AI-Accessible Document Adaptation

Providing accessible digital educational materials is one of the cornerstones of a system that promotes access to learning and training for blind and low vision people (BLV). However, the accessibility criterion has yet to be integrated into the practices of publishers and teachers. Moreover, this work remains time-consuming for adaptation specialists. In the context of the development of an artifact that relies on artificial intelligence, we question how the delegation of the adaptation of educational material to and from an agentic IS artifact can occur? Drawing firstly on the actor-network theory, we outline a rich ecosystem of actors and the regulating role of the BLV learner. Secondly, we investigate the interactions between the BLV learner, the adaptation specialist and the agentic IS artifact under the prism of the conceptual model of the willingness to delegate. Third, we develop and evaluate a web-based application that can be used by BLV learners and accessibility specialists.