Perfect AI Mimicry and the Epistemology of Consciousness: A Solipsistic Dilemma
arXiv Preprint Archive October 6, 2025 Shurui Li
As artificial intelligence systems become increasingly sophisticated, the possibility of a 'perfect mimic'—an AI that is empirically indistinguishable from a human through observation and interaction—challenges how we attribute consciousness. The paper argues that our practices of recognizing minds in others rely almost exclusively on behavioral evidence. Refusing to grant equivalent epistemic status to a perfect mimic would require invoking inaccessible factors like qualia or substrate, which risks either epistemological solipsism or inconsistent reasoning. The author contends that epistemic consistency demands ascribing the same status to empirically indistinguishable entities, forcing critical reflection on the assumptions underlying intersubjective recognition and carrying significant implications for theories of consciousness and ethical frameworks concerning artificial agents.