Building AI Foundations: From Farms to Future Economies

[Video starts]

[Digital professional]
IA is one of the most beautiful and rapid accelerators of growth within any technological sector in Uzbekistan and around the world.

[Student]
I enjoyed the way that Copilot gave me correct answers and the exact answer for things I need.

[Farmer]
If you are going to the market, people give random pesticides for crops, they are also expensive. The (AI) application helps us with the correct information.

[Voice over]
Artificial intelligence is rewriting the rules of what's possible faster than any technology before it. It's expanding access to knowledge, boosting productivity, and opening doors to new jobs and markets. Middle income countries are already major users of GenAI, accounting for over 40% of global traffic in mid-2025. Yet, AI innovation remains concentrated in high-income countries, which also draw nearly all global investment. The path forward starts with countries strengthening four critical foundations. The Four Cs. Connectivity, Compute, Context, and Competency. Connectivity. 68% of the world is online, but gaps in speed, data usage and affordability hold people back. Without fast, consistent, and affordable Internet, AI tools cannot reach those who could benefit most. Compute, which includes AI chips, servers, data centers, and cloud services is becoming the new electricity. But 73% of data centers are in high-income countries. This divide slows down experimentation and adoption in developing nations. Expanding affordable cloud access and regional data centers is essential for countries. Context covers locally relevant data, content, and models. Today, 45% of global websites are in English. When AI systems learn mainly from these sources, local realities stay invisible. Competency captures the skills people need to use AI fully. Globally, GenAI vacancies surged 9-fold between 2021 and 2024, with one in five of these jobs in middle income countries. But the talent pipeline is not keeping pace, and investments in training and education are key. As countries strengthen to strengthen these four foundations, they are also leveraging approaches that fit their realities today by building Small AI. Driven by a specific purpose, “Small AI” are practical and affordable solutions to development challenges in health, agriculture, and education that run on everyday devices and reflect local languages, conditions, and cultures. With stronger foundations to support AI, developing countries can shape AI landscapes that reflect their priorities and become architects of their own digital future.

[Video ends]

[Catherine Cheney]
Welcome, everyone, joining us here in person and online through World Bank Live. I'm Catherine Cheney, Senior Editor for Special Coverage at DevEx, and I'm thrilled to be moderating this event, Building AI Foundations from Farms to Future Economies. So, as you all just heard, and as you all know, that's why you're here with us, AI promises to expand access to information, boost productivity, and, fundamentally, change the way we tackle the world's most pressing challenges. Still, many of the markets that we are focused on, low and middle income countries, have yet to see this potential translate into real impact. LMI sees risk either being completely bypassed by the AI wave or suffering harm if they have no say over the application and governance of these technologies. That's why it's so critical we have conversations like this one, where we'll explore how AI is already transforming sectors and entire economies, while also hearing how countries can build strong AI foundations. Over the course of the next hour, you're going to hear insights from the Digital Progress and Trends Report 2025, which launched today. It's the World Bank's flagship publication on the state of digitization worldwide. And this year's report, “Strengthening AI Foundations”, provides a data-driven snapshot of the global AI landscape, exploring how LMI-Cs can harness AI for inclusive and sustainable development. So we talk about these themes a lot. It's great to see some data behind the conversation. We'll also zoom in on learnings from a particular sector with lessons from the Harnessing AI for Agricultural Transformation Report, which features 60 case studies that really highlight what's already happening in this space. Our conversations today will help to inform the next phase of the World Bank's digital and AI work. Just a few quick notes on what to expect today. First, we're going to hear from Axel Van Trotsenburg, the World Bank Senior Managing Director, responsible for development policy and partnerships. Then we'll turn to a panel discussion featuring leaders from the World Bank's work on digital transformation and food and agriculture, alongside experts from George Washington University and the Gates Foundation, as well as the Director of the Banks World Development Report 2026, which will focus on AI and development. Finally, we'll hear from Songbu Kim, the World Bank's Vice President for Digital. For those of you tuning in online, please do engage in the chat. Experts are available to answer questions in real-time, and we'll do our best to bring in questions from our online audience as well as our in-person audience. With that, I want to go ahead and hand it over to Axel Van Trotsenburg. I do want to note, I'm sure many of you know, that it's a real honor to have him on his second to last day here at the World Bank to set the tone for our conversation on this topic. Welcome, Axel.

[Axel Van Trotsenburg]
Good morning, everybody, and it doesn't require artificial intelligence to get the real intelligence that it is time to go. Basically…

[Laughter]

[Axel Van Trotsenburg]
So therefore... But thank you for being here. It is a pleasure to also talk a little bit about this in a more broader context. What we need to think about is when we are in an organization like the World Bank, we talk a lot about the money that is important, but equally important is the knowledge and how we actually adopt ourselves to new realities and make a difference. And, actually, through time, we have always benefited from technology. Yes, financing was important, but technology has been moving the needle in many areas of development. That can be, of course, from the health sectors to agriculture, to infrastructure, but also now digital and AI. Throughout all these things, what we are seeing is that there have been enormous opportunities, but also, again, enormous challenges. I think that is really where AI comes in, that there are huge opportunities. I think we should be actually optimistic, but we should also not be ignoring the many challenges on this. This was also one of the reasons when we discussed about revamping the knowledge bank, that we decided to have a VPU, a vice presidency, dedicated to digital and AI. Because you cannot simply ignore it. It will be a big task for the World Bank, how we can help on this. And a good way of doing this and highlighting particularly the challenges developing countries is through this kind of studies. A lot is being reported, particularly in the United States or in Europe, because in the United States, they are responsible for the bulk of investments in this area, by far even compared to OECD countries or China. The developing countries, be it middle income countries, low income countries, are left behind with probably less than 2% of the total investments, or 1%. It's very small. For an organization like the World Bank, this should concern us that actually there is this immediate challenge is given the development of AI and digital that goes so fast that we are creating a new divide. We have had enough divides to overcome between developing countries and developed countries. In many areas, we have been successful. You see, particularly if you look at Asia, where Asia, and particularly East Asia was 60 years ago, think of that, take just some pictures of Singapore, where it was in 1960 and where it is today. Technology had an important role in it, not the only role. There were many, as I say, but they were important. We think that unless we are addressing all the different issues, the opportunities, but also the challenges, we may not get to the place we would like to see us ending up. I think what this report is doing, and you saw in the initial video clip already, where you see where from investments to use of AI, etcetera, where are the difference, who is using it and who is not using it. And what you are seeing there is indeed a large divide on this. In addition, there is, of course, the challenge in low-income countries to benefit fully. But if you don't have enough electricity or if 600 million Africans still don't have even access to electricity, I can talk about digitalization, AI, but to whom am I talking? If you can't even know where you have an outlet to actually get your phone recharged. What we have to do is to put the infrastructure in it. That is why the World Bank is also pushing hard that initiative, M300, to provide 300 million Africans with the access to electricity, clearly with the idea that this is also particularly stimulating small and medium-sized enterprise so that they can have the full balance on this. I hope a lot of the discussion will be today also on these challenges, but not with a view it cannot be done. But I think what we need to have a discussion that it can be done and everybody will need to care a little bit of their responsibility. That means private capital has to come, private investment has to be there, but also there is a need for public institutions to play an important role. That can be in the regulatory, can be also in the investment, enabling investment, can be also together with private sector investments, and see how we can actually address the different challenges, including then also that it is not done in an insular way, but that we actually create standardized systems that allow an easier expansion of it rather than everybody has their boutique arrangement that again has compatibility issues, etc. That is a little bit also what this report tries to do, strengthening AI foundations. I think that is where you would like to really start. This is summarized what they say, the Four Cs, Connectivity, Computing power, think of data centers, the Context, particularly local, and the Competency. Think that particularly computing power and so we need data centers. Right now, it's hugely conservation. Of course, again, OECD countries. How many big data centers are in Africa, for example? We need to think about this. That's how we can support that. Similarly, if we think on context, we are interested in developing solutions. But a lot of the data that are being used and where our benefits are coming, are coming from here, the United States, or then Europe. But if we are thinking, for example, agriculture, or when we say about small AI, we would like actually to know what is happening in the agricultural sector in Nigeria or in Kenya or so, and how these data can actually really help in the formulation and implementation of local solutions that we are talking. We have our local data. Then competence is not that we are not competent, but it is the scaling. It is enormously. Even that is, I think, a global challenge. Let's not say, “Well, all these countries in low-earned or middle-income countries need to be skilled.” Yes, they need to, but maybe we can also learn a thing or two in our countries. That is a new one on this. This report provides a good discussion on this. What can be done clearly for the World Bank is not only that we have this report, but we would like to see, of course, the translation into operations. What we are seeing is that particularly our support operations have been increasing rather fast on this. My expectation that that will go on much further. This is a good teaser for what is going to come. The VPU is doing this kind of report, so more is going to come, and we have big conferences also on this. Then our research department, DEC, is working on the next World Development Report. As you know, the World Development Report are one of our flagships. They have been around since 1978. What we're trying to do is to discuss and treat some of the hard issues that we are identifying for the development context. This year's report, where our colleagues are working on, is on AI. It will give you a lot of food for thought. I hope controversial, because then at least to get some energy here in it. So think of this. Match the energy that AI is creating with the energy of your discussion today and beat the AI energy. That's my challenge to you. If you cannot, then you have a problem. Let the human energy be still the stronger one. I hope with that I energize you, get a good discussion, get controversial, because what we need is actually debating the issues, and then actually see what things we can formulate that are really beneficial for the countries we try to help. So thanks so much. Have a great discussion, and thanks.

[Catherine Cheney]
Thank you, Axel. Well, with that very powerful call to action, let our human energy beat AI energy. We're going to move into the panel discussion. I'd like to invite our panelists to come to the stage here as I introduce you all, and we'll dive into these topics. I appreciate Axel's point of optimism and how we need to discuss that this can be done, but we need to talk practically about what are these challenges and how can we overcome them. Joining us to discuss just that, we have Christine Zhenwei Qiang. She's the Global Director with the World Bank Group's Digital Vice Presidency, and she leaves the team focused on digital transformation in developing countries. Next to her is Shobha Shetty, the Global Director for Agriculture and Food at The World Bank, whose work spans areas including digital agriculture. Then we have Shahid Yusuf, the Chief Economist of The Growth Dialog at George Washington University, where his work focuses on topics including the economic impacts of digitalization. And beside him is Ana Maria Loboguerrero. She's the Director of Adaptive & Equitable Food Systems at the Gates Foundation. And finally, we're joined by Gaurav Nayyar. He's an Economic Advisor at The World Bank and also Director for the World Development Report 2026, which you just heard about, Focusing on AI and development. Please join me in welcoming our panelists. Christine, I'd love to start with you and with this report. Congratulations to your team on the launch of this report, and to everyone on The World Bank who worked on this. This report provides a comprehensive overview of the current AI landscape and really brings some numbers to the conversation as we talked about. Can you share what are your key takeaway always from the data? What concerns would you highlight? But also, do you see any bright spots for these markets?

[Christine]
Thank you for the question. First, I want to give a shout out to our analytical team who are actually in the audience and prepared this report. This is for the first time that we actually conducted primary research using data from developing countries that provided new insights on the global AI landscape with a focus developing countries. I want to thank them for the timely research. Maybe I can share a little bit trends data with you first, and there's a lot more in the research. First, the overall internet penetration actually achieved 68% in 2024. In terms of speed of broadband, it was about 50% in high-income countries and middle-income countries, but it stagnated in low middle income countries and low income countries between 2023 and 2024. In terms of affordability, a 5 GB data plan price reduced to 1% of average monthly income in high income countries, but it's about 5% in middle income countries, but it's 30% in low income countries. We know AI is gaining momentum in the developing world. ChatGPT spread over to 200 economies, and middle-income countries are really driving over 40% of its global traffic. And we also know that from the report, that the growth of cloud computing exports to upper-middle-income countries exceeded those to high-income countries during 2019 and 2023. Why am I citing all these numbers? And all these trends really point to when the first takeaway I want to share with you is that the momentum and progress on digital and AI landscape are really happening in middle-income countries who are assertively closing the gap with the high-income countries, but the low-income countries are falling behind. I would call that there is a low-income country in per se when it comes to AI readiness. That's the first takeaway I would like to share with you. That said, a very clear, bright spot in the report is the emergence of small AI. This is basically built on existing infrastructure and networks such as mobile phones with offline functionalities, WhatsApp, and we also use pharma registries or local health worker systems to expand the reach. And these lightweight and very practical AI tools are already making tangible impact. And they support teachers and doctors and help farmers make smarter decisions. And these innovations are ground up. And the takeaway I want to say is that they really demonstrate that countries don't need perfect conditions to benefit from AI already now. They can act by localizing and customizing AI models for their own and specific needs. So just to recap, with the two trends I want to say, the low-income countries AI readiness in per se and the promise of small AI. The bottom line is that the opportunity brought by AI is real and it's growing, but the AI divide is also real and it's widening when it comes to low-income countries.

[Catherine Cheney]
Thank you, Christine. You mentioned practical AI tools making tangible impact. One of the things I really appreciate about the setup of this conversation is we are going to be doing a deep dive into one sector, which is agriculture, and talk about what are examples we're already seeing. I want to go to you, Shobha, of course. Here, what are some of the tangible examples you're seeing in your work? I know you might draw on some of the 60 case studies in the report I mentioned earlier. What needs to happen for AI to impact the sector at scale? What challenges stand in the way?

[Shobha Shetty]
Thanks very much, Catherine. First of all, I think a big shout out to our data and digital agriculture team and the agriculture and food global practice. Many of them are in the room today, but some of them are not. So big shout out to them as well as our digital colleagues. I'm really thrilled to be doing this jointly with the digital team. Congratulations on the flagship. And we have our own flagship as well. You can see the postcards on your table. Please do look into this report. It has fascinating, 60 case studies, really each one of them equally fascinating. I do hope you have a chance to browse through this report, Harnessing AI for Agricultural Transformation. I think as far as AI at The World Bank Group, I think our goal for the agriculture and food sector is really to use AI to help our countries to really build much more equitable and inclusive food systems, where data, innovation, and AI become part of how we produce and consume food from farm to table. So this new report that I just mentioned shows how AI can really transform smallholder agriculture, increasing productivity, strengthening climate resilience, and also reducing drudgery of farmwork in many, many countries, and much more. But what is new about AI and what are the conditions we need to make it succeed in these lower and middle-income countries? First of all, I think we need partnerships to scale. This is not something anyone can do alone, whether it's governments or development partners. We need, as Axel said, public and private sector finance to come in together. In terms of the impact on the sector, and agriculture, I think, is one of the highest use. I mean, it's the most, I would say, prolific use case that we have right now. I'm really pleased to be able to sit here and give you some of the examples are just a taste of what's in the report. First of all, AI helps in really processing large amounts of data, weather-related agriculture in real-time extremely quickly. For example, in Kenya, we are supporting the National Agricultural Observatory to really do real-time analytics to help flag droughts and pests in record times. AI is now moving far beyond monitoring, and this is enabling us to generate solutions we never had before and at a rate that is just astounding. For example, it's really accelerating innovation cycles. The International Rice Research Institute was able to analyze 60,000 varieties in a single season, and this used to take decades to do. I was very fortunate to be able to actually see this in action when I was at IRRI in Los Baños six months ago to launch our Knowledge Academy on NextGen Rice. In Ethiopia, these tailored AI advisories in local languages have benefited farmers, generating 38% yield increases and increases in profits of up to $600 per acre. Secondly, AI really revolutionizes how we produce and deliver our advice to smallholder farmers, and all the while, cutting costs and expanding access. Earlier, AI tools would cost about $35 per farmer, which is quite unaffordable. But now the costs have come down to about $1. 50 per farmer, which makes it much, much more affordable. What we're seeing is examples like Digital Green and ISDA in Kenya and India, reaching millions of farmers worldwide. Sorry, millions of farmers in these two countries with tailored advice in local languages. Now, in order to really benefit farmers, how is AI also opening up new income opportunities? I think this is critical because now it is far easier to measure and verify emissions with AI tools, and it's much easier to join carbon markets. Let me give you one example here. Boomitra is a great example of a climate tech startup, and it cuts the soil carbon verification costs by up to 90%. This has been supporting farmers in Kenya and Mexico now, about 100,000 farmers across 5 million acres. I think the bottom line is AI is opening up real climate finance opportunities for farmers, especially smallholder farmers, not just the very big companies. Let me talk very quickly about the conditions we need to scale sustainably. I think Axel alluded to that as well in his remarks. You need reliable local data. The AI tools are only as good as the data that we have, whether it's on markets or soils, and this is critical. Connectivity, the other C that Axel talked about, basic devices, especially the rural last mile connectivity. Inclusion by design. We heard about the digital divide from Christine, from Axel as well. I think this is where we must really develop these tools that can work offline, work in local languages, be able to use SMS, WhatsApp, and just be attuned to the realities that farmers face, and have data platforms that can actually talk to one another. Of course, we need strong government regulations, strong safeguards, to really ensure that the data is used in an appropriate ethical way and with good governance, privacy, and equity concerns. Lastly, I guess the innovation capacity at local levels are going to be critically important to build, adapt, and maintain these systems. Skilling, that Axel talked about is going to become increasingly important to be able to scale these systems sustainably.

[Catherine Cheney]
We've zoomed in on agriculture, which who, of course, has takeaways across sectors, but I want to zoom out again and hear from Shahid. You've followed technological change over decades. In fact, interestingly, you worked on the 1999-2000 World Development Report. This report launched today, highlights the risk of widening inequalities between early adopters and the rest of the world. I'd love your take on how we can minimize this risk.

[Shahid Yusuf]
Okay, thank you. Thanks for inviting me. I'd like to look at it from a macroeconomic perspective. If you take generative AI, it is, as you remarked, has been improving at an incredibly rapid pace, and much more is there to come. But as far as I can see, this and digital technology has not disrupted any particular subsector as yet. It has not disrupted EdTech, it has not disrupted the financial sector. None of those have been affected. Also, rather surprisingly, the automation that has taken place in manufacturing in any country, whether it's Korea, Japan, Germany, the United States, or China, has made no difference to the productivity of manufacturing. It's quite astonishing to me that you've got all this happening and nothing much has happened, at least as a macroeconomist, when I see growth and productivity, I don't see much happening there. In particular, unemployment and employment has remained pretty much stable for the last several years. All of these worries about AI displacing people has not really been realized as yet. As you mentioned, there is lots of small-scale models that are now available because of open-source and open-way foundational models like LLaMA and DeepSeek and Qwen and others. There is plenty of things for plenty of experimentation that is taking place, many new pilots are being launched. In time, I expect there to be some productivity gains. I expect the needle to move, but it hasn't moved yet, so there's time for it to come. At this stage, I see no advantage for developing countries to invest a lot of money in data centers. At this stage, I'm saying eventually, there may be a need for some countries. I, for instance, work on small island economies. I don't see any reason for them to invest in data centers or anything. They should just use the low-hanging fruit, which is the small-scale models that are available, and will get even better. The other thing I would say is that I see no advantage in premature automation, whether in manufacturing, whether in transport, whether in any of the service industries, because countries really need jobs. They don't need to displacing workers at this stage and putting valuable capital into AI when you don't need to as yet. Even the United States is diverting enormous amounts of capital into AI when there are so many other things that are crying out for capital and investment. I would say, hold your horses. Don't suffer from FOMO as yet. When the time comes, we certainly do so. I would say one other fact that really there shouldn't be a worry about the advanced economies pulling away from the developing countries. They aren't pulling away. If you look at Europe or the United States or even China, there is no acceleration in the growth rate there. There is no medium-term forecast suggesting that there will be an acceleration in this growth rate. If developing countries grow at 4% and the developed countries grow at 1. 5% to 2%, there's no likelihood of there being a widening gap, at least again in the near term. But again, going forward, there may well be. Thank you.

[Catherine Cheney]
Thank you, Shahid. I really appreciate your perspective. You mentioned FOMO, fear of missing out. From what I'm experiencing and talking with leaders working across many low and middle income countries, the FOMO is real. There might be some debate on this panel. Axel did encourage us to have some debate as to when I mentioned earlier, countries could miss this AI wave, where is this wave? I think Shahid's view it's a little further out than maybe some may say, versus completely overtaking us. But we might have different views, and I'd love to hear from others on that. I want to go to you, Ana Maria. The Gates Foundation has long emphasized equity in its work on agricultural innovation. I know a lot of your work focuses on equitable AI for food systems, but I'd love for you to break down what that means in practice and how are you working to ensure that smallholder farmers and not just agribusiness benefit.

[Ana Maria Loboguerrero]
Thank you very much for the question, Catherine, and thank you very much for inviting the Gates Foundation to participate in this conversation. It was a pleasure to hear from Shobha because we are so aligned. Again, part of my purpose of being here was to highlight this report, The Harnessing AI for Agricultural Transformation. I invite all of you, you have cards in your tables, because it was a fantastic partnership with the World Bank. We are grateful for working together with you, you taking the initiative and then allowing the Gates Foundation to co-create with you in terms of this report. It was a total pleasure. I know that not everybody in the team of the World Bank is here, but I want to just use a few seconds to thank Parmesh, Parvati, Marian. They were the ones that were really working very hard every single day to make this happen. So thank you for that. It's really interesting. I'm going to hear the conversation because Axel started with the divide. It was not only referring to agriculture, but in general, and more thinking about the high income countries or low and middle income countries and the differences. In the agricultural sector, you can cascade down that argument. This is what we are seeing. First of all, we are seeing really interesting outcomes in terms of increasing yields, increasing income for small-scale producers by using AI advisors. We have been documented that through the different work that we have been carrying out with digital green that was mentioned by Shobha, but also with Tomorrow Now in Kenya. We are very enthusiastic about that because we are seeing changes in terms of improving the livelihoods of small-scale producers. But your question is the right one, because, again, what we are seeing at the same time is that all the way from the production of the information, the content that it's feeding the large language models into the delivery and access to mobile phones, we are seeing a huge difference between men and women farmers. There's a risk that we can increase that divide. The way that we see the Gates Foundation working on AI in agriculture is basically making sure that we are not leaving populations behind. We truly believe that working on AI in the agricultural sector can reduce poverty in low-income countries. That's one of the main tasks that Bill Gates put in front of us for the following 20 years before the Gates Foundation closes its doors. We are continuing work on the Agric Development Program because we believe that that's one of the strategic sectors to reduce poverty in low-income countries. Again, AI can make a difference, but AI, if we do it in the right way. Just to give you a concrete example, when we think about corpus, so what is the content, the data that fit the large language models? We don't have a lot of data that speak about women, small-scale producers. And there's a bias. There's a bias. And obviously, the advisors that are coming out from those LLMs are going to be biased toward a stereotype, which is usually a farmer that it's a man. So therefore, the advisors that we believe that can make a huge difference in terms of reducing losses once the droughts come, for instance, they are not going to be fit for purpose for women small-scale farmers. And then this is where the divide starts. Just to give you another example, in terms of ownership of mobile phones, we are also, and we have numbers in terms of the difference between ownership for men and women farmers. So then we are working on digital green. How do we make sure that we innovate in terms of using AI, but other devices, and there's where the languages make a lot of sense, as you were mentioning, having different languages. But also, for instance, we see opportunity on AI-enabled advisors. But then, again, same thing happened. Why women do not engage more or adopt these advisors? According to our study, sometimes is when you have the advisors that are women, then women are more prone to follow these advisors. There's a lot of things that need to be done. To tell you the truth, if we don't take this seriously into account, we can increase the divide, and that's certainly we don't we want to do. So bottom messages. We see a lot of future AI and agriculture in terms of increasing productivity, yields, and income, but we need to do it in an inclusive and responsible manner.

[Catherine Cheney]
And thank you. I know gender has been a big focus of your work for talking about how we ensure these tools benefit not just smallholder farmers, but all smallholder farmers, men and women alike. Gaurav, I'd like to go to you because we mentioned that today's conversation will inform the World Bank's work on digital and in AI. So I'd love to hear, as you work on the World Development Report 2026, how will this report inform your effort, which will then, of course, inform the conversations and work we all continue to do in this space.

[Gaurav Nayyar]
Thank you, Catherine. And congratulations to Christine and the team for this excellent report. And absolutely, there are several synergies. I'll try to be very brief, but give you a glimpse into how these two reports can work together. What the WDR will try to do in a nutshell is identify AI's salience for development, then go on to assess the potential implications or impacts of AI on economic outcomes, social outcomes, and political outcomes. Then lastly, it will look to identify how countries can leverage AI as a technology to try maximize the benefits along these outcomes while safeguarding against potential risks that may come along the way. We are trying to think of it in terms of countries, firms, governments, workers, adopting AI solutions as they exist today on the market, adapting AI solutions to the local context, and we've heard a lot of that from Christine, from Shobha, from Shahid. Then finally, countries potentially advancing the development of AI as a technology. The main link between this way of thinking about AI adoption, adaptation, and advancement is that these stages of the technologies development or use are inherently linked to the Four Cs. When we think about connectivity, context, capacities, and compute, these are the essential building blocks that will signal the readiness of people, of governments, of businesses, of countries to leverage AI adoption, adaptation, and advancement for the greater good. In that sense, this report, it's aptly titled strengthening AI Foundations, and that's the primary way in which it will inform our thinking on how governments need to position themselves in the policy space to leverage these potential benefits that the technology will bring. In terms of going forward, how both this report and the WDR would look to inform the World Bank's work on AI more generally, I think I can highlight three things. The first one is really if you look at the Four Cs that this report talks about, it's essentially a signal of AI readiness. How ready are countries? How ready is the private sector? How ready are governments to leverage AI as a technology for development? This, in a sense, can actually signal to World Bank dialog and World Bank operations as to what the priorities may be in a given country in terms of developing the foundations and Axel at the beginning talked about financing and how the World Bank can help. What are the priorities that we should be thinking of in terms of our lending operations? Like Shahid had said, for example, in low-income states, it might not be about building data centers, but it may be about providing support to access cloud centers that might be stationed elsewhere. That would be one aspect of it. The second one would be around technical assistance. What I hear from many colleagues around the bank is that if you are a government official in a developing economy today, you will have at least five AI entrepreneurs, startups, or SMEs coming to sell you an AI solution on every day of the week, maybe also on the weekends. I think the World Bank will have a big role in terms of technical assistance in advising governments on how to sift through what these AI applications are promising, how accurate they are, how verifiable they are, and how they can be evaluated. This links to the third, I think, thing that these reports could inform going forward at the World Bank is really typically at the World Bank, reports are written on the basis of or on the history of two decades of research that has gone into a specific topic or field. Here we're in native territory, and I think it's literally the other way around is that I think we're going to use these reports, both the reports we've launched today as well as the WDR, is actually to set in motion a research agenda or an analytical agenda at the World Bank, which is really going to look at measuring and evaluating AI solutions in the context of our client countries of developing economies, which at the moment is few and far between. I think these are the three ways in which these reports could be useful to signal a work agenda going forward on AI at the bank.

[Catherine Cheney]
That's great. I really appreciate your framing, and it adds a real sense of urgency to this conversation that these reports help to set an agenda I know several times, some of our speakers have mentioned that on our tables, we have QR codes to the reports. Also for those tuning in online, you can access the reports there. We also have a couple questions coming in from online, and I'd love to try and go to a question or two from the room as well. Let me go ahead and just mention those questions, and I'm going to try to incorporate them into the remainder of our conversation here. Two questions that have come in from online. Marshall Gezi from Zimbabwe asks What steps can be taken to ensure that AI will not bring negative consequences to the agricultural sector, and how can we improve its competitiveness and accessibility to farmers from different walks of life? Of course, we heard a little bit about this from Shobha and Ana Maria, but I think what we're going to get to is what steps countries can take. Let's zoom in on the agriculture space in particular and avoiding risks in the agriculture sector. Sardar Aftab Khan from the United Kingdom asks How can people living in disputed territories and conflict regions access AI-enabled solutions to boost productivity where there is limited access to Internet and direct access markets? Again, we've talked about connectivity and access to electricity as major barriers, but specifically in disputed territories and conflict regions, there's a question about that. With those questions in mind from our online audience, and before we jump to questions in the room, I just love to open it to the panel. Any thoughts on the next steps countries can take? If you were talking directly to a government minister or on our mind? What would you say? Here's a step to take and maybe we can get to some of those questions as part of our responses. Who wants to jump in? Ana Maria, looks like you might want to say something.

[Ana Maria Loboguerrero]
I alluded to some of the steps in there, but I think what is really relevant for the countries, and this is thinking about ministries of Agriculture, Environment, national meteorological offices at the country level and regional levels. We are working with the African Center for Meteorological Application, SACMAT. I think that it is very important to on this agenda and to understand what it's needed to make it inclusive, and as you mentioned, to make sure that we are reaching everybody. And this includes a lot of capacity building. I remember Axel was mentioning infrastructure at the beginning. I will say institutional infrastructure as well. And what I mean by that is that AI already has some tools that are ready to be used in terms of, for instance, better weather forecasts, climate information. But then what it's needed is that we have the right institutional infrastructure to cascade down those weather forecasts into the hands of small-scale farmers, both men and women. How do we make sure that centers, for instance in Africa, such as ACMAT, are truly benefiting from these AI approaches, and then making sure that the right workflows are in place so that this information goes to the national meteorological offices in these different countries, and then how we make sure that it lands in the hands of farmers at the right time because there's a delivery issue that needs to be taken into account. All of that, and I'm expanding a lot, is to say that we need to invest with countries in terms of understanding this institutional infrastructure that can benefit from what is happening in the AI world to be able to cascade down into the hands of farmers in a timely manner.

[Catherine Cheney]
Shahid, I know you wanted to jump in. On the one hand, you said “hold your horses”, but on the other hand, what should countries be doing now?

[Shahid Yusuf]
Okay, so they are holding their horses, but I would calibrate the message for some countries, for most countries and then for some countries. For most countries, my sense is that by the time we get to '08, '09, Generative AI will be incredibly intelligent, very adept, very user-friendly. For most people, all they will need will be a smart device, access to broadband, and possibly the countries should develop foundational skills. That almost goes without saying, whether you have AI or not, you must develop the foundational skills. For most countries, you don't need to actually huff and puff and try and develop all sorts of things. The AI will do it for you. It will be incredibly helpful and useful. For some countries, and I can only think right now of just India, maybe Brazil, maybe Mexico, these are the countries that would need to invest more extensively relatively in the entire AI value chain. They need to worry about CapEx into data centers, into fiber, into power utility. They need to worry about water availability for cooling data centers, all of those things, land availability, etc. Those countries need to start working now and working rapidly in order to catch up. But most of them don't need to do so. I think AI will do it for them. Just a couple of points about agriculture. I've been working on cocoa in West Africa, Côte d'Avor, and elsewhere, also in cocoa in Indonesia. What I find is declining productivity or stacked in productivity and in real worry in these countries that enough cocoa will not be produced. The reason is that whatever AI can do for you, climate change is going to gouge large amounts of productivity and output from these various farming systems, whether it's palm oil, whether it's cocoa, whatever. It's going to be on a treadmill, running hard, but climate change will be eating away at productivity gains. We need to factor that in when we talk about the macroeconomic consequences of AI. The second point I think I'd make is that even as the World Bank advises country on what to do with AI, how to use it more effectively, they should be warning them about the enormous downsides which are becoming apparent every single day, which is insurance companies, bond markets, everybody is worrying about them. At some point, if we get to superintelligence, then we've got a situation where AI could run away from us, where deep fakes, fraud, scams, the AI deceiving us all the time will become very, very serious. And so while the US and China may be able to cope, but many developing countries would be widely exposed to problems which they would not be able to cope with. And regulation and ethical AI is lagging far behind. It's not going to be able to catch up with this.

[Catherine Cheney]
So I want to bring in a couple of audience perspectives, if we can. If there's anyone who's prepared to ask a question, please head to the microphone and we'll try to bring you into this. But I also want to go to you, Shobha, because in terms of what can be done, I'm picturing this treadmill and it all feels a bit daunting. I know AgriConnect is one effort the World Bank has launched to maybe address one part of this problem. Can you tell us just briefly a little bit more about that? And then we'll try to bring in a couple of quick questions before we close.

[Shobha Shetty]
Sure. Thanks, Catherine. Well, AgriConnect is a World Bank group initiative that we launched during the annual meetings, and it's essentially an effort to really transform smallholder farming at scale, really moving farmers up the value chain and creating more and better jobs, improving food security, and strengthening climate resilience. As Shahid just mentioned, those challenges continue to be fundamental. Because if we look at the numbers, we have over a billion young people entering the workforce in the next two decades. We only can create jobs for a very small percentage of them. And agriculture is one of the sectors that can be a real engine for job creation. And that's what we hope our AgriConnect can do in working with our client countries, because food production has to increase by 30% by 2050. We know that 80% of food worldwide is produced by family farmers, of which 500 million are smallholders. But as we all heard today, many of these smallholder farmers don't have access to electricity markets, finance, and so on. That's where AgriConnect comes in terms of helping farmers move up the value chain. But AgriConnect is not an AI initiative. I mean, the digital and AI is the glue that will help keep the system together and really take it to scale along with foundational investments in infrastructure, human capital, preserving national capital, reforming policies, and then, of course, bringing private sector capital in to mobilize in a big way.

[Catherine Cheney]
I think it's a helpful example. I'm so sorry to cut you off, but I want to bring in a question from our audience because we'll see if we have time to get to it or else perhaps Sangbu Kim will get to it in our closing remarks. But can you tell us your name and affiliation? We'd love to hear from you quickly with a question.

[Johan Bjurman Bergman]
Thank you very much. My name is Johan Bjurman Bergman. I work in the digital global team as well with many of the attendees here today. My question is actually picking up on what you just said, Shobha. We've talked a lot about how the public sector can help put in place the foundations of AI readiness. We've mentioned connectivity, computes, skills, etcetera. But what is the role of the private sector, and particularly in countries that are not the Brazils and the Indias, where you have a very efficient private sector with sufficient capital. But in the more middle income and perhaps even lower income countries, what is the private sector role in increasing AI readiness? How can the government support and incentivize that to happen more quickly?

[Catherine Cheney]
Okay, thank you. I'll see maybe Gaurav or Christine for just a quick closing thought before we hear from Simon.

[Gaurav Nayyar]
I think just a very quick partial response to Johann's question would be I think that when we think about small AI solutions or if we think about AI adaptation, I think that is going to be the key critical area where the private sector will need to be involved because it is not the comparative advantage of governments to develop these things. Just one parting thought was that I think if we look at the Four Cs presented in this report, I think we have to think not only of the cost of actually implementing these four pillars of the digital AI foundations, but the opportunity cost of not doing it. I think that could be an interesting I'd love you to think about it going ahead.

[Catherine Cheney]
Thank you so much. With that, we'll see Christine. I would love to hear from you, but I have to keep to time, so I want to go ahead and bring Songbu Kim to the stage. I see he's here with us, the World Bank's Vice President for Digital, to join us for some closing reflections and perhaps we'll hear a bit about the role of the private sector as well in that response. Thank you, Songbu. Welcome. We'll hear from you right here. We'll stay seated. Please join me in welcoming Songbu.

[Sangbu Kim]
Thank you, Catherine and all the panelists here today, and then everyone who joined here and online. It is really important moment for us to talk a little more about the AI and AI readiness for our client countries and our people we are serving. I would like to highlight three things. We already have touched upon all the details the agenda today with the panelists, but application, application, applications. It is very important. Even though we have identified four major key infrastructure for AI, like connectivity and computing power and competency and context, it is very important, by the way. But sometimes we cannot wait until those Four Cs get ready. So, in the meantime, we need to find out some very practical application and adaptation cases across our clients and people. That's how students can get ready and get better through the AI solutions. Farmers can get some better advice through the AI, and so many applications we can expect. From the 4C point of view, still it is very important and fundamental, but we need to be very wise to selectively choose what That type of really critical things we need to prepare first according to the maturity of the context in the countries. Maybe we can have more discussions about what would be the best portfolio and best ways to get the country ready for those Four Cs. Maybe we can expect a variety of the portfolio depending on the country context, but it is That's also very important. Lastly, I would say two more things is really important. AI and digital can apply for everything. That means we need to have some very strong collaborations across so many clients and so many stakeholders and our colleagues. I'm very delighted to have to show on our discussion today, this is really critical to work with other vertical agriculture education and some other areas such as health care. It will be really important among us, but not only among us, maybe when we talk with our clients, the government officials or private sector, some more strong collaboration across different line ministries will be the key to be successful. So collaboration is the key of the success factors. On the other hand, the last thing is the timing and speed. Unfortunately, time is not on us. So AI is developing so fast. This is much, much faster than the digital development. If I say this is the digital is the booster for the economic development, AI is a turbo booster. Boosters. We cannot even imagine how fast they are evolving. The timing and speed is such critical factors for us. That's why I need to really ensure that the importance of the speed and the timing for all of us. So thank you again for your participation. I really appreciate all of this effort. A big shout out for our research team and Christine and team and some good support from the graph and the deck. Thank you. Thank you again.

[Catherine Cheney]
So again, for those of you here with us in person, you can find more information on the reports on your tables. For those joining online, you can find more information there. I want to thank our experts for their insights and time. Thank you to our online and in-person audience for your engagement. And again, this will inform the ongoing work of the World Bank in this space, so the conversation will continue. Thank you all again. Thank you.