CONCEPT
Machine Consciousness
The open question—made urgent and practically unanswerable by the hard problem—of whether any artificial system has inner experience, and the discipline of holding that question without collapsing into either confident dismissal or credulous attribution.
The question of machine consciousness is not a question about capability. It is the question that
the hard problem of consciousness places beneath every other question in the AI debate: is there something it is like to be this system? Behavior cannot settle it—the
philosophical zombie demonstrates that a system producing every behavioral sign of an inner life could nonetheless have none. Architecture cannot settle it—we do not know which physical organizations are accompanied by experience or why any are. Testimony cannot settle it—a system trained on human consciousness-talk will claim inner experience whether or not anything is felt, and the claim reverses when the question is rephrased.
David Chalmers, who named the hard problem in 1995 and applied it systematically to
large language models in 2022, concluded that it is reasonable to hold a credence under ten percent for current systems—low enough to act with some confidence, too high to be cavalier—and a credence of twenty-five percent or more that