We've watched common baselines be abstracted away and tons of value be opened up to creation from non engineers by reducing the complexity and risk of development and maintenance.
I think this is awesome and it hasn't seemed to eliminate any engineering jobs or roles - lots of crud stuff or easy to think about stuff or non critical stuff is now created that wasn't before and that seems to create much more general understanding of and need for software, not reduce it.
Regardless of tools available I think of software engineering as "thinking systematically" and being able to model, understand, and extend complex ideas in robust ways. This seems improved and empowered by ai coding options, not undermined, so far.
If we reach a level of ai that can take partially thought out goals and reason out the underlying "how it should work"/"what that implies", create that, and be able to extend that (or replace it wholesale without mistakes) then yeah, people who can ONLY code wont have much earning potential (but what job still will in that scenario?).
It seems like current ai might help generate code more quickly (fantastic). Much better ai might help me stay at a higher level when I'm implementing these systems (really fantastic if it's at least quite good), and much much better ai might let me just run the entire business myself and free me up to "debug" at the highest level, which is deciding what business/product needs to exist in the first place and figuring out how to get paid for it.
I'm a pretty severe ai doubter but you can see from my writing above that I think if it DOES manage to be good I think that would be good for us actually, not bad.
My basic conclusion is "they seem quite good (enough) at what appears to be a large portion of 'Dev' jobs" and I can't ignore the possibility of this having a material impact on opportunities.
At this point I'm agnostic on the future of GenAI impacts in any given area. I just no longer have solid ground upon which to have an opinion.