Your theory is wrong.

Someone will inevitably have to prompt AIs, CEOs and other executives are NOT going to be doing it themselves. The people driving those AI will have greater leverage as less and less people choose a career in tech.

Also, when an AI fucks up in a way only a human can fix, the human must be available.

What I see more likely is a future where software engineers do even less work but frustratingly you still need them around to fix problems whenever they come up. Kind of like firefighters.

Agreed, and the AI wranglers will be the equivalent of architects and staff+ engineers today, and they will be paid handsomely. BUT! They are a pretty small fraction of the current developer workforce. The remaining junior-to-mid level engineers will have to uplevel themselves while having no opportunity for hands-on experience to do so as they get laid off in bulk.

And note, this pattern is going to repeat across the entire white collar workforce, because the same pyramid scheme holds everywhere in knowledge work.

A new equilibrium will be found, but that will be years, maybe a decade+ away? That's the period of turmoil I am concerned about.

> Also, when an AI fucks up in a way only a human can fix, the human must be available

Actually its more “when an AI fucks up in a way AI that you need a human to take the blame, the human must be available”