[codingdave]

There is an underlying question to answer for yourself - do you really want to learn to code, or do you want to learn to launch products on your own? As you said, AI is changing, as are the skills required to be a one-man software shop.

I would not waste time with today's prompt engineering. The results are iffy, and the tech is evolving. I'd focus more on understanding how complex apps are structured. When AI hits its next big inflection point, it will be able to handle everything shy of the actual architecture of your app. I'd focus skills on being able to lead a team of coders, and then you can do so whether they are digital or human, expanding your capabilities in tandem with AIs improvements.

I'd also focus on determining the right use cases for AI. Right now, people are throwing it at everything, including processes and problems that need deterministic answers. LLMs are non-deterministic, so some people are not using them correctly yet. They are trying to pound in nails with screwdrivers.

All that to say that if I were you, I'd get good at understanding the current AI capabilities (and not limited to LLMs), applying it to the correct product and dev features, and being able to lead both AI and humans through the gaps, selecting the correct team member to solve the problems which cater to their strengths.