Ability Bias

A form of social bias encoded in artificial intelligence systems, particularly large language models, that reflects stereotypical or discriminatory assumptions about people with disabilities. Ability bias manifests through linguistic associations that link specific disabilities to negative attributes such as reduced independence, social difficulties, or diminished employability. Unlike explicit ableism, ability bias in AI systems can be subtle — for example, portraying individuals with disabilities as "inspirational" or "courageous" rather than ordinary, or assuming reduced capability based on disability status. This bias originates from imbalanced representation of disability in training data and can perpetuate real-world discrimination when LLMs are used in consequential domains like hiring, education, and healthcare.