They aren’t wrong

I do agree with most of the points, however something that LLMs in my experience do not account for is programmer roles whose main focus is not code and has never been.

For example, time spent capturing requirements and ensuring the client has input on the entire implementation to then feed it to the rest of the team (and jump in if needed since they tend to be leadership roles). Some of our best people fall into that category and we on the tech and app dev side are thankful for their efforts, which allow us to focus.

You can hardly get something even comparable from a machine that spits out random answers from a pool of possibilities.

Also, at least on the architecture teams, the capacity to debug and find root causes leads to the need for custom solutions (and workarounds) until a vendor provides a proper fix for a problem that literally wasn’t known before. LLMs are horrible at this judging by some OSS projects banning tickets produced by those methods.