The philosophical debate over AI consciousness intensified this week when science fiction author Ted Chiang published an Atlantic essay arguing that large language models fundamentally lack conscious experience. On LessWrong, a responding philosopher largely agrees—LLMs are probably not conscious—but argues that Chiang's proof is logically shaky. Chiang's central claim: if we wouldn't call a fictional Julius Caesar conscious just because an LLM generated his dialogue, then a chatbot persona generated the same way can't be conscious either. The rebuttal: this reasoning confuses the consciousness of characters with that of the author, much as conscious humans write fictional characters. Chiang's second argument—that consciousness requires a physical body to generate emotions—draws skepticism too: what counts as a 'body' for a digital system, and can information processing itself constitute a form of sensation? The thread reveals a deeper disagreement on what consciousness actually requires, even as both parties suspect LLMs fall short.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton