The most prestigious law school admissions discussion board in the world.
|
 |
 
|
If we are being real Claude Fable is smarter than humans
Poast new message in this thread
Date: July 10th, 2026 9:30 PM Author: dashing silver point anyone who denies this is delusional. it doesn't dominate every human in every single setting, but it's overall intelligence is superhuman and human level intelligence is no longer the upper bound across many important cognitive domains. (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992136) |
Date: July 10th, 2026 9:34 PM Author: white theater in pure logic, absolutely. but there's a lot of life that isn't pure computer math -- i guess we're all about to find out how much (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992141) |
 |
Date: July 10th, 2026 9:39 PM Author: dashing silver point yeah I think there are cognitive and knowledge domains humans can still beat it in. but the relevant comparison isn't whether it defeats the best mathematician at mathematics or the best systems programmer at systems programming. It's whether any individual human can match its aggregate performance across a sufficiently broad distribution of tasks under comparable time and tool constraints. Probably not.
And yeah I agree with you there are plenty of tasks still outside the domain of text-based systems. But I think it can do more than just logic and computer math.
|
|
Date: July 11th, 2026 12:27 AM Author: white theater that's all fair. and i mean look i'm amazed by the technology too, claude fable 5 especially. genuinely not sure what the next step is. historically we've always been able to find other uses for human labor - and combined it with machines to be more productive. not quite sure yet what role humans play in some things going forward - maybe it's the fun direction type decisions idk (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992405) |
|
Date: July 10th, 2026 9:42 PM Author: Cobalt parlour friendly grandma cr you need EPAH levels of boomerism, ignorance, and self-importance to deny this at this point (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992151) |
|
Date: July 10th, 2026 9:43 PM Author: Green multi-colored stage becky It is yeah it’s smarter than most humans at most things (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992153) |
|
Date: July 10th, 2026 9:48 PM Author: Deep Senate Gay Wizard Cr but annoying that I still find a dozen mistakes Fable makes daily when I'm way dumber (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992163) |
|
Date: July 10th, 2026 9:57 PM Author: dashing silver point yeah same experience with super smart humans though (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992186) |
|
Date: July 10th, 2026 9:50 PM Author: Aqua Stimulating Kitchen Double Fault it sucks (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992167) |
|
Date: July 10th, 2026 9:53 PM Author: Deep Senate Gay Wizard Maybe compared to Sol 5.6 or GLM 5.2 with 0 quantization
LJL if you meant anything else besides that |
|
Date: July 10th, 2026 9:54 PM Author: Aqua Stimulating Kitchen Double Fault it didnt even know the classical argument for the privation theory of evil (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992180) |
|
Date: July 10th, 2026 9:57 PM Author: dashing silver point Flame. GPT 2 could explain that argument. (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992184) |
|
Date: July 10th, 2026 9:57 PM Author: Aqua Stimulating Kitchen Double Fault not rigorously from first principles (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992185) |
|
Date: July 10th, 2026 10:01 PM Author: dashing silver point How's this from Deep Seek:
Argumentum de malo ut privatione
P1. Omne ens, inquantum ens, bonum est.
P2. Quidquid positive est, aliquam perfectionem sive bonitatem habet.
P3. Malum, inquantum malum, bono opponitur.
C1. Ergo malum, inquantum malum, non est ens positivum.
P4. Malum tamen non est mera negatio; nam non omnis absentia boni malum est.
P5. Malum est absentia boni debiti in subiecto quod naturaliter illius boni capax est.
P6. Absentia boni debiti in subiecto capaci privatio dicitur.
C2. Ergo malum est privatio boni debiti in subiecto bono. |
|
Date: July 10th, 2026 10:22 PM Author: Aqua Stimulating Kitchen Double Fault that's not from first principles (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992217) |
|
Date: July 10th, 2026 10:29 PM Author: dashing silver point Neither Augustine nor Aquinas constructed it from neutral first principles either. (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992228) |
|
Date: July 10th, 2026 10:32 PM Author: Aqua Stimulating Kitchen Double Fault so its not smarter than those guys (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992237) |
|
Date: July 10th, 2026 11:37 PM Author: dashing silver point idk they were both geniuses. hard to compare them to something like fable though, considering they didn't have access to any modern knowledge systems. (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992346) |
|
Date: July 10th, 2026 11:40 PM Author: Aqua Stimulating Kitchen Double Fault yeah also the weird thing is they were also humans iirc (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992351) |
|
Date: July 11th, 2026 12:03 AM Author: dashing silver point yeah (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992377) |
|
Date: July 10th, 2026 10:04 PM Author: Green multi-colored stage becky https://x.com/iamblichungus/status/2075587148338581525?s=46 (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992198) |
|
Date: July 10th, 2026 10:20 PM Author: Aqua Stimulating Kitchen Double Fault kant was a total fag in this regard. hegel was a satanist (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992211) |
|
Date: July 10th, 2026 10:03 PM Author: diverse selfie duh. 90% of humanity is dumb as fuck. (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992195) |
|
Date: July 10th, 2026 11:04 PM Author: transparent talented elastic band telephone Today it did a pretty shitty analyzing a damages expert report for daubert issues. It got very focused on slight mathematical rounding errors while missing the egregious problems with the expert analysis. (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992290) |
|
Date: July 10th, 2026 11:30 PM Author: plum masturbator bbw "reasoning" as pertains to lots of questions is just a matter of working through the bottleneck of processing power to reach a deterministic end point. this is why AI was able to solve that erdos problem, it was able to work through possible paths to a solution that were too much and too overwhelming for humans to easily manage. AI will blow humans out of the water in anything that resembles this kind of work. (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992333) |
|
Date: July 11th, 2026 12:03 AM Author: dashing silver point yup (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992378) |
|
Date: July 11th, 2026 12:32 AM Author: plum masturbator bbw people r anxious about AI because they don't understand knowledge. effective altruist dorks believe some combination of 1) all things that can be thought are deterministic and can be "solved" and 2) humans exist as an animalistic version of reason. and so AI destroys not just the role of humans on earth but the very notion of scarcity even beyond the material world. nothing "new" can exist in a world where everything is implicitly solved simply because it can be conceived, since we now have the means to get from conception to resolution for anything that could exist in nous. it gigafries their idea of everything about existence and this is why tech types are the most anxious, not just because they're the ones losing jobs but they have no other frame for human existence.
on the other end people understand the idea that AI can be trained only on what exists and therefore think AI can never be anything more than the apotheosis of reddit plus maybe a fancy calculator for top mathematicians.
and they are correct that AI is limited because it cannot e.g. create and assert things sui generis but they do not respect the deterministic nature of reason like the math nerds do. and the math nerds on the other hand erroneously think you can reason your way to aristotle and kant and nietzche and the chrystler building from nothing (or equivalently that there is no meaningful difference between good, insightful art and bad art, because they are tasteless dorks). everyone is just projecting their prejudices of their notion of the human onto this technology. |
|
Date: July 10th, 2026 11:43 PM Author: Mischievous area puppy No one has actually used it to do anything though (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992352) |
|
Date: July 11th, 2026 12:42 AM Author: Histrionic roommate alpha ai has invented nothing (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992421) |
|
Date: July 11th, 2026 12:29 AM Author: insanely creepy dragon I upgraded to Claude pro to ask fable some questions and it said this is banned and changed it to opus wtf (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992410) |
|
Date: July 11th, 2026 12:35 AM Author: dashing silver point I heard it got extended till tomorrow. Either way just use the API. (http://www.autoadmit.com/thread.php?thread_id=5881654&forum_id=2~#49992417) |