World Bank urges developing countries to build small AI for local needs

The World Bank is urging developing countries to build locally adapted “small AI,” with an ambitious argument: development that once took centuries might happen within decades. The immediate application is narrower. In comments published by the Guardian on October 18, chief knowledge officer Paschal Donohoe pointed to Indian farmers needing soil information in their own language.

That development claim is a statement of potential, not a measured economic result. Donohoe said the gains depend on adapting AI to local needs. He suggested that products could emerge through collaboration between the World Bank, governments and private companies, placing the emphasis on tools built for particular communities rather than technology alone.

Soil advice in a farmer’s own language

Small AI, as described in the Guardian’s account, means specialized models trained on narrower datasets for a specific purpose. The approach Donohoe advocates combines that limited task with practical access: tools available on mobile phones, usable with low connectivity and delivered in the community’s language. The proposal ties model design to where and how people would use it.

Donohoe’s example came from AgriConnect, the World Bank’s agricultural initiative. Indian farmers involved in the scheme needed access to soil analysis in their own language. He presented phone-based AI as a way to help meet those information needs where the supply of scientists and agricultural experts cannot stretch to every small farm.

“We argue that economic development in the countries that we serve and partner with that could have taken centuries to make happen, the potential is there for that economic development now to happen within decades,”

Paschal Donohoe, World Bank chief knowledge officer, speaking to the Guardian

Existing agricultural tools give the idea a foothold

In an earlier feature about farming in Uttar Pradesh, the World Bank described digital agricultural services connected to AgriConnect.

• UP-AGRIVERSE combines market information, advice and financial services. The bank describes it as the project’s digital backbone, using small-AI models that can work on basic smartphones and low-bandwidth networks. It cites partnerships with organizations including Google and Networks for Humanity.
• A separate partnership involving the World Bank Group, Maharashtra’s government and the EkStep Foundation supports an AI-powered, multilingual, open-source tool for real-time farming advisories. The bank’s earlier feature presents it as another example of the partnerships supporting AgriConnect in India.

Low connectivity is not the only access barrier

The World Bank’s own development work also documents basic infrastructure gaps. Its annual world development report, published in August, said almost a third of rural schools in sub-Saharan Africa lacked electricity. More than two-thirds lacked a consistent internet connection, according to the Guardian’s account of the report.

The proposed small-AI approach is explicitly designed for low connectivity; the bank’s infrastructure findings include the more basic obstacle of institutions without electricity.