This AI tool claims to pick the top 1% of preprints. Should researchers trust it?
Nature, Published online: 10 August 2026; doi:10.1038/d41586-026-02276-z QED Science says that its metrics reduce bias by assessing papers solely on the basis of their originality and validity.
QED Science, a Tel Aviv-based startup, has created an AI tool to evaluate the originality and validity of life sciences research manuscripts. The tool is trained on various data sources, including open reviews, user feedback, and synthetic data, and is designed to identify gaps in the work and assess the supporting evidence for claims made in a paper.
The AI platform provides an independent ranking of preprints based solely on its assessment of originality and validity, without considering the authors' identities, institutions, or publication venues. In a recent pilot, the tool ranked over 57,000 bioRxiv preprints and identified the top 1% as "hidden gems." An analysis comparing the tool's rankings to those of peer-reviewed journals found that QED Science rated 12.9% of the papers higher than the journals might have suggested.
Researchers have expressed concerns about the potential for AI tools to create a new measure of prestige and to reinforce a culture in academia that prioritizes metrics. However, QED Science argues that its goal is to offer free services to authors in a secure environment to help them improve their work before publication, rather than to create another metric or to replace the peer review process.
Written by urgent.news from Nature's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
