Agnosticism about Artificial Consciousness

The strongest version of the agnostic position. An 'epistemic wall' separates biological and artificial cases — our evidence about consciousness comes from organisms, and extrapolating to AI requires an unjustifiable leap. His most practically useful move: the consciousness/sentience distinction. Even if we can't know whether AI is conscious, we can design systems whose internal states wouldn't correspond to suffering *if* they were conscious. If you can't know whether the lights are on, at least make sure the room isn't a torture chamber.