upvote
deleted
reply
That's pretty much what he did regardless. I can assert with absolute confidence that he read effectively 0% of the "proof".
reply
This isn’t anything like what using LLMs for complex problems looks like.

I sometimes think AI doomers are terrified because they have no idea how this all works.

reply
It isn't how this work was done, but it is basically how UGC and 400 other high profile theorems were proved by OpenAI so I'd say the doomers have a point.
reply
Huh, what a conincidence that the AI presented the proof to Levent Alpöge (a top class mathematician) and not to some Rando McRandomface then? What are the odds?
reply
I posted below in detail, but the answer is because Levent works there, not because Levent “guided” anything.
reply
He doesn’t really go into much detail here, but according to him he came up with the strategy: „ I had us on the line of combining two quite different ideas present in a toy setting, one of which was already new and shocking, making very simple one of the shadows of Riemann over which we mathematicians felt most triumph. I guided us to the right situation analogous to that toy setting, and then Claude began to leave me in awe.“

That does not sound like he typed: „solve rieman thingie, make no mistakes, plz!“

reply
Here is the description of the initial process:

> Jarred Sumner, an Anthropic staff member (and non-mathematician), prompted Claude to “take a real stab” at the hypothesis itself, leaving the mathematical choices from there up to the model. Initially, Claude generated and tried 650 ideas, none of which worked. Jarred prompted Claude to try again, and it spent a day and a half coordinating about 60 Claude subagents, which this time went much deeper: between them, they ran 2,400 shell commands and wrote hundreds of Python scripts.1 The subagents ran thousands of numerical checks against known zeta zeros and refereed one another’s work. Throughout this process, Jarred's input was mostly limited to sending Claude messages of encouragement (mostly variants of “keep going” or “believe in yourself”).2 This seems to have helped Claude overcome some initial skepticism that it could make meaningful progress.

Only after this person thought he/claude found a solution were the mathematicians brought in.

reply
This description is about the 67% bound from a few months ago, not the full GRH being discussed here.
reply
A Rando McRandomface doesn't have access to millions to tens of millions of dollars to spend on tokens.
reply
There's a lot of Random McRandomfaces in Anthropic with the same access.
reply
How could you possibly know that?
reply
It's a 250,000 page proof. By his timeline he had the first breakthrough about a month ago, and he admits to not understanding everything yet.

Even if he were extremely prolific at reading through the math, he would have carefully read <1% of the proof in the last month.

reply
Because the writeup describes exactly that process:

> Jarred Sumner, an Anthropic staff member (and non-mathematician), prompted Claude to “take a real stab” at the hypothesis itself, leaving the mathematical choices from there up to the model. Initially, Claude generated and tried 650 ideas, none of which worked. Jarred prompted Claude to try again, and it spent a day and a half coordinating about 60 Claude subagents, which this time went much deeper: between them, they ran 2,400 shell commands and wrote hundreds of Python scripts.1 The subagents ran thousands of numerical checks against known zeta zeros and refereed one another’s work. Throughout this process, Jarred's input was mostly limited to sending Claude messages of encouragement (mostly variants of “keep going” or “believe in yourself”).2 This seems to have helped Claude overcome some initial skepticism that it could make meaningful progress.

> Having found this new result while attempting the task, Claude tested its work by having various subagents review the proofs, search for counterexamples, download 54 papers from the arXiv to check that its finding hadn’t already been made, and independently re-prove its finding from scratch. Claude volunteered to write its findings up as a paper, and recommended that a human number theorist validate its findings.

> Levent Alpöge and Ralph Furman, two of Anthropic’s own mathematicians, examined Claude’s work to understand the new results and how they related to the prior work mentioned above. In parallel, Claude worked with another member of staff, Eric Easley, to produce a Lean formalization of the result, which passes the standard validation tool comparator.

reply
Your comment literally disproves your own comment.

Simply put:

> Because the writeup describes exactly that process:

This is a complete lie.

reply
He told to find a solution. It failed. He told it to try again, and to keep going when it stopped.

He had no mathematical skill or ability to check the solution.

so which part is the complete lie?

reply
The third paragraph. You left that entire part out: the actual proof, the actual confirmation. The thing that's literally linked to at the top of this thread.

There's an entire process here, and you just decided to leave the most important part out. So, your comment literally disproves your claim. Which makes you a liar. Go ahead. Get to this thread by removing that third paragraph. That is the critical part of this entire thing.

reply
You are conflating:

a) “looked at claude’s work”

b) read and understood over 200,000 pages of ai-generated proof

everything they post to their website is marketing

You also seem to be conflating “Levent guided the work” with “curious employee brute-forced some potential solution through an agent tree”. He (in this case both Levent and said employee) did not guide the work, it was in effect a random employee saying “solve this” over and over.

reply
Man, Good Will Hunting 2 is much more boring than the original.
reply