I believe that I may have originally gotten wind of this from a less wrong post IIRC. Pretty interesting stuff. Imagine if we trained an AI on doing science and peer review, and set it loose on the suite of research findings and had it report back all the BS…