GENDER IDENTITY IN THE AGE OF AI AND VIRTUAL REALITY

This paper investigates how gender identity is being reconfigured in the contemporary convergence of artificial intelligence (AI) and virtual reality (VR). AI systems increasingly infer, classify, and generate gendered representations across domains such as visual media, health, education, and online platforms, often reproducing binary, Eurocentric, and heteronormative norms through biased training data and opaque model architectures. Recent empirical work documents systematic gender bias in large language models, recommendation algorithms, and text-to-image generators, including masculinized default outputs, stereotyped occupational depictions, and uneven error rates across gendered categories.At the same time, VR and social virtual environments afford intensified forms of embodiment and avatar-mediated presence, enabling queer, transgender, and gender-diverse users to explore, affirm, and experiment with gendered self-presentation in ways that may be inaccessible or unsafe offline. Studies of avatar gender transitions, VR Chat role adoption, and LGBTQ+ communities in social VR show that virtual embodiment can both alleviate and exacerbate dysphoria, providing spaces of euphoria, social support, and self definition, while also exposing users to harassment, exclusion, and platform level constraints. Drawing on an interdisciplinary review of scholarship in critical data studies, AI ethics, human computer interaction, VR research, and queer and trans studies (2020–2025), the paper develops a conceptual framework for understanding gender identity as co-produced by datafied inference and immersive embodiment. We map four key mediating processes-algorithmic classification of gender, generative representation of gendered bodies and voices, avatar design and customization, and socio technical governance of platforms-and analyse how they interact to stabilize, negotiate, or destabilize gender categories.