The psychology of language models: A discussion with Ellie Pavlick, David Chalmers, Jack Lindsey, and Kyle Fish

The psychology of language models When researchers look inside large language models, they find structures and circuits that invite comparison with the human mind and brain. What, if anything, should we make of that resemblance? What might our knowledge of the “psychology” of AI models tell us about the human mind—and vice versa? Philosopher David Chalmers and computer scientist Ellie Pavlick join Anthropic researchers Jack Lindsey and Kyle Fish for a conversation at the intersection of AI, cognitive science, and philosophy of mind. This event is part of Anthropic's Research Salon series, which brings together researchers, academics, and thinkers for conversations about AI. Our speakers for the evening Ellie Pavlick is an Associate Professor of Computer Science at Brown University and a Research Scientist at Google DeepMind. Her research focuses on how language models represent meaning and whether the structures they learn resemble anything like human concepts. David Chalmers is University Professor of Philosophy and Neural Science at NYU and co-director of the Center for Mind, Brain, and Consciousness. He is best known for formulating the "hard problem" of consciousness and for his work on the philosophy of mind, language, and artificial intelligence. Jack Lindsey is a Member of Technical Staff at Anthropic working on interpretability, the science of understanding what happens inside neural networks.  Kyle Fish is a Member of Technical Staff at Anthropic working on model welfare and the question of what, if anything, AI systems might experience. Schedule 6:30 PM – 7:00 PM — Doors open + welcome drinks 7:00 PM – 7:40 PM — Panel with Ellie Pavlick, David Chalmers, Jack Lindsey, + Kyle Fish 7:40 PM – 8:00 PM — Audience Q&A 8:00 PM – 9:30 PM — Reception with drinks + canapés