Some people think about AI in almost purely mathematical terms. For them, alignment sounds like a formula that can be solved, after which the system becomes fully reliable. Other people are much closer to practice: policy, competition, institutions.