It’s already hard to not write buggy code, but I don’t think you will detect them by just reviewing LLM code, because detecting issues during code review is much harder than when you’re writing code.

Definitely. That’s what I was trying to drive at, but you said it well.