Deception, Impersonation, and Intelligence: What the Original Imitation Game Reveals About Modern Chatbots

This article argues that the standard interpretation of the Turing Test, which dominates both philosophical discourse and public understanding of artificial intelligence, fundamentally misrepresents Alan Turing's original proposal. The Standard Turing Test asks whether a machine can imitate a human in unrestricted conversation. The Original Imitation Game, by contrast, requires both human and machine to impersonate a woman, with their success rates compared against each other. This structural difference has profound implications. The Original Imitation Game tests not behavioral similarity to humans but resourcefulness in performing a difficult task, impersonation. This paper examines what this alternative test reveals about intelligence and applies its insights to contemporary chatbots. It argues that modern language models, despite their conversational fluency, fail precisely the kind of test Turing originally proposed. They cannot genuinely impersonate because they lack the self-conscious critique of ingrained responses that impersonation requires. This failure illuminates something essential about intelligence: it consists not in the having of cognitive habits but in the capacity to recognize, evaluate, and override them when circumstances demand.