“This is more complex than it may at first appear, as references can be detailed by authors in a variety of different ways, often do not include DOIs, and simple tools to identify hallucinated references can produce false positives,” Graf told us by email.

It’s not complex when you have a fucking style manual which specifies exactly how you detail references so you don’t have this kind of problem.

[–] hoshikarakitaridia@lemmy.world 17 points 3 months ago (1 children)

I was gonna say, surely there's a database of published works you could literally query to check for validity. I mean literally traditional algorithmic verification. Why the hell would you need an AI for that. AIs are for black box problems but this has been solved way back. This tells me whoever is in charge is more concerned with optics than with things going well or they are highly incompetent.

[–] Sumocat@lemmy.world 5 points 3 months ago (1 children)

For medical articles, there’s PubMed, the medical research database mandated by the U.S. and E.U. For other fields, YMMV.

[–] hoshikarakitaridia@lemmy.world 6 points 3 months ago

Yeah so assuming Springer makes enough money (pretty sure they sell their copies and there's solid money there) they are absolutely able to get those systems in place before publishing. That's telling.

[–] FiniteBanjo@feddit.online 4 points 3 months ago (1 children)

The USA has several: MLA, APA, Imrad, etc but none of them explicitly require control numbers or serial numbers from publications.

Meanwhile, I knew a guy in the UK who did a dissertation on market economies and they literally did not specify any kind of format for him at all.

[–] thesohoriots@lemmy.world 2 points 3 months ago

Journals have “house styles” which will specify which one to use or any modifications, but they generally follow Chicago and can specify to require control numbers/serial numbers/DOIs if they wanted to.

[–] kamenlady@lemmy.world 3 points 3 months ago (1 children)

It’s not complex when you have a fucking style manual which specifies exactly how you detail references so you don’t have this kind of problem.

About that, it's kinda complicated

[–] thesohoriots@lemmy.world 2 points 3 months ago

Not especially. I’ve worked in the field and unless you get more well-known authors with egos who don’t care to follow the requirements — and there certainly are — you grab a current style manual while working and do some due diligence. Ignorance and fame are not an excuse.

[–] tae_glas@slrpnk.net 21 points 3 months ago (1 children)

the fact that following up about hallucinated references was met with a second list of hallucinated references is just insulting, imo.

surely needing to go through everything with a fine-toothed comb to make sure that a) a source hasn't been hallucinated and that b) cited sources actually say what is claimed, is more time-consuming than just doing the research & citations manually?

[–] Paradachshund@lemmy.today 13 points 3 months ago

That point about it taking more time applies to a lot of things with AI it seems like. It's only fast when you can trust it, and for important things you can't trust it.