AI’s potential to facilitate more agile development cooperation
Sinanoglu, Semuhi / Michael RollMitarbeiter sonstige (2026)
in: Anna-Katharina Hornidge / Axel Berger (eds.), Disruption and Reform: sustainable futures at stake? The case for politics of cooperation and perseverance, Bonn: German Institute of Development and Sustainability (IDOS), 63-67
ISBN: 978-3-96021-292-8
DOI: https://doi.org/10.23661/r4.2026
Two features of AI stand out among technological innovations: It is exceptionally powerful, and its (computing) power is increasing faster than any previous technology in human history. AI is not a tool that can simply be added to the existing arsenal just for the sake of task automation and increased efficiency (Sinanoglu, 2025a). Its consequences are so profound that using it to advance broader development objectives requires changes to some of the ways development cooperation works. In this chapter, we go beyond more general assessments (OECD, 2025) of the use of AI by governments (OECD, 2026) and of the global development risks posed by AI (United Nations Independent International Scientific Panel on Artificial Intelligence, 2026). We argue that successful development cooperation for sustainable futures depends on collaboration with others in iterative and adaptive ways to exploit opportunities for positive change more effectively in challenging contexts. To this end, current development-cooperation structures and processes should be reformed to facilitate and encourage organisational learning (Brandi & Büge, 2026) and adopt a more iterative approach to cooperation. To what extent does development cooperation currently deploy AI tools for such organisational learning? By “AI”, we mean generative and agentic artificial intelligence. We also distinguish between “AI in development” and “AI for development”. Whereas the former refers to how AI is used in middle- and low-income countries, the latter refers to how the development sector can use AI. We catalogued how the supply side of development cooperation (donors, ministries, agencies, development banks, United Nations (UN) entities, humanitarian implementers and funders) uses AI in its own operations, programming, advisory work and evaluations. The database comprises 172 AI deployments and, while not exhaustive, provides a comprehensive primary-source snapshot that lets us answer three questions empirically: Who is deploying AI? What for? And how far has it matured? (See Figure 9) Our data suggest that almost 60 per cent of recorded uses fall into just two functions: getting services to people in the field (programme delivery) and speeding up the back office at headquarters (internal operations). Together, the functions most closely tied to organisational learning – monitoring, evaluation, planning and coordination – account for only roughly 19 per cent of deployments. In short, the sector is now using AI to do its existing work faster, but not yet to change how it works and learns.
