I’ve been reading the new The World Bank Group World Development Report, “The Promise of Artificial Intelligence”, and one of the things that resonated most with me is how much of the AI conversation ultimately comes back to something much less new: the capacity of the State.
The report makes a strong case that developing countries do not need to be at the AI frontier to benefit from AI. But adopting AI is not simply about acquiring technology. It requires data, infrastructure, skills and institutions capable of adapting these systems to local contexts.
From my work on data governance, including through the Global Data Barometer, I would add another piece: the capacity to govern AI should develop alongside the capacity to use it.
AI does not fundamentally change the State’s obligations around equality, privacy, due process, transparency and accountability. But it does change the institutional capacities needed to fulfil those obligations when public decisions increasingly depend on data and automated systems. This was also at the heart of a recent class I taught as part of the AI and Human Rights in Public Policy course organized by Derechos Digitales and Universidad de Los Andes.
The WDR points to a familiar problem: government data remain fragmented across institutions, difficult to reuse and affected by weak coordination and uneven quality. This is consistent with what we have seen through the Global Data Barometer: governments may adopt increasingly sophisticated data governance frameworks while still facing significant gaps in implementation, data availability and interoperability.
I have also been exploring this question through research supported by the Datafication and Democracy Fund (Data Privacy Brasil), looking at Brazil, Colombia and Nigeria. One pattern that emerges is that governments are often much better at connecting data for administration and service delivery than for public scrutiny. Interoperability is not simply a technical matter: what gets connected, and for what purpose, is also a governance and political choice.
So perhaps AI readiness should not only ask whether governments have the data, infrastructure and skills to deploy AI. It should also ask whether they have the institutions needed to govern its purposes, data, providers and consequences, and to allow people to understand and challenge the decisions that affect them.
Ultimately, realizing the promise of AI will depend not only on the sophistication of the technology countries can access, but also on the quality and responsibility of the institutions that adopt, govern and use it.
Leave a Reply