Public Insists on Human Finality in Medical AI

A majority of U.S. adults are open to AI doing administrative and some clinical support tasks but prefer a human to do final checks on diagnoses, treatments and insurance decisions. Comfort varies by task: translation and scheduling score highest for acceptance, while claim decisions and autonomous clinical decisions score lowest. — This shapes likely political and regulatory outcomes: hospitals, payers and lawmakers will face pressure to codify human‑in‑the‑loop safeguards, transparency rules, and liability regimes before broad clinical AI deployment.