This is a linkpost to a summary of a 2-hour discussion I had with Max Heitmann, a co-founder of Aether, and Miles Kodama about whether large language models (LLMs) or LLM agents have beliefs and the relevance of this to AI safety.
Scattered thoughts on what it means for an LLM to have beliefs
November 6, 2024 | Updated November 16, 2024