This paper argues that contemporary AI systems are not sentient, not by appeal to biology or metaphysics, but by structural analysis of what sentience requires: first-person contact with constraint, normativity that binds from within, and the capacity for self-revision under cost. Current models can generate coherent outputs and simulate perspective, but their coherence is not constraint-coupled in the first-person sense; it is produced by optimization over patterns rather than lived answerability. The paper distinguishes resonance from evidence and fluency from contact, showing how coherence can scale without becoming experience. The result is a structural demystification: AI can be instrumentally powerful while lacking the binding conditions that constitute sentience as a human phenomenon.
Vladisav Jovanovic (Tue,) studied this question.