Maybe - humans have agency, they understand actions / consequences.

AI agents do not have agency(!), they have no understanding of consequences. They actually have no understanding. At all.