Using AI can lead to a sort of logical booby trap.

People ask AI for answers when they lack knowledge.

But knowledge is required to discern if the AI answer is just BS --- which is not uncommon.

So in a way, AI is often a case of the blind leading the blind using nicely articulated BS.

The logical approach is to verify everything but this can take a lot of time. Doing otherwise is where the trap comes into play.

You nailed the core of the de-skilling and cognitive offloading debate well. And, therefore, their suggestion about more prompting is just band-aid that may actually increase the vicious cycle.