World Bank backs phone-based AI tailored to developing countries
Chief knowledge officer Paschal Donohoe says compact models could serve local needs, including soil analysis for Indian farmers.
The World Bank is encouraging developing countries to pursue compact artificial intelligence tools that can operate on mobile phones despite limited connectivity, the Guardian reported. Paschal Donohoe, the bank’s chief knowledge officer, said adapting AI to local needs could help emerging economies grow faster.
Known as “small AI,” the approach involves training specialized models with narrower datasets to perform individual tasks. Donohoe said these tools could function in local languages and with weak connections, potentially compressing economic development from centuries into decades.
He cited farmers participating in the bank’s AgriConnect scheme in India who need soil assessments in their own language. There are too few scientists and agricultural specialists to provide that service to every small farm, he said. Donohoe suggested that collaboration among the bank, governments and private companies could produce such tools.
At World Bank and IMF meetings in Bangkok, IMF chief Kristalina Georgieva said broad AI adoption depended on electricity, digital infrastructure and safeguards against bad actors. The World Bank’s August development report found that almost a third of rural schools across sub-Saharan Africa lacked electricity.
Georgieva also warned that AI would likely alter or eliminate 40% of jobs globally over time. She identified young people entering the workforce as the most exposed because entry-level positions are disappearing.
