A solution or proof that nobody understands has no value.
It’s worthless until it’s been validated. That’s peer review. If it’s not, it’s just a claimed proof, or claimed solution. And there’s nothing that indicates that LLMs can operate with the consistent precision needed to do serious mathematics unless micro-steered by a mathematician.
Also, the hype about LLM solutions has a stench strongly reminiscent of Stephen Wolfram’s self-aggrandizing bullshit. And I’d say Wolfram has more of a claim to at least have an educational background that means he might occasionally know what he’s talking about, even if, in a broader view, he’s a charlatan.