The Making of a Global Problem: Artificial Intelligence Governance and the Politics of Object Constitution

Over the span of just a few years, artificial intelligence (AI) has developed into an issue of remarkable political salience. This includes the domain of international politics, where two seemingly contradictory developments can be discerned: First, as reflected in the emerging narrative of an “AI arms race”, governments as well as multinational corporations are increasingly competing for technological superiority, and have invested significant resources in AI research, development, and the acquisition of talent. At the same time, however, recent years have witnessed the parallel proliferation of global governance initiatives – partly driven by the very actors who compete for technological dominance – that promise to steer the development of AI in a way that its societal benefits are maximized and its manifold risks minimized. In my paper, I critically engage with this emerging debate on AI governance. In particular, I examine how actors with divergent interests and identities – ranging from researchers and civil society to governments and corporations – discursively construct “artificial intelligence” as an object of global governance. Drawing on sociological approaches to International Relations (IR), I theorize that while actors agree on the object of governance – “artificial intelligence” – they constitute this object in different ways. In other words, they diverge on their definitions of AI (what is to be governed), their demand rationales (why it should be governed), their conceptions of governance (how it should be governed), and authority (by whom it should be governed). My argument is that the broad nature of the term “artificial intelligence”, which subsumes a wide range of technologies, opens up the space for competing constructions of AI as an object of (global) governance – a space that different actors utilize for the strategic social construction of AI, by which they seek to establish, or reinforce, a particular governance logic that serves their interest. While the global governance of AI is still nascent, I argue that the underlying discursive structures of the governance debate matter, as they have a constitutive effect on how future governance options can be imagined and implemented. In the paper to be presented at the conference, I plan to outline the theoretical framework guiding my research and present first evidence from a number of case studies.