NeurIPS paper accepted with 5/5/4 scores, yet final review cites AI-use concerns
Rich-Mycologist-75 · reddit · 2026-09-25
A NeurIPS author shares a confusing outcome: their paper was accepted with initial scores of 5/5/4 and a positive meta-review, but the final justification was entirely negative, raising AI-use concerns and suggesting further investigation and reconsideration of the recommendation.
The flagged issue was trivial: one reference's author list had been copied over from an adjacent BibTeX entry — more a typo than evidence of AI-generated content. The author asks whether the final justification is written before or after the acceptance decision, given the mismatch.
The post highlights the risk of false positives from AI-detection tooling in academic peer review and inconsistencies within the review pipeline.
More from Research
- Sharpa's World Synesthesia Model for dexterous hands accepted at CoRL 2026 — jiqizhixin · 2026-09-25
- DeltaWAM cuts video model cost for bimanual manipulation, lifting RoboTwin success to 85.4% — Han Yan · 2026-09-25
- Amazon open-sources Rufus-Air: full 8-stage post-training recipe on GLM-4.5-Air-Base — amazon · 2026-09-25
- Researcher proposes agent-suggests-human-executes loop for real-world science experiments — suragnair · 2026-09-25
- MLPerf Training v6.1 adds first LLM post-training benchmark: agentic RL on a 397B model — TheKanter · 2026-09-25
- Human-in-the-loop or machine-executed: verifiable tasks turn agent traces into RL rewards — suragnair · 2026-09-25