Analysis of 250k ICLR reviews finds 65% AI-written, scores collapsing to middle
TuhinChakr · x · 2026-11-08
Pulling 250k ICLR reviews since 2025, GuyZys finds AI-written reviews are now the majority: 18% in 2025, 41% in 2026, 65% in 2027. Scores are collapsing into the middle — Weak Rejects jumped from 36% to 57% while clear accepts/rejects fell from 41% to 21%. AI-flagged reviews picked a clear side less than half as often as humans. Awkward slip-ups abound: 94 reviews this year called the paper "the provided manuscript" — chatbot speak for an uploaded PDF — raising concerns about peer review's integrity.
More from Research
- Near-identical neural net weights still diverge on ~3% of inputs after training — bravo_abad · 2026-11-08
- Removing negative-advantage samples drops robot zip-tie success from 86% to 60% — micoolcho · 2026-11-08
- DECIDE benchmark tests voice-agent decisions on 4,867 questions from 1,081 real calls, scoring 20 models — Aiden_Tech_Ai · 2026-11-08
- KV Cache Quantization Tested Across 35 Quants: 8-bit Is Essentially Free for Qwen3.8 27B — Hefty_Wolverine_553 · 2026-11-08
- YC amplifies demo of models handling semi-serious X-band phased-array design — ycombinator · 2026-11-08
- Israeli researchers report method letting AI teach itself without human data — windless12 · 2026-11-08