Capability Is Not Legitimacy: Human Purpose and Research-System Justification in AI-Assisted Inquiry

This paper introduces a purpose-sensitive legitimacy layer for AI-assisted research systems. The Reflexive Laboratory series has developed methods for candidate generation, artifact governance, canonicality, structural integrity, reviewability, and post-generation integration. Those layers remain necessary, but they do not by themselves answer a later question: why should a capable research system be trusted, used, relied upon, or allowed to shape research agendas? Using Magnifica Humanitas as an external normative-governance lens, the paper argues that capability is not legitimacy. A system may generate useful artifacts, preserve traces, pass integrity checks, and route outputs into review while still lacking an explicit account of its human purpose, public-interest relation, responsibility path, affected parties, contestability, and limits of use. The contribution is not a general AI ethics theory and not a theological argument. It is a Reflexive Laboratory methods proposal: AI-assisted research systems need a legitimacy layer that makes their ends, responsibilities, and conditions of justified reliance as explicit as their artifacts, traces, and governance states.