> and also risk over-relying on the AI's trial-and-error instead of developing a solid intuition of its functioning yourself.

Not something everyone values unfortunately.