Workshop : "Intelligence and Imitation: Mind, Mechanism, and Mimesis"
Juliette Sokolov, a member of the Digital Writing Research Lab, will speak at the conference Intelligence and Imitation: Mind, Mechanism, Mimesis organized by Johns Hopkins University in Baltimore. Her presentation, titled “Why Turing's First Imitation Game Has Never Been Played,” will explore a reconstruction of the Turing Test applied to an LLM as a thought experiment highlighting the tension between the ontology and performativity of intelligence.
Abstract
While the Standard Turing Test has become a dominant framework for AI evaluation, Alan Turing's first imitation game, the Gendered Turing Test (GTT) has remained largely theoretical. This proposal examines the tension between the performativity and the ontology of intelligence through a reenactment of the GTT. We argue that the marginalization of the original Turing Test is not accidental, as it unsettles intelligence as a stable ontological category.
In the GTT, intelligence is neither defined as a mental property nor a cognitive attribute but as a constrained linguistic task. An interrogator engages in text-based exchanges with two unseen interlocutors (a man and a woman) and must determine their genders. A machine may replace one interlocutor, tasked with impersonating a woman. We revisit this game through a reenactment conceived as an operational thought experiment. The test enacts a performative definition of intelligence evaluated through an interrogator's interpretative judgment.
Our results are paradoxical: women, men and LLMs all won games in equal proportions despite significant individual variation. Rather than modeling intelligence, this result reveals the normative and ideological commitments embedded in the test. Outcomes depend on participants' personal investment, gendered expectations and identity projections, demonstrating how definitions of intelligence reproduce cultural constructions rather than isolate cognitive capacities.
The illegibility of the test's result does not stem from intelligence being immeasurable, but from its impossibility of definitive abstraction. The GTT compares singular individuals against one another, producing heterogeneous outcomes and no discernible pattern validating the initial tripartite division (man, woman, machine) over alternative categorizations. Turing tests attempt to draw a boundary between human and non-human in order to reassert intellectual superiority. Intelligence constitutes not an elusive quantity, but an abstraction that resists generalization.
This structural indeterminacy explains why the Gendered Turing Test has never functioned as a stabilizing experimental paradigm. By foregrounding mimesis, interpretation and performativity, the GTT shifts the imitation game from a measure of mind to a mirror of the conditions under which intelligence is recognized in human–machine relations.