Consciousness tests for AIs are unreliable

Revisiting the Chinese‑room problem highlights that behavioural or input/output tests cannot reliably prove consciousness because we lack a mechanistic account of subjective experience even in humans. That uncertainty means legal, ethical and policy decisions that hinge on 'sentience' are likely to be contested and misdirected. — If tests can't settle whether AIs are conscious, policymakers should focus on concrete regulatory targets (capability, harms, provenance, liability) rather than chasing personhood determinations.