AI & ML · Jul 31
Caleb Robinson and Isaac Corley Found LLM Slop in 15 of 22 ML Submissions. The Fix Is Citation Auditing
A small reviewer sample shows why reference checks are becoming useful quality control for ML conferences.
- Audit references before submission, especially citations supporting central claims or related work summaries.
- Reviewers should treat fabricated authors and irrelevant citations as strong signals for deeper scrutiny.
- Citation checking is practical quality control, not a substitute for evaluating novelty, methods, and reproducibility.