Post

@petersuber sad that they used an (hosted) llm for detection of false positive. It means that their results and software toolchain are not easily reproducible 🫠

While I’m rather convinced that llm and the seemingly linked publication increase are polluting science, I wonder if the number of citation also increased on the 2023-2025 period. Couldn’t it explain the increase of fabricated citations ?

@petersuber

wow… whatever could have caused such a thing to happen? /s

@petersuber@fediscthat graph is a bit scary tbh. peer review really can't keep up with this paceience.org

@petersuber@fedisc that chart is pretty alarming. 12x growth in 3 years is not a blip, thats a structural problemience.org

@petersuber

No a i dea why?

@petersuber

And that's just the citation of non-existent papers. Citations to real papers but with a false claim about what the papers say will be much harder to quantify. And that's a real and very fundamental and deep problem.

@petersuber fortunately, we have this https://mastodon.social/@hyc/116494045809291067

@petersuber you get this when you let AI writing the paper

@petersuber zomg no way

@petersuber

Thanks Gen-ai! You are truly a gift to science and the world! 🙄

@petersuber

So much for LLM and AI, but of course authors used to make up things long before…

@petersuber
Potential incentized correction : Discovery of a fake citation during review leads to instant rejection.

@petersuber but can we trust this?

@petersuber Threshold of trust long since breached ...