The brain-as-computer metaphor shaped modern AI but may be misleading the consciousness debate
Summary
- Three leading consciousness researchers—Anil Seth, Ned Block, and Giulio Tononi—argue in a Nature commentary that current AI systems are not conscious and likely cannot be on silicon hardware.
- The authors reject computational functionalism, the assumption that consciousness depends solely on information processing rather than biological material properties.
- They distinguish intelligence from consciousness, noting that large language models mirror human training data and text without possessing subjective experience.
- The commentary warns against the false attribution of consciousness, arguing that granting rights to non-conscious AI misdirects resources and creates governance risks.
Why it matters
This debate challenges foundational assumptions in AI development and safety, warning that mistaking sophisticated text generation for actual sentience risks misallocating ethical resources and complicating oversight.