Discussion
One thing that other students in my year relied heavily on LLMs for was checking whether their citations were correct in their thesis papers.
You know, the thing that bibtex has automated away decades ago, unless of course you insist on writing a 50 page research paper in fucking MS Word.
@aesthr gods, I managed to use a reference manager in fucking 2004 (LiteRat, precursor of citavi). It made my writing process a lot better.
Contrary to some widespread opinions, it is possible to write academic texts in Word (or other word processing software), but people aren't taught the first thing about their proper usage. (Source: I copy-edit humanities documents for a living.) At least, the last few PhD students I worked with knew how to use Zotero.
@draco it's certainly possible. I mean, a majority of students does it outside of math-heavy degrees where TeX is common.
But I really wouldn't want to, especially with how unreliable newer versions of it have becomes
@aesthr I'm so glad I switched back to Linux last year, and I'm pretty impressed with LibreOffice.
For my own writing, Markdown and LaTeX are my formats of choice, so, yeah, I get it.
@draco I never had the opportunity or rather the need to write a complex document in libreoffice but it would be interesting to try
@aesthr If you're writing a 50 page thesis in Word, Zotero is fine (or it was last time I used it, may be bloated with AI by now)
@Bumblefish zotero is still pretty good, but integration into word always feels a bit unstable and weird. But I haven't used it extensively
@Bumblefish I use it mainly to get a bibtex bibliography from the collection of PDFs I have in there that I then throw in my git repository for the actual thesis. And for that it's perfect
@aesthr Since LLMs are notoriously reliable for citations? Heh
@mjj this isn't about generating new citations though and I don't know how well these things work for checking existing ones so I can't speak on that with confidence
@aesthr Right. I would expect them to reliably be able to say "yes, this has the formatting of af citation", but factually, less promising.
We're psychology students, the amount of LaTeX you need to write a thesis here is very small (unless you specialize in statistics like I did, but I'm the only one doing that).
You can totally learn how to write a psych thesis in LaTeX over a weekend. And then so many problems just go away.
Or you could spend your already scarce money on Claude tokens, I guess.
@aesthr I hope I'm not derailing too much, but these days I would advise that students try Typst instead of LaTex, especially if they don't need any special package.
@drmabuese I have tried that a few times when writing papers but I've always returned to LaTeX. To me it just wasn't there yet. Maybe in a few years
They keep talking about how "AI" solves problems for them that I just don't have.
They made themselves dependent on these services for nothing. And they're not done with their papers any faster either, or with less stress.
I'm writing my thesis with tools that are 20 years old or older. Heck, a central piece of it is a Makefile. And I feel a lot more chill about it than a lot of them tbh, because I have confidence in my tools and my skills.
[rants unintelligibly about OCR and these goddam lunatics]
@aesthr Or with more accuracy since LLMs invent papers to cite.
@aesthr When I got my psych degree many years ago, I wrote all my papers in MS Word, and MS Word--unless they removed the feature, I guess--supports citation references.
I can't imagine using a hallucination engine to "verify" the citations are correct.