Artificial Intelligence and Public Policy: Lessons for Policy Integration

Abstract

Public policies have increasingly evolved into specialized subsystems, each comprising issue-specific rules and regulations tied to domains such as health care or energy. The emergence of novel challenges—such as artificial intelligence (AI)—raises the question of how these issues can be integrated into existing sectoral frameworks. The literature offers contrasting perspectives: while research on policy design and instrument mixes suggests that governments can craft cross-sectoral responses in a coherent manner, studies on sectoral policies argue that top-down integration is unlikely to succeed, leaving policies fragmented and issue-specific. This talk addresses this tension by examining AI as a case in point. It advances five propositions: (1) governments have articulated comprehensive AI agendas through national strategies, which vary across countries; (2) emerging AI regimes differ internationally and are shaped by priorities such as innovation or citizen protection; (3) in multi-level governance contexts, AI policies complement one another despite persisting incoherence; (4) governments face a dilemma between promoting AI adoption and regulating its use, resulting in competing agendas; and (5) AI policy regimes are likely to remain incoherent due to trade-offs between technological optimism and data protection. Drawing on a review of the literature and original empirical research, the talk illustrates key challenges for policy integration and coherent policymaking in the face of disruptive technologies.