There’s no doubting the power and potential of artificial intelligence. But AI is not reliably moral. Indeed, its own inventors suggest that unchecked AI could lead to catastrophic consequences.
How do we control AI? There must be some mechanism that empowers human judgment and locates human responsibility—for example, requiring people who are accountable to build in guardrails. The frontier AI companies probably need to take this responsibility, overseen by an independent expert panel. Developing this oversight may require as much invention as AI itself.
But it’s not the only innovation required. The role of human judgment must also be resuscitated. We must abandon the myth that any complex legal system can achieve effective government.