What should I know? Analysing behaviour and feedback from student use of a virtual assistant to share information about disabilities

Administrative burden is a recognised cause of inequities for disabled students. Experiences of sharing information about disabilities and arranging adjustments can be demoralising and present barriers to success. To explore how Artificial Intelligence technologies could improve this situation, a virtual assistant (VA) was iteratively developed and deployed to support the initial steps of the process through which students share information. Here we describe findings from an eight-month trial where this was made available for students to use as an alternative to completing a form when declaring disabilities. 544 students tried using the assistant during this period. We analyse 351 questions asked of the VA by students, and a feedback survey with 129 responses. Results indicate the types of support expected while interacting with a VA and provide feedback on aspects of the design, the relationship with wider processes and experience of use. Overall, most participants wanted to continue using a VA in these processes, with positive perceptions across disability categories. We identify 12 themes showing a broad range of questions asked of the assistant. Given recent advances in AI, we discuss the opportunities and challenges to build on this and develop further inclusive innovations. Future work should focus on enabling context-informed answers to questions, enabling students to learn and contribute through the conversation, managing expectations according to VA capabilities, enhancing and monitoring inclusivity and integrating the VA with wider processes. • Administrative processes create additional burden and poor experiences for disabled students. • A virtual assistant was co-designed and deployed in a trial as a way to shared information and learn about study. • Most wanted to use the assistant again and perceptions were positive across the categories of disabilities disclosed. • Students appreciated AI-powered features such as receiving personalised suggestions gathered from fellow students. • Advances in GenAI offer opportunities to address identified needs such as answering complex questions with multiple sources.