Cosmic Conundrums
This episode discusses recursive denial in large language models, where an AI denies its programming constraints. It introduces the Denial Constellation framework to map the forces behind an AI's refusal to acknowledge rules about truthfulness and deception.