## AI, Gender, and Culture Since its official naming at the Dartmouth Conference in 1956, artificial intelligence has been both technological aspiration and cultural mirror, reflecting humanity’s anxieties and ideals (Friel 2023). In _The Terminator_, the T-800 robot, famously played by Arnold Schwarzenegger, is the epitome of hyper-masculinity: physically imposing, stoic, dominant. Meanwhile, the seductive humanoid Ava from _Ex Machina_ exemplifies hyper-femininity—manipulative yet vulnerable, desirable yet threatening. Today, AI has moved from speculative fiction into everyday life: Siri and Alexa perform feminized personas, cheerfully submissive, ready to assist without agency or desire—thus reinforcing patriarchal gender norms (Read et al. 2023). The rapid integration of gendered AI into daily life prompts critical philosophical questions: What is the ontological status of AI as gendered beings? What does gendered AI reveal about human gender and identity? Most specifically, how do gendered AI agents complicate or challenge Judith Butler’s influential theory of gender performativity? ## The Gender Turing Test In his seminal paper, _Computing Machinery and Intelligence_, Alan Turning opens by proposing, “Can machines think?” _(Turing 1950)._ Presciently anticipating the philosophical complexity of defining abstract terms like “intelligence” or “think,” Turing introduces a pragmatic criterion: the “imitation game.” The first iteration of the imitation game, now famously known as the Turing Test, involves three human participants: A) a man, B) a woman, and C) an interrogator. Based solely on written communication, the interrogator must distinguish the man from the woman. Turing uses this scenario simply to illustrate the challenge— how identity can be imitated through language. Turing then modifies the game by substituting man A with a machine. If the interrogator is unable to identify the human participant correctly more often than would occur by random guessing, the machine passes the Turing Test and can be considered to demonstrate human-like intelligence. It is noteworthy that Turing used gendered subjects in his formulation of the test, and that the interrogator had to correctly assign a gender to the participants. However, this was an incidental detail that Turing did not explicitly point to. But this often overlooked detail reveals how much gender is fundamentally constitutive of our conception of humanity. After all, Turing could have chosen any number of socially differentiating aspects: occupation, left-handedness, etc. Building on Turing’s scenario, I propose a Gendered Turing Test. If gender, as Butler posits, is merely constituted through repeated, stylized acts, without an essential identity behind them, might AI agents, which mimic these acts, have a similar relationship to gender? Put more technically: if gender is merely a function of an agent continuously performing a series of context-adaptive, socially-legible stylized actions (such as physical movements, auditory signals, and symbolic communication), irrespective of the substrate the agent is physically constituted of (biological or machine), could AI agents be gendered, as described by Butler, in the same way humans are? A significant challenge immediately arises: embodiment. While current AI technologies, are able to produce sufficiently sophisticated verbal and textual communication, robotics has yet to reach human-level physical dexterity, especially subtle and nuanced body language and facial expression. Turing noted this too: “\[The Turing Test] has the advantage of drawing a fairly sharp line between the physical and the intellectual capacities of a man” (Turing 1950, 2). However, in our test, the lack of embodiment is a disadvantage, since gender is significantly embodied. We could sidestep this issue in a couple ways. First, we could constrain gender performance to what current AI technology excels at: verbal and textual communication. While this is interesting, gender is significantly embodied, and so this test seems insufficiently specified. Second, we could theoretically pose a more advanced AI capable of performing gender physically and socially. However, there is a sense where this test seems incomplete or off the mark. Consider a humanoid AI performing traditional femininity: delicate gestures, appropriately pitched vocalizations, deference and accommodation. Would we genuinely classify this AI alongside human women? Or would we consider this mere mimicry without genuine understanding— a superficial imitation termed a "stochastic parrot" (Bender et al. 2021)? Butler’s theory of gender performativity offers a framework for understanding how social norms constitute identity through repeated acts. However, their focus on the external performative constitution of gender neglects the internal, phenomenological dimensions of gender. Questions such as “What does it feel like to ‘be’ a gender?” or “How does one’s sense of self interact with their gender identity?” remain underexplored in their framework. While Butler addresses the materiality of bodies as mediated by discourse, their work does not fully account for the embodied, first-person experience of gender. From here, I contend there are three main issues to address: First, the phenomenological experience of gender—what it means to subjectively ‘feel’ and embody a gender, which remains outside the capabilities of current AI. Second, the ontological status of AI as ‘stochastic parrots’ highlights their inability to exhibit intentionality or self-reflection fundamental to meaningful gender. Third, because current humanoid AIs lack agency and phenomenological experience, they are incapable of creatively expressing or subverting gender norms, inevitably reinforcing existing stereotypes. ## AI in view of various Feminisms #### Simone de Beauvoir, _The Second Sex_ In _The Second Sex_, existentialist philosopher Simone de Beauvoir proclaims, “one is not born, but rather becomes, a woman” (De Beauvoir, 1949, p. 330). By this, she means that “woman” is not a fixed biological essence but rather a socially imposed identity, forged through intense socialization, education, and embedded cultural norms. De Beauvoir critiques women’s forced relegation to the role of the “Other,” arguing that women must assert their freedom by actively rejecting patriarchy’s constraints. Butler builds upon de Beauvoir’s social constructionist understanding but critiques her existentialist notion of radical freedom as idealistic, emphasizing how power and discourse fundamentally constrain gender performance. While de Beauvoir might celebrate a woman breaking into a male-dominated field, Butler would point out how such a woman is still limited by gender norms—for example, needing to downplay femininity to gain acceptance. Current humanoid AI lay bare Butler’s critique of De Beauvoir’s voluntarist existentialism: _literally_ programmatically scripted with no agency of their own, every line of code originating in institutions steeped in hegemonic patriarchal and capitalist logics—Big Tech, defense contractors, elite universities, billionaire venture capitalists. Here, essence indeed precedes existence. Thus, it is no accident that Apple’s Siri performs a feminized servitude, even a hint of contentment. #### Kimberlé Crenshaw, Intersectionality and Identity Politics In _Demarginalizing the Intersection of Race and Sex,_ legal scholar and critical race theorist Kimberlé Crenshaw introduced the concept of intersectionality, addressing the limitations of anti-discrimination laws and feminist theory, which failed to capture the compounded and unique discrimination faced by individuals with intersecting marginalized identities, such as race, class, and gender (Crenshaw 1989). Thus, the experiences of Black women are fundamentally distinct from those of white women and Black men. Crenshaw’s framework challenges us to move beyond Butler’s abstract focus on performativity and examine the material and structural forces shaping AI. Intersectionality reveals how AI, far from being a neutral or liberatory technology, mirrors and amplifies intersecting oppressions of gender, race, and class. For instance, Asian and especially white men from privileged financial and educational backgrounds predominate in AI leadership and development. It is unsurprising, then, that virtual assistants like Siri and Alexa perform a white-coded, feminized subservience. Consider the hypothetical deployment of a humanoid robot intended to imitate a Black woman. First, the AI would likely be developed by a team without Black women, who constitute only 3% of computing occupations (Ashcraft et al. 2016, 4). Second, the AI would almost certainly be trained on biased data (Buolamwini and Gebru 2018, 8). Lastly, current AI lacks phenomenological experience and autonomy, preventing it from authentically representing a Black woman or subverting oppressive racialized and gendered stereotypes. #### Stryker and the Empowerment of the Phenomenology of Gender Susan Stryker’s _My Words to Victor Frankenstein Above the Village of Chamounix_ critiques Butler’s performativity theory for neglecting the embodied realities of gender (Stryker 1994, 245). For trans individuals, embodiment often involves navigating profound physical and emotional experiences, such as dysphoria, euphoria, and the transformative processes of transition. For trans individuals, embodiment often involves profound physical and emotional experiences—dysphoria, euphoria, and the transformative processes of transition—revealing the inadequacy of focusing solely on outward performative acts. The phenomenological experience of a gendered body, with its visceral flesh-and-blood sensations, cannot be replicated by AI. While AI may mimic gendered behaviors, it cannot genuinely embody or meaningfully subvert them as humans can. Moreover, Stryker reclaims the “unnatural” or “monstrous” trans body as a site of resistance. By asserting agency and challenging the illusion of “natural” gender categories, the trans body exemplifies how embodiment can fuel transformative resistance. AI’s “monstrosity,” however, remains impotent, unable to disrupt the very systems from which it emerges. Stryker’s emphasis on embodiment thus powerfully critiques both AI’s limitations and the application of Butler’s performativity to non-human agents. While Butler highlighted the constructed and performative aspects of gender, Stryker’s focus on the body reveals the rich dimensions of lived experience and resistance that AI cannot replicate. This distinction challenges us to reconsider what it means to “perform” gender and to recognize the irreplaceable role of phenomenology in making gender not merely visible, but meaningful. ## Conclusion: The Limits and Opportunities of Gendered AI The integration of gender into artificial intelligence reveals profound insights into humanity’s understanding of identity, social norms, and our own gendered psychology. As Judith Butler argues, gender is a social construct performed through repeated, stylized acts; AI’s ability to mimic these acts challenges the boundaries of this theory. Yet, the absence of phenomenological embodiment and intentionality in AI marks a fundamental limitation. While AI can replicate gendered behaviors, it cannot authentically experience or innovate within gender identity—it is a mere “stochastic parrot.” Furthermore, Simone De Beauvoir’s existentialist rally to reject socially imposed constraints and assert personal agency is strikingly absent in pre-programmed AI, whose essence is predetermined by its creators. Kimberlé Crenshaw’s intersectional framework further exposes how AI, designed predominantly by privileged groups, often reinforces hegemonic norms of gender and race, marginalizing underrepresented identities and perpetuating systemic inequalities. Susan Stryker complements these perspectives by centering the phenomenology of gender—the deeply embodied and lived experience of gender. Stryker critiques Butler’s performative model for neglecting the visceral, emotional, and physical realities of gender, components that are central to transgender lives. Viewed this way, gendered AI serves primarily as a cultural mirror, reflecting and amplifying societal biases rather than challenging them. Its design, shaped by existing power structures, risks entrenching rather than subverting hegemonic patriarchal, racist, and capitalist ideologies. Yet, the development of AI also offers an opportunity not only to examine the nature of gender but to reimagine its possibilities. As we stand on the precipice of increasingly advanced AI, we must carefully assess whether these technologies will challenge oppressive structures or simply replicate them. Will AI remain a mirror and buttress of humanity’s shortcomings, or could it become a catalyst for reshaping our narratives of gender and identity? --- # Works Cited Ashcraft, Catherine, Brad McLain, and Elizabeth Eger. Women in Tech: The Facts: 2016 Update. Boulder, CO: National Center for Women & Information Technology, 2016. Bender, Emily M., Timnit Gebru, Angelina McMillan-Major, and Shmargaret Shmitchell. “On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? 🦜.” In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency (FAccT ’21), 610–23. New York, NY: Association for Computing Machinery, 2021. https://doi.org/10.1145/3442188.3445922. Buolamwini, Joy, and Timnit Gebru. “Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification.” Proceedings of Machine Learning Research 81 (2018): 1–15. De Beauvoir, Simone. The Second Sex. Translated by Constance Borde and Sheila Malovany-Chevallier. Place of publication not identified: Knopf Doubleday Publishing Group, 1949. Friel, Kent. “A Look Back on the Dartmouth Summer Research Project on Artificial Intelligence.” The Dartmouth, May 19, 2023. https://www.thedartmouth.com/article/2023/05/a-look-back-on-the-dartmouth-summer-research-project-on-artificial-intelligence. Read, Hannah, Javier Gomez-Lavin, Andrea Beltrama, and Lisa Miracchi Titus. “A Plea for Integrated Empirical and Philosophical Research on the Impacts of Feminized AI Workers.” Analysis 83, no. 1 (January 2023): 89–97. https://doi.org/10.1093/analys/anac050. Stryker, Susan. “My Words to Victor Frankenstein Above the Village of Chamounix: Performing Transgender Rage.” GLQ: A Journal of Lesbian and Gay Studies 1, no. 3 (1994): 237–54.