ArXiv announces a ban on AI content and the responses are hilarious.

[a_cubed@mastodon.social]

@jamesmarshall @docpop
He's wrong. Not everyone does it. He's trying an argumentum ad populum. I'm on the publication ethics committee of a major Comp Sci academic society publisher and we would redact papers with multiple non-existent citations and ban the authors from submitting to our publications for at least a year.
You submit it as your work, you stand byit as valid. Jointly and severally, although we sometimes go easy on student joint authors with faculty (faculty should know better).

[dukeduke@mastodon.social]

@docpop That fucker belongs on Fux News. ""People are saying" means that's good enough for me to cite!"

[linkachus17@furry.engineer]

@docpop Clearly those people never had experience in making research papers or something like that. And that's bad

[jamesmarshall@sfba.social]

@a_cubed @docpop thanks for your comment, very informative. I assume your academic society publisher is international? I'm wondering if this is a US problem.

[heptasean@social.tchncs.de]

@adriano @docpop @FediThing Depending on what is usual in his academic field, his thoughts are maybe less “How the f is he a professor?!?” outlandish than the Fedi commenters here think.

For papers with lots of authors and quite strict division of labour for different parts of the paper and different sub-groups of authors, I can very well see that it maybe has not been usual up to now to know every other author and double check their work.

Still seems net positive to me to reinforce responsibility for things with your name on it.

[carl@chaos.social]

@docpop the responder does not understand Science

[heptasean@social.tchncs.de]

@marta Thank you! I'm also a bit disturbed by the comments here seeing Miller as being _completely_ out of line and ridiculing him for the points raised. His point is more reasonable than the “Why is that guy a professor?!?” mob thinks, but in the end still false in my opinion. Yes, even if you work in a field with papers with lots of authors and quite loose cooperations, you are and should be responsible for papers with your name on it. Laskowski investigating the data manipulation and retracting the papers is a good example for taking that responsibility. @docpop

[congusbongus@mastodon.gamedev.place]

@unchartedworlds @FediThing @docpop so all those Wikipedia nerds adding [citation needed] tags just want more reading material? TIL wow lmao

[liztai@hachyderm.io]

@docpop dafuq isn't that a basic skill

[onnob@mastodon.social]

@docpop jesus, you have to check everything AI writes. Citations or otherwise. And it can be a great help in finding sources, but you have to check everything. By hand. And brain. Yes, all of it.

[zjb@mastodon.social]

哈哈，这个反应也太真实了。学术圈其实很多人都在偷偷用AI写论文，现在arXiv一禁估计要炸锅了。

[heafnerj@mstdn.social]

@ftranschel @docpop Had to look up scnr then it made sense.

[heafnerj@mstdn.social]

@Darkphoenix @docpop There are sometimes physics errors in physics textbooks too.

[theorangetheme@en.osm.town]

@docpop I can't tell if AI made people stupid or if it just made pre-existing stupid people louder and more obnoxious.

[adipoeserpursch@troet.cafe]

@docpop ‍

[cuchaz@gladtech.social]

@jamesmarshall @docpop no, we don't make up fake citations. But also sadly, not every co-author checks every citation. We rely on trusting our co-authors to not commit academic fraud.

I'm very in favor of verifiable academic fraud leading to (impermanent) bans from preprint servers. If your co-authors are committing fraud and you find out by getting banned from a preprint server, that sucks, but it's a great signal to stop trusting that co-author. Nature will heal.

[gimulnautti@mastodon.green]

@docpop In a civilised age James Miller would go down in history as the fool of the year. Now he’s just a another hustlebro competing other hustlebros trying to get rich off abuse.. Of systems that literally hold our civilisation together.

[frutigeraero00@mastodon.social]

@docpop "So... We have to actually use our brains? "

[illuminatus@mstdn.social]

@DamonHD @FediThing @unchartedworlds @docpop There are universities and colleges and then there are "universities" and "colleges". Have you read about "Trump university"?

[emmecola@fediscience.org]

@docpop I think people tend to overestimate the percentage of scientists who actually read, in full, the papers they cite. This isn't about AI, it has always been like that!

https://arxiv.org/abs/cond-mat/0212043