Rebuttals at ML conferences increasingly AI-written, researchers debate broken review calibration
mmitchell_ai · x · 2026-09-25
Researchers are calling out problems in ML peer review: one proposal would give an extra point to rebuttals that clearly took real effort, since authors can spend days addressing a single reviewer with nothing gained. The quoted discussion adds that a huge share of rebuttals in the position paper track appeared 100% AI-generated, that reviewers of the same paper are poorly calibrated, and that borderline scores cluster because hedging reviewers gravitate to the middle — meaning track-wide calibration would likely reject many papers.
Related event: Researchers Find Many Rebuttals Now Written by AI(3 posts)→
More from Research
- Continual-ARC: lifelong ARC-AGI stream finds agent harness persistence rarely beats simple caching — unifyai · 2026-11-16
- DySENSE: MIT's GenAI framework synthesizes satellite imagery with dynamic city energy maps — MIT · 2026-11-16
- WebArena-Pro: a 300-task multimodal benchmark across 20 self-hosted apps for web agents — McGill-NLP · 2026-11-16
- Training-free page-sparse decode in llama.cpp: 1.72× faster 128k decoding on an RTX 5090 — topppggg · 2026-11-16
- Training Qwen3-8B to drive a simulated hand spawns a dedicated internal 'pressure' direction — repligate · 2026-11-16
- Google Research open-sources PaperOrchestra, a multi-agent framework for writing AI papers — google-research · 2026-11-16