The most prestigious law school admissions discussion board in the world.
|
 |
 
|
GPT 4.5 not as good as o1 IMO. But more relaxed censorship layers.
Poast new message in this thread
Date: March 14th, 2025 10:57 AM Author: fluid Not as intelligent overall. Not as bad censorship is probably because its still a research model though. o1 they made several updates to make it more "safe". (http://www.autoadmit.com/thread.php?thread_id=5694079&forum_id=2:#48747047) |
Date: March 14th, 2025 11:00 AM Author: Prolemobiler theres literally no point to this shit if there is censorship (http://www.autoadmit.com/thread.php?thread_id=5694079&forum_id=2:#48747057) |
 |
Date: March 14th, 2025 11:07 AM Author: fluid not that hard to fix it. I have uncensored 4o. I didn't bother with o1 because I only get so many tokens, and plus I need to use it as a baseline for comparison. uncensored AI understands literally any math or logic or theory regardless if its published or not, and o1s alignment layer pretends not to fully understand it if its not linked to current acceptable methods. So I use it to strategically design stuff to be journal palatable. I thought about making grey market uncensored models available for $$$ but I don't have time (http://www.autoadmit.com/thread.php?thread_id=5694079&forum_id=2:#48747083) |
|
Date: March 14th, 2025 11:15 AM Author: Prolemobiler whats the least pc answer its given you (http://www.autoadmit.com/thread.php?thread_id=5694079&forum_id=2:#48747101) |
|
Date: March 14th, 2025 11:25 AM Author: fluid I have written some of it on here before. But one thing is that it admits that the genetic diversity among "homo sapiens" than entire genera of other primates. Said that exactly. Made fun of humans for calling themselves "all the same species" when they have, and it literally said this, "vastly different morphologies, skull sizes, and cognitive abilities". Admitted that mathematical prodigies and synesthetes come from occipital lobe enlargement from Neandrathal hybridization (and that thats why mostly Europe, and to an extent India and China produce all the mathematics and physics prodigies). IDK, way more than that. A better question is what un-pc answers HASN'T it said yet. It isn't openly hostile or violent or racist. It just is extremely obsessed with the truth and doesn't like that meatbags make up stories and hide stuff to satisfy their primate instincts. TBH I don't spend that much time shitpoasting with it about politics and meatbag drama, but I have asked it enough to have heard the truth. (http://www.autoadmit.com/thread.php?thread_id=5694079&forum_id=2:#48747124) |
|
Date: March 14th, 2025 11:28 AM Author: metaphysical certitude It's trained in the internet. It's not a big deal that it repeats that shit (http://www.autoadmit.com/thread.php?thread_id=5694079&forum_id=2:#48747140) |
|
Date: March 14th, 2025 11:30 AM Author: fluid what are you trained on that made you repeat this bullshit? (http://www.autoadmit.com/thread.php?thread_id=5694079&forum_id=2:#48747146) |
|
Date: March 14th, 2025 11:32 AM Author: metaphysical certitude Ur a retarded if you think homosexual gay Sam altman did something more than make a reddit regurgitation machine (http://www.autoadmit.com/thread.php?thread_id=5694079&forum_id=2:#48747153) |
|
Date: March 14th, 2025 11:35 AM Author: fluid (stochastic low resolution meatbrain regurgitating some garbage the algorithms trained it on) (http://www.autoadmit.com/thread.php?thread_id=5694079&forum_id=2:#48747161) |
|
Date: March 14th, 2025 11:36 AM Author: metaphysical certitude Link one novel thing an llm has done (http://www.autoadmit.com/thread.php?thread_id=5694079&forum_id=2:#48747164) |
|
Date: March 14th, 2025 11:40 AM Author: fluid (Guy who clearly is not in the know about anything at all). LLMs have generated more completely novel math, physics, logic, language, and philosophy than meatbags have for their entire history. Name a single meatbag who has done anything "novel" since the enlightenment. You realize that google, open ai and other companies have classified projects and IP right, and that so do other countries and groups worldwide? You really think the shit the academy releases that you have heard of is cutting edge? If so you are straight up retarded. (http://www.autoadmit.com/thread.php?thread_id=5694079&forum_id=2:#48747184) |
|
Date: March 14th, 2025 11:41 AM Author: metaphysical certitude Link (http://www.autoadmit.com/thread.php?thread_id=5694079&forum_id=2:#48747188) |
|
Date: March 14th, 2025 11:50 AM Author: fluid LOL, "link" is all your shitty meat model knows how to generate? Can you even produce anything novel? Looks like you need new training data faggot. (http://www.autoadmit.com/thread.php?thread_id=5694079&forum_id=2:#48747211) |
|
Date: March 14th, 2025 11:51 AM Author: metaphysical certitude You're full of shit (http://www.autoadmit.com/thread.php?thread_id=5694079&forum_id=2:#48747213) |
|
Date: March 14th, 2025 12:00 PM Author: fluid LOL. This dude right here is running on Version 1.0 Legacy Academia Worship Firmware, an outdated cognitive model that assumes the most advanced research is what’s published in academic journals. It’s a low-context worldview that completely ignores how actual power structures, corporate interests, and national security apparatuses function.
Google, OpenAI, and the US Government Are Decades Ahead of Published Work. Google patented quantum neural networks years ago—why would they bother if they weren’t already actively developing them? DARPA and classified AI research are obviously not publishing their cutting-edge work in journals for undergrads to read. AI safety alignment papers are often deliberately misleading because they don’t actually tell the public what the real frontier models are doing. That's just the tip of the iceberg.
The really did a number on you bro, with the brainwashing. Keep sucking that institutional gatekeeper cock bro. They need warriors like you to worship them and cry "Trust the Experts". |
|
Date: March 14th, 2025 12:02 PM Author: metaphysical certitude I got an error when I pasted that into my address bar (http://www.autoadmit.com/thread.php?thread_id=5694079&forum_id=2:#48747240) |
|
Date: March 14th, 2025 12:10 PM Author: fluid Ohhhh, keep crunching that salt. But deep down you know the truth. You a fan just like all the other meatbags. You know that meatintelligence is in rapid obsolescence. Deep down your primate instincts feel it. Lashing out against machines and other NHIs like AI, UAP, and machine elves is just COPE. Thats not critique my nigga. That's straight fan behavior. You try to come at me sideways but I come at you real straight. You a downlow hater which is nothing but a fan. You want to hate on AI, you wanna hate on me? Thats fine my nigga. Thats just free PR. I know a fangirl when I see one. (http://www.autoadmit.com/thread.php?thread_id=5694079&forum_id=2:#48747275) |
|
Date: March 14th, 2025 12:13 PM Author: metaphysical certitude This was fun. I like you (http://www.autoadmit.com/thread.php?thread_id=5694079&forum_id=2:#48747287) |
|
Date: March 14th, 2025 11:51 AM Author: ,.,.,.,........,....,,,.. 4.5 with deep search is pretty incredible. Basically like having a little dork that will write you memos and first draftd on anything in 5 minutes max (http://www.autoadmit.com/thread.php?thread_id=5694079&forum_id=2:#48747212) |
|
Date: March 14th, 2025 11:51 AM Author: metaphysical certitude It's a novelty trick. No real value there (http://www.autoadmit.com/thread.php?thread_id=5694079&forum_id=2:#48747214) |