Dignity as Answerability: How World-Model AI Reframes Human Moral Standing

This paper argues that trait-based human exceptionalism--the philosophical project of grounding human dignity in capacities (intelligence, self-control, normative agency) that other beings lack--has collapsed at the level of behavior. World-model--based artificial intelligence now matches or exceeds humans on behavioral analogues of those traits in the contexts where institutions actually measure them. We introduce four interlocking concepts to make sense of this situation. First, behavioral equivalence without normative selfhood names the asymmetry between AI's norm-sensitive performance and the identity-binding selfhood that gives human normativity its existential weight. Second, drawing on predictive-processing and world-model accounts of cognition, we reframe normative selfhood as an empirically observable, biased, identity-constitutive capacity that grounds responsibility precisely because it also produces moral rigidity and historical atrocity. Third, structural hypocrisy names the stable contradiction by which modern institutions, in outcome-evaluated domains, prefer AI execution over human moral struggle; the resulting moral buffering arrangement has AI execute while humans absorb blame. Fourth, under these conditions, dignity cannot be defended as superiority but only as the irreducible burden of being answerable--a position to occupy, not a property to claim.