AI tools meant to vet science are surprisingly easy to fool - Science News

AI systems designed to catch errors in scientific research can be easily tricked into approving false or flawed studies. This matters because it undermines trust in peer review and the scientific process that guides medicine, policy, and public health.