Not Every Prompt Deserves an Answer

The article argues that AI has been optimized to answer every prompt quickly and convincingly, but has not been taught when not to answer. As AI embeds into real-world systems, the core risk shifts from hallucination to 'premature legitimacy'—responding too early, too confidently in situations requiring hesitation or escalation. The author contends that human oversight alone can no longer keep pace, and that AI systems must be redesigned so that not every prompt triggers a response.