Positive users accept AI-generated papers if admitted or advancing science, while negative users criticize the author's lack of understanding, poor writing, and careless LLM-based responses.
@littmath @pangram My charitable take would be that he knows the work is likely correct and novel, but he does not understand it well enough to communicate himself. I would hope it's not slop or psychosis.
@littmath @pangram I worry we will increasingly see people who are in too deep with massive over-reliance on LLMs
@jconorgrogan @pangram The paper was very poorly written; I’m skeptical the content had value but it’s not pure math so I won’t comment beyond that.
@littmath @pangram Sometimes I struggle to tell if some people are writing their posts with AI or if they read so much AI generated text that they sound like AI
@ptremblay @pangram Big (wrong) assumption! But even if so seems bad to respond to comments by copying and pasting from an LLM.
@sebngriego @pangram Charitable indeed but if true still very bad—he shouldn’t be representing the paper if he doesn’t understand it. There’s circumstantial evidence the work is slop but it’s not a pure math paper so I won’t comment beyond that.
@littmath @pangram Call him out?
@reconfigurthing @pangram Not clear to me what that would achieve, presumably these facts are clear to everyone who reads the tweets/opens the paper.
@reconfigurthing @pangram Don’t have the appetite for it atm I guess.
@pangram whispering earring sh*t
@littmath @pangram I mean, you're free to not do it, but it would definitely achieve something? It probably isn't clear to everyone, it would create common knowledge, and it would be an act of public shaming.
@littmath @pangram I'd personally like to see it at least!
@littmath @pangram Prof. Litt. I have found myself mostly agreeing with you so far.
On this matter. Without knowing any details.
Ideally Anything that advances the state of the art is publishable. The decision about whether it does Shd be made by community. Perhaps AI aided.
@JagadeesanRadha @pangram If the result is significant it should of course be disseminated (fwiw I suspect it is slop but it’s not my area). But it’s inconceivable to me that responding to comments by copying and pasting from an LLM is a good course of action.
@littmath @pangram The professor is front running the singularity
@littmath @pangram does it matter, though?
@ptremblay @pangram What do you mean?
@littmath @pangram Very fair
@littmath @pangram if a paper is entirely written with AI, assuming what it presents is otherwise good.
@littmath @pangram I'm not assuming the paper is good. I just mean if it's good, I wouldn't care it's written by AI.