LLMs Fail to Replicate Peer Review Standards for Top-Tier Journals
October 23, 2026
Testing four LLMs on 200 articles from Nature, Science, and Cell revealed extreme inconsistency and harshness. 75.6% of generated reviews recommended major revision or rejection, failing to align with the gold-standard validity of the original published works.
HOW THIS AFFECTS YOU
●
researcherYou cannot rely on LLMs to validate scientific literature or simulate peer review quality.
●
policyThis suggests significant risks in using automated systems for academic gatekeeping and quality control.