The Algorithmic Leviathan: Hybrid Intelligence and the Future of Public Governance

This chapter develops theoretical foundations for understanding hybrid intelligence in public administration through Herbert Simon’s bounded rationality theory, James March’s exploration-exploitation framework, and Michael Zürn’s global governance theory. The analysis demonstrates that humans and algorithmic decision-makers exhibit distinct but complementary forms of bounded rationality: humans face cognitive constraints, while AI systems face computational limitations stemming from training data constraints and an inability to process context. March’s organizational learning framework explains why overreliance on algorithmic exploitation without human-guided exploration creates organizational brittleness and competency traps. The chapter examines how recent judicial shifts—including Loper Bright’s elimination of Chevron deference and the revival of the nondelegation doctrine—both constrain and create opportunities for hybrid intelligence frameworks that satisfy constitutional interpretability and accountability requirements. Through an analysis of State v. Loomis and international implementations in Singapore, France, Japan, and South Korea, the chapter develops six core design principles: human authority over values and goals, complementarity over substitution, graduated automation based on risk, transparency by design, continuous learning, and participatory governance.