But that’s the problem.

The LLMs are still generating all that rote shit, that at some point must be maintained and read and understood. Potentially at a higher rate than any human could.

A lot of what AI coders are being praised for could be accomplished just as well with some good Lisp macros with much less technical debt.

I understand what you're saying. But in the programming environments I work in, the rote stuff that LLMs are automating is marginally better than the rote stuff I would write given the same parameters. Obviously, we could all do better if only we were writing the right Lisp macros. ;)