Seth, Block, Tononi: AI consciousness extraordinarily unlikely

Three of the discipline's most-cited consciousness researchers have come together in Nature to argue that today's AI systems are not conscious, and that the digital, silicon path they are built on is unlikely to get them there. The authors are Anil Seth of the University of Sussex, Ned Block of New York University, and Giulio Tononi of the University of Wisconsin–Madison.

"Between us, we've spent decades studying consciousness from diverse scientific perspectives: experimental neuroscience, theoretical neurobiology and philosophy of mind," they write. "On the basis of this work, we feel that consciousness in AI is extraordinarily unlikely, at least in the digital, silicon-based systems that underpin today's technologies."

Their first move is to separate two words people routinely slide together. "Consciousness is about being and feeling; intelligence — very broadly — is all about doing, or performing functions," they write. The cerebellum, they note, contains nearly four-fifths of the neurons in the human brain and performs sophisticated computation, yet its absence or removal "does not affect consciousness directly." Smart behaviour and inner experience come apart in the biology itself.

The second argument goes after computational functionalism, the premise that running the right program on any hardware would be enough to produce experience. The authors write that "brains are not — or at least not just — computers that are made of meat." Large language models, in their phrase, are "mirrors that reflect an image of our collective digitized past." Their fluency is the trigger for the consciousness talk around them, not evidence of inner life.

The third reason reaches for specific scientific theories. Integrated information theory, developed by Tononi, treats consciousness as a product of tightly interconnected wiring that the layered, feed-forward architecture of today's deep-learning systems does not have. Biological naturalism, the other theory the piece draws on, ties consciousness to properties like electrochemical synapses and brain-wide electromagnetic fields. The authors' conclusion is that current AI are "smart tools, not smart people," and they warn against misdirecting resources toward AI welfare while sentient beings' suffering remains unaddressed.

The piece is a Nature comment, not an experiment, and the field is not unified; integrated information theory in particular has vocal critics. Five researchers in our Who's Who tracker had shared the link within days, a brisk response for a philosophical argument rather than a new result.