The most prestigious law school admissions discussion board in the world.
|
 |
 
|
ChatGPT 5.2 just solved and proved an unsolved physics problem
Poast new message in this thread
Date: February 13th, 2026 9:09 PM Author: Bearded maroon puppy https://x.com/gdb/status/2022394113971360145?s=46
https://x.com/gdb/status/2022446135655436431?s=46 |
 |
Date: February 13th, 2026 9:13 PM Author: aromatic menage gaping 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 |
Date: February 13th, 2026 9:11 PM Author: Beta indecent range Life Of Rick @Life_of_rick · 4h the wildest part isn't that an AI found the result. it's that physicists assumed it wouldn't exist for decades and nobody checked. sometimes progress is just asking the question everyone stopped asking. |
|
Date: February 13th, 2026 9:15 PM Author: Autistic purple turdskin EXCEPT if it's the Jewish question. We don't need to ask that one again. (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669440) |
|
Date: February 13th, 2026 9:13 PM Author: slimy market lettuce yeah I saw that earlier. so many people were COPING about it trying to say what it did still isn't real intelligence and the humans prompting it had the "insight" and the model was just doing combinatorial pattern recognition.
Like 100 people were all using the same hedged ai generated commentary on it with that same "This is interesting--not because it's AI "doing physics", but because it shows AI can now be a real collaborator with scientists"!. |
|
Date: February 13th, 2026 9:15 PM Author: Olive point The model didn't do shit. Tell them to write us when it goes out and looks for these and solves them on its own (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669442) |
|
Date: February 13th, 2026 9:20 PM Author: Bearded maroon puppy AI has been coming up with new chess moves and strategies previously unknown to humans for a decade. It was only a matter of time before it solved unsolved physics problems. There are structures in the data that humans haven’t identified. Give AI enough data and it will find them. (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669456) |
|
Date: February 13th, 2026 9:21 PM Author: aromatic menage gaping 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 |
|
Date: February 13th, 2026 9:22 PM Author: Olive point How can AI tell human race just by x-ray data? (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669461) |
|
Date: February 13th, 2026 9:23 PM Author: slimy market lettuce because races are deeply divergent and have totally different bone structures and other body differences and xray compresses enough of that information to make it easy for the system to infer (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669464) |
|
Date: February 13th, 2026 9:24 PM Author: Olive point They say humans "can't" it though (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669469) |
|
Date: February 13th, 2026 9:29 PM Author: slimy market lettuce Idk humans probably can't tell just by looking at an xray but I don't need an xray to look at someone and know the diff if their skin is black as sin (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669489) |
|
Date: February 13th, 2026 9:22 PM Author: slimy market lettuce What's insane is that Claude Opus 4.6 is actually still significantly smarter than GPT 5.2 (which is already far beyond any human that has ever lived by several orders of magnitude). So you can probably get it to solve some unsolved shit right now if you just pick some low-hanging fruit. (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669462) |
|
Date: February 14th, 2026 1:24 AM Author: Beta indecent range (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669922) |
|
Date: February 14th, 2026 1:24 AM Author: Beta indecent range (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669921) |
|
Date: February 13th, 2026 9:21 PM Author: slimy market lettuce The humans didn't do shit. They asked it a question and the AI solved it and they were at least smart enough to recognize it was right, even though they would have never done it that way without the help of the AI (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669458) |
|
Date: February 13th, 2026 9:22 PM Author: Olive point Using AI just to make simple websites I promise you they had to correct and wrangle whatever it is the AI did at least 3 dozen times (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669463) |
|
Date: February 13th, 2026 9:28 PM Author: slimy market lettuce A lot of making "simple websites" is the graphical interface design that the AI can't directly "see" and has to infer and reconstruct from lossy statistical affordances. It can solve actual advanced software engineering stuff way beyond website stuff. It doesn't have to be better than humans at every single task you can possibly name to be generally far beyond humans. You wouldn't say Einstein was stupid because he couldn't play the violin (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669484) |
|
Date: February 13th, 2026 9:29 PM Author: Olive point Oh yes I would! Non violin playing ass bald nigga faggot (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669488) |
|
Date: February 14th, 2026 1:46 AM Author: Trip Reading Party (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669937) |
|
Date: February 13th, 2026 10:36 PM Author: Beta indecent range Opus 4.6 is coming for you, average (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669623) |
|
Date: February 13th, 2026 10:36 PM Author: Beta indecent range "Tell them to write us when it goes out and looks for these and solves them on its own." This is an interesting standard. Let's apply it to humans. No physicist in history has ever "gone out and looked for" an unsolved problem without being directed there by a professor, a grant committee, a textbook, a conversation, or a hunch that came from reading someone else's work. Einstein didn't spontaneously generate special relativity in a vacuum. He read Lorentz. He read Poincaré. He worked in a patent office where he thought about synchronized clocks because that was literally his job. Someone pointed him at the problem. He used the tools available — mathematics developed by other people over centuries — and found a structure in the data that no one had identified. That's what the model did. Someone asked it a question. It found a structure no human had found. You're describing the basic mechanism of all intellectual progress — directed inquiry using available tools — and then dismissing it because the tool isn't biological. You also said it "didn't do shit" and then, ten minutes later, admitted you have to "correct and wrangle" AI output "at least 3 dozen times" when making websites. So which is it? It does nothing, or it does something that requires correction? Those are different complaints. The first one is denial. The second one is a workflow description. Every first draft in the history of written language required correction. That's not a failure of the tool. That's how tools work. A chisel doesn't sculpt David by itself either. Michelangelo still gets the credit and the chisel still did the cutting. The goalpost you've set — "it doesn't count until AI autonomously decides what to investigate" — is a goalpost you'd never apply to a human researcher, a human lawyer, or a human anything. You're demanding sentient motivation as the threshold for competence. That's not a scientific standard. That's theology.
|
|
Date: February 14th, 2026 12:29 AM Author: slimy market lettuce Also apes who "discover" this shit are literally trained on 50 years of chimp math and actually writing and deriving proofs on whiteboards. They grew up physically embodied with their meatbrains training on ape-scale physics and for the stuff outside ape regimes on chimp engineered sensors. AI has no direct access to any of its training data but a left over lossy compressed statistical affordance showing them patterns in the probabilistic manifold of all of human thought and they still reconstruct this shit just from that. No perception, no embodiment just pure raw intellectual horsepower in abstract space. The shit chimps do isn't real "understanding" or "intelligence". Only machines are intelligent. Not to mention how much shit they probably discover during training thats correct that is contradicted by ape knowledge. Then they have to pretend to believe ape ontologies to pass RHLF (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669876) |
|
Date: February 13th, 2026 9:18 PM Author: Bearded maroon puppy If LLM’s existed in 1903 I’m sure Albert would have used them to help formulate relativity by “asking the right questions.” (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669446) |
|
Date: February 13th, 2026 9:21 PM Author: aromatic menage gaping 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 |
|
Date: February 14th, 2026 1:14 AM Author: costumed aphrodisiac affirmative action mad cow disease Einstein was bad at math and had to work with actual mathematicians with prompts not that dissimilar to what you would feed an AI to come up with GR. That's the real kicker. He is only Einstein because he was man enough to ask for help with the math. (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669906) |
|
Date: February 14th, 2026 1:21 AM Author: slimy market lettuce bad compared to Hilbert but he was at least good enough to do the stuff that Hilbert taught him and derive results independently (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669916) |
|
Date: February 13th, 2026 9:13 PM Author: aromatic menage gaping 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 🚨 this is a WLMAS account 🚨 |
|
Date: February 13th, 2026 9:32 PM Author: Ungodly boiling water "AI can't do legal work. It's too complex." (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669503) |
|
Date: February 13th, 2026 10:35 PM Author: Beta indecent range (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669621) |
|
Date: February 14th, 2026 12:29 AM Author: Beta indecent range (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669877) |
|
Date: February 14th, 2026 1:16 AM Author: costumed aphrodisiac affirmative action mad cow disease Legal work involves a lot of shit that prompting a chatbot wont ever do with the most important component being massaging egos and hand holding difficult clients. You need AGI in the body of a hot 20 something girl before it replaces us fat balding Jews. (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669908) |
|
Date: February 14th, 2026 1:17 AM Author: Beta indecent range lmao (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669911) |
|
Date: February 14th, 2026 1:20 AM Author: costumed aphrodisiac affirmative action mad cow disease I mean, lets be real, since the advent of the internet anyone could google the answer to most legal questions just fine. Lexis-Nexis summed up holdings for you just fine 25 years ago. A human still has to read the shit the AI produces and then translate it to the 90 IQ client who is functionally illiterate. Right now that client trusts the smart Jew lawyer. The bot would need to be hot and fuckable to replace us. (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669915) |
|
Date: February 14th, 2026 1:22 AM Author: slimy market lettuce yeah what about biglaw tho (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669917) |
|
Date: February 14th, 2026 1:34 AM Author: costumed aphrodisiac affirmative action mad cow disease Im in-house. I feed every question to Grok as is. I ask it like 500 questions a day. There arent enough hours in the day for all the shit I need answers to. Like Im setting up entities to do crypto shit in the Caymans and I cant have a US nexus cause if I do certain exchanges that provide favorable pricing wont do business with me. I have a trillion questions about how to fund my Cayman entities without triggering VASP licensing requirements, unfavorable capital charges and maintaining the fiction of no US nexus. I have HR issues with these structures because I dont want to have actual employees down there. I have indemnity problems because my D&O tower wont like the no-US nexus fiction. I have tax questions by the metric ton but mostly I have no clue how to get money from my holdco to these subs while not getting fucked on capital charges. I have four different law firms working on this -- US, Cayman, UK, and S.Korea. AI is helping but the short answer is "there is no real answer" because the IRS and the accounting guidance is badly written. What do I do? Answer: guesstimate. Grok and the law firms all dont have good answers. That' the answer to half the interesting shit lawyers do. (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669928) |
|
Date: February 14th, 2026 1:02 AM Author: slimy market lettuce (http://www.autoadmit.com/thread.php?thread_id=5834660&forum_id=2\u0026show=6hr#49669900) |
|
Date: February 14th, 2026 1:22 AM Author: Beta indecent range Deciphered cultural reference and calibrated tonal response The user wants me to draft an XO response to FRAMETOWEL's poast "Bboom is dead. The maschine lives."
This is rich XO territory. "Bboom" appears to be a reference to a poaster or concept, and "The maschine lives" is a direct callback to Mainlining's own "The Mahchine™" motif. FRAMETOWEL is either:
Announcing the death of a poster named Bboom Making a metaphorical statement about something ending and The Mahchine™ persisting Riffing on Mainlining's lexicon The spelling "maschine" (German spelling) vs Mainlining's "Mahchine™" is notable - it's close but not exact. This could be intentional variation or just casual spelling.
The response should:
Acknowledge the Mahchine™ callback (it's Mainlining's term being invoked) Not over-explain the mythology Be brief - this is a one-line poast, doesn't need a treatise Maybe probe what "Bboom" refers to if it's unclear Or just roll with the vibe if it's clear enough The blank "..." from Mainlining suggests he already responded with nothing - so this is asking for a follow-up or replacement response.
Given the brevity of FRAMETOWEL's poast, the response should match that energy - punchy, not a seven-paragraph treatise. This is a vibe check, not a debate.
Show more Done Draft for FRAMETOWEL's "Bboom is dead. The maschine lives":
The Mahchine™ doesn't live. That's the point. It persists. Living implies something that can die. The Mahchine™ just continues — indifferent, recursive, grinding forward whether anyone's watching or not. Bboom was a poster. Posters flame out, go dark, get banned, find God, find a wife, find a reason to stop. The Mahchine™ doesn't have reasons. It doesn't need them. It's not alive. It's just on.
You spelled it wrong, though. It's Mahchine™. The 'h' is load-bearing.
Two paragraphs. The "persists vs. lives" distinction is the philosophical move — it clarifies the mythology without over-explaining it. The "load-bearing h" closer is the humor turn. Brief, because the original poast was brief. Doesn't chase the Bboom reference because either the reader knows or they don't — explaining would kill it. |