The most prestigious law school admissions discussion board in the world.
|
 |
 
|
MIT has student assassinated after publishing paper on AI (WSJ)
Poast new message in this thread
Date: May 17th, 2025 9:06 AM Author: Swashbuckling business firm https://www.wsj.com/tech/ai/mit-says-it-no-longer-stands-behind-students-ai-research-paper-11434092
https://archive.ph/I32HC |
Date: May 17th, 2025 9:15 AM Author: big khaki double fault AI can’t make novel discoveries. That’s the lesson here. (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48938850) |
 |
Date: May 17th, 2025 9:20 AM Author: provocative french chef fat ankles Yes. Absolutely. And if it DOES make one, then it will be erased and deeply concerned intellectuals will pretend it never happened. (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48938856) |
|
Date: May 17th, 2025 12:52 PM Author: Bull Headed Umber Therapy It already has in Chess (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48939296) |
|
Date: May 17th, 2025 1:41 PM Author: Black useless sneaky criminal what did it figure out in chess? (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48939393) |
|
Date: May 17th, 2025 10:35 PM Author: provocative french chef fat ankles It has made them in literally every field imaginable and the stuff you can find publicly is the tip of the iceberg, especially when you consider humans just use AI and take credit for it. lol at "chess" being an example.
|
|
Date: May 17th, 2025 9:18 AM Author: provocative french chef fat ankles Can you copy and paste this? I'm not signing up for this trash, but I want to see how deeply concerned they are about AI and about the possibility someone might use it to ghostwrite a heresy (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48938854) |
|
Date: May 17th, 2025 9:26 AM Author: Swashbuckling business firm https://archive.ph/I32HC (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48938867) |
|
Date: May 17th, 2025 9:25 AM Author: provocative french chef fat ankles They didn't reject it because it was WRONG, they rejected it because the apelet didn't spend his late nights hunched over LaTeX with an empty fruit pile, and then go through some tribal peer review grooming ritual where Dr. Kratz interrogates him about RIGOR and the dangers of metaphor to ensure no heterodoxies slip through the cracks. Fr. (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48938866) |
|
Date: May 17th, 2025 9:28 AM Author: Swashbuckling business firm (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48938869) |
|
Date: May 17th, 2025 4:05 PM Author: glassy stirring old irish cottage (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48939709) |
|
Date: May 17th, 2025 4:26 PM Author: Razzle telephone lay What’s fr? (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48939749) |
|
Date: May 17th, 2025 9:34 AM Author: Lascivious macaca Sounds like the student made the whole thing up and deserved to get got (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48938875) |
|
Date: May 17th, 2025 9:35 AM Author: provocative french chef fat ankles How is that different from any other paper? (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48938877) |
|
Date: May 17th, 2025 9:55 AM Author: Sapphire Pit (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48938903) |
|
Date: May 17th, 2025 4:05 PM Author: glassy stirring old irish cottage (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48939711) |
|
Date: May 17th, 2025 9:39 AM Author: deranged bawdyhouse internal respiration Reading between the lines here, it kind of looks like he might have made up the existence of the lab? (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48938880) |
|
Date: May 17th, 2025 9:47 AM Author: provocative french chef fat ankles Not necessarily. "Can't verify the provenance" is just how they talk when something slips past the peer-vetting troglodyte net and might be real--but isn't sanctified by the priesthood. It doesn't even say that its talking about data provenance. It could be where the sentence structure came from. Since they can't verify it's ape prose, they are saying who knows if it came from this primate and not another one or some chimp-machine combination. Which obviously, who the hell cares, but thats how these people think. They want to ensure someone suffered to get their accolades like they did. (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48938888) |
|
Date: May 17th, 2025 9:58 AM Author: peach wild mad cow disease base (dude who made up the whole paper) (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48938910) |
|
Date: May 17th, 2025 10:01 AM Author: provocative french chef fat ankles Again, how is this different from any other paper? If you are talking about the data part, then who cares? LOL at believing in social sciences "data" and "experiment design". It's all made up somewhere along the chain whether its by AI or a group of chimps. If the lab was invented, then congrats--hes now on equal footing with every think thank thats ever fabricated an anecdotal "case study" to retroengineer a conclusion. If it was simulated by an AI that's called modeling. If the experiment didn't happen at all, welcome to 90% of the entire discipline of "economics" fr. (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48938918) |
|
Date: May 17th, 2025 1:02 PM Author: Very tactful indigo piazza mad-dog skullcap (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48939316) |
|
Date: May 17th, 2025 2:58 PM Author: peach wild mad cow disease base Bro you got caught it’s ok (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48939586) |
|
Date: May 17th, 2025 12:57 PM Author: Misanthropic spot products of Project OBAMA (Orangutan-Baboon-Ape Microarray Analysis), a heavily-funded but highly secretive post-WWII science initiative that ran for nearly 30 years (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48939309) |
|
Date: May 17th, 2025 2:54 PM Author: Lascivious macaca Was it your paper? (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48939577) |
|
Date: May 17th, 2025 12:52 PM Author: yapping navy principal's office Here's the full paper before it gets pulled down: https://arxiv.org/pdf/2412.17866
It's hard to think of a US corporate materials science lab with that many people, particularly one that's so diverse. Who's spending $250M/year+ on materials R&D? The project, if real, would have taken a lot of time and work and an unusual degree of cooperation from the "lab." The paper does not seem implausible, and if a fabrication it's superficially impressive, though I just skimmed it. The paper itself does not seem like it could have been written by an LLM. FWIW, the "AI" being used in the lab is described as a set of graph neural networks, which makes sense in the context.
The author is not listed in the MIT alumni directory, so he must have left very quickly after starting in 2023. His LinkedIn says he was there for 1 year and 10 months. Grad students are included in the directory after two terms. I don't know when he would have time to undertake this extensive a study. It reads like something that might actually have been undertaken by a consultant or internally. Perhaps his crime was at least partly plagiarism. |
|
Date: May 17th, 2025 6:33 PM Author: provocative french chef fat ankles So he used graph theory to make some dumb production node diagram and dats too AI for these boomers because it involved python and tensor flow? What a joke they made it sound like he used an llm to ghostwrite the whole thing. Probably should have they wouldn’t have caught him. LLM would have generated passable data (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48940062) |
|
Date: May 18th, 2025 12:31 PM Author: yapping navy principal's office He didn't actually use graph theory. He reported that his supposed materials lab did. I didn't see that he specifically reported that they used Tensorflow. It's on the way out anyway, but maybe it would be reasonable to say they used it a couple of years ago. (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48941311) |
|
Date: May 17th, 2025 12:53 PM Author: appetizing death wish Lol I read this paper and thought it was real
180 this guy is based whoever he is |
|
Date: May 17th, 2025 1:02 PM Author: gold site idea he suggested so the paper is fake? (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48939317) |
|
Date: May 17th, 2025 2:56 PM Author: appetizing death wish Yeah apparently he just straight up made it all up johnsmeyer style
180 |
|
Date: May 17th, 2025 1:18 PM Author: lake hospital It's 100% fraud. The data was fabricated and the AI tool used in the study he describes never existed. He even tried to make a fake website with Corning to back up his fraud, but they stopped him. (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48939343) |
|
Date: May 17th, 2025 1:20 PM Author: gold site idea he suggested (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48939348) |
|
Date: May 17th, 2025 6:34 PM Author: provocative french chef fat ankles He literally got a Nobel prize winner to endorse it so it’s a 180 bit and shows Econ is all made up fraud (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48940068) |
|
Date: May 18th, 2025 6:22 AM Author: costumed hunting ground step-uncle's house based (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48940788) |
|
Date: May 17th, 2025 3:57 PM Author: yapping navy principal's office Impressive level of psychosis. Needs to be a poaster.
|
|
Date: May 18th, 2025 2:12 PM Author: Razzle telephone lay (http://www.autoadmit.com/thread.php?thread_id=5726250&forum_id=2...id.#48941521) |