I don't think there is much crossover unless your model responds well to being anthropomorphized.

I find LLM's to be very easy to manage since they at least to me, appear more rational the humans. Myself included.

I tell it "believe in yourself" and task goes from unsolvable to trivial.

It's human ;)