Investigating AI-supported Reading for Scholars with ADHD
Under review, CHI 2027.
Overview
Scholarly reading is an open-ended activity involving both self-regulation and interpretation. We investigated how 15 scholars with ADHD manage this work and experience AI assistance through interviews and reading activities with an interactive research probe. The probe offered summaries, category-based highlights, source-linked question answering, and editable AI-drafted margin notes. Participants encountered the features with a common paper before reading a paper relevant to their own research.
Participants described ongoing work to give reading structure and direction, maintain its connection to a purpose, and preserve context for returning later. They valued AI support for this work while setting boundaries around the interpretation they wanted to develop themselves. Yet assistance could make reading manageable partly by supplying judgments about what mattered and how much to read, complicating that division of work.
Using and assessing assistance could also reintroduce demands for maintaining purpose and recovering context. We articulate these intertwined benefits and difficulties as design tensions, offering a situated account of how AI can support scholarly reading while reshaping the work researchers with ADHD seek help with.