Knowledge Tree
Henry Shevlin
Facts (20)
Sources
AI Sessions #9: The Case Against AI Consciousness (with Anil Seth) conspicuouscognition.com Feb 17, 2026 19 facts
perspectiveHenry Shevlin expresses skepticism regarding the prospects of consciousness science as a falsifiable research programme.
perspectiveHenry Shevlin argues that for artificial intelligence, determining the necessary conditions for consciousness is more relevant than determining sufficient conditions, because ruling out consciousness in artificial intelligence systems clarifies the ethical situation.
claimHenry Shevlin notes that the classification of dreamless sleep and general anesthesia as examples of losing consciousness is contested in debates around consciousness.
claimHenry Shevlin notes that the administration of amnestics as part of general anaesthesia protocols raises suspicions about whether patients are truly unconscious during the procedure.
claimHenry Shevlin notes that the argument that simulated weather systems do not get anyone wet is originally John Searle's point, which serves as a restatement of the disagreement regarding functionalism rather than a refutation of it.
accountHenry Shevlin recounts the case of Hisashi Ouchi, a Japanese nuclear researcher who received a lethal dose of radiation that destroyed his chromosomes and halted cell production, yet remained conscious and in pain for eighty-three days while kept alive through medical interventions.
claimHenry Shevlin identifies the danger of anthropomorphism and anthropocentrism as a major ethical issue in AI, noting that humans may develop highly dependent relationships with social AI, leading to phenomena like AI psychosis.
perspectiveHenry Shevlin defines computational functionalism as the view that mental states are individuated by their computational role.
claimHenry Shevlin argues that if consciousness is computational, it must be substrate-invariant, similar to how games like poker or chess, and money, remain the same regardless of the medium (coins, banknotes, or digital balance sheets).
claimHenry Shevlin asserts that AI systems have achieved human-level performance on a wide range of verbal reasoning tasks and can produce high-quality fiction, suggesting that the attribution of cognitive abilities to AI is not entirely a result of pareidolia.
claimHenry Shevlin defines the "specificity problem" as the difficulty of applying existing theories of consciousness to non-human systems because those theories are often too underspecified.
claimHenry Shevlin identifies Integrated Information Theory (IIT) as a theory that avoids the specificity problem, though he notes it leads to extreme predictions.
accountThe session 'AI Sessions #9: The Case Against AI Consciousness' features hosts Dan Williams and Henry Shevlin interviewing neuroscientist Anil Seth.
referenceHenry Shevlin authored a response to Anil Seth's paper published in the journal Behavioral and Brain Sciences (BBS).
claimHenry Shevlin asserts that while computational functionalism is one path to concluding that AI can be conscious, there are other types of functionalism that also support this conclusion.
claimHenry Shevlin observes that arguments claiming AI consciousness research diminishes human dignity parallel historical arguments against Darwinian evolution, which claimed that viewing humans as animals diminished human dignity.
perspectiveHenry Shevlin argues that the case of Hisashi Ouchi challenges the necessity of autopoiesis for consciousness, suggesting that if consciousness persisted despite the cessation of autopoietic processes, it undermines the claim that autopoiesis is necessary for consciousness.
perspectiveAnil Seth agrees with Henry Shevlin that viewing humans as continuous with the rest of nature is a beautiful, empowering, and dignifying perspective.
perspectiveHenry Shevlin argues that viewing humans as continuous with the broader tree of life and as part of a broader space of possible minds does not strip humans of their dignity.
Consciousness and AI - Open Encyclopedia of Cognitive Science oecs.mit.edu Feb 5, 2026 1 fact
claimHenry Shevlin (2021) argues that it is questionable whether evidence for neuroscientific theories of consciousness, which is largely derived from studies on humans and primates, supports their extension to AI systems, particularly because these studies do not specify how similar features must be to suffice for consciousness.