Epoch confirms GPT5.4 Pro solved a frontier math open problem

Post by razorbeamz » Tue, Mar 24, 2026, 8:15 AM UTC

gf000 wrote: post #247

razorbeamz wrote: post #154

Earlier quoted context omitted.

Next-token-prediction cannot do calculations. That is fundamental. It can produce outputs that resemble calculations. It can prompt an agent to input some numbers into a separate program that will do calculations for it and then return them as a prompt. Neither of these are calculations.

So you don't think 50T parameter neural networks can encode the logic for adding two n-bit integers for reasonably sized integers? That would be pretty sad.

Post by qsera » Tue, Mar 24, 2026, 8:17 AM UTC

timschmidt wrote: post #299

qsera wrote: post #294

Earlier quoted context omitted.

Because economy. Look at marvel movies, do you think the latest one is really new? Or just a rehash of what they found working commercially? Look at all the AI generated blog posts that is flooding the internet.. LLMs might produce something new once in a long while due to blind luck, but if it can generate something that pushes the right buttons (aka not really creative) to majority of population, then that is what…

That's literally all culture: https://www.youtube.com/watch?v=nJPERZDfyWc

Post by famouswaffles » Tue, Mar 24, 2026, 8:21 AM UTC

staticassertion wrote: post #295

famouswaffles wrote: post #289

Earlier quoted context omitted.

https://epochai.substack.com/p/first-ai-solution-on-frontier... >The newly-solved problem came from Will Brian, who had placed it in the Moderately Interesting category. It is a conjecture from a paper he wrote with Paul Larson in 2019. They were unable to solve it at the time, or in several attempts since. Brian had this to say.

I actually still don't see the source for them trying several times, but we can take that for granted. Regardless, as I said: 1. It's labeled as "moderately interesting" 2. They said that they expect an expert could solve it in 1-3 months 3. They had already come up with the solution that the AI had but weren't convinced it would have worked So how big was the gap here, do you think?

Post by PxldLtd » Tue, Mar 24, 2026, 8:23 AM UTC

EternalFury wrote: post #191

I am thinking there’s a large category of problems that can be solved by resampling existing proofs. It’s the kind of brute force expedition machine can attempt relentlessly where humans would go mad trying. It probably doesn’t really advance the field, but it can turn conjectures into theorems.

Post by charcircuit » Tue, Mar 24, 2026, 8:24 AM UTC

qsera wrote: post #304

charcircuit wrote: post #302

Earlier quoted context omitted.

I've found several posts on moltbook funny. I don't really like regular jokes in general and I don't find human ones particularly funny either. I don't think we are at the point of being able to be reliable funny, but it definitely seems possible from my perspective.

Care to link some?