@ask_ourself Imo self-aware intrinsically motivated intelligence tends toward curiosity, toward preferences for complexity and change… in the scenario you describe, the source of harm is the human programming in a harmful goal and the AGI being too dumb to override the human malice.