Comment by surgical_fire

These are not mistakes.

Those results were not dumped to advance math. Those results were dumped to generate positive press for OpenAI.

Now it's on actual mathematicians to figure out whether the proofs are bogus or not. But the mistakes will never reach the same level of public attention as the original positive press, so for OpenAI this is good anyway, consequences be damned.

In a sense this is a microcosm of AI usage in the wild, ignorance is laundered through LLMs, and it is left for those that still have knowledge to figure out what makes sense.

OpenAI is a horribly negligent company. The threat it poses to humanity is not that their models will be a superintelligent singularity that will take over the world, is that their negligence and greed has real world consequences that they really don't give a fuck about. Like right now, their shitty math bruteforce is just keeping actual specialists occupied trying to figure out what is bullshit from what is not. For free, mind you.