Biglab Researchers Read Almost Zero Papers, See Top AI Conferences as Full of Fraud
ahandvanish · x · 2026-08-05
Former Google researcher Keller Jordan shared his blunt views on the current state of AI academic conferences. He claimed that most biglab researchers now read almost zero papers and understand top venues like ICLR, ICML, and NeurIPS to be mainly full of overclaims and fraud, though there are a few exceptions.
This was in response to a researcher complaining about an ICLR 2026 paper that suspiciously achieved unrealistically high results by using test set labels for parameter selection.
Related event: Ex-Big Tech Researcher Slams AI Conferences for Hype(4 posts)→
More from Fun
- Proposed AI Game: Write Like an LLM to Fool AI Detectors — Afinetheorem · 2026-08-05
- 'I run my AI locally' meme goes viral: k3 local run sparks discussion — NERDDISCO · 2026-08-05
- Claude Opus Solves Olympiad Math But Blindly Believes User Is a Shark — Enough-Piano-2362 · 2026-08-05
- User Complains Claude is Lazy: Admits 'I Didn't Read' Emails — kimmonismus · 2026-08-05
- Satirizing AI Alignment: A Viral Copypasta Mocking Super Alignment Jargon — voooooogel · 2026-08-05
- User Complains Claude's Output is Painful to Read, but Notes it Understands Sarcasm — shekitup · 2026-08-05