> How? I'd imagine that most typically means continuing to program by hand.

I think the use of LLMs is assumed by that statement. The point is that even experienced programmers can get poor results if they're not aware of the tech's limitations and best-practices. It doesn't mean you get poor results by default.

There is a lot of hype around the tech right now; plenty of it overblown, but a lot of it also perfectly warranted. It's not going to make you "ten times more productive" outside of maybe laying the very first building blocks on a green field; the infamous first 80% that only take 20% of the time anyway. But it does allow you to spend a lot more time designing and drafting, and a lot less time actually implementing, which, if you were spec-driven to begin with, has always been little more than a formality in the first place.

For me, the actual mental work never happened while writing code; it happened well in advance. My workflow hasn't changed that much; I'm just not the one who writes the code anymore, but I'm still very much the one who designs it.

Yes, I've seen many people become _too_ hands-off after an initial success with LLMs, and get bitten by not understanding the system.

Hirers, above, are more focused on the opposite side, though: engineers who try AI once, see a mess or hallucinations, and decide it's useless. There is some learning to figure out how to wield it.