In my experience, the generalist end up running things, and the specialists end up forever ICs or on narrow specialized projects.
The one area generalist tend to fall down is ending up as the fix-it person. Moving from one fire to the next, building up a lot of system specific knowledge that isn't transferable. Don't end up the fix-it person.
Find the highest value capability and build it simply. Then do that again. The simply part is super important if you're always interested in 10 things at once. And always trying to code yourself out of a job.
If you want to do AI, just do a couple side projects with it.
I found myself in a similar spot recently. I'm making more money than I need as an exec of a VC-backed startup, and the work needed from me in the current company doesn't fill my time.
I've started pouring myself into charity work and mentoring younger executives, engineers, and data scientists. It's different, but I love it. It's really fulfilling.