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

Three of the most prominent names in consciousness research have co-signed a bluntly worded commentary arguing that today's AI systems are not conscious, and probably never will be. Neuroscientist Anil K. Seth of the University of Sussex, philosopher Ned Block of New York University and neuroscientist Giulio Tononi of the University of Wisconsin–Madison write in Nature that intelligence and consciousness are different things, that substrate matters, and that the leading scientific theories of consciousness do not fit the architecture of today's large language models.

Their first move is to pry the two concepts apart. They point to the cerebellum, which contains nearly four-fifths of the neurons in the human brain and performs complex motor computation, yet contributes nothing to subjective experience. A lot of smart machinery, no lights on. From there they take aim at computational functionalism, the premise that running the right program on any hardware is enough to produce experience. "Brains are not — or at least not just — computers that are made of meat," they write.

The third reason is theory-specific. Integrated Information Theory, Tononi's own framework, treats consciousness as a product of components wired together in tightly interconnected loops, a pattern the authors say is absent from the layered, feed-forward architecture behind today's deep-learning systems. Biological naturalism, closer to Block's and Seth's position, pins consciousness to electrochemical and metabolic properties specific to living nervous systems that no current silicon implementation reproduces.

The target is not purely scientific. The authors note that figures like Geoffrey Hinton and Richard Dawkins have floated the possibility that AI already has, or soon will have, some form of inner life, and that Anthropic has published work describing "computational signatures of a form of consciousness" in its Claude model. Against that drift, the trio is explicit: "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." Their verdict on current systems is one phrase long. "Smart tools, not smart people."

Seven of the researchers we follow in our Who's Who tracker circulated the piece, which fits a commentary this field was waiting to see on the record from these three names together.