During a recent trip to Kazakhstan, I was struck by people’s enthusiasm for AI. Virtually everyone I encountered, including academics, policymakers, and entrepreneurs, seemed convinced that the technology would help solve thorny challenges, from diversifying the economy away from dependence on natural resources to expanding access to critical services, particularly for remote populations.

I had expected the diffusion of knowledge about AI to be slower, but perhaps their positivity should not have surprised me. After all, the rapid development of AI implies important opportunities for emerging economies.

As the latest United Nations Human Development Report showed, emerging-economy populations are not only well aware of these opportunities but are also more optimistic about the technology than their developed-economy counterparts. In advanced economies, conversations about AI tend to turn immediately to fears of excessive automation, job losses, and labour-market disruption. Emerging-economy populations also worry that AI will bring a wave of automation, but they anticipate even more augmentation and human-machine collaboration.

One might be tempted to highlight the risk of an “AI digital divide,” with high-income countries benefiting disproportionately from the technology, and low- and middle-income countries falling further behind. But such concerns focus on one dimension of the AI revolution. The development of an expanding set of powerful tools to be deployed, say, to advance scientific discovery, increase productivity, generate new products and services, or automate (via agents) complex tasks that involve planning, sequencing, and integration of steps.

Since few countries can fulfil the associated requirements about scale, investment, and infrastructure, such activities are currently taking place largely in the United States and China.

But model-building is not the only game in town. The AI revolution also entails the querying, adaptation, fine-tuning, and deployment of existing tools to solve context-specific problems and accelerate learning. The costs of these activities are much lower, and with the expansion of open-source models, many of which are developed in China, they will continue to decline. As a result, the field is wide open for innovation across a broad range of countries.

While this dimension of the AI revolution is far more accessible, it does require some basic infrastructure, especially a reliable electricity supply and mobile internet connectivity. Accessibility is key, and fast internet as well as affordable data plans are needed, not least to ensure that generating the data that fuels many of the most significant AI applications and use cases. To make a difference, these data should be mobile and shared, so well-designed regulatory structures, which allow for secure data mobility (subject to individual permission or control), are also essential.

India’s Unified Payments Interface, which facilitates secure payments and financial-data exchanges, offers a useful model.