A New Digital Strategy
In October 2026, the World Bank Group unveiled its first-ever comprehensive digital strategy, a plan aimed at tackling the global digital divide. A core component of this strategy is the promotion of 'small AI'. Paschal Donohoe, the Bank’s chief knowledge
officer, suggests this approach could compress centuries of economic development into decades for emerging economies. The strategy acknowledges a stark reality: in 2025, over 2 billion people remained offline, and internet use in low-income countries was just 23% compared to 94% in high-income nations. Instead of waiting for perfect infrastructure, the plan focuses on deploying practical AI tools that can function now, under existing conditions.
What Exactly Is 'Small AI'?
Small AI refers to lightweight, specialized artificial intelligence models that require minimal computing power and energy. Unlike the giant, all-purpose large language models (LLMs) from big tech, these are designed to perform specific tasks on basic devices like smartphones, even with weak or intermittent internet connectivity. The concept is to create affordable, purpose-built tools that are trained on local data and operate in local languages to solve specific community challenges. This approach, sometimes called edge AI, allows for offline functionality, which is crucial for privacy, cost, and availability in areas without reliable cloud access. The focus shifts from doing everything to solving a defined problem efficiently and accessibly.
Real-World Impact on the Ground
The World Bank points to several pilot programs as proof of concept. In India, the AgriConnect scheme uses small AI to provide farmers with soil analysis and market information in their own languages. One service, UP-AGRIVERSE, combines this information with financial services on basic phones. In Ethiopia, a tool called AI-Coach gives teachers on-demand feedback on their lessons, delivered through an app that works with spotty internet. Other applications being tested globally include helping healthcare workers screen for diseases like tuberculosis without broadband and providing students with AI tutors. The goal is to use AI to support frontline workers and deliver essential services where they are most needed.
From Proposal to Practicality
The World Bank is not just publishing papers; it's creating a framework to make small AI a reality. Its new strategy aligns its public lending, private investment, and risk guarantees to support digital development in a unified way. A key component is the AI Infrastructure and Readiness Initiative, which will help emerging markets in Africa and Latin America access shared, pay-as-you-go computing power (GPUs). This offers a practical alternative to building expensive national data centres. The plan also includes partnerships with manufacturers and mobile operators to lower the cost of smartphones and data plans, addressing another major barrier to access. The Bank has allocated significant funding to support pilot projects and scaling efforts in dozens of countries through 2028.
Hurdles on the Horizon
Despite the promise, significant obstacles remain. The most fundamental challenge is the lack of basic infrastructure. The World Bank's own 2026 development report noted that nearly a third of rural schools in sub-Saharan Africa lack electricity, let alone internet. Digital literacy is another major hurdle, as deploying tools is useless if people don't have the skills to use them. Officials like IMF chief Kristalina Georgieva have emphasized that for AI adoption to be democratic, governments must first provide fundamentals like power, digital infrastructure, and cybersecurity. The 'small AI' proposal is designed for low connectivity, but it cannot function with no connectivity or no power at all, highlighting the need for parallel investments in basic infrastructure.
