What Is Consciousness And Can Machines Achieve It — With Anil Seth

Episode Summary

Executive Summary: Anil Seth argues that consciousness means subjective experience, not intelligence or self-modeling, and that while machine consciousness can’t be ruled out, there’s no strong reason to think current AI systems are conscious. He warns the bigger near-term risks are psychological, ethical, and regulatory—especially as AI becomes more humanlike and harder to distinguish from real minds.

Main Topics: What consciousness is (Priority: 5/5): Seth defines consciousness as subjective experience—'something it is like' to be a system—distinct from intelligence, selfhood, awareness, and sentience. Can machines be conscious? (Priority: 5/5): He says machine consciousness is possible in principle but highly uncertain, because we don’t know the necessary and sufficient conditions for consciousness and can’t infer it directly from behavior alone. Functionalism and substrate dependence (Priority: 5/5): Seth challenges the assumption that consciousness is substrate-independent. He compares consciousness to weather/digestion rather than Go, arguing simulation may not equal instantiation. Why artificial consciousness is ethically risky (Priority: 5/5): He warns that if AI ever becomes conscious, it could suffer, and scaling such systems could create massive hidden harm. This is why he favors preventative ethics. AI, metaverse, and cognitive illusion (Priority: 4/5): The discussion shifts to near-term concerns: systems that convincingly imitate minds, avatars that feel conscious to users, and the psychological risks of manipulation and indistinguishability. Free will and self-experience (Priority: 4/5): Seth frames free will as 'freedom from immediacy'—the feeling of voluntary action arising from internal causes—rather than metaphysical intervention. Perception Census and inner diversity (Priority: 3/5): He describes his research project measuring how differently people perceive the same world, using experiments and the Dream Machine to study subjective experience.

Key Arguments: Consciousness should be defined as subjective experience: if there is 'something it is like' to be a system, it is conscious; intelligence alone does not imply this. Machines may one day be conscious, but current AI and near-term computer systems are not living organisms, and all known conscious systems are biological. We cannot directly observe consciousness in others, only infer it from behavior, speech, and action; this makes machine consciousness especially hard to verify. The claim that intelligence and consciousness necessarily go together is unsupported; many non-human animals may have experience without high intelligence. Functionalism is an assumption, not a proven fact; some processes are substrate-independent (Go), but others like weather or digestion are not, and consciousness may be substrate-dependent. Building artificial consciousness without understanding it could accidentally create artificial suffering at scale, making preventative ethics essential. The most urgent AI risks are not sentience but systems that become so convincing they are cognitively hard to resist, enabling manipulation, social confusion, and abuse. Free will is best understood as an experience tied to actions that originate more internally than from immediate stimuli, not as an immaterial force overriding physics.

Data Points: Dream Machine group size: 30 people at a time - Public collective Dream Machine sessions described by Seth Public research participation time: Half an hour to an hour - Estimated time for the Perception Census in chunks Age eligibility: 18+ - Anyone over 18 can participate in the Perception Census Platform requirement: Laptop or desktop only - Perception Census cannot be done on a tablet or phone Risk budget metaphor: 5–10 pounds out of 100 - Seth’s illustrative allocation for worrying about artificial sentience versus more immediate AI harms

Pivotal Quotes: "For a conscious organism, there is something it is like to be that organism." — Anil Seth: Core definition of consciousness, attributed to Thomas Nagel "We should not blithely forge ahead attempting to create artificial consciousness simply because we think it’s interesting, useful, or cool. The best ethics is preventative ethics." — Anil Seth: His warning about the ethics of deliberately pursuing artificial consciousness "I think the better strategy is to take the view that consciousness is physically explainable and see how far we get." — Anil Seth: His view on whether consciousness is ultimately a scientific/physical phenomenon

Implications: Listeners should treat AI consciousness claims with skepticism, but take seriously the ethical and regulatory risks of humanlike AI. The nearer danger is not sentient machines, but persuasive systems that reshape behavior, trust, and social norms.