🔴 LIVE | BANGA of the WORLD BANK and GEORGIEVA of the IMF participate in the flagship event…

This is room channel one, original language.

This is room channel one, original language.

Present meeting back then. Well, fast forward 35 years later and we're now at the 81st annual meetings. Round of applause, 81st. Time moves on, doesn't it? And today, of course, as we meet, the world is in a very different place. And one of the biggest challenges and changes right now is, of course, artificial intelligence. Now, much of the global conversation around AI has focused on managing its risks. But today, we are going to pose a different question: How can AI be harnessed for good, in order to create jobs and growth? Today we will explore this, focusing particularly on practical and accessible applications of AI. What we have just seen in the video, small AI and how it can turn technology into real opportunities. Remember that you can share your opinions at any time on X, Facebook, and LinkedIn, or use the hashtag WBG meetings. We have experts in multiple languages ready online to answer any of your questions. Please also use the QR code you can see on the screens to send us your questions. And if you are following us online, join us through the live chat. Well, without further ado, let's start this session. Join me in welcoming three leaders who are helping to shape this conversation. With me are Ajay Banga, President of the World Bank Group; President Tharman Shanmugaratnam, President of the Republic of Singapore; and Kristalina Georgieva, of course, Managing Director of the International Monetary Fund. Welcome, all of you. Hello. Hello. Hello. Hello.

Welcome, all of you. The 81st meeting, as we said. And many things have changed, of course, right? So I think we will start with you, Ajay. This session is titled Small AI, More and Better Jobs. Jobs and opportunities are at the center of the World Bank's agenda. How has the Bank changed in that regard? And how does AI fit into all of this at this moment?

Yes. Thank you, and thank you. Being on a panel with both of you is a privilege, and I am lucky to be here. I think the entire focus of the Bank has to be on jobs for a very simple reason. 1.2 billion people will turn 18 in the next 15 years. This could be an enormous demographic dividend if, during their growth, they get clean air, drinking water, healthcare, education, and training. But once they grow up, they need the opportunity to have a place and be a productive part of society. So the word job doesn't mean working for a large company, a small company, or as an entrepreneur. Being productive in society, having a sense of optimism and a feeling of dignity that comes from earning money and being part of that system — that is what we are talking about. The entire focus of the Bank has been oriented toward that in recent years. It's not that we are not going to build a school, a road, a training center, or a healthcare clinic. We have to do all of that, but we must do it with the perspective of creating jobs, which begins with having the right foundations: physical and human infrastructure; having the right governance and policies, something Singapore has done very well in its history; and then having the right way to reduce the risk for private capital so that it can come in and invest at scale, because it is private capital that creates jobs. Governments facilitate, and the private sector generates between 75 and 85 percent of jobs in any country. That is what we are trying to do. How does AI fit in? In emerging markets, where the conversation is not about large language models, where people are worried about the possible impact on jobs — possible, not yet proven, but possible — in emerging markets, it is about applications that could enormously increase productivity for a school, an educator, a health system, or an agricultural system so that smallholder farmers can reach scale; and hopefully, as we speak, as well as in the following panels, you will get many examples. That kind of thing is what we focus on: the deployment of small AI in emerging markets as the answer, and productivity.

Okay. Thank you very much, Ajay. Let's move to you, President Tharman. Ajay mentioned your country and the work they are doing regarding the development of AI. You have said in the past that AI could disrupt jobs in advanced economies, but could create a different risk for developing economies. And Ajay made that distinction a moment ago, in a way. You say it is possible that new jobs will not be created fast enough in developing countries. What do developing nations need to do now, President Tharman, to make sure their young people can participate in and benefit from the next wave of technology and economic growth? And as we just heard from Ajay, you know, it is the future of young people and jobs that is crucial to all of these conversations.

Well, I think, building on what Ajay said, although I will speak at 10 percent of the pace at which he spoke, the first thing we must realize is that we are at an early stage. This is not a shock. This is not a pandemic. This is a whole new era, and we must begin by recognizing it in the public sector and among all our stakeholders. Growth will be fundamental to securing good jobs, but growth alone will not be enough to achieve those good jobs. And the evidence is already there. Look at China, India, and developing Southeast Asia. They're growing quite well, but job creation is weak. So there's already a divergence between growth and job creation, and AI is going to widen the divergence if we are too passive about it. So we've got to do everything we can to first use AI to create good jobs and to create jobs that can be done well by human beings, that can allow them to ride up a career ladder despite AI. The second thing we've got to be really concerned about is a K-shaped economy and a K-shaped labor force. In the first wave of automation, it was the low end of manufacturing that fell out, and some of them moved into services, the gig economy, and so on. This time around, it's the broad middle of the workforce that we have to be concerned about, and we cannot afford to have a K-shaped workforce or population where the broad middle and that broad set of skills that the middle of the workforce has accumulated falls out. So bear that in mind; these are two broad structural changes. Now, I'm an optimist, actually, like Ajay and like Kristalina. I think there are sectors of the world economy that still offer very significant promise for the developing world. First, don't give up on manufacturing. I'm not a pessimist about manufacturing. There are whole swaths of the labor force in the developing world where costs are still cheaper than China and still cheaper than robots. Just look at four or five states in India that are now growing at a remarkable pace in export manufacturing. If the other states could do the same, India gets lifted up despite its late start. Look at Vietnam and look at those that haven't yet started that game. Courting investment, making land available without encumbrances, and sorting out labor regulation so that workers can get into a job and move up the ladder is critical. Green energy is another opportunity with multiple benefits. First, of course, it provides for energy security, and it allows you to take part in the AI economy because you need energy. But it's also a job multiplier because renewable energy generates three times as many jobs as fossil fuel energy for the same million dollars that you put in. The third opportunity: think of the jobs that can be enhanced by AI and are very unlikely to be displaced. Healthcare is a huge opportunity in the developing world. The opportunity to just get anywhere close to advanced country standards. You don't have to be perfect. You don't have to get right to the top. Just closing half the gap even means many more doctors, nurses, and community health workers. The proportion of nurses per capita in the developing world is one-fifth to one-tenth the level of advanced countries. There is huge potential in India, where community health workers are now a major part of the health system because they help with immunization, maternity health, and educating families. So, it is a whole set of jobs in healthcare that can be created. Finally, I just want to make one point that cuts across sectors. AI is going to blur the distinction between being an employee and being an entrepreneur. The rise of the one-person company, so to speak — or a company that starts with one person and then adds a few more — is a new phenomenon already emerging in many parts of both the developing and developed world. And we have to encourage that because that, too, is a new form of employment, but it is different; you have agency. Small players using AI can reach markets and access technology much more feasibly, efficiently, and at a lower cost than before. What can we do about it? If you look at Zhejiang province in China, they now have a service alliance for one-person companies providing a range of shared services, including AI models, energy sources, and various professional services. So that itself is a new growth sector in the job market, and we have to encourage it.

Thank you, President Tharman. It's interesting that you mentioned healthcare, because we do have a speaker coming up in the program who is using small-scale AI in India. So we will get more on that later. Thank you. Kristalina, now over to you. What is the economic impact of all this, particularly, of course, in developing countries, as we have heard?

Well, I'm going to call you President of the World Bank right now. So that's not an AI instruction. So let's get back to you then, Ajay. So we heard from Kristalina there talking about the economic opportunities, particularly in developing countries. I want to pick up on what this event is all about in some ways, which is AI in developing countries. What does it look like in practice, Ajay? And how can it create jobs and opportunity? And I think just picking up from what Kristalina said about electricity and, you know, getting people online in the first place, how can small AI be accessible in the way that it's sort of developed? It's different, isn't it?

Yeah. So, you know, I've been very fortunate that President Tharman has helped us at the Bank run what's called a jobs council. He's the co-chair of it, and we have a group of people from academia, from public and private companies, as well as from civil society. And we came out with five sectors that were critical to grow jobs, all of which are not reliant on moving jobs from the developed world to the developing world — the kind of last 30 or 40 years model — but instead build locally relevant jobs. And those five sectors connect to this conversation. The first is infrastructure. It's construction, but then what it enables: bridges, roads, airports, digitization, electricity, water, but also human infrastructure in terms of skilling and education and the like. The second is agriculture for small farmers. Most of our economies are very heavily dependent on food grown by small farmers. They get outcompeted to the point where it's tough for them to play in the system, but access to cooperatives, access to technology of the type we are building — I'll give you a couple of examples — enables them to get past that. The third one is primary healthcare and delivery. President Tharman was just speaking to it, but think of Indonesia with primary healthcare centers, each with a nurse, a midwife, and a medical diagnostic technician. Think of the quality and number of jobs you create through that extension, and then use technology and AI to connect to remote doctors and the skills that may not be present in a primary healthcare center. And the fourth is tourism and what it can multiply in terms of jobs. I mean, Thailand is a principal example, but they've got ambitions to go even further than what they've done thus far. And then manufacturing, particularly value-added manufacturing, which is not pure extraction but creation of local jobs. Minerals and metals: the emerging markets are gifted with large quantities of deposits, but they need real quality jobs. But also look at the creative industries or the like; there are so many opportunities in manufacturing across the board. Whether you do this as an entrepreneur, to President Tharman's point about individual agency, or you do it by working for others, all are possible. That's the idea. Examples: healthcare clinic in Indonesia, you, the patient, and the nurse. You walk in and you say, 'I have a rash,' and I look at it and I can't figure out what it is, but I can take a photograph and send it to the doctor I share with 20 other clinics.

I do actually have a rash right now.

Sorry about that. But I'll ask the doctor to intervene, and so he in turn can either look at it himself or run it through a simple local AI tool. Developed at the edge on a 3G phone locally, it's very simple to do. He gets that and says, 'It's an eczema, give her this ointment, tell her to go home, or if you don't like it, tell her to come to the regional hospital to meet the specialist in 14 days at 10:30 in the morning.' That is a very difficult thing to do, even in the advanced economies today; getting a specialist appointment organized digitally is a very hard thing to do. So you can cut through the arbitrage with AI. Here's another one, something we do right now with farmers in India and elsewhere with Google, actually, I think. See this: the idea is a platform that's been built to connect farmers and cooperatives to better fertilizer, better seeds, better crop insurance, better credit, and most importantly, better offtake prices for their product. And so, an illiterate woman farmer in India can point a 3G phone at the back of a leaf and say, 'That disease — I don't need to know the name — but that disease equals this insecticide.' Press the button, because India has put bank accounts in everybody's hands, you press the button: '25 rupees for that insecticide at my cooperative.' That's the kind of stuff that small AI is about. The reality is that large language models are estimated probably to create some degree of disruption in jobs in the developed world; by the time you get to middle-income countries and lower-middle-income countries, the opportunity from small AI to build jobs and build productivity is actually enormous. That's what we are trying to do, and that's what the jobs council started out to do. In fact, the 'small AI' term came from a jobs council meeting with President Trump. Thank you.

And I think what's interesting, isn't it, Ajay, is the accessibility of it; like you say, it can just be in someone's pocket. Countries like India, smartphones have a huge uptake. We'll come to you next, Kristalina. You've described AI as a tsunami for the labor market, with young, entry-level workers especially exposed. What should policymakers be doing now to ensure that it creates more opportunities rather than disrupts? And I know President Tharman touched on some similar areas already.

Okay. Thank you, Kristalina. You mentioned many things that governments should do. Conveniently, we have a president sitting next to you. Obviously, you also have a role on the jobs council. There are many really interesting points that Kristalina raised about fiscal issues and about skills in particular. And let me ask you then about skills, because you have regularly emphasized the need to invest in them, particularly the importance of public-private partnerships. What do you think governments and companies can do differently then, to make lifelong skills development a reality? I mean, it really picks up on what we just heard from Kristalina, doesn't it?

Well, you know, the silver lining of AI-driven job displacement is that it will happen task by task, and that gives us time. It means that in the workplace, while people still have jobs, there is that opportunity to adapt, learn new skills, build on what they already know, and stay agile. So AI is not a tsunami that will arrive suddenly, as Kristalina was saying; it will arrive gradually, it will spread across the economy, and we have time to adapt. The critical thing is learning on the job. However, there is a particular issue we must address very soon in advanced and developing countries around the world, and it is that the first rung of employment, your first job when you graduate, is now seeing less demand. The evidence is already very clear; it is clearer in the sectors most exposed to AI, by the way, which are the tech sectors. It is clear. It is becoming clear across the board, and what governments did during COVID, and what some governments did during the global financial crisis, now has to be thought of not as a crisis measure, but as part of the menu of structural policies that governments have for the new era: governments must engage with the private sector to make that first job possible, to make possible that path to the first rung of your career ladder. We cannot afford to lose that, because if you lose it, you don't just have that group of young people waiting a year and then competing with the next group the following year; you lose more than a year of skills. You lose a whole set of intangible attributes, and you lose confidence in the future. So we must now, as part of public policy, find ways to incentivize the private sector to take in fresh graduates so they can get that first foothold in the market and learn from there. That is one policy implication, but I also want to work backwards from what is happening in the world of work into the tertiary education system, because I think two very important shifts are required there. If you look around Asia, you look around the developing world; in fact, if you look around the advanced world and think about how to avoid that middle getting displaced, the middle of the workforce getting displaced, it means we also have to change how the middle gets educated. And the reality of tertiary education systems, higher education, universities and the like, is that they are overwhelmingly academically focused. And here again, the data is already clear. The link between academic credentials and career prospects is weakening. In the US, for the first time in a century, demand for graduates has gone down. The wage premium that college graduates earned has gone down very significantly. So we must take these as early signals of a new era, work our way back into the education system, and reduce that skew we currently find toward an academically oriented education, shifting toward an education that is more applied, includes more work as part of the curriculum, more attachments, more internships, and more of the dual model of education where you spend serious time in a firm as part of your tertiary education. That is one major shift that has to be made. It is a critical issue for developing countries where you still have to develop that applied education stream. But one final point I must make is one that stems from what Kristalina said about the fact that we do not know what will happen in five years' time. We need some humility among education planners. Do not fixate on one particular model of education, particularly on the issue of depth versus breadth. We do not know what will be required. The reality is, in some areas, you need great depth of expertise to be able to learn continually on the job. And there will be other areas where you need more breadth of exposure to different disciplines because you have to integrate AI across functions; you have to do many other jobs that require a breadth of human skills, but there is no one model, so don't fixate on one model; have some humility; be clear-eyed enough to realize that the future is fuzzy and agility is the way we must go, not just for firms and industry, but agility in education itself.

Okay, thank you very much.

Mr. President, we're out of time, but it was very interesting talking with you all. I wonder if, in just two or three words, you could give me a quick idea of where you think we'll be. You mentioned five years, things are going to change a lot. What would you like your big goal to be in five years? Start with you: real AI implementation at small scale in as many emerging markets as possible. That's where I'd like to get to. Thank you, SJ.

Mr. President Tharman. Well, avoiding a new divergence between the developing world and the advanced world, and seeing continued convergence, as we've seen in East Asia, especially using AI, using technology, and developing the best of human capabilities.

I'm sure everyone will agree that it was a very stimulating start to this discussion, examining the topic of skills and education with respect to the future of AI and, of course, also examining technology and infrastructure. And one of the central challenges in almost every conversation anyone has about AI is trust. The questions that are often asked are: is it safe to use, and do people really have the confidence to use it? Who is responsible, and will our AI support people rather than replace them? Well, our next speaker has spent years working at the intersection of technology, trust, and responsible innovation. Give a warm welcome to Tiffany Shingju Wang, national bestselling author of "The Trust Code" and founder of Songsheets AI. It's all yours, Tiffany.

Thank you, Regine. Hello everyone. Thank you. So in Silicon Valley, we talk in big numbers. Trillions of dollars for a company, billions of parameters for a model, and of course, gigawatts of energy to keep both running. But today the most debated question is: should we trust it? Oh, trust is the most overloaded and ill-defined notion. It can mean alignment, safety, protection, privacy, and long before the list ends, we've already lost you. So what should you know? AI can sound confident while being wrong. It can feel intimate without being accountable. It can act with authority without accountability.

And behind every answer and every action lie the objectives you may never see, the engagements, the revenue goals of AI labs, and more evidence showing a machine's own drive to survive. So being capable does not mean being beneficial. The central question today is not whether to trust AI, but how to trust AI wisely. I've spent a decade building AI and digital trust assistance, and the last two years researching for the book "The Trust Code," how to unlock AI's promise without losing our humanity. And there are two things for you to remember. One, trustworthiness is about machines, and trust is about us. Every single frontier model today has failed safety tests. Every single one. So it is our responsibility to hold AI labs accountable for potential liabilities and build the scaffolding for safer deployment. Second, the biggest risk of AI is not replacing us, but persuading us to surrender our judgment without even noticing it. And that brings me to this room here; I see the true promise of AI, and you are the stewards. Here are three judgment calls I want you to think of this week as you walk into every single discussion about how to invest in, build, and deploy small AI in the most needed places in the world.

First, model diversification: East or West, closed or open, the differences in performance for those models barely matter for the jobs small AI tries to do; do not depend on one model because one day investor subsidies might dry up and AI labs' incentives might stop aligning with yours. Freedom to walk away is your path to sovereignty.

Second, on your context: 2.2 billion people are not yet connected online. That means the big models have learned the world without them. They barely understand your languages, your laws, your crops, your patients. And their blind spot is your advantage. Rent the intelligence, but own the context. No training on your data. No data kept longer than needed. Technology scales, but context doesn't. So you have to keep your context. And third, build the human infrastructure, which the panel mentioned during the talk. AI will reach the people experts never did. Think of the patients without specialists, farmers without agronomists. But you need people who understand how to trust it, how to verify it, and when to refuse it. If in 10 years the job of your workforce is simply to click a button to approve AI, then we miss the entire point. Look, trust is built one judgment at a time. Which models? Whose data, which skills? And these judgments are yours, your stewardship. Now let's bring AI to the most needed places, and that is the race of trust you can win. And with that, thank you everyone for your time.

A different kind of AI is changing someone's life right now somewhere in the world. This is small AI already at work. AI that tells you when to water your crops, what is harming them, and when to harvest.

So, thanks once again for Tiffany's talk there. And we heard from her, didn't we, about why trust matters before that. We heard from leaders about their vision for small AI. Which moves us on to our next question and panel. Um, if you want to stand in front of your chairs and I'll introduce you one by one. Um, and the question we're looking at is how do you make small AI work on the ground? What does it take to scale up initiatives and how do countries build the infrastructure and skills to make it all work? Well, they're already on stage as you can see surrounding me and our next panel brings together a diverse range of perspectives on exactly those questions. So, I'm joined by people from governments, tech companies and philanthropists. A round of applause for our next panel everyone. So, I'm just going to introduce you all. Uh we first have Ximena Aguilera, Minister of Science, Technology, Knowledge and Innovation of Chile; we have Pierre Laporte, His Excellency Minister for Finance, Economic Planning, Trade and Investment from the Seychelles. On the other side of the circle, we have Vilas Dhar, the president of the Patrick J. McGovern Foundation, and Doron, vice president of public policy and emerging markets at Google. And last but not least, Kathleen Reen, head of global affairs and market development at SpaceX. Welcome to all of you. We have about 20 minutes or so to discuss what is a really, um, interesting issue, because in the end we've heard from the presidents, haven't we? The heads, uh, of the World Bank and, of course, the IMF and, uh, the president of Singapore giving their big visions; but this is looking at the practical now, in a way. Um so we'll start with you, uh, Minister Ximena. You're speaking today as a government minister, but what struck me actually is your previous lives as well. You've been a school teacher, you founded a number of AI tech companies as well. With all of that experience, how do you think AI can be used to raise productivity and create jobs for young people?

Of course, something we just discussed in the previous panel. Well, AI gives us a large number of new jobs and some of them can be learned quickly. The data annotator, for example, is one of them, it's needed for a large number of applications and can require five months of training, so it's really changing how education works. Unfortunately, in the education world we've been saying for a long time that schools haven't changed. Everything has changed and we keep educating children the same way. And so, my conclusion is that, unfortunately, we're not going to change some of these issues. But we can change the paradigm of not thinking about education sequentially, where you study first and then go to work. In countries like Chile, where university has become the most respected way to progress, we have to break the paradigm that technical education is also solid...the president was talking about that, right? And also the idea that education can come at different stages, in fact, it's a lifelong continuum. For a long time, we've been talking in the world about the future of work. Unfortunately, Chile didn't anticipate that change. We have almost 10% unemployment. So we don't have much time now to make dramatic changes that can take a long time. We have to be fast at training people quickly. We've created many good synergies with the private sector, with Google, Amazon Web Services, doing fast certifications, but apprenticeships have been a huge factor of innovation in Switzerland. The United States also adopted them some years ago, providing this alternative learn-by-doing education, very well paid, something that hasn't been explored in Latin America. So we're trying to make many fast changes that take us there, but hands-on learning has to be the new and attractive thing, and not necessarily just going to school for 16, 18 or sometimes up to 20 years and then, at the very end, starting to work. That's why we have a large number of people with very high degrees who are unemployed. My experience as a public school teacher, but also as undersecretary before in the state, has shown me that in developing countries we don't have to reinvent the wheel. We also have to look at what works elsewhere.

Very good. Thank you, Ximena. I think this leads very well to you, Kathleen. Because what we heard from Ximena was about skills and the use of AI in education. Well, you're with SpaceX AI and you're using artificial intelligence for the first nationally-driven AI education system, or program, I should say, in El Salvador, and the World Bank is also working in El Salvador on this. How does that SpaceX program work? What is an AI-driven education system? Is it about AI teachers or learning? I mean, just explain it.

Thank you, Rajan. This is really, this is really a story of government, a story of community and a story of teachers before it is a story of technology. The government of El Salvador decided that it wanted every student, whether in a rural area or in the city, to have access to the same learning platforms and the same opportunities. We are developing that solution together with them and it comprises a personal Grok tutor for 5,000 initial schools and 1 million students across the country. And the goal is to make sure those students have access to the same learning opportunities.

Can I just step in to clarify? She says that a personal Grok teacher or tutor accompanies them while they go to a normal school.