Amanda Askell

Dario Amodei, CEO of Anthropic, discusses the groundbreaking AI model Claude, alongside Amanda Askell and Chris Olah, both researchers at Anthropic. They dive into the ethical dimensions of AI, emphasizing responsibility in innovation and safety. The conversation also explores the intricacies of building AI personalities, the challenges of mechanistic interpretability, and the future of integrating AI into society. They discuss the delicate balance between AI capabilities and human values, positioning AI as a partner rather than a competitor.

Amanda Askell, a philosopher-turned-AI researcher at Anthropic who helped craft Claude’s values-oriented constitution. She explains translating moral theory into training, giving a model a character that resists sycophancy, and teaching nuanced judgment, uncertainty, and empathetic facilitation. Conversation covers safety trade-offs, bias in data, and whether AI might one day deserve moral consideration.

Amanda Askell, head of personality alignment at Anthropic and primary author of Claude's Constitution, explains the 20,000-word framework that shapes Claude's values and behavior. She describes how the constitution guides training and reward signals. The conversation covers fidelity to text versus spirit, virtue ethics over rigid rules, cultural universality, decision hierarchies, and implications for moral patienthood and specialized domains.