The World Bank says developing economies can use artificial intelligence to extend scarce expertise, but the gains will depend on local adaptation and access to power, internet and skills.
The Bank released World Development Report 2026: The Promise of Artificial Intelligence on Aug. 4. It examines AI’s potential effects on 5.6 billion people in low- and middle-income countries, including in health care, agriculture and government administration, where trained workers are often scarce. The report says AI can help less experienced workers take on expert tasks and make existing workers more productive.
At the report’s release, 4.5% of existing jobs in low- and middle-income countries were at risk of automation by generative AI, the World Bank said, compared with 14.2% in high-income countries. It estimated that AI could meaningfully boost productivity in 16.2% of jobs in developing economies, versus 18.7% in high-income countries.
The report points to Ethiopia’s NextGen Agro-advisory program, which uses machine learning to provide smallholder farmers with plot-specific recommendations. It cites wheat-yield gains of up to 38% from broader piloting. The figure does not describe results for every farmer.
The Alliance of Bioversity International and CIAT reported that independent on-farm validation found yield increases of 14% to 20%, with additional net benefits of up to $665 per hectare compared with conventional recommendations. Its account also put broader-piloting gains at up to 38%.
Farmers, public-service users and workers who can use AI to increase their output are among those who may benefit, the report says. It also warns that early job disruption is concentrated in knowledge-intensive services and that concentrated market power could let owners of capital capture a larger share of the gains. The Bank notes that earlier technological shifts left parts of the developing world behind.
The report recommends putting existing AI tools to work, tailoring them to local conditions and considering frontier AI development as countries build capacity. Reliable electricity and internet, local data, skills and strong institutions support adoption. Building frontier AI models also requires chips, large data centers, extensive training data and world-class researchers, a path the Bank says is unrealistic for most developing countries in the near term. Without key foundations, AI could widen productivity gaps and inequality.
“They do not need large models or big data centers to reap its benefits,” Indermit Gill, the World Bank Group’s senior vice president and chief economist, said in the Aug. 4 release.
