This is a great use case of LLM.

This is impressive. The real game-changer will be when AI creates an entirely new, significant branch of mathematics.

Which only it understands?

I would argue we're crossed that point recently. This is a comment from twitter that I appreciated:

> "Already, there are very few mathematicians qualified to verify OpenAI’s new results. As progress continues, that number will approach zero."

Not that I'm good enough at math to have any uniquely formed opinion, but after reading commentary from people who are, my impression is that these new results are bamboozling the humans due to using tools from so many disparate areas.

At least, we can say that there isn't a single human who is smart enough to understand all ten proofs, even if there is a collective sense in which all proofs are understood.

The "power shell" can only run a single command and then stuck.

Looks like authored by AI. With quite some reasoning, but no real data to back up main point.