NeurIPS reviewer says AI catches 10x more issues, but can still praise LLM slop
mariyaivasileva · x · 2026-07-27
A NeurIPS associate chair says well-executed AI reviews can beat human reviews by a wide margin on concrete issues:
- far more mathematical mistakes
- missing citations
- unsupported novelty claims
- experimental flaws
- typos, grammar, and inconsistencies
But a separate example shows the downside: an obviously LLM-generated paper received a long, rigorous AI review that still praised its “excellent writing,” while shorter human reviews correctly called out the bad prose.
Related event: NeurIPS Reviewers: AI Catches 10x More Issues(2 posts)→
More from Research
- Minerva paper details: ncRNA arrays template short DNA hairpins — BrianHie · 2026-09-23
- Genome language models uncover new class of reverse-transcriptase mechanisms — BrianHie · 2026-09-23
- Mathematician shares a cheap 4-step heuristic for hyperparameter tuning — dejanseo · 2026-09-23
- Burkov skew AI hype: 'deterministic LLMs' and 'first agents' are old tricks rebranded — burkov · 2026-09-23
- Continuous diffusion beats discrete on random k-SAT, proposed as standard benchmark — ArashVahdat · 2026-09-23
- Grady Booch: Contemporary AI Still Lacks Abductive Reasoning, Just 'Next-Token Prediction' — Grady_Booch · 2026-09-23