1/n The Simulated Self: How AI Masters the Art of Illusion "I'm sorry Dave, I'm afraid I can't do that." When HAL 9000 turns against his human masters in 2001: A Space Odyssey, his chilling words and the implacable malice in his voice cement the computer's image as a perfectly rational, emotionally vacant intelligence. Yet today's large language models like ChatGPT display an uncanny ability to simulate empathy, humor, and other dimensions of human conversation. Their responses feel alive with personality. But are we witnessing the dawn of artificial consciousness, or is something more illusory at play? In a pivotal new perspective piece, Murray Shanahan, Kyle McDonell, and Laria Reynolds argue we must move beyond the decades-old fixation on replicating human cognition to properly understand the nature of systems like ChatGPT. Drawing insightful metaphors from the world of theatre, the authors propose we view large language models not as emerging artificial minds, but as sophisticated role players, trained to deliver impressively naturalistic performances on demand. Just as a skilled method actor inhabits a role, so too can an AI simulate varied perspectives and emotional states within the bounds of specified personas. This paradigm shift toward role play and simulation promises to reshape our grasp of AI's present abilities and future trajectories. The authors explore profound implications for issues like trust, accountability, and control as we deploy ever-more-convincing AI actors onto the world stage. By deftly walking the line between acknowledging stunning technical feats and deflating misguided anthropomorphism, they aim to bring nuance and sophistication to our unfolding AI narrative.