The World Bank wants small AI built for local needs

Start with a farmer’s problem

Donoughue, a former Irish finance minister, pointed to farmers in the World Bank’s AgriConnect programme. They need soil analysis in their own language, he said, but there are not enough agricultural scientists and specialists to advise every small farm in India. A small AI model on a phone could help fill that gap.

These systems are trained on narrower datasets for specific tasks, rather than built as general-purpose products. Donoughue suggested the World Bank, governments and the private sector could develop them together.

That is a more grounded vision than the race to build ever more capable consumer AI. It also shifts the hard question from what a model can do in a demonstration to whether it can provide useful advice in a farmer’s language, on a phone, with a weak connection.

Infrastructure is the constraint

AI’s effects on jobs and economic growth are central topics at this week’s annual meetings of the World Bank and the International Monetary Fund in Bangkok, Thailand.

At a Monday discussion, IMF managing director Kristalina Georgieva said governments need to provide electricity, digital infrastructure and public readiness to use new technologies. They also need to protect people from bad actors. If those basics are in place, she said, AI could spread widely, much like mobile phones.

The World Bank’s annual World Development Report, published in August, set out potential benefits of AI alongside a stark infrastructure gap: nearly a third of rural schools in sub-Saharan Africa lack electricity, and more than two-thirds lack reliable internet access.

Georgieva also warned that AI could eventually change or eliminate 40% of jobs worldwide, with young people entering the workforce especially exposed. Early-career workers often start with routine, repetitive tasks and move up from there. If those jobs disappear, she said, a coffee shop is unlikely to be the career future they want.

The financing story behind the pitch

Under president Ajay Banga, the World Bank has focused on bringing private financing into emerging economies. Donoughue said the bank mobilized $120 billion in private capital over the past year, about £90 billion, using guarantees and other financial instruments. That came as many countries, including the United States and the United Kingdom, cut aid budgets.

The bank wants to show how private capital can translate into jobs and economic development. I think that is the more consequential claim here: small AI is not just a model-design choice, but part of an effort to connect technology investment to development outcomes.

What remains unclear is who will maintain these tools, verify their advice and pay for them once they reach farmers. A phone-based model may lower the cost of access, but it cannot substitute for the infrastructure and institutions that make its answers useful.