Three reasons why AI is not conscious and perhaps never will be Large language models can mimic thought and feeling with uncanny skill. But evidence from neuroscience suggests that intelligence alone is not enough for subjective experience. By Anil K. Seth, Ned Block & Giulio Tononi

https://www.nature.com/articles/d41586-026-03313-7?utm_source=Live+Audience&utm_campaign=b691f2a608-nature-briefing-daily-20261020&utm_medium=email&utm_term=0_-33f35e09ea-50432164 Large language models (LLMs) can mimic thought and feeling with uncanny skill — but they are not conscious, and deeming them so could hinder attempts to regulate them, argue neuroscientists Anil Seth and Giulio Tononi and philosopher Ned Block. LLMs beguile us by aping human language, the authors argue, noting that other impressive systems — such as AlphaFold, the protein-structure-prediction tool — aren’t considered to possess consciousness. And the mind isn’t merely a carbon-based computer — the living tissue of the human brain is essential, according to the leading theories of consciousness. Silicon-based computers are very unlikely to be sufficient for consciousness, and don’t satisfy certain crucial properties that conscious brains do, the authors write.