What do Rene Descartes and Rick Deckard have in common? More than you'd imagine.

For centuries, we’ve dreamed of building minds in our own image, from Descartes’ mechanical automata to the replicants of Blade Runner. It has only been in the last few years that science fiction now seems closer to science fact than ever.

Today’s large language models feel, to many, much closer to that dream of conscious machines, not because they think, but because they so effectively perform the things we associate with thought: emotion, personality, vulnerability, even care.

This talk explores the psychology and design of anthropomorphised AI. It discusses why we instinctively project humanity onto machines, how modern LLMs exploit those cognitive seams, and what that means for trust, safety, and user experience.

Drawing on academic and practical research into human-like prompting, deception detection, and a concrete taxonomy of anthropomorphism, we’ll examine the subtle cues that make an AI “come alive”. We’ll also discuss the ethical edge where connection becomes manipulation, and how to design systems that are engaging without being deceptive.

Part philosophy, part cognitive science, part practical AI design, Descartes’ Daughter traces the line from the first chatbots to today’s emotionally fluent models, asking a simple but unsettling question: when the machine feels real, is that because it progressed, or because we did not?