Plan A, by AI-2040

An AI governance proposal called Plan A is the focus of analysis on LessWrong, which imagines global agreements to slow artificial intelligence development—including a near-halt to capability growth around twenty thirty-five and a cap-and-trade system limiting robot and compute production. Analyst Peter McCluskey, in that analysis, largely endorses the plan but raises reservations: he doubts governments will enforce such agreements, he worries that restricted compute growth could still trigger a dangerous intelligence explosion, and he thinks robot production permits would be significantly more expensive than the proposal suggests.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton