Artificial Intelligence; Visionary entrepreneur and philanthropist.
– Area of Expertise: AI for social impact, digital public infrastructure, health and education solutions. [S7]
– Moderator
– Role/Title: Moderator of the India AI Impact Summit (referred to as “Ashwani” in the opening).
– Area of Expertise: Event facilitation, AI policy discussion.
Additional speakers:
– Demis (referenced by Nandan Nilekani, likely Demis Hassabis) – mentioned in discussion about weather models; no explicit role or title provided.
– Ashwani – addressed by James Manyika (“Thank you, Ashwani”) and appears to be the moderator’s first name; no further details.
The moderator opened the India AI Impact Summit by stating that real-world AI impact depends on “population-scale” transformation of education, health-care and agriculture, and that such impact can only be achieved through a built-in coordination layer [1-3]. To frame the discussion, James Manyika – Google’s senior vice-president for research, labs and technology in society and former co-chair of the UN High-Level Advisory Board on AI – was introduced as the first speaker [4-7].
Manyika began by affirming Google’s belief that universal access to AI is essential for expanding innovation capacity worldwide [10-11]. He highlighted the rapid technical progress of AI, citing AlphaFold’s breakthrough in solving the 50-year protein-structure problem and the open AlphaFold database now used by more than three million researchers in 190 countries, with India ranking fourth in adoption [14-16]. He argued that to “fully take advantage of this potential” access must be expanded from the outset, warning that the “digital divide must not become an AI divide” [18-19]. According to Manyika, digital public infrastructure (DPI) and open networks provide the coordination layer that translates human intent into real-world action [20-21]; India’s UPI payments system and the Bhashini language-network exemplify this [21-22]. Google’s partnership with the Indian Institute of Science on Project Vani, now in its second phase, has released speech data for over 100 Indic languages – including 20 languages previously undocumented – as a free resource [23-26]. He noted a recent $10 million Google.org grant to the Networks for Humanity Foundation, which is building universal tools such as asset-tokenisation and open-network standards across innovation labs from Singapore to Switzerland [38-41]. Concrete AI-enabled interventions were described: multilingual agents for 1.4 million frontline health workers, AI-driven pest-surveillance for national crops, and an education platform already reaching ten million learners with a target of 75 million students and two million educators by 2027 [43-48]. Manyika illustrated the power of agents with an energy-trading example, showing how a farmer with rooftop solar can sell excess power through a simple AI-driven commerce interface, eliminating the need to understand complex market mechanisms [84-86].
Following Manyika, the panel was introduced: Nandan Nilekani (co-founder and chairman of Infosys and co-founder of Networks for Humanity), Sang-Boo Kim (World Bank vice-president for digital and AI), Kiran Mazumdar-Shaw (chairperson of Biocon Group) and Sunil Wadhwani (co-founder of the Wadhwani Institute for AI) [52-67].
Nilekani framed AI as a “general-purpose technology” whose fastest diffusion requires open networks that allow many innovators to build applications at the edge [73-80]. He used India’s Unified Payments Interface (UPI) as a prototype of an open architecture that grew into the world’s largest payments system [77-78] and argued that AI agents on such networks “remove complexity for the user”, enabling inclusion for farmers or small-scale electricity producers in their own language [81-82]. He stressed that multilingual capability is the “holy grail” of diffusion, noting initiatives such as the government’s Bhashini, AI-for-Bharat and Google’s Project Vani that address code-mixing (Hindi-English-Tamil) and aim to eradicate language barriers [84-95]. He warned that if a single AI query costs even 500 rupees, the model would be unaffordable for mass-market use, underscoring the need to drive inference costs down dramatically [242-244].
Kim described the World Bank’s AgriConnect as an “open-stack, farmer-oriented” platform that delivers coherent services through an open network [102-106]. He positioned the shift from supplier-centric to user-centric services as a hallmark of the AI era and suggested that the same open-standard architecture could be extended to health and education, becoming a “universal network” for the future [107-108]. Kim likened the World Bank’s role to that of a sommelier, curating and recommending the most suitable “wine” (i.e., AI-enabled solutions) for each country’s needs [312-315].
Mazumdar-Shaw outlined a vision of a national “health stack” built on consent-based, secure data sharing analogous to UPI [122-124]. She explained that India is aggregating phenotypic, genomic, demographic, radiological and treatment-outcome data, which can be risk-profiled at population scale and linked to insurance products – a task that only AI can perform efficiently [125-128]. She also highlighted the potential of AI-augmented biology: learning from the energy-efficient, distributed computation of cells, re-programming cells, and creating virtual cell models to move from hospital-centric to preventive, community-centric care [133-140]. She warned that data-sharing reluctance outside India remains a barrier [288-292].
Wadhwani explained that DPI supplies two essential functions for AI in the public sector: (i) data pipelines that feed training models, and (ii) distribution channels that deliver inference at scale [170-174]. He illustrated this with a TB-control programme that uses the NIXA patient-management database to (a) diagnose TB from a cough sound on a smartphone, raising case detection by 25 % nationally [196-200]; (b) automate lab results for same-day reporting [202-204]; and (c) predict treatment non-adherence to focus the work of 2 000 caseworkers [204-206]. In education, a 20-second speech-based reading assessment costing five paise per child has been piloted for three million students and is being mandated for eight million more, with plans to reach 75 million by next year via the Rakshak DPI platform [208-218]. He noted growing interest from governments in the Global South to adopt these AI platforms, citing recent engagements with African and Latin-American ministries [321-326].
The conversation then turned to the economics of scaling. Nilekani warned that “the cost of AI inference has to drop dramatically” because per-query fees of even a few rupees would block mass adoption [246-248]. He illustrated how an open network enables plug-and-play of improved models: integrating Google’s latest weather forecasts into AgriConnect would instantly benefit ten million farmers [250-254]. Manyika added that India’s infrastructure allowed Google’s Neural GCM monsoon model to reach roughly 38 million farmers, demonstrating the power of a ready-made coordination rail [260-263].
Across the panel there was strong consensus that open, interoperable DPI is the foundational coordination layer enabling AI to convert intent into action at population scale [1-3][19-21][73-80][170-174]. Speakers agreed that multilingual AI agents are essential for inclusive diffusion [80-82][83][94-95]; that inference costs must be driven down to enable billions of daily queries [246-248][252-254]; and that open standards and open networks together allow new models to be “plugged in” without rebuilding the whole stack [105-106][252-254]. The panel also concurred that India’s open-network blueprints – from UPI to AgriConnect, TB diagnostics and reading-assessment tools – constitute replicable models for other low- and middle-income countries [102-108][229-236][321-326].
While Nilekani highlighted cheap inference as the primary bottleneck, Wadhwani emphasized the equally critical role of DPI-provided data pipelines and distribution channels; together these perspectives outline the full stack of requirements for population-scale AI deployment [246-248][170-174]. A secondary nuance concerned emphasis on “open standards” (Kim) versus “open networks” (Manyika) as the key technical foundation for user-centric services [105-106][19-21].
In the closing “lightning round”, Nilekani called for massive diffusion of AI applications to millions of farmers worldwide [301-308]; Mazumdar-Shaw urged the creation of a sustainable, universal health-care standard built on AI-driven risk profiling [311-313]; Kim pledged to disseminate successful use-cases globally so that developing-world actors can grasp AI’s potential [328-330]; and Wadhwani reaffirmed India’s role as a model for the world, noting the Prime Minister’s call to “develop in India, deliver to the world” [317-326].
The moderator wrapped up by reiterating that AI’s true benefits “can only be realised when we build for everyone using open networks” and invited participants to apply for the Google.org Impact Challenges (AI for Science and Government Innovation) and to visit Google’s exhibition booths [340-345]. The session concluded with applause and a group photograph [346-358].
Key take-aways – Open DPI and interoperable networks constitute the essential coordination rail that lets AI translate intent into real-world outcomes; AI-powered multilingual agents remove language and complexity barriers; inference costs must fall dramatically to achieve billions of daily queries; consent-based, open health data stacks enable risk profiling, insurance integration and preventive care; India’s sector-specific pilots (AlphaFold, UPI, Bhashini, Project Vani, AgriConnect, TB cough-analysis, reading-assessment) provide a scalable blueprint for the Global South; and the convergence of biological intelligence with AI promises future breakthroughs in regenerative medicine.
Overall, the discussion was upbeat and collaborative, with senior leaders from Google, the private sector, the World Bank and academia aligning on a shared vision of inclusive, open-network-enabled AI for population-scale transformation, while recognising the need for continued work on inference economics, global data-sharing norms and standard-setting for cross-border replication.
Because we believe that AI’s true potential lies in its ability to deliver population -scale impact, transforming education, healthcare, and agriculture for every citizen. However, that impact can only be possible when there’s coordination that’s built into the system. And therefore, today, we are here joined by global leaders to explore how open networks and digital public infrastructure can create a global, interoperable coordination rail, powered by AI to translate intent into action across borders. To set the stage, it’s my honor to introduce James Manika. James is the Senior Vice President at Google, leading research, labs, and technology in society. He also served as the co -chair of the UN’s High -Level Advisory Board on AI. James, welcome. The floor is yours to set the stage.
Thank you, Ashwani. Good morning, everyone. It’s a real pleasure and privilege to be back in India and to join all of you here at the India AI Impact Summit. At Google, we believe that access to AI is essential for unlocking opportunities and expanding the innovation capacity for people everywhere. The rapid technological progress that we’re seeing in AI’s development is really quite breathtaking and represents an extraordinary opportunity to solve problems and empower people, power economies, advance science, and tackle some of society’s greatest challenges. Indeed, we’re beginning to see the impact of this, so it’s not just in the future, but we’re already starting to see some of these benefits. benefits and impacts materialize today. Take science, for example.
Five years ago, our AlphaFold system, which is our Nobel Prize winning innovation, solved the 50 -year grand challenge of protein structure prediction. And since then, the freely available AlphaFold protein database has been used by more than 3 million researchers in over 190 countries. And in fact, India is actually the fourth largest adopter and user of the protein database, where people are working on a variety of problems, everything from neglected diseases all the way to even breeding resistance, soya beans, and a whole range of things that are incredibly beneficial to people in India and beyond. But to take full advantage of this potential, we need to collectively expand access right from the beginning. As you may have heard our CEO Sundar Pichai say yesterday, we need to ensure that the digital divide does not and AI divide.
Digital public infrastructure and open networks are an important part of making this possible. They provide the coordination layer that allows AI to translate human intent into real -world action. And India has been leading the way with systems like UPI and Bashini network and infrastructure, bringing the capabilities of AI into the daily lives of people across the country and at population scale. At Google, we’ve been a very committed partner in this journey by helping to build the foundations that help to scale it. For instance, our collaboration with the Indian Institute of Science, and in particular on Project Vani, has now completed its second phase, where we’ve been covering every Indian state, making speech data for over 100 Indic languages available for free.
And we’ve been able to do this through the government of India’s Bashini mission. In fact, this includes 20 languages that had never been spoken before. been recorded before digitally that we’re now building onto these systems in ways that truly try to attempt to reflect India’s true linguistic and cultural richness and diversity. And we continue to build on our commitment to drive scaled impact at the grassroots level. This commitment to scaled impact is reflected in our recent partnership with the World Bank, and I’m sure we’ll talk about this later today. Together we’re taking a blueprint born right here in India and scaling it by localizing it across the globe to countries from Brazil to Nigeria, Ethiopia, and Kenya.
And the heart of this blueprint began with our partnership with the government of Uttar Pradesh. There we piloted a Gemini -powered open network for agriculture that provides farmers with multilingual AI agents to facilitate everything from credit to crop prediction. By taking the lessons we learned in Uttar Pradesh, where digital tools drove real… measurable impact. We’re proving that a small holder farmer can compete and execute on the value that they create rather than the platforms that they’re on. This isn’t just a regional success. It’s now a global architecture and a model that can be taken everywhere for global digital inclusion. The success of these networks depend on a single fundamental principle. It must remain decentralized and open.
This is the driving force behind our support for the Networks for Humanity Foundation. Again, one of the things we’ll talk about this morning. And through a $10 million Google .org grant that we announced last year, the Network for Humanity Foundation is building the universal tools for tomorrow from the FinInternet, for asset tokenization, to the BEK and open networks. And by establishing innovation labs from Singapore to Switzerland, they’re ensuring that the that the infrastructure of opportunity is a global standard and not just a local exception. Having this type of infrastructure in place is what will allow all of us to collectively achieve population scale change. That’s why we’re supporting change makers like Wadwani AI through Google .org grants that try to embed intelligence directly into the digital rails for millions of Indians to be able to use.
In healthcare, for example, this means empowering something like 1 .4 million frontline workers with multilingual AI assistance, providing early warnings to combat child malnutrition across the country. In agriculture, it means integrating AI into the national pest surveillance system to protect India’s most important crops at a national scale. And in education, it means integrating AI into the national pest surveillance system to protect India’s most important crops at a national scale. And in education, it means delivering high -quality learning experiences through AI -led transformation of government government -owned education and development platforms. And this is an initiative that’s already reached 10 million students and educators with the goal of empowering as many as 75 million students and nearly 2 million educators by the end of 2027.
Ultimately, to fully capture AI’s beneficial potential, we must be bold and responsible and be committed to building all of this together. We must pursue AI’s most ambitious possibilities while ensuring that we build the coordination layer necessary to bridge and close the AI divide. With that, it is now my great pleasure and honor to welcome an extraordinary group of incredible leaders and innovators to the stage. We’ve been doing this for an extraordinarily long time with incredible impact. First, I’d like to invite Nandan Nilikani. Nandan is the…
Thank you.
Nandod is the co -founder and chairman of Infosys. He’s a global leader in digital public infrastructure and the co -founder of Networks for Humanity, an initiative building open, interoperable digital infrastructure for the intelligence age. I should say I’ve known Nandod for a very long time. When he first told me what he was working on 15 years ago, I’m not quite sure I quite believed him, but here we are. Next, joining us is Sang -Boo Kim. Sang -Boo is the World Bank’s vice president. Sang -Boo is the World Bank’s vice president for digital and AI, leading efforts to drive digital economy growth in developing countries by strengthening infrastructure, cybersecurity, data privacy, while modernizing government services and also touching many areas like health, education, and more.
Our third guest… is Kiran Mamzouma -Shaw. As chairperson of Biocon Group, Kiran is a pioneering biotech… Kiran is a pioneering biotech entrepreneur, health care visionary, and a passionate philanthropist committed to expanding access to health care through affordable innovation. And finally, please welcome Sunil Wadhwani. Sunil is a visionary entrepreneur and philanthropist who co -founded the Wadhwani Institute for Artificial Intelligence to drive systematic and systemic social transformation through AI solutions and innovation in the public systems across health care, education, and agriculture. So we’re now going to have a conversation. I can’t wait to do a conversation with these extraordinary leaders. Thank you. Nanda, let me start with you. You’ve been championing digital decentralized ecosystems for a very long time, building open networks, taking things to extraordinary scale in India, and recently with Bakken and Finantech.
And obviously you bring a lot of credibility to both users of these systems and to regulatory bodies. How do you see AI as a multiplier or a factor as you think about open networks and the kind of transformational change you’ve been pursuing?
No, I think AI is very fundamental, and I’ll explain how open networks and AI come together. I think what some of us have been thinking about is if AI is a general purpose technology, what is the fastest way of diffusing the use of AI in a productive way for people? And, you know, ultimately all this, it only makes sense if you can do it. people’s lives improve. And I think we have a lot of experience with open networks. I mean, in some sense, UPI was an open network for payments, and the open architecture led to the massive growth and became the world’s largest payment system. So a lot of those principles are embedded in Beckon, and we have other examples.
But I think open networks allows many actors, many innovators to build applications on the edge using AI. And I think we keep talking about agents, but I think the real power of agents is in removing complexity for the user. So if a user is there who is a farmer or somebody who is producing a little bit of electricity, if they can very easily transact with somebody else through an agent, which is in their own language, then suddenly this is inclusion at massive scale. So I see really AI… agents on an open network as the fundamental construct for massive diffusion of technology.
And also the importance, as you mentioned, of doing that in languages, in local languages.
Oh, totally. I think you talked about what you’re doing at ISE. I think there are many initiatives in India which essentially are driving to make language completely accessible. Because language is not just pure language. I mean, the way Indians speak, they mix the English, Hindi, and Tamil in one sentence. So how do you deal with that? How do you recognize that? So I think all that is getting addressed. There are many initiatives, voice AI, there’s Bhashini of the government, there’s AI for Bharat, there’s the Google project. So I think there’s lots of stuff. But fundamentally, I think language as a barrier will go away. So if you combine language, so a person talks to the agent in their own language, and then the agent does some transaction with hiding all the complexity behind it, then, you know, that’s the holy grail.
We can get everybody on the system, and that’s how AI will get diffused.
And then speaking of, you mentioned farmers and agriculture. Sangbo, let me come to you. I mean, the World Bank recently launched AgriConnect initiative. First of all, I’d like you to describe that a little bit. I think it’s intended to make what smallholder farmers do much, much more efficient and scalable. But I’m curious, what has that work taught you so far about the type of global standards that are going to be needed to scale local solutions?
So, if you just go to look about the AgriConnect in Uttar Pradesh for now, so it is a very farmer -oriented approach to provide very coherent and consistent services at the same time with open stack, open network. That means, if you think about the previous day, from the… computer innovation, mobile innovation, now we are seeing AI innovation. I would interpret this evolution from the supplier -oriented service environment to the customer, user -oriented environment. In that sense, some open standard and open network is a really crucial part to make sure user -centric service. So it is very efficient and affordable solutions for an AI era to fully provide quality of service to the user. In that sense, this AgriConnect project is really important, but it is not only for agriculture project.
With that, we are really looking forward to expanding to another sector like healthcare and education. So it can be a very… universal network in the future.
That’s pretty powerful. In fact, speaking of health care, you mentioned you’re taking this to health care. Kieran, you’ve been an incredible advocate and innovator when it comes to thinking about medicine as a whole. And you’ve talked about this idea that we need to move beyond the industry of medicine. And tell me, say more about what you have in mind about what we need to move to, and in particular, how you can connect what’s going on maybe with AI and data sets with fundamentally transforming medicine.
So I think I have to answer this in two parts. The first part is how do we basically leverage what Nandan refers to as the digital stack to a health stack. That is the first big opportunity we have. And I think India is a country that can uniquely create a global reference model when it comes to… The use of AI in the kind of… health data that we are collecting. So, for instance, India is beginning to collect a lot of health data in its health stack, and it’s phenotypic, it’s genomic, it’s demographic, and radiological data, and, of course, treatment and treatment outcome data. Now, when you start collecting this data, I think the whole objective, again, which is a holy grail, which is universal healthcare delivery at scale in a sustainable way, and how do we reduce the disease burden, I mean, and increase lifespan, all these are big challenges and a very complex set of solutions.
But I think this is a starting point where you get this huge digital stack of health data. And because India has this open source and the consent -based kind of secure data share, already established in UPI. I think we should quickly apply this to healthcare. And when you do that, you will start risk profiling your population at a demographic level, which I think is very exciting and at scale. And if you can integrate insurance into that, that will be even more powerful. That can only be done by AI. So AI has the opportunity to risk profile very fast, to try and find interesting insurance models, to see how we can marry the risk profile with the insurance instrument.
Not easy, but I think it’s a good challenge because AI can be given a lot of exclusion -inclusion criteria, which it can adopt. So I personally am very excited with what AI can do for health, digital health, and the whole universal healthcare delivery model. Additionally, of course, India has this unique model of ASHA work, because if ASHA work is not done, can be empowered with AI that is even more powerful. So I think you know deploying AI for the common people, the common man is very important to both Nandan and Samguth’s point of view. Now coming to my excitement about your second question about what is it that I’m looking beyond this. Beyond this I’m looking at advancing medicine using AI.
Now biology on its own was limited because it didn’t have the kind of power of technology to get deeper insights. AI like what you’ve just done, alpha fold, alpha genome is going to give it immeasurable opportunities to understand biology and to me biological intelligence is just amazing. And if you combine it with artificial intelligence and convert bring that convergence I think we are in for huge And when I look at biological intelligence, when I just look at cell biology, how cells signal, how cells create circuits, how cells regulate, how cells connect and disconnect. I mean, the human body, living systems, have distributed data centers. And these data centers are connecting, disconnecting with sips of energy, not gigawatts of energy.
And they’re actually translating that into instant information and decision making. If we can learn that and apply it to AI, I think it’s going to be transformational. I am really looking forward to reprogramming cells, right? That’s the holy grail. How do you convert a cancer cell into a non -malignant cell? How do you basically look at regenerative science? How do you look at lifespan? I mean, your biggest question today. Right. How do we shift from… from hospital -centric care to primary and community care. That can happen with AI, with predictive and preventive medicine. That’s, I think I’ve said enough. Yeah.
Well, it would actually take us, in effect, Kiran, it would probably take us from kind of treating diseases to preventing diseases. And I like it. You and I were talking earlier. You and I and Demis were talking about this idea of someday we should be able to try to build virtual cells, models of virtual cells, and be able to do kind of cell -based biology, basically.
Absolutely. That’s one of the exciting things. It’s very exciting. Yeah.
We’ll come back to that. But I want to come to you, Sunil, which is, you know, I’m curious, Sunil, as you think about what you’ve been doing, what role does DPI and open networks pay in developing and scaling the kinds of solutions to some of society’s most pressing problems? I mean, you’ve been thinking about this for a very long time. I mean, from way back. When you set up the AI. Institutes, way before most people were thinking about these things. But I want to hear what your perspectives and experiences have been.
Thanks, James. Good morning. Just so we’re all clear, there’s a lot of intellectual horsepower on the stage, and it’s all on this side of me. I’m basically here for my good looks, so just so we manage expectations. But when I set up Vadbani AI back in 2018, and Prime Minister was good enough to come and inaugurate that, basically we had a huge benefit, which is the following, to your point, that over the last 20 years, the government of India has developed a set of DPI, Digital Public Infrastructure, that is broad and that is deep. And this DPI are basically digital building blocks that are going to be used to build a system of infrastructure that connect policy, program implementation, public service workers, and citizens.
in the country. And I’ll give you a couple of examples. But these, again, these digital public, this DPI provides two key, very practical, down -to -earth functions and benefits, as I see it. Number one, they provide data and data pipelines. And for AI, you couldn’t build AI for the social sector without the kind of data and the data pipelines that they provide. Secondly, this DPI provides distribution channels so that once your inference models are ready, these AI platforms, again, developed and managed by government, provide a distribution channel to get our AI models out at scale. Without these, trust me, the usage of any model in the public sector, in the social sector, would cost incredibly more and wouldn’t scale anywhere near what we see.
So, two quick examples. One in healthcare, one in… education. So one of the challenges, one of the national health priorities for the government of India for the last several years has been the elimination of tuberculosis, TB. It’s the largest infectious disease killer in the world, kills close to 2 million people a year. It’s the largest infectious disease killer in India, kills close to half a million people a year in India. And for each person that unfortunately dies, 20 other survivors live miserable lives and it impacts their life, their ability to earn a living. So the government asked us to come in and see what we could do. And we identified with the government what are the three or four key pain points in the patient’s journey.
First one is diagnosis and diagnosing TB in economically vulnerable communities isn’t easy. X -ray machines, sputum analysis, etc. These are all challenging, they’re expensive, they’re time -consuming, they’re tedious. So that’s challenge number one. Secondly, if you do sputum analysis, these have to go to 64 government labs around the country. There’s throughput time, and by the time the patient gets the results back, you’ve lost some time in initiating treatment for the people who have TB. And finally, there’s a huge problem that there’s a subset of TB patients who stop taking their medication because it’s a very toxic regimen of medications, which has a very toxic effect on the body. The problem is once you stop taking them, you develop drug -resistant TB, then the mortality rate goes up dramatically to 50%, and then you infect a lot more people and so on.
So fortunately, the government has a DPI called Nixia. It’s a very large data platform. It’s a patient management system that has data on all of the TB. They detected TB cases in the country. Government gave us access to that database. We developed a range of models. And… And to address these challenges that we saw in the patient care journey, for diagnosis, we’ve come up with a way of diagnosing TB from the sound of a cough into a smartphone. It’s instant. It’s quick. It shows you what the risk is of that patient having TB, and so government workers can focus on those patients. In the one year or so that this program has started getting rolled out, the diagnosis part using the sound of a cough, detection of TB patients has gone up by 25 % nationally.
You may think that’s bad news because now there’s more TB cases, but they were there. But now we can make sure they get treated. On the labs and the turnaround time, we’ve come up with an AI way to automate a lot of this testing, so now literally you get the results the same day. Patient and the doctor finds out instantaneously, and then they can start treatment. On the issue of patients who develop drug -resistant TB, we’ve come up with algorithms that predict. predict which TB patients are likely to fall off their medication regimen. And so the 2 ,000 or so TB caseworkers in the country, which is a small number for the 4 or 5 million TB patients that we have, but now they can target their time and bandwidth on their subset of patients that really needs help.
But all of that was enabled by this DPI called NICSHA, this database, which enables all of this. One other quick example in the education space. In the global south, including in India, there’s a very high dropout rate of students in early grades, grades 1 to 5. So we got a call from a large state government in India about a year back saying, can you help? We took a look at it, and it turns out that the single key reason for this very high dropout rate in early grades, grades 1 through 5, is the inability of these young children to read proficiently in that environment. And so we’re going to have to do a lot of research. if you can’t read properly it will affect how you do in all your subjects geography, history, science, everything you start failing you get frustrated, your parents say come back home, work on the farm work in the kitchen, etc.
and that affects the rest of their life so we’ve come up with a system to diagnose within 20 seconds for each child by just speaking into a phone into our model in 20 seconds we figure out exactly where they are struggling what words what phrases, what sentences and what will help them get on the right track and this is being done at the cost of 5 paise per student I think cost is another very big part of scaling that doesn’t get discussed too much but cost is very important so 5 paise per student while this was in pilot the suite of solutions we came up with so we’ve got a way like I said of assessing in 20 seconds where the patient where the student is struggling where the patient is struggling We come up with a diagnostic and then come up with a remediation plan and exercises for each student to practice at home to improve their reading.
The state was so impressed with the pilot, they made it mandatory for all 3 million kids of that age in school. Three or four other states, including the state of Rajasthan, just made it mandatory for all 8 million kids over there. And by the way, all of this, again, enabled by DPI. So Rajasthan has a state -level DPI called Rakshak. Our system, our models sit on top of that system. It reaches 400 ,000 schools, 8 million students, and it’s spreading. So now the government of India, by the end of next year, wants to make this standard across the country. All 75 million children of that age group will get their reading improved and strengthened through the systems that we have. Bottom line, all enabled by DPI.
No, I mean, those are very… APPLAUSE Thank you. Those are incredibly powerful examples. In fact, the case of TB is actually one that’s super important because something like 40 % of people in the world go undiagnosed with TB. And, in fact, most of them are in the global south. But it also just brings me back to maybe a question of scale. I mean, in what you’re doing, you mentioned a few countries, but I think your goal is to get with some of these education and health solutions to, like, 25 countries or more. How are you thinking about kind of taking that to kind of multi -country scale? And what are some of the ideas you have about how you do that?
achieve the same you know academic goal within six week which usually takes more longer than a year long process so that’s one course one example in Nigeria also not only the TV some very small handheld ultrasound device can scan the the pregnant woman and then easily diagnosis some some problem for a baby and then it drastically reduced the birth that the baby death rate so it is one another example how can we scale this up another good example as you said we are expanding the current India model to other three African countries and in Brazil we added one more in Philippine okay and then the one of the way is that you how can you find some very standardized and scalable model, but this is not easy.
But from the one concrete example, like in India case for Agricultural Connect, we are figuring out what would be the best way and lighter model we can quickly replicate to other countries. This is our role. The World Bank is trying very hard to figure out what that means and then how they can really replicate this model to other countries. And what would be the really critical component to be replicated. So this is our role. So we are just working on what would be the best simple model is. So in that sense, usually I’m not sure it is really right analogy. I’m using this analogy as a sommelier. We are not innovation creator. There’s a bunch of really good wine producers, but I would say that this is a very good model.
customers is not really aware of which wine is really fit for their taste. So as a sommelier, the World Bank is trying very hard to understand the wines and then find some better, recommend better wines for our customers’ taste.
Yeah, I think on this question of scale, I mean, those are great ways to… Do you need any help on quality control with these wines? Yeah, but I think on this question of scale, I mean, Nandan, I’ve heard you say that, you know, we won’t get to true population scale unless we actually scale things like inference in AI and how we do that at massive scale. And I’m just curious for you to expand on that a bit more, but also what lessons and implications it might even have for people like us who are building frontier models. Say more about why the inference part of this really matters.
No, I think broadly speaking, I think, especially in the global sub… the cost of AI inference has to drop dramatically because if you’re serving a customer with one query and that costs, you know, 500 rupees or something, it’s not going to work. So we have to make it really, inference has to be, which I think, you know, you’ll do that because, I mean, there’s a lot of focus today on the training side, you know, getting bigger and bigger models and launching them. But I think as that sort of stabilizes, I think the focus will shift on the inference side to make inference cheap. But I think the, I’ll give you an example of, a very tangible example of open networks, even AgriConnect.
Yesterday I was talking to Demis. And Demis was saying that Google is improving its weather models. Yeah. They’re making it better, more efficient, more predictive, more granular in where it, you know, area by area and so on. Now, if you had an open network, if you had a network for Agri, like AgriConnect, which suppose it has millions of… then all we need to do is just plug in the latest weather model of Google into that open network, and suddenly 10 million farmers have access to the latest weather data. That’s a good example of why this open network thing is important, because it allows you to plug new models, new sources of capability, new ideas, and so on.
And I think that’s what we’re doing. And to give you another example of how it reduces the complexity, there’s a very interesting demo here of energy trading. Now, we never thought of energy as something that you traded because you bought it from the utility. But today, with millions of people producing energy, they’re able to – somebody has a rooftop solar, there’s some extra energy they can sell it to somebody else. But how does a farmer in UP learn how to sell energy? It’s a whole new concept. It’s only by an agency. It’s a classic commerce interface that is simple. So I think low -cost inference combined with – with agents that hide complexity is the key to massive diffusion.
Yeah, and in fact, I like the example you brought up on weather because in some ways, thanks to the, quite frankly, the forward thinking of part of the Indian government and the Ministry of Agriculture, they’ve set up that infrastructure. And in fact, last year, they used one of our models, Neural GCM, which predicts monsoons. And we’re actually able to deliver, I think, to something like 38 million Indian farmers predictions about the monsoons. But that only worked because the Indian government actually set up that kind of infrastructure where you could plug in these models. Kiran, I want to come back to you because, I mean, in some ways, you raise some more foundational questions here about the future of biology and health overall.
And I’ve heard you say, for example, that, in fact, AI doesn’t replace biology, that biology is much, much more fundamental and foundational. Say more about that. that and what can AI learn from biology and vice versa and what do you imagine what needs to happen to fully take advantage of?
Yeah so I think first and foremost biology works through distributed data centers okay and when it wants to build intelligence retrieve memory and inference data it does so with sips of energy not with gigawatts of power that our data centers. So we could learn something from biology. So I think we can learn something from that. I think more fundamental to that is that biology also has generational learning you know if you think about how our DNA stores generational memory I think that’s fascinating how does the Arctic turn fly out of its nest for the first time travel 70 ,000 kilometers to the Antioch and then travel to the Arctic and then travel to the Arctic and then back it has navigational intent embedded in its DNA.
How does that work? So I think we have to learn a lot from biology and use AI to learn that biology. Because without AI, you cannot have any insights into biology. So I just feel that the future is going to be about the convergence of biological intelligence and AI. And that is going to be a very powerful transformative process. Because biology has a lot to teach AI in terms of how to do it with less energy, how to do it rapidly, and how to multiplex multimodal data very rapidly. Now that is something which I think is very exciting. And I think to go back to what you’ve just been discussing with Nandan and others, I think what makes it very exciting right now is the volumes of data you can collect.
And I think that is the reason why we are so excited to be here. And I think that is the reason why we are so excited to be here. And I think that is the reason why we are so excited to be here. Thank you. on open networks. I think we have to talk about, I mean, I know I work with a lot of, you know, organizations around the world in my field. There’s a huge reluctance to share data. You know, there’s a lot of wariness about IP being, you know, fragmented. And therefore, I think, except for India, there’s a lot of this resistance to share data. Now, when you don’t share data, you’re going to silo it.
India has this unique opportunity because of its, you know, this open networks and the public digital infrastructure to share volumes of very important data like Nandan just illustrated in terms of the, you know, the environmental data that he was talking about, the climate data, and farmers taking huge advantage of that. That is what we have to really, really, really focus on because India. is uniquely positioned in terms of its open networks. And if we can actually keep generating data and then make
I’m being told that we’re going to have to wrap this up. But before we do that, though, I want to just see if we can do a quick lightning round, so to speak. I mean, this summit has been extraordinary. The example that India is setting for the world, quite frankly, is extraordinary. If you could say, each of you say, one thing you’d like to see happen in the next 12 months, particularly with this idea of open networks and change at population scale, what would that be? I don’t know who wants to go first.
I think I’d like to see massive diffusion where all these applications that are just rolling out on open networks reach millions of farmers and farmers in the world. I think that’s going to be a big deal. I think that’s going to be a big deal. I think that’s going to be a big deal. I think that’s going to be a big deal. I think that’s going to be a big deal. I think that’s going to be a big deal. I think that’s going to be a big deal. and so on, and actually show the world that AI is a force of good. I think we have an obligation to show
That’s good.
Yeah, I definitely want to see a sustainable standard of care, high -quality universal health care coming out of this AI effort and the health stack.
That’s preventative, presumably.
Absolutely. Diagnostic, preventative, predictive, and precision, because you can’t do away with treatment. But how do you basically stage it up front?
Sunil or Sangut?
Yeah, so yesterday when the Prime Minister spoke at Bharat Mandapam, you know how he has been saying for years, make an India for the world. He said, in the age of AI, let’s develop in India, and let’s deliver to the world. So in our case, is just one little example at Vadvani AI. I’ve given you two or three examples of what we’ve done. But we’ve developed over 25 AI platforms in India in education, healthcare, agriculture, which are scaling up. What’s interesting is over the last year, we’ve had an incredible amount of incoming interest from governments throughout the global south, in Africa, Asia, and so on, who are hungry for these solutions. And they’re looking to India to provide these.
In fact, when PM Modi launched our institute back in 2018, you know, he was saying the U .S. is so far ahead, China is so far ahead. I said, Mr. Prime Minister, we can set the example in India for how AI can be used for societal transformation. No one else is doing that. We are showing how it can be done.
Samgul?
So for the next 12 years, I really want to work more to disseminate the really good use cases to the world. To our country and people. One of the reasons is that… some big challenge for the people in developing world they do not clearly know what they can do with AI even though it can provide really much affordable and easiest way to expand their capability and productivity and intelligence so you know in a very easy way compared to the old old day so once again they can get to know oh this is a real important opportunity for them and then I can believe that they will really find some really good way to fully utilize this one in a very affordable way
no no thank you I mean it is what what I’m taking away is that it’s not just the example that India is setting for India and the world but also quite frankly the example that each of you is setting because all of you in your work and your organization your teams through your initiatives but also your and, quite frankly, insight. I’ve done a lot to show what leaders can do. So I appreciate the examples that you’re setting and the example that India is setting. Please join me in thanking my panelists here. Thank you, and I think with that we’ll draw to a close. Thank you.
Just request you all to be seated 30 seconds more. First of all, could we just please have another round of applause for our esteemed panelists? Very insightful. Thank you very much for coming here. The true benefits of AI, the discussion shows, can only be realized when we build for everyone using open networks. Very insightful conversation. Thank you, James, for moderating it. And to further help drive population scale impact, we invite changemakers and researchers to apply for the two Google .org Impact Challenge. One is in AI for Science, one is for Government Innovation. There’s a QR code for you to learn more. And I encourage you all to visit us at Booths 3 and 4 in Hall 5 to see firsthand how Google AI is delivering a real -world impact.
And finally, I just request all the panelists to please join Center Stage for a photograph. Thank you, everyone. Thank you. Thank you. That was great. Thank you so much. I wish you that more this morning. No, no, no. For me personally, this is inspiring. Thank you. Thanks. Okay, all right. Okay. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you.
“Real‑world AI impact depends on “population‑scale” transformation of education, health‑care and agriculture, and can only be achieved through a built‑in coordination layer.”
The knowledge base states that AI’s true potential lies in delivering population-scale impact in education, healthcare and agriculture, and that this requires coordination built into the system [S7] and [S9].
“James Manyika – Google’s senior vice‑president for research, labs and technology in society and former co‑chair of the UN High‑Level Advisory Board on AI – was introduced as the first speaker.”
James Manyika is listed in the knowledge base as Senior Vice-President at Google-Alphabet and Co-Chair of the UN Secretary-General’s High-level Advisory Body on AI [S8].
“The “digital divide must not become an AI divide.””
A knowledge-base entry warns that without proper digital public goods we risk creating an AI divide that could be even more dangerous than the existing digital divide [S97].
“Digital public infrastructure (DPI) and open networks provide the coordination layer that translates human intent into real‑world action.”
The source describes DPI and open networks as the coordination layer enabling AI to turn human intent into real-world outcomes [S9].
“India’s UPI payments system and the Bhashini language‑network exemplify this coordination layer.”
The knowledge base cites India’s UPI and Bhashini (spelled “Bashini” in the source) as leading examples of digital public infrastructure that provide such coordination [S9].
“A recent $10 million Google.org grant to the Networks for Humanity Foundation is building universal tools such as asset‑tokenisation and open‑network standards across innovation labs from Singapore to Switzerland.”
Google.org announced a $10 million grant for nonprofits to help them integrate AI, but the source does not specify the Networks for Humanity Foundation nor the geographic scope described in the claim [S102].
“UPI is an open‑architecture prototype that grew into the world’s largest payments system.”
The knowledge base highlights UPI as a leading digital public infrastructure and an example of an open architecture, but it does not explicitly state that it is the world’s largest payments system [S9] and [S71].
There is strong, cross‑speaker consensus that open digital public infrastructure, multilingual AI agents, and ultra‑low inference costs are the foundational pillars for delivering AI at population scale across agriculture, health, and education. The panel collectively envisions a coordinated, open‑network ecosystem that can be replicated globally.
High consensus – the alignment across senior leaders from Google, the private sector, the World Bank and academia suggests a shared strategic direction, which bodes well for coordinated policy action, investment in DPI, and multi‑country scaling of AI solutions.
The discussion was largely collaborative, with participants unified around the vision of using open digital infrastructure to deliver AI at population scale. The main points of contention revolve around which technical component—low‑cost inference versus robust data pipelines, or open standards versus open networks—should be prioritized to unlock that scale.
Low to moderate disagreement; the differences are strategic rather than ideological, suggesting that coordination among stakeholders can reconcile these perspectives without major conflict, facilitating progress toward the shared objective of scalable, inclusive AI.
The discussion was driven forward by a series of pivotal remarks that moved the conversation from high‑level optimism about AI to concrete mechanisms for achieving population‑scale impact. James’s opening example of AlphaFold established the value of open, shared AI resources. Nandan’s articulation of AI agents and inference cost framed the technical and inclusion challenges, while Sunil’s DPI explanation and TB case study grounded the dialogue in practical, low‑cost deployments. Kiran’s health‑stack vision and biological‑intelligence analogy expanded the scope into new domains and offered a deeper, interdisciplinary perspective. Sangbu’s emphasis on a universal network tied these strands together, showing how sector‑specific successes can be replicated globally. Collectively, these comments shifted the tone from abstract promise to actionable infrastructure, highlighted the necessity of open, interoperable platforms, and underscored the role of government‑backed digital public infrastructure in turning AI potential into real‑world, equitable outcomes.
Disclaimer: This is not an official session record. DiploAI generates these resources from audiovisual recordings, and they are presented as-is, including potential errors. Due to logistical challenges, such as discrepancies in audio/video or transcripts, names may be misspelled. We strive for accuracy to the best of our ability.
Related event
