Walk into a hospital in Nigeria, where one doctor serves every 9,000 patients–fifteen times the WHO-recommended ratio. Every minute that a doctor spends on administrative work is a minute taken from a patient who may have waited hours to be seen.

A Nigerian startup decided to do something about it. They built a voice-based tool that transcribes and structures clinical notes in real time, offline, on a small computer box installed right in the hospital ward. Doctors get time back. Patients get more of their attention. The startup did not adapt a tool built for a hospital in Boston or Berlin. They built it for the conditions on the ground in the language people speak. The result attracted investors and has a path to sustainable growth.

That story is not the exception. It is the blueprint.

The Gap Is Real and So Is the Opportunity

AI is advancing at a remarkable speed. But the benefits are not reaching everyone equally. By early 2025, nearly a quarter of internet users in high-income countries were already using generative AI. In low-income countries, that figure was less than one percent, and the gap is widening.

This is not simply a connectivity problem. It is an ecosystem problem. Most AI tools are designed for conditions that do not exist across much of the developing world: reliable electricity, fast internet, and large proprietary datasets. But the inverse is also true: the problems that AI could solve most powerfully, in health, agriculture, education, and small business, are concentrated precisely in the emerging markets that the global AI industry is not building for.

That is where the opportunity lies. And it belongs to local entrepreneurs.

Small by Design, Transformative by Impact

What we call ‘small AI’ is not about smaller ambition. It is AI purpose-built for the task at hand, efficient enough to run with limited energy, flexible enough to work with intermittent internet, and grounded in local languages and local realities. It asks a different question: not “how do we bring the world’s most powerful model here?” but “what is the most useful thing AI can do for farmers, teachers, and small business owners in developing economies — given what’s actually available?”

The results are already remarkable. In Ghana, the World Bank Group supported the deployment of a WhatsApp-based math tutor that delivered a full year of learning gains in just eight months, at five dollars per student. In Kenya, a startup embedded their AI solution in the M-Pesa application to turn transaction data for small merchants into business insights, helping them increase earnings without using a new app, or interface. In India, an AI app for cough-based tuberculosis screening has tested over 120,000 patients, increasing correct diagnoses by up to 16 percent.

The technology is not the constraint. These entrepreneurs have already proven that. What they need next is the ecosystem to match their ambition.

The Missing Piece is the Ecosystem

The hardest question in AI for development is not whether the technology works. It is why so few of the startups with working solutions reach scale. What is missing is the ecosystem that allows them to go from a promising idea to a company that can scale, attract capital, and last. 

Building that ecosystem is how we turn this technological moment into an ongoing economic transformation. 

For many countries, that means updating policy and investment approaches, shifting from laying the foundations to building the ecosystems that help AI startups launch and grow. Governments will need to create the conditions for responsible AI growth, including data infrastructure, digital public goods, and regulation that supports responsible innovation. International partners can help de-risk early investment in markets where commercial capital is still hesitant. And local AI startups need stronger links to the mentorship, partnerships, customers, and financing that help them move from pilot to scale.

We are working with partners to develop a global acceleration program to support AI startups across emerging markets from upstream to bankability and commercial scaling. Lessons from this program will also inform countries and multilateral development banks, helping them shape investment and policy to strengthen local AI ecosystems. 

Trust Is the Foundation

As AI tools reach more people, ensuring they are safe and trustworthy is essential to drive adoption and sustained use that unlocks benefits. People need to know their data is protected, that the tools informing decisions about their health or money are secure, and that someone is accountable if things go wrong.  In places where these protections are still being built, getting it wrong does not just slow progress — it excludes people from participating at all.

A Development Opportunity of Historic Scale

The countries that get their approach to AI right will do more than help their citizens benefit from AI. They will empower entrepreneurs to build and scale solutions that create jobs, raise productivity, and solve local challenges. In doing so, they can develop new industries, expand exports, and ensure that the gains from AI translate into broader prosperity.

The World Bank Group is committed to helping countries build the foundations for this future—expanding digital access and affordability, strengthening skills, mobilizing investment, and supporting the innovators who are turning AI into practical solutions for everyday development challenges. The opportunity is not simply to adopt AI, but to adapt it in ways that create growth, jobs, and opportunity for millions of people.  

As global leaders gather at AI for Good in Geneva to chart the course of AI development, the decisions made now—about who builds AI, who governs it, and who benefits from it—will determine whether this technological transformation delivers on its promise for everyone, not just the few.

Source : World Bank