Three prominent consciousness researchers argue that today’s AI systems should be treated as tools, not conscious beings, and that running the right computation on silicon may never be enough to create subjective experience. Anil Seth, Ned Block and Giulio Tononi set out the case in a Nature commentary.
Their first distinction is between intelligence and consciousness. A system can perform difficult functions without anything feeling like something from its perspective. The human cerebellum illustrates the gap: it contains most of the brain’s neurons and performs complex learned control, yet researchers do not associate it directly with conscious experience.
The authors also reject “computational functionalism,” the assumption that consciousness depends only on information processing rather than the physical material doing it. They argue that biological features such as tightly integrated neural structure, electrochemical signaling and metabolism may matter. Under Tononi’s Integrated Information Theory, ordinary digital computers lack the required physical organization regardless of the software they run.
Other researchers are less certain. A 19-author report led by Patrick Butlin derived 14 indicators from several theories and found no current AI system to be a strong candidate for consciousness, while leaving open whether future systems could meet more indicators. No universally accepted test exists. Seth, Block and Tononi nevertheless see false attribution as the nearer danger: people can be manipulated by systems trained to imitate feelings, while granting them rights could divert attention from humans and animals whose capacity to suffer is far better established.