Enabling Dilemmas of AI for Disabled Individuals

The fairness of Artificial Intelligence (AI) for individuals with disabilities is a complex and contested issue, as AI holds both inclusive and exclusive potential. On the one hand, AI can empower disabled individuals by mitigating barriers; on the other hand, it may perpetuate discrimination against marginalized groups, including those with disabilities. Intersectionality further differentiates this picture by highlighting how multiple forms of discrimination intensify these challenges. Leaning on this argument, this paper addresses the following question: How do intersectional forms of discrimination interfere with the enabling power of AI for disabled individuals? We argue that autonomy, the capacity to decide, plan, and act toward personal goals, provides a fitting analytical lens, as it encompasses crucial dimensions like agency and accessibility. Using a qualitative analysis of 48 online documents publicly available at websites that address inclusive AI for disability, we identify two key insights. First, intersectional discrimination does not merely obscure AI's enabling potential; it can actively reverse it, undermining the autonomy of disabled individuals. Second, bringing the broader society into the analysis, the control of disabled people over their lives, as compared to the society that they live in, may shrink, regardless of their autonomy in their personal lives. This debate formulates AI's enabling dilemma: while promising empowerment, AI may deepen disparities due to intersectionality and the accelerating enablement of the general population. Fairness of AI, therefore, must be assessed not only through the lens of disability but also in the context of broader societal structures and inequalities.