Research Daily Summary
Research Daily Summary: What Happened on October 24, 2026
AI reviews made fewer errors than human reviews in a 124-paper comparison
AI-generated reviews contained about one-third as many incorrect claims, and former journal editors preferred them in most of their judgments.
Today in brief
Gary King and Steven Worthington compared AI and human reviews of 124 papers, finding fewer incorrect claims in the AI reviews.1 Three former journal editors selected the AI review in 26 of 30 judgments.1
AI peer-review comparison
In a comparison of reviews for 124 papers, AI-generated reviews contained about one-third as many incorrect claims about the papers as human reviews, and included no fabricated citations.1 Three former journal editors chose the AI review in 26 of 30 judgments.1
The study authors said their findings did not address using reviews to help editors decide what to publish.1
Also published
- On GDF-15 neutralisation and cachexia, with CatalYm pharmaphorum