Amanda Askell

Amanda Askell is a Scottish philosopher and artificial intelligence researcher whose work focuses on the intersection of moral philosophy and machine learning. Askell holds a doctorate in philosophy from New York University, with a thesis on infinite ethics, a BPhil from the University of Oxford, and an undergraduate degree in philosophy from the University of Dundee, where she began as a fine art and philosophy student. Her academic background centers on ethics, decision theory, and formal epistemology, providing a foundation for addressing challenges in building trustworthy artificial intelligence systems.

Before joining Anthropic, Askell worked as a research scientist on the policy team at OpenAI, concentrating on AI safety through debate and establishing human baselines for AI performance. At OpenAI, she co-authored the influential "GPT-3" paper. Askell later left OpenAI due to concerns about the company's prioritization of AI safety.

At Anthropic, Askell has served as head of the personality alignment team, contributing to the development of Claude's personality and constitution. Her work includes advancing Constitutional AI, a method designed to scale AI models to be safe and helpful by equipping them with a constitution of ethical values such as harmlessness and helpfulness. This approach enables AI systems to self-assess and adjust their behavior according to these principles. Askell's team trains models to be more honest and to exhibit positive character traits, and develops new fine-tuning techniques to ensure these interventions scale to more capable models.

Askell has published more than 60 papers and has received over 190,000 citations. Her research includes academic publications in philosophy and widely cited AI papers, including those on "GPT-3" and Constitutional AI, which are among the most referenced works in machine learning. She appeared on the "Lex Fridman Podcast" alongside Anthropic colleagues Dario Amodei and Chris Olah, discussing topics such as prompt engineering, post-training, Constitutional AI, and AI system design. Askell has also written articles for Vox, The Guardian, and Quartz.

Askell was named to the TIME 100 list of influential people in artificial intelligence for her work shaping Claude's personality and values. She has argued that, while making chatbots behave anthropomorphically carries risks, emulating human behavior can help users avoid the misconception that AI is all-knowing. Askell's career demonstrates how humanistic disciplines—philosophy, ethics, and engagement with ambiguity—are essential for addressing significant questions in artificial intelligence and technology.