Creating Digital Pathways for All: Understanding AI's role in global advancement

DIALOGUES is a platform for diverse perspectives and candid conversations on AI, technology, and society — and our collective responsibility to get it right.

At the World Bank’s 2025 Global Digital Summit, James Manyika, Senior Vice President of Research, Labs, Technology & Society here at Google & Alphabet sat down with Ajay Banga, President of the World Bank for a wide-ranging discussion on the opportunities, challenges, and pace of innovation when it comes to AI technology in both developing and developed markets. This conversation has been condensed to highlight key discussion points and edited for clarity. Listen to the full conversation, moderated by Niala Boodhoo, here.

NIALA (MODERATOR): It's clear that artificial intelligence has the potential to revolutionize our world. But the big question is: how do we not repeat the mistakes that led to the current digital divide when it comes to AI? As the World Bank itself has said in its inaugural Digital Trends and Progress Report, this calls for a new playbook and a new level of cooperation between stakeholders like policymakers and the private sector.

Ajay, last year the World Bank announced the launch of a new digital vertical. Can you give us an update on that and how it fits into the bank's overall mission and vision?

AJAY: The idea of creating a separate digital vertical was not to say that “digital is not an important part of human capital, or prosperity, or development of infrastructure” – it is. But if it gets lost in each of those, it doesn't get the attention that it deserves, given the outsized implications it has for each and every aspect of development. More germane to today's discussion is how do we help get AI to become embedded in countries in the emerging markets? How do we help them meet the challenges of using it? Of not getting left behind? Of creating the right governance and the right rules and the right systems around it?

I see enormous pluses with [AI] in health and agriculture and other areas. The issue is, do we have the right skills and infrastructure in these countries to enable that to happen? On the other side, there is the issue of governance and transparency and data protection and regulations. A friend of mine once said to me that [AI] is so important that not regulating it would be a big error, but not regulating it in the best possible way would be an even bigger error. The difference between this and prior innovations in systems is the speed of which it’s happening.

NIALA (MODERATOR): James, do you want to comment on the speed of innovation, and what you’ve seen in terms of advancement of AI?

JAMES: I did my PHD in AI and Robotics almost 30 years ago and I never could have imagined the progress that we've seen in the last decade. First, the models are getting better, they’re more capable, they’re more natively multimodal (which means they can work with content of any kind, not just text). They’re also becoming what's called agentic, which means it's more than just inputting things and getting outputs out; they can actually take actions on your behalf and be connected with other tools.

The second thing I’d highlight about the pace of progress is that because AI is a general purpose technology, it's starting to impact other adjacent technology. Take for example, robotics. Robotics is going to have massive acceleration. I live in San Francisco, and at Google we introduced driverless cars with Waymo. If you live in San Francisco or LA or Phoenix, you’re probably already riding in these cars. In fact, a couple weeks ago, we reached a milestone where in those three cities, we’re doing 200,000 paid rides in those cities per week

One of the things that's underappreciated is how in the last couple years especially, these models have become a lot more efficient, both in terms of the compute and energy required. We're starting to see huge gains in efficiency.

NIALA (MODERATOR): What are the different challenges and opportunities you’re seeing in developed and developing economies when it comes to AI adoption?

JAMES: The opportunities range from economic, scientific, addressing societal challenges, healthcare, education - AI is transforming sectors that matter. AI is even helping governments deliver public services to their citizens. All of these areas can be transformed by this technology. However, none of these gains are automatic or even guaranteed. That's as true for developed countries as it is for developing countries. The particular challenges that I see in developing countries include, as Ajay pointed out, access to what I think could be considered the ingredients for AI. That’s access to compute, access to data, and even access to the models themselves - that’s fundamentally important.

Then you’ve got some of the basic infrastructure challenges. That includes digital infrastructure, but even basic electricity and connectivity can be challenging.

Third, there’s also talent and skills challenges that are very important. This includes ensuring that the people with the training and expertise are able to capitalize on the technology and also skilling the workers that are going to work alongside the technology.

A fourth challenge that often gets left out, is the creation of vibrant ecosystems. What’s going to be very important in developing countries is that there are entrepreneurs, innovators, researchers who are developing applications for their communities and their countries in order to transform and create both business and societal opportunities. This is a challenge for developed countries too, but especially developing countries.

I think it's going to be quite important to have what I think of as enabling policy – policy that is focused not just on addressing the complexities and risks, but also enabling the innovation that's needed to fully capitalize on this technology. The opportunities, especially many that are in the public interest, are not going to happen by themselves. I think Ajay’s comment was correct that policy should also enable the kinds of opportunities and innovation that are going to benefit societies.

AJAY: I think there's an important point there which is if you get ahead of yourself with the policies you could stifle the growth of this innovation. If you're too far behind, you will end up

with enormous challenges in reigning in what could have already left the stable. But if policy follows innovation quickly and nimbly, then I think we're in the right place.

JAMES: I would also add that because these challenges are so enormous, we've been making investments to try to partner with countries both in the developed world [here in the US] but also in developing countries. For example, we're spending an enormous amount of time investing and trying to improve access to languages. This technology enables that advancement. For example, Google Translate can now translate about 250 languages – we added 110 languages just in the last year.

When it comes to infrastructure, Google has been building data center infrastructure in a variety of countries. Whether it's in Latin America or in Africa or in Asia. We've also been investing in cable and broadband connectivity in different countries. When it comes to skills, we've now trained over 100 million people in digital skills and last year, we announced an additional new commitment to spend another $120 million in skilling.

NIALA (MODERATOR): Let's talk about those skills. I was struck by this one data point: 60% of employment in the United States in 2018 was attributed to job titles that did not exist in 1940. Ajay, I know in your upcoming [World Bank] Spring Meetings, jobs are going to be a huge priority and point of discussion. How does that fit into what we're talking about here?

AJAY: In the emerging markets, 1.2 billion young people are going to become eligible for a job in the coming 12 to 15 years. If young people do get a job, whether it's as an entrepreneur or working for a small business or working on a small farm or for a medium-sized business – then they are productive, they have hope, they have optimism. But if they don't get a job, then the downsides are quite severe. I think if we are not careful we run a very severe risk in the coming 15 to 20 years of allowing this demographic bulge to go from a dividend to a liability.

We think there are five areas where jobs get created. In a place like Africa, for example, the first one is in infrastructure. We talked a little bit about electricity and digital, but I'm talking about bridges, roads, airports, agricultural warehouses, and railways.

The second is in agriculture. Making small holder farmers viable and a business in themselves so that every young child of a small holder farmer does not feel that their future is better in an urban area trying to find a job, which today is the model in a lot of countries. We've got to change that to enable the small holder farmers to feel that this is somewhere they can be. That's why we made a commitment to spend $9 billion a year on helping small holder farmers through cooperatives and technology and AI and better fertilizer, and better farming techniques.

Third is in the delivery of primary health care. Not just because people will become healthier – of course that's important. But because you will employ nurses and medical diagnostic technicians and PPE manufacturers, not just doctors. And when you do that you create the ladder of jobs that every country and every society needs.

The fourth category where things will be very important, will be in manufacturing for local consumption. Africa has more minerals and metals for the energy transition which we will all need, but if all we do is extract the mineral and bring it to our country and process

it here, we don't create enough jobs there - that's a different model of job creation.

And the fifth one is tourism, and tourism enabled by skill institutes as part of the infrastructure. So whether you do this through companies or entrepreneurs, you're going to need these five. If you look at these five they will all be impacted by AI.

NIALA (MODERATOR): There is experimental evidence that generative AI can actually boost labor productivity and expand job opportunities, especially for lower skilled or less experienced workers. Can you talk about how you're seeing that?

JAMES: There's now a lot of recent empirical and experimental evidence that shows that in many occupation types, when workers are using generative AI technologies in an assistive way, it actually lowers the barriers of expertise and experience. Less skilled workers are actually able to perform at a higher level of productivity, which reduces barriers not just to jobs and opportunities for those workers, but also increases their output in ways that contribute to productivity.

For example, it's quite striking that many developing areas lack sufficient primary care. These technologies can expand the number of primary care providers. For example, we're using technology in East Africa to assist less-skilled individuals in providing maternal health primary care. Additionally, 30 to 40% of people with TB in the developing world go undiagnosed. Our year-long study in Zambia and India showed AI tools can diagnose TB as accurately as expert doctors. This means that even less-skilled individuals with this technology can provide primary care, which is extraordinary. This assistive technology not only helps providers deliver primary health care but also creates jobs for less-skilled people, especially where primary care physicians are limited.

I think whether we're discussing developed or developing economies, research indicates we'll see three main effects on jobs. Some occupations will decline overtime because the constituent activities and tasks are going to decline. Simultaneously, some jobs will experience growth due to increased demand. However, the most significant effect will be job transformation and assistance, as we've discussed. I think that’s important to keep in mind. However, the research emphasizes that to achieve such a substantial assistive effect, extensive skilling and capability building are essential.

NIALA (MODERATOR): I wanted to end by asking both of you what you're most excited about when you're thinking about the potential for AI.

JAMES: I'm excited about the societal impacts, both through the advances in science, which are going to benefit everybody, and also through the ability to tackle the most pressing global challenges. Whether it's food shortages, food security, or the effects of climate change, I think this technology is going to help us address these things in a huge way. I'm also excited about how this will enable people to have access to information and education – being able to learn in your own language, being able to get customized training, and being able to get personalized, contextual learning. That's now possible, and I think that's going to be extraordinary. I'm still amazed by the number of people who come to tools like Google Search or YouTube to learn. Think about how much better that will be when it's personalized and assisted through AI. I grew up in southern Africa, and information and education were things that I didn't have access to.

I should say that as excited I am about these possibilities, I don't think these gains are automatic. It will require significant investment and focus from everyone, including countries and policymakers, to realize AI's potential. That's why I think it's important to highlight the critical challenges that exist. We all have to work hard to make this happen.

AJAY: I think of AI through the lens of young people and this demographic bulge coming through the emerging markets. To me, young people are our future. That's my optimism. That's what I come to work for and that's what I care about.

NIALA (MODERATOR): Thank you James and Ajay for being with us.