The World Bank has urged governments and development partners in low- and middle-income countries to invest in what it calls “small AI” — machine-learning models designed to run on mobile phones and operate with minimal connectivity. Paschal Donohoe, the bank’s chief knowledge officer, said these compact systems could help emerging economies capture the benefits of artificial intelligence without relying on large-scale, resource-intensive infrastructure.
Small AI, as the bank describes it, is distinct from the complex, cloud-dependent systems pursued by many Silicon Valley firms. These lighter-weight models are intended to function on devices commonly used across developing regions and to be adapted to local languages, contexts and constraints, Donohoe said. He argued that, properly deployed, the technology could significantly accelerate economic and social progress.
The World Bank frames the approach as a way to broaden access to AI-powered tools where internet connectivity, data infrastructure and computing resources are limited. By designing models that can run offline or with intermittent access, nations could use AI to deliver services, information and analysis closer to communities, the bank said, potentially shortening long-term development timelines.
At the same time, the institution cautioned that realizing these benefits will require deliberate policy choices and investment. The bank highlighted the need for local adaptation of models, capacity building, and governance arrangements that address data, privacy and equity concerns — areas where a one-size-fits-all commercial approach may fall short.
Donohoe contrasted the small-AI approach with the drive by major technology firms to develop increasingly large and sophisticated consumer products, saying that emerging economies stand to gain far more if AI is oriented toward local problems and constraints. The World Bank’s appeal seeks to prompt governments, development agencies and private actors to begin creating and deploying these compact models as part of national development strategies.
The bank did not offer a single blueprint, but its message underscores a growing conversation about how to make AI inclusive and relevant beyond high-connectivity, high-income settings. For development practitioners, the small-AI concept raises questions about where to direct funding, how to develop local talent, and how to regulate new tools so they support equitable growth.