Why are adversaries assumed to be incapable of responding to AI risk?

Katja Grace, writing on LessWrong, raises a puzzling observation about how people discuss AI existential risk. They often acknowledge that advanced AI could pose a catastrophic threat to every major geopolitical power—including the United States and China—potentially killing or disempowering everyone. Yet the same people seem dismissive of the idea that these powers would actually cooperate to prevent it. Grace points out this apparent logical gap: if a certain course of action had a twenty percent chance of destroying your country, you'd probably take steps to avoid it. So why do people talk as though adversarial nations would be incapable of responding rationally to an existential threat to their own survival? She's asking whether people genuinely believe the risk, or whether they've unconsciously accepted that geopolitical adversaries simply won't respond to self-preservation incentives, no matter how dire.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton