As shown in the latest UN Human Development Report, people in developing countries are not only well aware of these opportunities, but are more optimistic about this technology than those in developed countries. There, discussions about AI usually turn immediately to fears of over-automation, job losses, and labor market disruptions. People in developing countries also fear that AI will unleash a wave of automation, but they are much more optimistic about human empowerment and human-machine collaboration.
It’s tempting to emphasize the risk of “digital AI inequality,” where high-income countries benefit disproportionately from the technology and low- and middle-income countries fall further behind. But these concerns relate to only one aspect of the AI revolution: the development of a growing set of powerful tools that will be used, for example, to facilitate scientific discovery, improve productivity, create new products and services, and automate (through AI agents) complex tasks that require planning, sequencing, and integrating actions. Few countries can meet the relevant requirements of scale, investment and infrastructure, so most of this work is now being done in the US and China.
But creating AI models is not “the only game in town.” The AI revolution also involves finding, adapting, fine-tuning, and implementing already existing tools to solve specific problems and accelerate learning. The costs of these activities are much lower, and will continue to fall thanks to the proliferation of open source models (many of which are being developed in China). As a result, the field is wide open to innovation in a variety of countries.
However, while this side of the AI revolution is much more accessible, it requires some basic infrastructure, most notably reliable power and mobile internet. Affordability is key: you need fast internet and inexpensive data plans, among other things, to ensure that you generate the data you need for the core applications and uses of AI. To make a difference, you need that data to be mobile and accessible, so well-designed regulatory structures are required to ensure mobile data is secure (with permissions or user controls in place). A useful example is India’s Unified Payment Interface (UPI) for secure payments and financial data sharing.
When these conditions are in place (they are already in place in many developing countries), the opportunities for implementing digital and AI-based solutions are almost limitless. These include financial services for previously underserved people and businesses.
People who have few assets, or have a poorly measurable financial and commercial history, or – as in some cases – lack all documentation, access to finance through traditional channels is prohibitively expensive or simply impossible. But today, digital tools offer an inexpensive way to bridge these information gaps. As the cash economy gives way to digital payment systems, and households and small businesses set up bank accounts and e-wallets, data hoarding (if well managed) addresses the old problem of anonymity.
AI-assisted credit assessment could form the basis of sustainable, scalable, and profitable micro-lending options, allowing businesses to grow and hire more people. Meanwhile, e-commerce platforms are opening up small businesses to a larger and (with AI) more targeted market, also promoting growth, dynamism, and potentially innovation.
The prospects are not limited to finance and commerce. In healthcare and education, digital applications (many of them AI-based) are being used to increase access to services, especially for those who live far from areas with a high density of available services.
AI can also help acquire knowledge and skills and accelerate this process (which is the foundation of human capital), for example through digital learning assistants. Sustained growth in the quality of human capital is a key component of all development success stories. Not everyone has access to a school classroom or tutor, but the right infrastructure will allow everyone to communicate with generative AI models that have read and (to some extent) understood the vast digital literature that exists on any topic and in any language. This will directly impact productivity, growth and development.
In addition, in some jobs, AI will reduce training time and increase worker productivity. For example, customer service. AI assistants can mentor inexperienced human employees based on their experience, which will speed up the learning curve and allow employees to start providing better assistance right away. This effect can be achieved in activities ranging from nursing to software development.
While developing countries are not leading the way in creating AI models, they can utilize this technology to achieve their economic and social development goals. Fortunately, they realize this. And that’s exactly what they are about to do.
Michael Spence,
Nobel Prize-winning economist, professor emeritus of economics.
and former dean of the Stanford Graduate School of Business.
(with Mohamed El-Erian, Gordon Brown and Reed Lidow)
Permacrisis: A Plan for Repairing a Fractured World (Simon & Schuster, 2023).
© Project Syndicate, 2025.
www.project-syndicate.org
