3 ms·

Elisabeth Bik has a twitter where, at least in the past, she exposed a bunch of questionable things in papers (lots of manipulated images or images copied from

by bagels 2y ago

Elisabeth Bik has a twitter where, at least in the past, she exposed a bunch of questionable things in papers (lots of manipulated images or images copied from other papers) https://x.com/MicrobiomDigest https://x.com/MicrobiomDigest https://x.com/MicrobiomDigest/status/902016709019672577 https://x.com/MicrobiomDigest/status/902016709019672577 These efforts are just going to accelerate with AI applied, new and exciting frauds will be discovered.

etrautmann 2y ago
Yes, but AI tools also make it easier for bad actors to generate false images that are harder to detect as false going forward. The ones that get caught now seem to be incredibly simplistic manipulations.
- bagels 2y ago
  That is also true. There are decades of fraudulent papers out there to find from the before times, so it'll still be useful.
- theGnuMe 2y ago
  Ehh maybe, the images will be too perfect. I don't think they can model every source of bias, that would be a huge training problem.