I think AI is just allowing everyone to speed-run the innovator's dilemma. Anyone can create a small version of anything, while big orgs will struggle to move quickly as before.
The interesting bit is going to be whether we see AI being used in maturing those small systems into big complex ones that account for the edge cases, meet all the requirements, scale as needed, etc. That's hard for humans to do, and particularly while still moving. I've not see any of this from AI yet outside of either a) very directed small changes to large complex systems, or b) plugins/extensions/etc along a well define set of rails.
I have been meaning to put up a blog ...
Essentially there's a delta between what the human does and the computer produces. In a classic compiler setting this is a known, stable quantity throughout the life-cycle of development.
However, in the world of AI coding this distance increases.
There's various barriers that have labels like "code debt" where the line can cross. There's three mitigations now. Start the lines closer together (PRD is the current en vogue method), push out the frontier of how many shits someone gives (this is the TDD agent method), try to bend the curve so it doesn't fly out so much (this is the coworker/colleague method).
Unfortunately I'm just a one-man show so the fact that I was ahead and have working models to explain this has no rewards because you know, good software is hard...
I've explained this in person at SF events (probably about 40-50 times) so much though that someone reading this might have actually heard it from me...
If that's the case, hi, here it is again.