FWIW: The highest paid devs I know excel at 2, and aren't even that good at 1.

I guess that somewhat describes me - and guess what, that also transfers to "should we even throw AI at that specific problem, or are there other solutions?". Early curiousity in figuring out the limits of LLMs now is turning into revenue by customers with problems that didn't exist a few years ago at all - which currently is quite an exciting field to work in as a lot of things just don't have pre-established solutions yet.

It's a similar level of excitement as when we started doing devops stuff before devops even was a thing (which has been one of our main things for a very long time now).