Remix.run Logo
ninjagoo 2 hours ago

At first I thought this was a Nature paper. Turns out, it's a feature article.

The true test for this would be a blind test that involves human doctors - primary care since that's where something like this fits - exposed to the same data (fake papers), as well as LLMs.

Isn't it interesting that the fake papers made it onto science preprint servers? I didn't think that they were open to posting by random authors and had some basic checks in place. Currently these papers are showing as "withdrawn" on their DOI links [1] [2].

[1] https://doi.org/qzm4 [2] https://doi.org/qzm5