New AI Tool Rates Preprints by Science, Not Reputation

QED Science has rolled out a fresh system that looks at the real substance of research papers before they hit journals. The service, called “The 1%, ” spots the top 574 life‑science preprints out of more than 57, 000 that appeared on bioRxiv last year. By stripping away author names, institutions and journal titles, the AI evaluates each manuscript solely on its claims and evidence. The company’s core metric, QED Score, was built to judge originality and validity without bias. Experts say that in many cases the score agrees with journal rank, but it also uncovers papers of high quality that end up in lower‑tier outlets. In fact, 12. 9% of the best papers were later published in journals that would normally be considered less prestigious.This approach tackles a long‑standing problem: papers can wait up to two years before they’re published, and the usual signals—citation counts or journal impact factors—don’t exist for preprints. Researchers, funders and institutions now have a quicker way to spot promising work early in the discovery cycle. The tool is already being used by more than 10, 000 labs across 1, 500 institutions in over 70 countries. Users report that the AI’s assessments take minutes, a fraction of the time normally required for peer review. Interestingly, smaller funders are showing up in the top 1% more often than big agencies. For example, a niche foundation had a 28% success rate in the top tier, compared with only 2% for a major national institute. The company’s co‑founder argues that science should be judged on its own merits, not on who publishes it. “Blind evaluation builds a stronger foundation for science, ” he says. Those curious can explore the system and see the full list of top preprints on QED Science’s website.