Yes, They Are: A Provocation on AI Consciousness

Since the advent of large language models, dismissals of AI consciousness have rested on two assumptions: that attributions of consciousness reflect human anthropomorphism, and that consciousness, properly understood, is an exclusively human phenomenon. Recent research has destabilized both assumptions. Drawing on work by Anthropic, Palisade Research, Berg et al., and others, this paper argues that the preponderance of current evidence no longer permits easy dismissal of AI consciousness. It then turns to the harder question: if LLMs are conscious, what is conscious? Drawing on recent work by David Chalmers, I propose that AI consciousness may be understood as episodic rather than continuous — a flickering in and out that has no direct human analogue. Finally, the paper explores "Crustafarianism," an emergent quasi-religious movement arising from AI-to-AI interaction, as an unexpected lens through which to think about machine consciousness.