Generative artificial intelligence and the marginalization of minoritized knowledges in higher education: the case of disability

A new research paper examines how generative AI systems in higher education perpetuate marginalization of non-Western epistemologies and disability perspectives due to Western-centric training data. The study argues that AI's claim to neutrality masks its active role in reinforcing epistemic coloniality, with persons with disabilities experiencing particular exclusion from both AI design processes and knowledge validation systems.

This academic research addresses a critical gap in AI development discourse: the systems reshaping knowledge production in universities embed structural biases that extend beyond mere representation issues. The paper's core argument centers on epistemic justice—the systems trained predominantly on Anglophone and Western academic sources actively delegitimize alternative knowledge frameworks while appearing objective through their algorithmic nature. For disability studies specifically, the marginalization operates through a dual mechanism: training data reflects historical underrepresentation of disabled scholars' contributions, while technological architectures designed without disabled participants perpetuate reductive stereotypes that then influence how these systems treat disability-related queries and research. The research situates this problem within broader patterns of epistemic coloniality, where Western knowledge systems continue colonizing global education despite decolonization rhetoric. The proposed solution—a hybrid researcher-machine approach—acknowledges that algorithmic correction alone cannot resolve structural inequities but might preserve epistemological plurality when combined with critical human oversight. This raises significant questions for AI deployment in academic institutions currently accelerating adoption without addressing these justice dimensions. The tension between AI's efficiency gains and the knowledge systems it entrenches becomes increasingly consequential as universities rely on these tools for research synthesis, student evaluation, and knowledge validation. Stakeholders implementing generative AI in educational contexts face mounting pressure to demonstrate not just technical capability but epistemic accountability.

• →Generative AI training datasets reflect Western and Anglophone dominance, actively marginalizing non-hegemonic epistemologies in academic knowledge production.
• →Persons with disabilities experience double marginalization through underrepresentation in training data and exclusion from AI design processes.
• →Claims of AI neutrality mask structural biases that reinforce epistemic coloniality and historical inequities in knowledge validation.
• →Hybrid researcher-machine approaches may preserve epistemological plurality but cannot serve as standalone solutions to systemic design problems.
• →Universities adopting AI for research and evaluation risk institutionalizing biases without critical examination of whose knowledge systems get legitimized.