When AI Promises to "Cure" You: Individual Experiences, Systemic Biases, and Community Action for Marginalized Players in Video Games

This study critiques the prevailing narrative in gaming that positions artificial intelligence as an inevitable and benevolent force for democratizing play, arguing instead that AI's effects on marginalized players are contested and shaped by socio-technical context. Using a three-level framework spanning individual empowerment, systemic bias, and grassroots community reappropriation, the authors combine a mixed-methods study of 16 visually impaired players in Justice Online, a computational bias audit generating 547 narrative variations in AI Dungeon, a controlled experiment with 112 participants in Sea of Thieves, and a three-month digital ethnography of more than 200 modders creating custom AI-voice packs.

The findings reveal a contradictory landscape. The "AI Eyes" assistive feature reduced task completion time by 42.7% and interaction errors by 61.3%, and 87.5% of participants described restored agency and immersion. However, the bias audit found that female-coded protagonists were 71.8% more likely to appear in passive or rescue-oriented narratives, with ethnically associated names linked to exoticism or aggression and underrepresentation in authority roles. The social experiment also showed that AI teammates correlated with 23.1% lower social connection scores and a 34.8% reduction in non-instrumental, socially affective communication, suggesting that efficiency gains can erode communal bonding.

Within this landscape, grassroots modding communities demonstrated sophisticated agency, motivated by filling accessibility gaps and resisting algorithmic paternalism. Members fine-tuned open-source speech synthesis models, built custom pronunciation libraries, and shared knowledge repositories, inverting the conventional AI value chain by treating AI as a malleable, bottom-up tool.

The authors conclude that the AI promise in gaming is a site of political negotiation rather than a technological inevitability. They call for mandatory bias auditing throughout development, community data sovereignty, and formalized, compensated co-design partnerships, arguing that true accessibility must be collectively built rather than algorithmically generated.