Governing AI decison-making: balancing innovation and accountability

This article explores the growing use of algorithmic models to make or inform decisions within the public sector. Amidst a climate of accelerating investment, expanding system applicability, and rapid technical progress, it concentrates on how key jurisdictions, most prominently the “digital empires” of the United States, European Union, and China, construct the problems associated with such algorithmic systems, and how these constructions impact governance. Drawing on an example from the legal sphere, it highlights both the potential efficiency gains and the increasing tensions concerning automation and fairness. This article then adopts aspects of Carol Bacchi’s Foucauldian-inspired “What’s the Problem Represented to Be?” framework to trace how divergent problem framings, ranging from the United States’ emphasis on an “innovation gap,” to the European Union’s “trust deficit,” and China’s “stability risk,” have produced distinct regulatory trajectories. Yet, despite these divergent framings and national strategies, this article argues that a common post-2024 trend emerges, revealing a general shift toward regulatory softening, one that privileges innovation over precautionary safeguards. This convergence raises critical questions about the future direction and resilience of “algorithmic decision-making” governance.