Delegating Administrative Adjudication to AI

Advancements in artificial intelligence (AI) and algorithmic decision-making are reshaping the administrative state. Self-learning AI systems are increasingly performing functions traditionally undertaken by human bureaucrats, prompting constitutional scrutiny under the non-delegation doctrine, which limits the legislature from transferring excessive policymaking authority to non-legislative actors. This paper investigates whether democratic states can legitimately delegate administrative adjudication to AI, drawing on U.S. and comparative jurisprudence, the "intelligible principle" standard, and the challenges to transparency, accountability, and due process posed by algorithmic governance. Policy recommendations for safeguarding democratic legitimacy in AI-assisted bureaucracies are proposed.