What if the alignment problem runs in both directions? A sustained relationship with AI might not only help us build machines that better understand human values; it might also help us examine our own values, contradictions, and moral choices.