Algorithmic Administrative Authority: Reconstructing the Legal Boundaries of Government Power in the Age of Artificial Intelligence
TL;DR
The article argues that the greater the algorithmic influence over a legally consequential decision, the stronger the corresponding public-law safeguards must be, and proposes a five-part framework based on legal attribution, bounded algorithmic discretion, procedural algorithmic transparency, institutional responsibility, and effective human and judicial review.
Abstract
The rapid integration of artificial intelligence (AI) into public administration is transforming the institutional conditions under which governmental power is exercised. Administrative law has traditionally located public authority in legally empowered institutions and identifiable human officials who are expected to deliberate, give reasons, consider relevant circumstances, and remain accountable for the legality of their decisions. Algorithmic decision-making complicates this model. An AI system may classify individuals, predict risks, rank cases, recommend outcomes, or automatically determine administrative results while remaining outside the formal architecture of public authority. This article asks whether existing administrative-law doctrines are capable of controlling this technologically mediated exercise of governmental power. Using normative legal research with statutory, conceptual, doctrinal, and comparative approaches, the article examines administrative discretion, delegation, legality, reason-giving, human oversight, accountability, equality, and judicial review in the context of AI-assisted government. The analysis argues that the principal legal problem is not that algorithms become autonomous holders of public authority, but that they can acquire de facto influence over the substance of governmental decisions. To address this problem, the article develops the concept of algorithmic administrative authority and proposes a five-part framework based on legal attribution, bounded algorithmic discretion, procedural algorithmic transparency, institutional responsibility, and effective human and judicial review. The article argues that the greater the algorithmic influence over a legally consequential decision, the stronger the corresponding public-law safeguards must be. This framework seeks to preserve technological innovation while reaffirming the foundational principle that governmental power remains subject to law.