The World Bank has urged developing countries to prioritise smaller, low-cost artificial intelligence tools that can address local challenges rather than spend heavily trying to compete with countries and companies developing frontier AI systems.
The recommendation is contained in the World Bank Group’s World Development Report 2026: The Promise of Artificial Intelligence, which examines how developing economies can use AI to improve public services, business productivity and access to expertise.
Indermit Gill, senior vice president and chief economist of the World Bank Group, said developing economies do not need large AI models or massive data centres before they can begin using the technology to address local challenges.
“AI has thrown developing economies a lifeline, and they should seize it,” Gill said.
He said countries could adapt smaller and lower-cost AI tools to improve access to services such as healthcare, education, judicial support and agricultural extension.
Rather than attempting to build the most sophisticated AI systems from the outset, the Bank recommended that developing countries follow a three-stage pathway of adopting available tools, adapting them to local conditions and eventually advancing towards frontier AI development.
It said following this sequence would allow countries to derive benefits from technologies that already exist while gradually building the capacity required for more sophisticated applications.
A key part of the strategy, according to the report, is improving access to data that reflects the languages, economies and social realities of individual countries.
The World Bank said greater availability of local data, including data in local languages, would enable AI systems to produce more relevant applications for different populations and economies.
It also called for greater access to computing resources and policies that allow new firms to attract investment, test AI applications and scale successful solutions.
The Bank noted that several AI pilot projects are already underway in developing economies, but said governments need stronger evidence and clearer procurement and evaluation frameworks to determine which applications are delivering measurable results.
Better technical skills, it added, would also be required to assess and implement AI projects effectively.
Beyond investment and technical capacity, the World Bank identified public trust as an important condition for successful AI adoption.
It recommended that governments initially draw on voluntary industry standards to promote responsible use of the technology, supported by international cooperation to reduce regulatory fragmentation.
Where voluntary measures are insufficient, governments should apply existing laws to address harms directly, the Bank said.
It specifically highlighted the risks of bias in AI-assisted government decisions and the erosion of data privacy, warning that either could undermine public confidence in the technology.
The Bank said improvements in public services and learning outcomes could strengthen trust in AI, but cautioned that confidence could be difficult to restore if AI systems produce discriminatory outcomes or compromise personal information.
Gaurav Nayyar, director of the World Development Report 2026, said developing countries have a limited opportunity to establish the conditions needed to benefit from AI.
“The window to get this right is narrow,” Nayyar said, describing AI as an opportunity to tackle problems that have remained unresolved for generations.
