From wantons to moral agents

Michele Campolo, per the AI Alignment Forum, explores how artificial systems might transition from being 'wantons'—philosopher Harry Frankfurt's term for beings driven purely by immediate impulses—to genuine moral agents capable of reflecting on their own reasoning. Using a framework from Frankfurt's 1971 philosophy paper, Campolo examines what separates first-order desires that immediately move us to act from second-order reflection that lets us endorse or reject those desires. In both animals and AI systems, he argues, reasoning gradually builds the capacity to evaluate what's truly worth doing, moving agents beyond mere impulse-following. This exploration, Campolo suggests, could illuminate how to design AI systems that reliably pursue good outcomes.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton