Conversational AI for Digital Accessibility: An Experimental Study Involving Blind and Low Vision Users
Andrea Esposito, Rosa Lanzilotti, Maristella Matera, Antonio Piccinno, Ludovica Piro, Emanuele Pucci, Grazia RagoneAbstract
Digital inclusion is a fundamental right and a “must-have” for access to knowledge, education, and work. Many individuals living with disabilities or limited digital skills face digital barriers that determine inequalities and discrimination. Accessible digital services are therefore essential to guarantee everyone’s right to information and full societal inclusion. Nevertheless, the Web remains largely a visual experience, inadequate for many users. Assistive technologies, such as screen readers, can help, but not without problems. Conversational Artificial Intelligence and Large Language Models have emerged as technologies for inclusive interaction with digital services. This article presents an experimental study involving 30 BLV participants to explore the benefits a new AI-based paradigm would bring to the browsing experience of BLV people compared to screen readers. This aligns with a human-centered view of AI, which prioritizes users’ needs and trust over the powerful, autonomous tools that the users do not perceive as safe.