(Infra)structural Problems of the Technoscientific State? Rereading Claus Offe’s Critical State Theory in the Age of AI Capitalism

Artificial Intelligence (AI) is not only a versatile general purpose technology; since the term itself was coined it recurrently serves as a potent “technological promise” (Hirsch-Kreinsen 2023) and operative “techno-economic paradigm” (Berman 1992) that currently forces strategic choices about building AI capacities (data, computing power, AI talents) from state actors as well as private companies in response to big tech’s infrastructural concentration of power (Jacobides et al. 2021, Verdegem 2022). I want to argue that in order to refine and connect current debates circling around a paradigm shift towards addressing the underlying geoeconomic importance of technological infrastructures (Abels and Bieling 2023) and a potential rebirth of industrial policy (Aiginger and Rodrik 2020, Mazzucato 2021) with critical discussions on AI capitalism (Verdegem 2022) and technoscientific capitalism (Birch 2020) in times of global polycrisis, a relational approach to state theory as provided by German late capitalism scholars during the 1970s might be a promising point of (re)departure. In the aftermath of 2008’s financial crisis those theories, alongside other heterodox economic approaches treating crises as rather inherent than external to a global capitalist dynamic, have seemingly regained their appeal and once again fueled academic interest (Klenk and Nullmeier 2010, Borchert and Lessenich 2016, Hausknost 2020, Rapić 2023). A contested, but still strong point of neo-marxian late capitalism theories in contrast to most institutionalisms is their unique consideration of state and market economy as intertwined yet separate, potentially conflicting economic and democratic imperatives, which together with the interest of the state in itself comprise its logic of action (Jessop 2001, Lessenich, 2020). In this regard, Claus Offe (1976) delivered a profound critique of state interventionism, which aims at determining the limits of the policy-making capacity of the state, paving the way for a critical institutionalism (Buchstein 2019). Despite obvious shortcomings to his approach (Geis and Strecker 2005, Strecker 2013) its ambiguity offers at least some considerations for assessing structural limitations of state interventionism and policy formation in the field of AI policy. Considering the current German AI interventionism from this perspective, I thereby offer an analytical framework that adheres to the presumption of contradictory state imperatives, the state’s need for both the creation of legitimacy and the stabilization of accumulation processes in order to secure its very own existence, yet also gauge relevant dimensions and alterations of state interventionism in regards to their technoscientific and geoeconomic requirements in the policy field of AI. The research draws on semi-structured expert interviews with policy makers, public servants, AI researchers, venture capitalists, representatives of business associations, unions and NGOs and other experts (n~43) as well as document analyses. The empirical analysis focuses on three dimensions of state actions: the specific legitimating subsystem, the political-administrative state actions and strategic fiscal resources within the scope of AI technology policy. The results suggest that “AI made in Germany” in the context of EU infrastructural policy, while addressing relevant dimensions of AI capacities, poses a threefold stabilization project of technization, disregarding the infrastructural dimensions of technoscientific AI innovation and power concentration.