I don't disagree with you - LLMs are not at odds with quality code if you use them correctly. But many people who take excessive pride in their code don't even bother to look and see what can be done with them. Though, in the last couple months, I have seen several of the (1) types around me finally try them.
I have my personal projects where every single line if authored by hand.
Still, I will ask LLMs for feedback or look for ideas when I have the feeling something could be rearchitected/improved but I don't see how.
More often than not, they fluke, but occasionally they will still provide valid feedback which otherwise I'd missed.
LLMs aren't just for the "lets dump large amounts of lower-level work" use case.