5 Likes
Students using llms to make up research articles that donât exist is a g d epidemic this term
11 Likes
That seems like the easiest thing to verify and get caught on for academic misconduct.
2 Likes
Yep! Very weird and sad. I have told them multiple times that this will happen!
3 Likes
So I was watching a recent Major League Wrestling show and the guy I was watching it with swears they used a LLM to come up with the show theme song, so I guess this is another thing I have to keep an eye (ear?) out for now.
1 Like
And here I was having funâŚand then this ai slop vomits on my faith. Fuck this!
Fuck this shit so fucking hard!
3 Likes
Sometimes I wonder if the countless false christs of Revelations are all competing AI models of Jesus.
4 Likes
A worse woe upon the flock, I cannot imagine.
2 Likes
8 years ago
3 Likes
gut reaction to this is, as usual, we live in hell, but on second thought i actually feel a little bit more ambivalent about this.
one reason i am less skeptical is that i do think LLMs may have more practical value for this kind of application when restricted to a very particular corpus. this way you can create something you have a little bit more control over rather than just presuming the statistical probabilities of the open internet are enough to generate truth. or, anything of value, really.
but i still just donât understand⌠why this is necessary. âwe put all the facts into a textbook so students can focus on discussion insteadâ is the goal of basically every humanities course in any university.
on the other hand, i also frequently encounter a problem of no textbook providing exactly the right kind of context i need to introduce the primary sources that are at the core of the unit/meeting/whatever, so i can kind of understand the instructorâs desire to create something that is a bit more bespoke for the goals of the class without having to go to the trouble of writing a new textbook about it.
but thereâs also the parallel issue that students just⌠donât read. itâs very hard to offload any kind of important detail into readings that one intends to have been prepared well prior to the class meeting, and i donât know if handing them a stack of LLM generated text (likely to be extremely dry) is really the solution to reigniting a passion for that kind of text-based learning. it may respond to a common complaint students have about reading (the textbook is not just a list of facts i need to memorize to get an A in this classâŚ) but in this case it feels like an attempt to âmeet them in the middleâ that will leave everybody unhappy.
9 Likes
âBecause the course is a survey of literature and culture, thereâs an arc to what I want students to understand,â said Stahuljak, a professor of comparative literature and of European languages and transcultural studies. âNormally, I would spend lectures contextualizing the material and using visuals to demonstrate the content. But now all of that is in the textbook we generated, and I can actually work with students to read the primary sources and walk them through what it means to analyze and think critically.â
I canât make sense of this at all. What is she even saying? Now theyâve got fake pictures to look at on their own time?
But unlike ChatGPT and other public large-language models, Kudu will draw information only from the resources Stahuljak has uploaded.
No it wonât. Iâm not quite sure what Kudu is or how it works, but thatâs not how LLMs work. Thereâs not enough text in her corpus to get a coherent bot out of. My guess is itâll be âprompt-engineeredâ to only answer queries that pertain to the coursework i.e. every query will be prefaced by some âYou are a helpful professor of comparative litterature high quality 4k answering studentsâ questions bla blaâ hocus pocus.
Anyway, incompetence or payola you decide.
9 Likes
The sad thing is, it doesnât even need to be payola⌠People in education love to jump on the tech hype bubble of the day, just out of deep insecurity over the relevancy and prestige of their profession and institution. The same thing applies to libraries. This is the 2024 version of âwe built classrooms in Second Life and weâre holding lectures there!!â
14 Likes
If you do it before anyone else does, you just might get recognized for your contributions to the âdigital humanitiesâ
10 Likes
oh i was just thinking about this the other day!!
iâm a little bit of a boy on the floorâŚ
the closing monologue still kind of hits at points⌠âiâm sorry, i donât like him. but i can go home, and be so bad, and iâll love himâ
2 Likes
yeah, iâve thought about this for more than 5 minutes now and have concluded that it is unambiguously 100% bad and dumb lol
i do think it is theoretically possible for llms to have some kind of practical value in humanities research, but almost certainly not for teaching. the dream is still to be able to have some kind of perfectly intuitive search and retrieval function for information within large sets of documents, the memex finally realized
but anyway, part of the âphilosophyâ behind this is a not particularly new buzzword in pedagogical bullshit, the âflipped classroom,â basically that all in-person classes should be discussion/expression focused and, rather than in lectures, all information dumping (donât know what else to call it) should be done by students alone at home, either reading or watching prerecorded video lectures or something. i donât think this is necessarily a bad idea, i mean, itâs basically just how literature courses have always functioned, but i also donât think it is some kind of perfect solution to the problem of (most) students not really getting much out of lectures. unfortunately, the solution is just smaller class sizes, but of course nobody wants to hear that so the quest for better mass audience pedagogy through technology continues
anyway additional things that have soured this for me are a) the website for the platform they are using is just marketing bullshit, with literally no information about what it is or how it works b) it is yet another attempt to reinvent the large undergraduate intro survey class, something that literally no student likes taking and no teacher likes teaching, but universities keep insisting we offer (see my point about large class sizes above) and c) as i just read someone point out, even if this is a kind of utopian system whereby a limited corpus of lecture notes and presentations is magically transformed into a wonderful textbook through the power of the LLM, those lecture notes are likely to have been derived from lots of other texts. using this kind of thing to prepare lectures is not a big deal*, but as soon as you put it in print it suddenly feels a lot more like plagiarism, or at least very sloppy citation management
and yeah on the âdigital humanitiesâ frontâas someone who has briefly experimented with topic modeling and like network analysis as part of literary studies research, part of me always bristles when people use âdigital humanitiesâ pejoratively, just as like an emblem of how the Academy has strayed from the path. it just starts to feel exactly the same as when people decided that âCritical Theoryâ was the bogeyman assaulting the fine traditions of idk close reading and philology and polluting them with foul nonsense. just an example of constructing an object of critique out of various anecdata /fears/presumptions and then assuming that represents reality (âmaking up a guy to get mad atâ). e.g., i think thereâs lots of productive research that falls under the umbrella of âdigital humanitiesâ that is doing stuff perfectly in line with a lot of different overarching goals of humanities/literary research, but also a lot of stuff that feels sort of half baked, failed experiments, unconvincing arguments, meaningless digressions, etc., just like there always is in all research all the time (no subdiscipline is exclusively capable of producing arguments everyone agrees with, thatâs⌠the point). likewise, the pedagogical half of âdigital humanitiesâ is both a) more or less unrelated to the research, most of the time, and b) a combination of the genuine work of figuring out how to use technology in a way that makes sense (was the first person to type a research paper on a word processor rather than a typewriter or in long hand a âdigital humanistâ?). thatâs not to say that every piece of technology/software is instantly going to revolutionize everything we do, but if we really wanted to exist in opposition to this it would mean picking some arbitrary point in the past and locking on to it. we have always been digital humanities⌠the fact that there are also shitloads of startups and other corporations making slick deals with admin to get their products adopted is also a tale old as time (see textbook publishers). calling all of that stuff âdigital humanitiesâ in order to dismiss it all doesnât seem particularly helpful. but i agree that a lot of it sucks!
(*iâve started to put âfurther readingâ slides at the end of my lectures with a list of most of the sources iâve used to write them, not for necessarily because i think it would be plagiarism otherwise, but it does still feel kind of weird to not cite my sources, probably because iâm still new enough at this that i donât really trust just presenting myself as the authority on any topic. and yet it feels odd to stop midway through and just be like⌠donât take my word for it⌠then rattle off a list of names no one in the room has heard of before. also, it may be helpful for a student who for some reason wants to know more about something i address in the lecture, may save both of us an emailâŚ)
9 Likes
will cop to my snark about âdigital humanitiesâ being overly aggressive and possibly actually outmoded. thereâs a lot of cool and valuable stuff being done in it/under that banner, and thereâs a lot of lame and useless stuff being done outside it. my three years doing an MA seemed to align with a particular flourish of enthusiasm in âthe digital humanitiesâ and it semed to me that a lot of money and attention was going to projects that felt like cop-outs, either adding the thinnest veneer of âtechâ to a research project to sex it up or amassing grants for pointless and unusable âdatabasesâ that required no real application of intellect, training, or analysis
that is to say, points well taken, and itâs probably time to stop painting with that broad a brush lol
4 Likes
9 Likes
corporations treat michael clayton like a tutorial video
9 Likes
I wonder if Bill Burr has an alibi for this one
1 Like