Artificial intelligence holds the potential to transform economies and improve living standards, but history shows that the benefits of technological change are not automatically shared equally. The World Bank Group is now shaping its approach around ensuring that AI serves the public good, particularly in developing countries where practical, low-tech applications can make the most immediate difference. This focus comes as the global economy faces a structural slowdown and as 1.2 billion young people in emerging markets are expected to reach working age by 2035, with only about 400 million jobs projected to be created in the same period.
During recent visits to Mozambique and South Africa, World Bank officials observed how digital technology is already improving lives in modest but meaningful ways. A farmer can use a smartphone to photograph signs of disease in his cassava crop and receive an immediate diagnosis and treatment advice, protecting his harvest without waiting for a specialist. Nurses in rural clinics use a digital manual app to navigate clinical procedures step by step, helping them deliver better care to more patients. These examples, while seemingly low-tech, illustrate the kind of practical AI solutions that can address real development challenges.
The World Bank argues that for many developing countries, the greatest opportunity will not come from large-language AI models that require enormous amounts of electricity, water, data, and computing power. Instead, the biggest gains will come from what the institution calls «small AI» solutions designed to work within local realities. These tools use local data, local languages, and everyday devices with existing infrastructure. They can support frontline health workers, assist teachers in classrooms, and provide small businesses with market information and access to credit that helps them grow.
To make these opportunities widely available, the World Bank emphasizes the need to invest in the foundations of digital opportunity. This includes reliable electricity, affordable connectivity, stronger digital skills, modern regulatory frameworks, and the infrastructure needed for innovation to flourish. Without these basics, the benefits of AI risk being concentrated among a few, widening existing divides rather than closing them. The institution describes the solutions it is pursuing as replicable, scalable, and capable of compressing decades of development into days or even minutes.
The discussion about AI is also fundamentally a discussion about people and their trust in technology. Trust will depend on whether AI is reliable, safe, and built on strong data governance. The World Bank’s latest Global Economic Prospects report highlights AI’s potential to become an important source of productivity growth and investment at a time when the global economy is facing a structural slowdown. Yet the real test, according to the institution, is whether AI expands opportunity and supports job creation, especially for the 1.2 billion young people in emerging markets who will reach working age by 2035.
Behind these numbers are young people seeking the chance to work, support families, contribute to their communities, and build better lives. The choices made today about infrastructure, affordability, skills, data governance, and inclusion will determine whether AI widens existing divides or helps close them. With the right policies, AI can reduce inequalities and expand opportunities. When researchers, innovators, policymakers, and practitioners gather at the AI for Good Global Summit in Geneva this week, their goal should be to ensure that the benefits of AI reach everyone: the farmer in Mozambique, the nurse in a rural clinic, and the young people seeking new skills.
Years from now, the success of AI will be measured not by the power of its models or the efficiency it unlocked, but rather by whether these models earned our trust, expanded opportunities, advanced the public good, and improved millions of lives. The World Bank’s message is clear: the technology exists today, and the question is whether it will be widely available to many or only a few.
