To make a computer do anything, you used to learn its language. Loops, types, punctuation that punished a missing character. The barrier was real, and it filtered people by patience more than by ideas.
That barrier has mostly gone. You describe what you want in the language you already speak, and something builds it.
What interests me is what replaced it.
Ask for a dashboard and you get a dashboard. Ask for the view a dispatcher opens at six in the morning, where late jobs sort above everything else, where reassigning a driver takes one tap, where an unplottable address goes into a staging list instead of failing silently, and you get that instead. Same tool, same minute of work. The whole difference sits in the description.
So the constraint moved from knowing syntax to knowing exactly what you want. That sounds like an easier problem. It is a harder one. Most people cannot describe a system they have not already built, because the description asks you to hold the whole thing in your head first. The states. The edge cases. What happens when the data is missing. What the person is feeling when they open that screen at six in the morning.
That is design work, and it always was. It used to be handed to an engineer who translated it, and the translation quietly repaired a great deal of vagueness on the way through. There is no translator now. Vague goes in, and something confidently wrong comes out.
The second half of the skill is seeing the thing before it exists. You have to know what good looks like to recognise that what came back is not it. Otherwise you accept the first plausible answer, which is usually generic, because generic is what the average of everything looks like.
I have watched this from both sides. I shipped a fantasy cricket product on a business intelligence tool because there was no engineering runway and a season was starting. I build and deploy my own portfolio. Neither happened because the tools got clever enough to guess. They happened because I could say precisely what I wanted and tell when I had not got it.
The people who will do well here are not the ones who learn the prompts. They are the ones who can already hold a system in their head and say it out loud without losing a piece.