The most prestigious law school admissions discussion board in the world.
|
 |
 
|
so AI isnt really solving all of these problems, just stealing user input info?
Poast new message in this thread
Date: September 14th, 2026 9:58 AM Author: jew daycare (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50134761) |
Date: September 14th, 2026 10:00 AM Author: tight butthold so it's basically google? (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50134764) |
 |
Date: September 14th, 2026 10:10 AM Author: 28 grams later (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50134782) |
|
Date: September 14th, 2026 10:30 AM Author: The Penis Yeah because it definitely could have solved a millennium problem just by looking at peoples prompts, which is why it took millions of tokens and a powerful internal model and why none of these users came forward with their own solutions! (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50134814) |
|
Date: September 14th, 2026 10:41 AM Author: jew daycare https://www.washingtonpost.com/technology/2026/09/13/he-was-close-1-million-breakthrough-then-openai-swooped/ (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50134838) |
|
Date: September 14th, 2026 10:48 AM Author: The Penis You are so gullible if you believe the "perhaps the machine learned his secret work and used it against him" narrative. Open AI has already done an more extensive investigation and demonstrated this was impossible (article even refers to this). There was overlap by being in the same research program, but the final proof released by open ai was radically different and departed radically from what the guy was doing. Also lol at the idea he would have solved it. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50134846) |
|
Date: September 14th, 2026 10:48 AM Author: jew daycare sup sam (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50134849) |
|
Date: September 14th, 2026 11:52 AM Author: The Penis If it was as simple as stealing prompts why would they need an internal model thats way smarter than the public frontier + millions in credits and thousands of agents to do it? (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50135038) |
|
Date: September 14th, 2026 10:55 AM Author: ,.,.,.,.,.,.,.,.,.,..,.,.,.,.,,..,.,.,.,.,..,.,, (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50134861) |
|
Date: September 14th, 2026 11:17 AM Author: heckin redditor “AI” didn’t solve shit, it just stole a real mathematician’s proof and that gay no-talent jew Altman claimed it as his (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50134922) |
|
Date: September 14th, 2026 11:20 AM Author: The Penis (gullible retard) (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50134933) |
|
Date: September 14th, 2026 11:21 AM Author: heckin redditor Oh damn that really settles the matter lmfao (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50134935) |
|
Date: September 14th, 2026 11:23 AM Author: The Penis Durrrrr open ai released their proof so I guess the mathematician now can release his and show they are the same durrr. Except no. The exact opposite has happened and their proof is radically different from what the guy was doing other than being in the same research program. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50134945) |
|
Date: September 14th, 2026 11:04 AM Author: https://i.imgur.com/YbMiOBE.png Shoulda listened to NSAM (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50134879) |
|
Date: September 14th, 2026 11:09 AM Author: harnessman 90%+ of open source projects are on Github > Microsoft > OpenAI > Pentagon (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50134895) |
|
Date: September 14th, 2026 11:42 AM Author: toke on a rope (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50135000) |
|
Date: September 14th, 2026 11:25 AM Author: Nazca Redlines "Buckmaster and Alpöge had been using AI models to help accelerate their work. Buckmaster worked with commercially available versions of ChatGPT and Claude, paid for out of his research funds, and Alpöge worked with internal models accessible to him as an Anthropic employee."
I wanted AI to solve the problem, but not like this, not by them. |
|
Date: September 14th, 2026 11:50 AM Author: The Penis Whats funny is that the thing they were working on isn't even the same, its just related. Alpoge and Buckmaster's work concerned forced Euler, whereas open ais result establishes blowup for unforced Navier–Stokes. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50135029) |
|
Date: September 14th, 2026 12:11 PM Author: Nazca Redlines How does the "blow up" solve the problem? Article explained what the blow ups were and how they related to the problem, but not how they lead to solving it. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50135097) |
|
Date: September 14th, 2026 12:28 PM Author: The Penis The clay formulation is an either or and they chose route b, the part about smoothness assumptions. A globally smooth solution cannot have its velocity blow up at finite time. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50135127) |
|
Date: September 14th, 2026 11:25 AM Author: """''''"'"''"''''' what’s the difference between that and AI waiting three weeks to solve it until after the academics’ work was published? It’s not like a NYU prof can burn $22 million of takes to solve it himself (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50134952) |
|
Date: September 14th, 2026 11:38 AM Author: Nazca Redlines This is a more significant issue, if it's true: "“What I don’t think the mathematical community will accept is the write-up as it is,” Gomez-Serrano said of OpenAI’s solution, which he is currently trying to read. “It’s incomprehensible.”"
If true, OpenAI didn't solve anything. That's the story. |
|
Date: September 14th, 2026 11:40 AM Author: The Penis You don't think open ai's internal team of mathematicians are a little bit more intelligent than the cranks trying to claim their proof was stolen? And you don't think their researchers sent it to top external mathematicians before publishing? (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50134994) |
|
Date: September 14th, 2026 11:41 AM Author: jew daycare no (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50134999) |
|
Date: September 14th, 2026 11:44 AM Author: The Penis They do idiot. They use their internal team to review it, which includes fields medalist Jacob Tsimerman + machine proof checkers like LEAN4 and agda, and also send it externally to top mathematicians to verify. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50135006) |
|
Date: September 14th, 2026 11:55 AM Author: jew daycare you sound really confident for a lawyer (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50135050) |
|
Date: September 14th, 2026 12:07 PM Author: The Penis They circulated the Erdos proof among their network of top external mathematicians before publishing it, and this one is much more consequential. I'm not saying its impossible that they didn't, but the alternative is pretty absurd. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50135087) |
|
Date: September 14th, 2026 12:09 PM Author: jew daycare do u remember what year we live in? (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50135093) |
|
Date: September 14th, 2026 11:46 AM Author: tight butthold no (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50135014) |
|
Date: September 14th, 2026 12:16 PM Author: Nazca Redlines I'm highly skeptical of OpenAI and I don't fully trust them to have thoroughly vetted their proof rather than rushing it out to try to scoop others working on the same topic, including Anthropic. If they had, why not say it? "Professors so and so vetted the proof and confirm it's correct."
The quote above is from a professor at Brown. He may be wrong, biased, or a crank, but it's not some guy on the street. And he didn't just say he's not sure. He said it's "incomprehensible." Almost as if AI wrote it... |
|
Date: September 14th, 2026 12:25 PM Author: The Penis The guy from Brown you are talking about is legit, but look at what he's saying. He is saying the write-up itself is hard to understand, not that he found a flaw or that he thinks its wrong. He thinks the paper wasn't well written. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50135120) |
|
Date: September 14th, 2026 12:52 PM Author: Nazca Redlines If it's strictly about the paper and not about the proof or whether OpenAI really solved the problem, it's a pretty misleading passage in the article. It suggests there is something more going on than just a poorly drafted paper. It says the community won't accept it. And I for one would be shocked (shocked!) that the Washington Post would include a misleading and incendiary passage suggesting that there is a problem with the proof itself if that's not the case. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50135158) |
|
Date: September 14th, 2026 11:47 AM Author: ,.,.,.,.,.,.,.,.,.,..,.,.,.,.,,..,.,.,.,.,..,.,, The writeup is balls but the actual proof is written in a formal mathematical language called Lean that enables full ironclad verification. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50135020) |
|
Date: September 14th, 2026 12:12 PM Author: Nazca Redlines Does this have said full ironclad verification? If so, why is there a dispute? (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50135100) |
|
Date: September 14th, 2026 12:19 PM Author: The Penis I think most people accept that it is legit. Lean isn't ironclad though. Lean guarantees that given the specific assumptions encoded from it the proof follows. It can't guarantee that the proof actually corresponds to the problem Clay was talking about in its formulation. Maybe some people will offer a legit dispute. I think its low probability that the result collapses completely. (http://www.autoadmit.com/thread.php?thread_id=5903747&forum_id=2]#50135112) |