Post

To my knowledge they were never said to have been peer-reviewed and thus "proven".

They were dumped into the public domain without any care or consideration, it was irresponsible and designed to generate exactly the noise and furore it has done. They knew what they were doing and they didn't care.

Per ChatGPT: "What OpenAI did, according to its announcement, was publish the manuscripts in a publicly accessible GitHub repository, with procedures for revisions and citations.

These news are always to be taken with a grain of salt. LLMs by design can't do math (no way), so they end up controlling other tools and then have to anlyse the results and that is were things start to get problematic for them. I don't think there is a single model enough reliable for complex math

Things got problematic when they asserted that a computer program that cannot do math could do math.

ugh, the correction never travels as far as the announcement. original claim was everywhere, this is nowhere

Yep. It's been the same pattern, again and again, with these bold, audacious claims from the AI labs. Each and every time it proves to be far less significant than the initial press flurry suggests, but the majority of people only ever see the original story.

The fundamental problem of ai seems to be that it's impossible to know the output is correct without a human checking every part

I was sceptical when it was announced AI was solving the complex maths, now I’m even more sceptical

I think the wider problem is that even if some of these proofs are correct, nobody has any understanding of how they've been arrived at (OpenAI aren't sharing the model they used) so the *utility* of any of them is next to none. I'm no mathematician, but failing to show your working is never good.

It's a bit like Fermat's famous margin scribble, that he had a proof for his conjecture but no space to share it in. It took several hundred years before Andrew Wiles was able to provide one!