A new paper from Google DeepMind argues that disagreements over whether AI is conscious could spill into politics, law, and public institutions.