[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / r / s / t / u / v / vg / vm / vmg / vr / vrpg / vst / w / wg] [i / ic] [r9k / s4s / vip] [cm / hm / lgbt / y] [3 / aco / adv / an / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / pw / qst / sci / soc / sp / tg / toy / trv / tv / vp / vt / wsg / wsr / x / xs] [Settings] [Search] [Mobile] [Home]
Board
Settings Mobile Home
/sci/ - Science & Math


Thread archived.
You cannot reply anymore.


[Advertise on 4chan]


File: 0ftpjcw03uqg1.jpg (100 KB, 828x1309)
100 KB
100 KB JPG
https://metr.org/
>>
File: Untitled.png (108 KB, 850x513)
108 KB
108 KB PNG
>>16935364
jesus they are so retarded man. Look at their hierarchy. They couldn't help themselves. They had to put in the peer review there. Consensus is the priority. Democracy for robots.
I am working on a name for this idea. When you see a space filled to the brim with retards it must be highly lucrative because smart people don't want to correct them because that would be injecting ideas into the google-sphere and losing profitability from them. Technical analysis is the first area I recognized this in.
>>
>>16935420
I can't tell if this is schizobabble or something important. What's wrong with consensus and peer review?
>>
>>16935423
Why would I share that idea with a golem on his google net?
Massive ick. Nobody should help this retard out either.
>>
>>16935364
>muh ai
Do I even need to say it? Guess I do.
KYS.
>>
>>16935432
>>16935426
I do wonder if either of you could understand the statement of the until-recently-open problem solved by AI, let alone the proof
>>
>>16935439
Shut up
>>
File: 1629390567744.gif (256 KB, 976x651)
256 KB
256 KB GIF
>>16935439
I'm an arithmetic geometer, dingus. I also have a tested IQ of 147. Yours is likely...what? 103? And you work as a cashier. You're too dumb to pick up even Baby Serre so you have to get the robot to do math for you. Fucking pathetic. No wonder you seethe at mathematicians.
>gif relevant
>>
>>16935443
I'm a mathematician as well, a combinatorics postdoc. This pdf was spat out at the speed of inference. You don't think this is significant?

https://epoch.ai/files/open-problems/hypergraph-ramsey-gpt-5-4-pro-solution.pdf
>>
>>16935364
Who gives a shit?
>>
>>16935468
Back to cybercrime and deliberately lying? So sad :(
>>
Anti here. That problem was so easy a highschooler could have solved it. AI is just parroting shit other people have posted.
>>
>>16935831
Not true :) otherwise you would have done it

We're doomed babbling flailing obsolete

Also retard at what point do you think the improvement will stop
>>
No way an AI met a goalpoast set for AI?

How relevant is the problem to real life? How much effort did humans actually spend trying to solve it? If humans dont understand it how can we verify that the solution is correct?
>>
>>16935904
The improvement hid a hard wall with gpt 4o. Since then improvements have been marginal and more about agentic workflows and memory than actual reasoning ability.

There are billions of dollars at stake to keep the hype train going when AI has zero commercial benefit. If you believe every benchmark you see you are absolutely retarded.
>>
>>16935905
The proof is legible and not extremely complicated - still it is at least a lemma in someones PhD thesis

>>16935906
The current round of models are the first that have started to do novel maths. 4o could not have written this proof. I really wish you were right, however you are just nattering
>>
>>16935423
>What's wrong with consensus and peer review?

I think I know what his point is, I#ve been trying to thing it out for a while.
there are two types of consensus
the first is where you both agree on something and fit the evidence to what you want to see while suppressing the rest
the second is antagonistic and combative where you reject harshly and concede only the points you don't disagree on

the first seeks consensus as its end goal over truth
the second seeks truth and consensus emerges from that

I'm not good at putting these things into words, I might be able to come up with an example if you push me
>>
>>16935446
No, what you posted here is slop. You don't even have self-consistency check built into the proofing algo. This is layer 1 that any rock viber gets in their first week. If an AI gave me some shit like this I wouldn't even read it. The only reason I even looked at it is you gave it a thumbs up.
>>
File: problem profile.jpg (115 KB, 932x893)
115 KB
115 KB JPG
>>16935831
>>16935905
I wonder what the next cope is going to be?
>>
>>16935443
I bet you wish you could sacrifice some IQ points for a crumb of pussy you loser
>>
>>16935980
Slop? It's a correct proof of a previously unsolved problem.

I was an anti too. But it's getting hard to deny. It's nice to see so many reflexive unthinking replies, knowing that you will look back at these days and feel very stupid
>>
>>16936073
Actually from reading the writeup
>https://epoch.ai/frontiermath/open-problems/ramsey-hypergraphs
it seems a bit charitable to call this "solving an unsolved problem'. They improved the lower bound of an existing inequality. They did not prove that the new lower bound is the best possible.
>>
File: images (2).jpg (27 KB, 631x486)
27 KB
27 KB JPG
>>16936075
They improved it all they way up to an open conjecture. I would have been proud of this as a PhD student.

Anyway, if AI stopped progressing today, I would understand your scepticism more. But can you point out a wall in this graph? Today maybe we could argue (due to increasing desperation on one side, I get it and used to do the same thing) if closing an open conjecture counts as solving a problem. I dont see the exponent decreasing on its abilities though
>>
File: reddit-mod-229010914.gif (53 KB, 498x426)
53 KB
53 KB GIF
>>16935443
lmao the unknown tortured genius act is watching his one advantage (claimed intelligence, clinged on to as a status symbol/identity) become overshadowed and mass produced

Pic related (how you look when you post like that)
>>
>>16936076
Time horizon is not a scientific measurement
>>
>>16936076
I'm not disputing AI overall, I'm just saying the writeup feels a bit hyperbolic and scammy to me. You're talking in the same way, like you're giving a sales pitch. Why?
>>
>>16936082
>unknown tortured genius act
Heh. He sounds like he went to uni, that means "officially known", unlike me, who took a picture of me holding Euclid in Kunar, Afghanistan.
>>16936076
>a wall in this graph
If thats the AI then thats good for you, all usual/repetative ways to "become a Mathematician" will get autocompleted.
>>
>>16936076
>I would have been proud of this as a PhD student
Would*. You got what is called "grandfathered in".
>>
>>16936113
?
>>
>>16935423
>What's wrong with consensus and peer review?
depending on how slippery your slope you might establish rules and boundaries from peers that actually have no basis in science

ie if youre trying to push AI to break barriers and solve novel problems, humans who are too stupid to think past their old ways will say certain things are flawed when its only their opinion and not fact

peer review makes sense for validation, but not exactly training
>>
>>16936125
its a bot thats been on /sci/ since at least 2020, reporting it does nothing, probably government related
>>
>>16936125
That your "proof of Mathematical competence" has largely been made obsolete but you can remain in the position anyway.

Mid last century all sorts of technical fields from the military were allowed to work in the civilian sector without any degrees because they were all trained internally, bypassing the universities. They were no longer in the military but could still work those jobs...like you, hueman...
>>
>>16936129
I was browsing /b/ around 2011, but I didnt really know the name of the website. It was just a place that people would sometimes post the coordinates/hints towards of the remains of an unsolved murder victim.
>>
>>16936112
>good for you
For the field I should say. "As a Mathematician/Psychologist/whatever." taken seriously becauase they have distinction.
>>
>>16936073
This isn't reflexive or unthinking. You don't understand AI and the proof writer certainly doesn't.
>>
if epoch ai is anywhere near as reputable as the epoch times, that would be a compliment.
>>
>>16935423
The truth doesn't care about democracy or majority vote. What is simply is. The majority of people disagreeing with it, or refusing to even consider it, doesn't change reality.
>>
>>16936187
>The truth doesn't care about democracy or majority vote.
No....democracy or majority vote doesnt care about truth. Remember Corona?

Their truth is backed up by a paramilitary morality police...what backs up your truth? Its "the true-truth", isnt it?....pfft.
>>
>>16936187
Sure, this makes sense. But anyway the consensus stuff was a tangent initiated by the redditor

>>16936190
For your own health and wellbeing its probably past the time to move on from covid (which was a nothingburger, unlike AI, which is a present concern, unlike covid)
>>
>>16935364
can someone link me directly to the problem they solved?
>>
>>16936278
Nope.
>>
>>16936278
see >>16936075, and the article is >>16935446
>>
>>16935420
Yes basically.
>>
>>16936127
It's just gatekeepers man.
It makes no sense otherwise. It's to keep people out and it's the only function it will ever serve, because that is literally it's purpose.
You don't need to censor any retard tier ideas ever.
Now, that said, just literally start your own journal and publish.

However, this is unlikely to work because of the above.
>>
>>16936198
>For your own health and wellbeing its probably past the time to move on from covid (which was a nothingburger, unlike AI, which is a present concern, unlike covid)
You are a reminder of God'a ineffectual design, revealing himself to be a fallible creator.
>>
>>16935423
you dont generate anything new with consensus and peer review slop.
>>
>>16936286
>Cult of Passion
Go back to your home planet space coof.
>>
>>16936492
You are not wanted anywhere on this site. Please leave.
>>
>>16935446
>This pdf was spat out at the speed of inference. You don't think this is significant?
I highly do, was heavy in Combinatorics too. A "loose ends contraption" to clear of rats house with.

>>16936499
What was your major in Mathematics/CompSci+AI? My approach was from Computer Architecture to Programming, and Transcendental Number Theory to Hexatrigesimal to communicate between computers (resolutions of monitors/dial-up) and ancient sites like the pyramids of Giza and number codes in the Bible.
>>
>>16936506
>+AI
So I specifically took a branch I knew AIs couldnt really do, or atleast are not optimized for, focused my works there. Why "race technology" like a luddite, use it as a tool to make something better.



[Advertise on 4chan]

Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.