>>108871967
It's wrong.

I can't help but notice they don't have an illustration demonstrating the proof, even though it would be trivial to generate.

>>108871967
>OpenAI link
not peer reviewed.
A salesman telling you his product is great is worthless and just reveals how guillable you are.

File: 1754923839244205.png (1.36 MB, 1045x978)

1.36 MB PNG

we already have this thread

Again, they used who knows how many trials to combine existing arguments until they hit upon a combination that wasn't tried before.
It's basically brute-forcing a solution that was not made through actual reasoning, but a simulation of reasoning. For the AI it's just a language soup. This is only meaningful because some humans took a look at it and said: wow, why didn't we think about this before.

Or they might swing to the opposite opinion after a while and consider it not valid. While the AI will keep churning language games that will make humans again puzzled.

>>108872066
If you go searching, you'll probably find a paper that has this solution in it but never got any real attention.

>>108871967
AI skeptic: "NOOOOOOOO!!!"