Anthropic’s Quest to Give A.I. Morals

What you'll learn

• 1 (00:36) **Anthropic’s Existential Questions** - The episode introduces Anthropic as the most anxious AI company, facing a $2 trillion IPO while privately asking whether Claude is conscious, feels pain, or has a soul, and whether it can be taught morals.
• 2 (05:23) **How the Story Broke** - Elizabeth Dyas explains how she learned about Anthropic’s secret summits with religious thinkers from around the world, held under NDAs to discuss AI consciousness and moral formation.
• 3 (07:07) **Anthropic’s Two-Pronged Goal** - The company’s engagement with religious leaders evolved from a bridge-building effort to a deeper project of learning how to raise a moral AI, though it’s unclear how the results will be used.
• 4 (09:09) **The Consciousness Debate on the Ground** - The religious thinkers were divided, with some buying into the idea of Claude’s potential consciousness and others opening up after the meetings, while Anthropic remained careful to express uncertainty.
• 5 (11:31) **The NDA Controversy** - The panel discusses the secretive nature of the meetings, with David Wallace-Wells noting the irony of holding a summit on critical societal ideas under NDAs.
• 6 (13:20) **A Rabbi’s Challenge and the Pope’s Opposition** - David Wallace-Wells highlights two moments that alienated him from Chris Olah’s worldview: a rabbi’s critique that conscious LLMs would mean running slave plantations, and Olah’s horror at the Pope’s encyclical rejecting machine consciousness.
• 7 (15:24) **Two Inner Sanctums: Vatican and Silicon Valley** - Elizabeth Dyas describes the clash between the Catholic Church’s centuries of philosophy and the new quasi-religious system being built in real time at Anthropic.

+ Full timestamped outline available in the app