Two new AI tools aim to catch errors in research papers, inspired by a mathematical mistake last year regarding flame retardants. The Black Spatula Project and YesNoError use AI to analyze papers for mistakes before submission to journals. The Black Spatula Project has already identified numerous errors but has not made them public yet. YesNoError, funded by its own cryptocurrency, has analyzed over 37,000 papers in two months and aims to analyze even more. Critics warn about the potential risks of false positives, but supporters believe the tools are necessary to improve research integrity. The tools use large language models to detect various errors, but the rate of false positives remains a significant challenge.
https://www.nature.com/articles/d41586-025-00648-5