As someone who learned programming the hard way I hated it even when I am really good at it. It got on the way of me doing useful products and services that I wanted to create.

Because of that I learned Lisp so I could do metaprogramming, because manually it would take multiple humans lives to be able to create what I want before I die, even (or specially) controlling a small group of people.

We use Claude code and personally I love it. It is like the electric sawmill instead of humans cutting manually, sweating and being exhausted after half an hour of work.

After decades programming I know how to tell the AI what I want, and how to control/log/assert/test/atomize everything so it behaves.

You can use AI to teach you programming, the problem is that you need to tell her what you want, and if you are not experience you don't.

So do small projects and let the AI do 80% of the work, spend the remainder 20% finishing by hand. Usually LLMs are amazing for approaching valid solutions but are really bad at making something perfect. They fix something and destroy something else. You do that manually.

But "learning programming" in abstract is like "learning to drill", why do you want to drill? What do you want to drill? Where do you want to drill?

You need to make it specific into specific projects, with specific timelines. "I want to play the piano" is abstract. "I want to play this version of Feather theme in 3 months" is specific.