Governing AI in the economy and the state
How AI is changing jobs and firms, how governments with limited capacity can regulate it and use it in public services, and how they can respond to concentrated market power.
Draws mainly on areas ACD
Consensus, Evidence, Capacity and a Shared Research Agenda
GDN's Inclusive Digital Transformation Hub launches its AI strategy, "AI and Development in the Global South", at GDC 2026 in Tokyo. It opens a twelve-month path from practical projects to a shared AI research agenda, leading to GDC 2027 in the last quarter of 2027.
Working theme. GDC 2027 is planned for the last quarter of 2027. Venue and dates will be confirmed.
The conference could be organised around four themes. Each draws on several of the five focal areas of GDN's AI strategy, so that communities working on education, languages, research systems and public policy can each find their priorities within one programme.
How AI is changing jobs and firms, how governments with limited capacity can regulate it and use it in public services, and how they can respond to concentrated market power.
Whether and how AI can help narrow learning gaps along the whole education chain, rather than one cycle at a time, and how education systems can use AI to evaluate their own progress continuously.
How AI can work in the languages people speak and learn in, what it would take to build and govern language data as a shared resource, and what this means for schools, public services and research, and how data ecosystems need to evolve to build locally relevant AI.
Whether AI widens opportunities for Southern researchers and universities, or deepens their dependence on tools, data and publication systems controlled elsewhere.
These themes are a working proposal and will be refined with researchers, policymakers and partners over the coming year. Each is open to partners who wish to support or co-design it, whether through a dedicated session at the conference or through research on that theme in the year leading up to it.
AI is already changing how economies, governments and workplaces operate. For low- and middle-income countries the stakes are uneven. It could speed up development, but it could also widen inequality, hollow out the jobs that have carried countries up the income ladder and concentrate economic power further.
The evidence that exists is fragmented, drawn mostly from a handful of rich countries, and focused on individual applications more than on wider economic and social change. A policymaker in Dhaka, Cotonou or Lima has to reason by analogy from contexts that look little like their own, and donors have little basis for judging what their AI investments can achieve.
The constraints also differ. AI adoption in the Global South trails the North, computing capacity and access to cutting-edge models remain concentrated there, and linguistic diversity affects how well models built mostly for Northern languages work. The ambition goes further than closing an access gap. Countries want AI to advance inclusive development and better public services, an agenda they are already leading through the AI Impact Summits.
Governing AI with the world majority means bringing the countries where most of the world's people live into decisions on AI, alongside the governments and companies that develop it, not in place of them.
Several major concerns recur in this agenda across many countries of the Global South. One is whether AI can help close learning gaps, within countries and between them, at a time when many education systems are being reformed. Another is whether AI can work in the languages people speak and learn in, which bears on schooling, public services and research alike. A third is the quality of local data, on which reliable AI depends in every sector, from health to agriculture and education. GDC 2027 places these within the wider question of how AI is governed, and of who takes part in governing it
How AI is changing economies and societies, who gains and who loses, when it helps development, and how countries can steer it towards more inclusive and sustainable growth.
What the strategy sets out to understandHow AI will affect productivity, employment, firms and growth across very different economies
Who gains, who loses, and how AI might reshape existing inequalities
How infrastructure, markets, data and public institutions shape what AI does
How countries can generate their own evidence and use it in decisions and investment
The AI strategy extends a research and partnership network that already spans research, policy engagement and programme delivery across the Global South.
How investment in digital public infrastructure affects livelihoods, productivity, financial inclusion and access to public services.
A cross-country framework for measuring how inclusive digital ID and digital payment systems are, with national research teams and governments.
The financing, governance, incentives and market dynamics behind digital public infrastructure, and a Global South-led research agenda on them.
See the programme › Open Transaction NetworksEconomic research capacity to assess open transaction networks in sectors such as health, agriculture, mobility, commerce and energy.
See the programme ›Five areas where GDN's way of working could add something distinctive. The agenda is not fixed. Final questions will be set with researchers, policymakers and partners, and the list is expected to change as the technology does.
For example, following firms and workers in business-process outsourcing in several countries at once
For example, following an AI triage tool in clinics or adaptive tutoring in schools beyond the pilot
For example, extending inclusivity measurement from digital ID and payments to AI-mediated public services
Builds on the work on DPI and Open Transaction Networks. For example, how regulators with limited capacity oversee AI in financial services, gig platforms or hiring
For example, testing with GDN grantees how AI tools change the speed and quality of social science research
Studies comparable across countries where feasible, drawing on household and firm surveys, administrative and operational data, experiments and quasi-experimental methods. Policymakers and practitioners help identify the questions and interpret the findings.
The progression is deliberate. The Co-Creation Labs in Tokyo are one of the routes through which research priorities and partnerships for 2027 are being identified.
The Co-Creation Labs bring together project hosts, technical and research specialists, practitioners and potential funders around one concrete obstacle per project, several of them related to AI. Each table drafts a two-page scale-up concept, and together they form the Tokyo Portfolio, the starting point for follow-up, partnerships and new research questions.
GDN follows selected concepts with hosts and prospective partners and documents what happens, including what does not work. In parallel it sets up the Theory Unit's AI work and launches the first country studies, drawing on questions that recur across projects and countries.
Where available, initial findings from the first studies are presented, together with progress on projects from the Tokyo Portfolio. Researchers, project teams, policymakers, technical partners and funders then discuss which questions, methods and projects should come next.
The questions that matter emerge in local contexts and during implementation. Following projects beyond the pilot stage shows what works, for whom and under what conditions, including why promising initiatives struggle to scale. The Theory Unit turns recurring questions and lessons into shared frameworks and methods, building evidence that reaches beyond individual projects.
GDN works with researchers in more than 100 low- and middle-income countries. It can ask the same question in many places and see how the answer changes, linking local institutions, economists, other disciplines and policymakers around shared questions and methods while each country team tells its own story. As a neutral convenor, GDN connects research institutions across Africa, Asia and Latin America with policymakers, practitioners and funders.
Partners do not need to fund the same things. Because the research base is shared, different interests can sit side by side within one coherent programme.
So it can set its research agenda and mobilise teams around the priorities it identifies.
Research on a specific question or sector within the wider agenda.
The development of methods and reusable research tools.
To data, expertise and implementation partners.
Support exchanges between researchers and policymakers, so that evidence gets used.
Funders, research institutions, companies and agencies interested in shaping this agenda are welcome to get in touch. The GDN team will get back to you to discuss, and nothing here is a commitment on either side.
GDC 2027 is planned for the last quarter of 2027. Further details, including the confirmed venue and dates, will follow.
Read the concept note (PDF) Read the AI strategy (PDF)