The world is being quietly rearranged by people who write very long documents.


The title they went with DeepReviewer 2.0: A Traceable Agentic System for Auditable Scientific Peer Review Noisy translates that to

Scientific peer review can now be done by an AI that shows its work


Researchers built an AI system that can review scientific papers and explain how it reached its conclusions. This system finds more major problems in papers than human committees do in blind tests.
Peer review is slow and often misses critical issues. This paper shows an AI can help make the process faster and more thorough, while also making its reasoning clear. For years, a major concern with using AI in review was that it would be a black box, making decisions without explanation. This system directly addresses that by forcing the AI to show its work, making its judgments auditable.
Watch for major scientific conferences or journals to announce pilot programs using this system or similar auditable AI tools for initial paper screening.

If you insist
Read the original →