I appreciate your nuanced position. I believe that any developer who isn't reading more code than they are writing is doing it wrong. Reading code is central to growth as a software engineer. You can argue that you'll be reading more bland code when reviewing code generated with the aid of an LLM. I still think you are learning. I've read lots of LLM generated code and I routinely learn new things. Idioms that I wasn't familiar with, or library calls I didn't know existed.

I also think that LLMs are an even more powerful accelerant for senior developers. We can prompt better because we know what exists and what to not bother trying.