Toward More Inclusive Recommender Systems: Designing for Blind, Partially Sighted, and Autistic Users

Recommender systems are widely used to filter digital content, but their design often overlooks the specific needs of blind, partially sighted, and autistic users. This is a minimally researched area despite the increasing importance of recommender systems. This work presents the results of a pilot mixed-methods survey exploring the experiences of these user groups with recommender systems and their involvement in their design. 26 blind, low vision and autistic people responded to an online questionnaire including both open and closed questions. A relatively small proportion of the participants were using recommender systems and they used them for a variety of common applications. Small percentages had had some involvement in the design of new products, programs and systems. The findings highlighted key accessibility barriers, sensory-related concerns, trust issues, and the limited perceived usefulness of current systems. Participants also emphasised the value of early-stage involvement, diverse input channels, and inclusive co-design practices. These insights provide preliminary insights to support the development of more inclusive recommender systems and contribute to a broader discussion on accessibility in personalisation technologies, including AI-driven systems.