Strong disagree.

How many times have you written a boring CRUD endpoint that returns a list of results?

Half the stuff I lose time on is the boring boilerplate code (despite using frameworks). Now I can focus on the fun stuff like determining an optimal algorithm or even getting the LLM to walk me through using kubectl on the terminal

Now I know some kubectl commands off by heart, which I didn't before.

- kubectl get namespaces | grep "foo"

I think you're coping. A lot of AI addicts come off like smokers who think their coworkers can't smell them.

Every time you do not write boring api code, some of your skill atrophies there. And, perfectly honestly, maybe that's OK.

I had an AI write some Vimscript for me and I lost out on that skill building. And I was fine with that.

But it is loss.

Or, in your case, do you think you'd still be building the same algorithm skills if you punted those to the AI, as well?