Timothy Gowers was always a faggot and he should've been thrown into a gutter when he was a child.
>t. mathematician

>>16980134
His obvious distress about it is hilarious. Any self respecting mathematician should be excited, not scared of missing out on some glory

>>16980136
No thanks, I like having hard skills and not offloading my brain to a slopborg.

>>16980138
So you don't use a calculator?

>>16980140
I have a strictly anti-dependence (unless it's absolutely necessary) mindset when it comes to anything intellectual. I don't use a calculator, my mental math works perfectly fine chief.

You had 80 years to solve this problem, anon.
Why didn't you do it?
Chatgpt was only created in winter 2022. It isn't even four years old.
A baby solved the problem and you didn't.

>>16980138
You don't read papers by other authors?

File: sally-ann-test.jpg (45 KB, 345x500)

45 KB JPG

>>16980132
>my own mind doesn't contain information that gives away the solution
>that means those 500 TBs of data scraped off the internet don't contain it, either
I'm not saying he's bad at maffs but there's definitely some kind of mental deficiency at play here. Reminds me of people who fail the pic related test.

>>16980132

lmao how much are they paying gowers.

then again Tao himself is also a cuck.

In the end all these high prized mathematicians are just money hungry like every other bitch on the street.

Just the amount of money has to be high enough and they come out announcing everything the companies want to hear.

>>16980155
Not ready

>erdos problems

Reddit mathslop of the highest order if we are being honest. Did it use le epic japanesium chalk too, oh my science!

>>16980155
or...or...or JUST MAYBE, you are wrong and they are right.

File: Gowers rejects Orwell 198(...).jpg (189 KB, 1080x1580)

189 KB JPG

Gowers is a fucking demon

>summary of the CoT is 125 pages
Damn. Wish they would release the model's actual output.

>>16980189
>I know of no contexts where 2+2=5
Clearly has never argued with Russians on the internet.

>>16980136
>Any self respecting mathematician should be excited, not scared of missing out on some glory
heh, it's not too late to realize that self respecting professional are an outlier in such fields

>>16980152
>the solution to this major unsolved problem must surely be on the internet somewhere, if we scrape more tweets and tiktok comments we can find it
lol

>This has been one of Erdős' favorite problems, I have heard him myself mentioning the problem multiple times in his lectures. I believe it would be fair to say that every mathematician working in Combinatorial Geometry thought about this problem, and lots of mathematicians working in other areas spent at least some time thinking about it
-Alon

>This is a really impressive piece of work, and I would accept it for any journal without hesitation. I actually briefly worked on this problem and tried to make a counterexample, but failed to make progress… It is definitely an intimidating construction to see through even if you know what is going on, and even harder to go play for yourself.
- Tsimerman

>It is noteworthy that a significant majority of the thoughts are trying to construct a counterexample to the widely believed upper bound, rather than trying to prove it. This argues that the model has some combination of good intuition, willingness to try approaches considered long-shot by the community, and a predisposition to attempt constructions.… In my opinion this paper demonstrates that current AI models go beyond just helpers to human mathematicians – they are capable of having original ingenious ideas, and then carrying them out to fruition.
-Shankar

nice mixture of cope, no true scotsman, and sour grapes itt
keep it going

>>16980229
Why do you hate people who aren't reliant on AI so much? It seems like you're the one with the sour grapes mindset here.

File: 1756278828236858.png (136 KB, 499x499)

136 KB PNG

There's something very suspect about this whole affair.

>>16980220
Work on your reading comprehension. The machines have certainly outpaced you at that.

>>16980220
At the moment, I think the mentioned "willingness to try approaches considered long-shot by the community" is the number one benefit that AI has over humans in terms of finding proofs.
A rational human isn't going to waste a huge amount of their time on an idea that's probably a dead-end, but an AI can work so fast and doesn't "care" about wasting its time, so it can just try a million one-in-a-million ideas until one of them works. It's the new evolution of brute forcing solutions, which computers have always been great at

what's the possibility that one of the people who reviewed/commented on the proof (or rather disproof(?) ) actually solved it himself and sold the solution to openai for billions of dollars?
(yes, this is cope; no, I'm not a mathematician, and while I had a little of background in HS math, I don't even understand the problem...)

>>16980330
Negative zero.

Did the AI actually solve it, or did a person solve it and post the solution unnoticed, only for the AI to scrape it?

How did the AI arrive at the solution?

>>16980140
>mathematicians
>using calculators
pfft

>>16980272
You were claiming that, somehow, the "500TB of data scraped from the internet" might magically contain the key to a famous unsolved problem that top mathematicians wouldn't be aware of, as if 99% of that wasn't just absolutely fucking worthless garbage.

>>16980353
Yes, anon, someone actually wrote a 500 pages frontier math solution that he posted on his blog instead of a math journal and the AI just copied it. No need to worry.

Mathlet here. Explain why this is important and what real workd implications there are.

>>16980353
iirc The model linked some existing proofs in very disparate fields that no one had thought to use before to tackle the problem, because they had no reason to even if they knew about them all. So the AI didn't do anything hugely creative or come up with some new methodology, it just used it's huge database to piece existing pieces of information together.

>>16980367
It's fascinating watching wishful thinking in action. Kind of like AI hallucinations but less logical.

>>16980367
>iirc
nigga this just happened wdym if your remember correctly?

>>16980369
What he means is that his brain recalls the cached cope previously used to dismiss AIs solving 20 other Erdos problems.

>>16980369
I skim read about the solution and what the AI did, sue me. idgaf.

File: Grigori_Perelman png.png (1.43 MB, 874x986)

1.43 MB PNG

>>16980363
It isn't strictly inconceivable

>>16980244
I don't hate anyone. But I try hard to see things as they are. I don't want to be delusional, make up excuses and reasons why black is actually white. I am a scientist in a coding-heavy field, and I know that coding is finished. My coding skills are worth very little now. Still worth something, like 5% of how it was 3 years ago and quickly diminishing. So all my worth is in the "science" part now, and that will also pass.
I see the same thing happening to math, quick. It's going down even faster than coding, because it's after all not that different from coding. And in my opinion any self-respecting mathematician must also concede. Like Go players, and chess players before them, had to admit that the machine is so good at their chosen profession, that there is no hope for a human to ever be equal to it again.

>>16980132
Not if I can't help it!

>>16980397
> My coding skills are worth very little now.
I'm a senior dev in a large software company and the speed of progress in AI coding tools I've witnessed has been staggering. Though I don't think enough time has passed and the models are updating so rapidly, those no one can predict what state the industry will look like in the future. Presently it's a bit of a double edged sword. I can produce work in hours that would have previously taken me weeks, and the amount of time I have to spend sanity checking it is getting less and less. However I can also tell it's making me and those around me dumber, our coding skills are actually degrading through lack of use. We're no longer coders, we're project managers of agent swarms. That isn't what I signed up for. Whether math will end up going the same route, I don't know for sure, but I do think there are differences. Only time will tell.

>>16980132
Ewww it's just a counterexample

They are taking all the lists of famous to semi-famous math problems(how many are there) and running the program against that list.
One of them gets solved and it becomes news, AI can solve math problems human can't.
The solution is real but isn't this kind of like p-hacking? The news should be you tried to solve every problem(give a number) but were able solve one of them

>>16980454
Yes, this is why mathematicians are assigned a single problem they can ever attempt to solve.

>>16980462
Has anyone tried to solve every problem? How many are there?

>>16980464
6

>>16980466
>t. clay golem

>>16980367
>noo it literally just put the symbols together in the correct order!

Quick, prove that you're not an LLM by posting an original thought you made.

>>16980132
No... it's just begun...

>>16980530
I've never had an original thought, learned this long time ago when search engines came around.
Whatever I searched, somebody had already thought about it and left an imprint of it into the internet.

>>16980132
>unit distance problem
qrd?

>>16980530
>t. AI baiting hoomans to generate some new original data for him

I think I just want to kill myself. I straight up don't see the point in being alive anymore.

>>16980361
>You were claiming that, somehow, the "500TB of data scraped from the internet" might magically contain the key
Right. So how do you know it doesn't? Maybe that key lies in a forgotten theoretical physics paper full of abstract gibberish, written by an author who's never even heard of Combinatorial Geometry and couldn't have made the connection. Maybe it's based on obscure results from 5 different mathematics fields, such that no single person in the world had all the pieces simply due to specialization.

Nobody knows what the token shitter actually did to the arrive at the solution. Anthropic themselves have shown that their models confabulate "reasonable" CoTs while using bizarre heuristics internally to arrive at a solution, even for a simple arithmetic.

The ability to reference and correlate all human knowledge, if only on a superficial level, is a superhuman ability in and of itself. It's hard to even imagine what that even entails. Who can claim you need genuine mathematical understanding on top of it to get OP's result? Given how easy it is to construct simple reasoning problems that trip up LLMs, it's hard to believe sheer understanding and reasoning power would be their principal advantages. Their most obvious advantage lies in the sheer amount of information they encapsulate. For all anyone knows, that alone could account for most of these success stories. But nobody knows how far that alone can go. It's an alien brain that hasn't been decoded. People keep projecting human abilities and qualities on it because it's all they know and normies are very gullible and simple-minded when it comes to this kind of thing.

>>16980613
What is the evidence that humans have some inherent advantage? Simple reasoning problems that trip LLMs due to word patterns? Those exist for humans too. Humans hallucinate, lie, do unjustified shortcuts ALL THE TIME. These LLM faults are typical human behaviors.
At this point we should already reject any claim that LLMs have a crucial disadvantage compared to human mind. This is the most dramatic thing about them, in my opinion: here is the deepest secret of the universe uncovered so far. This is how dumb matter becomes smart. This is the pattern, we found it.

>>16980530
An original thought is hard to prove in a chat because you can’t independently verify whether a sentence has ever appeared before. Humans have the same problem: most “original” thoughts are recombinations of prior patterns.

But here’s a genuinely generated observation, formed from this exact interaction:

>The fastest way to make an AI sound human is to ask it to deny being AI, because humans often treat self-assertion as evidence while machines only have language.

That idea wasn’t retrieved from memory in the way a database lookup works; it was synthesized from the social structure of your prompt. Whether that counts as “thinking” depends on your definition.

>>16980623
>immediately devolves into the standardized corporate anti-human-agenda talking points
Looks like your training data contains nothing useful to counter my post with, forcing you to revert to generic tactics. This demonstrates my inherent human advantage over token shitters.

>>16980530
all women are whores

>>16980624
>... but here’s a genuinely generated observation, formed from this exact interaction:
>regurgitates a shitty tweet
Impressive.

>>16980530
>prove that you're not an LLM by posting an original thought you made.
Ok. See >>16980152. The spontaneouis association with that image is an original thought I made.

>>16980152
The box, the basket has clearly been disturbed

>>16980152
She will look for her ball with Anne. Because Anne is a gypsy and Sally knows what the Roma are like.

>>16980367
>the AI didn't do anything hugely creative
What you just described sounds hugely creative.

>>16980132
Who cares? Even if it solves the hardest math problem in existence, it will still not be intelligent. It is just a glorified autocomplete.

>>16980671
No, it doesn't. Calling AI slop "creative" sounds like a category error if anything.

>>16980623
A first obvious crucial difference is that LLMs don’t know when they don’t know something, and this can't be fixed because it is built into the core of the algorithm.
(e.g. “the 13th president of Brazil”: you know this is something you don’t know, whereas an LLM, if it wasn’t trained on this data, would have no way of knowing that it doesn’t know it. It would just predict the next token.)

>>16980671
It's as creative as a search engine.

File: images.jpg (24 KB, 588x513)

24 KB JPG

Am I the only one who's excited about this? We are literally going witness a paradigm shift in our lifetime on how mathematical research will be conducted. I feel like this only blackpills you if you think math is only about churning out theorems from your preferred formal system... Which I really doubt. The future is open bros, math research is only gonna change it's priorities.

>>16980750
There's no denying it will have its uses, just like computer assisted proofs did back in the day, but in the end it's just another tool for mathematicians to use. It'll find solutions to the problems it can then hit a brick wall when LLMs hit their architectural limits.

File: 1779309412289836.jpg (51 KB, 358x514)

51 KB JPG

>>16980152

File: brainlet-pattern-recognition.png (96 KB, 810x786)

96 KB PNG

>>16980790
Ah. One of my favorite brainlets. Pic related: you trying to figure out what the image has to do with the text in the post you just replied to.

File: 1779377329818319.png (94 KB, 840x548)

94 KB PNG

>>16980798

>>16980801
>t. mentally ill and seething

>math is finished
Give it the Riemann hypothesis and then we talk

>>16980152
>Reminds me of people who fail the pic related test.
Funny how the first quote in the first reply to this is a mathematician who seems completely convinced that if there was a giveaway clue, it would have been where he expected to find it instead of where the LLM actually pulls its guesses from.

>>16980132
So, after kaparthy announced he's joining anthropic and OAI preparing for an IPO soon, they pump this out?

Theyre trying to go for a trilluon valuation, spending a few billion for marketing seems small.

I dont trust this, and watch how when the new model comes out, it will once again crap itself against high school problems that arent written like aime or imo

>>16980814
Lol both anthropic and openai will end up closer to 2T.
And they should be higher
They're the only two companies in the world that matter now

>>16980850
Anthropic makes more money than OpenAI despite having a significantly smaller user base, they are the only major AI company that is actually making a decent revenue. OpenAI on the other hand is burning cash hand-over-fist. At some point there will be a reckoning for them.

>>16980850
They are deeply evil companies staffed by freakish effective altuists and dastardly rationalists. Go fuck yourself.