Researchers push back on AI-audited scientific literature: LLMs lack judgment

khademinori · x · 2026-11-19

Responding to floating ideas about using AI to audit scientific literature, rbhar90 argues it's a terrible idea: LLMs lack judgment, the subtleties of good science make error-detection prone to witch hunts, and today's models aren't smart enough — yielding only make-work. khademinori adds that LLMs' inability to write proper scientific papers without strong supervision is itself a valuable signal.

Related event: Google Brain Researcher Warns Against Using LLMs to Audit Scientific Literature(2 posts)→

Original post →

More from AGI Musings

AGI Musings channel →