@atax1a I don't intend to defend it. If it's bad theorems, obviously that's a problem.

The issue appears to me that the theorems are plausibly sound but in real human world nobody actually gets rewarded for verifying work, only for creating work, And as consequence letting the AI do the theorem creation part crashes a significant portion of the self-image of purpose of the project.

It feels analogous to online advertising crashing print media advertising which really threw off the news industry because it turns out that generating and reporting news, while valuable, had become a loss leader for putting advertisements in people's hands.

It is definitely a problem that news media didn't recover from this. I'm trying to wrap my head around what the analogous look problem looks like in the mathematical community.