The end of programming as we know it: https://www.oreilly.com/radar/the-end-of-programming-as-we-know-it/

Feel like a point that is missing is the loss of expertise. Users of intepreted languages (mostly) don't understand the what their code boils down to, and so solutions are generally scuffed.

Hardware performance gains allowed this. Without those gains, much of what is done today where I work wouldn't be feasible.

I think many of the efficiency gains of AI for programming might not materialise if AI produces even worse performing code and hardware doesn't continue to speed up.

The loss of expertise will become a problem. Or maybe its a transfer, from underdtanding how the computer works to understanding how AI works - people building solutions they don't understand.