A cognitive bias wherein individuals, after identifying clear errors or limitations in AI-generated content within their own area of expertise, nevertheless continue to place undue confidence in the same AI’s output on topics outside their expertise. This phenomenon reflects a selective skepticism: the observed unreliability of AI is compartmentalized rather than generalized, leading to inconsistent trust in the technology.
Jared, a software engineer, noticed that the AI couldn’t correctly format a for-loop in Python. Ten minutes later, he trusted it to draft a legally binding employment contract. Classic Smith-Sauer AI Amnesia Effect.
by Geoff Carr April 24, 2025