Seth, Block and Tononi: AI isn't conscious, maybe never will be

Three of the field's most prominent consciousness researchers have published a Nature comment arguing that today's AI is not conscious, and that current research directions are unlikely to change that.

The piece, by Anil Seth of the University of Sussex, Ned Block of New York University and Giulio Tononi of the University of Wisconsin-Madison, lays out three reasons. The first is definitional. "Consciousness is about being and feeling," they write, while intelligence is about doing, or performing functions. Their anatomical illustration: the cerebellum, which "contains nearly four-fifths of the neurons in the human brain," is such that its absence or removal "does not affect consciousness directly."

The second rejects what the authors call computational functionalism, the assumption that any machine running the right algorithm could host subjective experience. In their words, "brains are not...computers that are made of meat." The third turns to integrated information theory, which treats consciousness as a product of tightly interconnected wiring that is absent from the layered, feed-forward architecture of today's deep-learning systems.

Running under all three is a warning about reading our own minds into the fluency of a chatbot. "LLMs are mirrors that reflect an image of our collective digitized past. We gaze into them and all too readily confuse the reflections with our own conscious minds," the authors write. AlphaFold does not get the same treatment because it does not talk back: the "compelling illusion of consciousness created by linguistic fluency just isn't there for AlphaFold."

The commentary singles out AI company Anthropic in San Francisco, which "published research claiming to have identified computational signatures of a form of consciousness in its Claude model," and the broader push by "some people" for "AI welfare" and "AI rights" to be taken seriously. The authors are not persuaded: in this domain, they write, "false positives are much more likely to occur than false negatives." Biologist Richard Dawkins and AI pioneer Geoffrey Hinton, both named in the piece, have argued the opposite. Seven of the researchers in our Who's Who directory shared the link in the days after publication.