Advances in artificial intelligence (AI) and machine learning (ML) affect disabled people both directly, as potential users of AI/ML products and processes, and indirectly, through the broader societal changes. Despite the rapid expansion of research on AI and natural language processing, the representation of disabled people in AI systems remains largely uninvestigated, even though disabled people constitute at least 12.5% of the global population.

Although AI technologies hold the potential to create more inclusive environments in many spheres of life, they may also reflect and perpetuate existing biases. AI-related harms can affect people with disabilities in concrete ways. For instance, voice-activated systems often fail to recognize input from users with speech impairments. Mainstream AI fairness approaches are insufficient for addressing these harms. They tend to address one marginalized identity at a time, rely on well-defined trait categories, and frame marginalization relative to specific assumed desires (e.g., employment). This produces standards that are unjust across many areas of life.

Addressing these gaps requires an approach that goes beyond technical fixes and engages with disability as a structural and intersectional phenomenon. This doctoral project therefore examines the ableist bias in large language models (LLMs) through an interdisciplinary lens that combines technology assessment, human-computer interaction (HCI), and disability studies. The dissertation spans from empirical investigation to conceptual development, probing how disability is represented in AI systems and benchmarking practices, how people with disabilities experience these systems in practice, and how the methodology of TA can be enriched through engagement with inclusive design and disability scholarship.

The project aims to build a bridge between scientific approaches to disability studies and the potentials and risks of AI. In doing so, it aims to move beyond techno-deterministic framings and develop conceptual perspectives for AI-based technologies that can genuinely support and enhance the lives of people with disabilities, taking into account their diverse experiences with disability.

Tel.: +49 721 608-26813
E-mail