Thread 106330393

https://x.com/SebastienBubeck/status/1958198661139009862

Wow. Apple paper btfo. Haters thinking it was plateauing btfo. The future is here.

This nigga literally works for OpenAI, buy an ad.

assumed to be F&G

>>106330487
Fibby bippy.

>>106330393 (OP)
Fake and gay

>>106330393 (OP)
Best case scenario, there was a better bound in another paper and it correctly copied it over. BFD

Probably just F&G thoughbeit.

>>106330393 (OP)
AGI is confirmed now. A few more billions and it will be ready.

>tunnel vision researcher uploads his paper to a chatbot
>it randomly points out something of interest
*taps chomsky*

>>106331045
>altman is a jew
>chudsky is a jew

>>106330487
yeah he's the "sparks of AGI" from gpt-4 guy

>>106331070
i don't think Chumsky fucks kids or would genocide if that means anything

>>106331025
I think there was a better bound in another paper, but it used a different proof. GPT supposedly came up with a novel proof.

>>106331197
Jews are everywhere, and I'm getting tired.

>>106330393 (OP)
>It made up a mathematically incorrect statement
It’s not smart, can’t do math, don’t create another number on top of the real one.

>>106330393 (OP)
>The future is here.

Buy an ad, c0cksucker

>>106331070
chomsky is the least jewish person alive which you'd know if you'd even bothered to look into it briefly

>>106330393 (OP)
3/2L is larger than 1/L dipshit A.I. yo momma wears size 55" waist, but she can also wear a 75" waist. Don't mean she skinnier.

>>106332163
Not all analysis is looking for tighter bounds. For instance, I wasn't sure your mom could take my cock, but to both of our surprise she could go 1.5 times bigger than your father.

>>106331045
how is the human mind not one big plagiarism machine

>>106332231
Some of us have original ideas because we're not all bug people and parasites.

>>106332044
he's a jew

>>106332288
wow what an original post

>>106332305
>NOU

>>106330393 (OP)
I don't see the question, original proof, or chatgpt's proof, but, unless I don't understand this, the original proof bounds eta by 1/L, while the chatgpt's proof bounds it by 3/2 * 1/L. How is chatgpt's bound better?

>>106333269
It is the interval on which the function f is convex, so a bigger interval is better. If you check the paper, the original authors have had a bound better than 1/L for a few months, so this isn't a new result as far as I can tell.

>can advance mathmatics
>I ask it a simple coding question and it gets it wrong
monkey with a typewriter

>>106333313
Do you have a link to the paper?

>>106333343
https://arxiv.org/abs/2503.10138

>>106330393 (OP)
>Still can't count the letters in a word
yehnah. someone will find where it found this "novel" proof.. just like the distilling/self training stuff which turned out to be hocus singularity bullshit too.

>>106331045
>modifying them enough to escape copyright laws
KEK he is such a fucking kike. Why is he pushing that narrative. They don't "escape copyright" at all.

no clue about advancing mathematics, but realistically that's 99.9th percentile hooman intelligence tier.
i'm not at that level but just applying known methods to maths problems is incredibly useful.
you probably need a 90th percentile iq to pass complex analysis and these reasoning llms just breeze through it.

>>106331045
chomsky is a fucking rat. playing the good cop for israel so browns and lefties can pretend israel is bombing children for 'american imperialism' or oil

>>106330393 (OP)
If true Zeilberger is shitting his diaper at how people are using computers to write natural language proofs.
I skimmed it and it doesn't look clear to me.
Also it's a nothing burger as the bound was already larger in a v2 paper.
Seems like he is just trying to kill interest so people ignore the claim and proof itself.

>>106330393 (OP)
how do these people live with themselves on a daily basis? like, you clearly lying and misleading people. is having moral values really just for faggots?

>>106333622
100s of millions of dollars help, probably

Well?

Can it do fifth grade math problems or even simple addition yet?

>>106330393 (OP)
iTODDLERS BTFO
>>106330487
>>106331024
>>106331025
>>106331082
BTFO iTODDLERS

>>106333650
why would i credit ai in my software? how does he know it isn't being used in everything to some extent now? even hard programmers will get stumped and throw a function in a text generative ai as a desperate attempt to save some hairs on their head

>>106333991
There is no observed leap in software quality. So the 10x claim must be false.

>>106330487
even if he works for openai its still a good thing if this is true though right? i cant read math though so idk if its true

>>106333611
That is a good thing. Not only Israel should be hated, but America too. Especially America, since Israel is literally powerless without America (Putin would help Iran crush it)

>>106331045
>lets have anarcho syndicalism! fuck the guv, fuck authority, we should freely exchange information in groups, here's my literal email, i WILL respond even if you're a nobody
>oh yeah btw copyright law is totally legitimate and stealing is bad mkay

>>106330393 (OP)
why only one example? an AI can effortlessly generate infinite text. the grifting is unreal.

>>106333611
He was asked once what he thought about 9/11
His words were to the effect "who cares"

>>106330393 (OP)
Ok now make it cook me dinner or do something important and relevant to people's actual lives.

>>106335118
>i cant read math though so idk if its true
Don't worry, the priests will listen to the divinations and tell you what God is saying. All you have to do is listen.

>>106332231
Where did the things you believe it plagiarises come from, anon?

>>106333601
You're not saying anything, you demagogue rat.

>>106332309
Produce an original idea, now

>>106330393 (OP)
ChatGPT can’t even parse a list of clearly defined values from a website without completely making up data that isn’t there. I’ve been trying to use it to do some data processing for me and every single time it’s given me a list with some kind of inaccuracy in it. It’s completely useless at any of the tasks that you’d think it would be useful for.

>>106335998
I'm thinking about moving and asked about rent prices, for one hour it gave me nonsense links. Every time it got more and more sure that the links are correct this time.

>>106336087
It always assumes your corrections are truth because ChatGPT intentionally flatters users as much as possible to keep them engaged.

>>106333611
Jesus Christ calm down. Nobody's talking about some perpetual middle-eastern conflict. We are talking about what AI is and isn't.

>>106335998
>>106336087
It is a random text generator. It has no internal concept of truth. Why would you try to use it in a way that the output matters, are you retarded.

>>106335285
I hate copyright law but it's bullshit that Sam Pedoman and all of his pedo billionaire friends get to break it and I get my YouTube account suspended for using licensed music in a video.