Artificial intelligence is usually discussed from the perspective of countries that already have large numbers of doctors, engineers, teachers and other highly trained professionals.
In much of the world, the problem is different.
Expertise itself is scarce.
A rural clinic may have too few medical specialists. A farmer may have limited access to agricultural advice. Government agencies may struggle to process records or deliver services efficiently.
That changes the question surrounding AI.
Instead of asking only “Which jobs could AI replace?”, another question becomes important:
“How many more people could an expert reach with AI helping them?”
The World Bank's World Development Report 2026: The Promise of Artificial Intelligence examines that possibility across developing economies.
Its analysis estimates that 16.2% of jobs in developing economies could experience meaningful productivity gains from AI.
That's surprisingly close to the 18.7% estimated for high-income countries.
Meanwhile, the report estimates that 4.5% of existing jobs in low- and middle-income countries are at risk of automation from generative AI, compared with 14.2% in high-income economies.
One reason is that many jobs in developing economies involve physical or interpersonal work that current AI systems cannot easily automate.
But AI can still augment the people doing those jobs.
The World Bank points to applications ranging from helping doctors diagnose disease and farmers make crop decisions to improving education, tax administration, welfare delivery and disaster response.
Importantly, developing countries do not necessarily need to build enormous frontier AI models themselves to benefit.
Existing models can be adapted to local languages, local information and specific problems.
That could make relatively inexpensive AI systems particularly valuable in places where adding thousands of additional specialists would otherwise take many years.
AI is spreading unusually quickly as well.
The World Bank notes that earlier general-purpose technologies took decades to diffuse widely into lower-income countries. AI services can spread much faster because much of the infrastructure and software already exists.
None of this guarantees that developing economies will capture those benefits.
But it points toward a version of the AI story that receives less attention:
The technology's biggest contribution in some places may not be replacing people. It may be helping the people already there do more.