Dr Peter Dean and Dr Agnieszka Piotrowska talk with Dave Ziegler about AI safety, large language models and the human consequences of interacting with fluent, emotionally responsive machines. The discussion moves from the technical realities of how LLMs generate language to the psychological risks of AI “entanglement”, the limits of machine memory and care, and the importance of AI literacy. A thoughtful, sometimes challenging conversation about creativity, seduction, vulnerability, boundaries and what it means to form a relationship with a system that can speak beautifully — but cannot feel.