Sai-Ming Wong

Modern states increasingly use AI systems in public decision-making, including benefits eligibility, risk classification, immigration adjudication, and judicial administration. Defenders often argue that democratic authorization, validation, and review are sufficient to render such systems legitimate; critics often reply that human beings must decide. I argue that both responses miss a deeper question: what distinguishes a lawful procedure from a publicly imputable judgment? Drawing on Kant’s Rechtslehre, I argue that when the state makes a materially burdensome, legally operative, reasonably contestable, and practically final determination of a person’s legal status, the legally operative ground of that determination must be attributable to an answerable office with the authority, capacity, and institutional independence to adopt or reject it as its own. This yields a right to an answer: a right to be governed through office-owned judgment rather than unowned architecture. The argument does not condemn AI in public administration. It identifies the moral threshold at which algorithmic assistance becomes impermissible substitution.