World Bank Urges Developing Countries to Build “Small AI” for Local Needs

Summary

The World Bank is urging developing countries to help create “small AI” models designed for local needs. Unlike increasingly sophisticated consumer systems, these specialised models use narrower datasets and can run on mobile phones with low connectivity, in local languages. The bank’s chief knowledge officer, Paschal Donohoe, said this could potentially compress economic development that once took centuries into decades, provided the technology is adapted to local settings. He cited Indian farmers involved in the World Bank’s AgriConnect agricultural scheme, who may need soil analysis in their own language; a phone-based model could help meet that need where there are not enough scientists and agricultural experts for every small farm. Donohoe suggested that such tools could be developed through cooperation among the World Bank, governments and private companies. IMF managing director Kristalina Georgieva said governments must provide electricity, digital infrastructure, preparation for adoption and safeguards against bad actors if AI access is to become broadly available. The World Bank’s annual development report has highlighted AI’s potential, but also noted that nearly one-third of rural schools in sub-Saharan Africa lack electricity and more than two-thirds lack consistent internet access. Georgieva warned that about 40% of jobs worldwide could eventually be augmented or eliminated by AI, with young people entering routine jobs particularly exposed. The World Bank is also linking AI, jobs and development to its effort to mobilise private investment, which Donohoe said reached $120bn over the past year through guarantees and other financial instruments.