World Bank urges locally adapted ‘small AI’ for developing economies

The World Bank is urging developing countries to develop ‘small AI’ tools suited to local needs, according to the Guardian’s 18 October report from the World Bank and International Monetary Fund annual meetings in Bangkok. Chief knowledge officer Paschal Donohoe said tools that work on phones with limited connectivity and in local languages could widen access to expertise and help accelerate economic development.

Donohoe described small AI as specialised models trained on narrower datasets for particular tasks, the Guardian reported. Their design would allow people to use them locally, including where connectivity is poor. He presented the possibility of economic development taking decades rather than centuries as a potential gain, rather than a measured result.

How small AI could help Indian farmers

The World Bank’s event page describes small AI as practical, affordable technology built for specific tasks and local needs. It includes tools designed for basic devices, limited connectivity and local languages. Those characteristics matter where a service dependent on a constant internet connection or a more capable device would be harder to use.

What wider adoption would require

The event page identifies infrastructure, skills, local data, governance and partnerships as factors that can help countries move beyond pilot projects toward locally relevant AI tools. The bank’s case therefore extends beyond making a smaller model: people need ways to build, deploy and use a tool suited to the task and community it serves.

Donohoe suggested that such products could emerge through collaboration among the World Bank, governments and private companies, the Guardian reported. His proposal focuses on making AI useful in settings where language, connectivity and access to specialist advice constrain existing services. The bank has described a route for developing those tools, but Donohoe’s projected development gains remain a possibility.