Rethinking Productivity with GenAI: A Neurodivergent Students' Perspective
Hira Jamshed, Mustafa Naseem, Venkatesh Potluri, Robin N. Brewer · 2025 · Proceedings of the 27th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS '25) · doi:10.1145/3663547.3746329
Summary
This qualitative study examines how 19 neurodivergent university students (including those with ADHD, autism, OCD, dysgraphia, and auditory processing disorder) use LLM-based generative AI tools like ChatGPT in their academic work. Through semi-structured interviews, the researchers explore students' motivations for adopting these tools, the challenges they encounter, and their vision for how GenAI could better support their learning. Students described using GenAI tools in four primary roles: as a personalized on-demand collaborator for brainstorming and explaining concepts, as an information decomposer that breaks content into digestible pieces aligned with their learning preferences, as an unbiased coach that provides judgment-free support (particularly valued by those who feel anxious about asking repetitive questions), and as a tedious-task worker that handles repetitive or non-essential work like formatting and drafting routine emails. The study draws on Alison Kafer's concept of crip time — which recognizes that disabled people experience and navigate time differently due to bodily, cognitive, or systemic factors — to critically examine how GenAI tools interact with neuronormative expectations of productivity in higher education. The researchers used reflexive thematic analysis to identify themes across students' accounts of motivation, use, refusal, and future visions for these tools.
Key findings
While students found GenAI tools genuinely helpful for managing their academic workload, the study revealed significant tensions. Students reported challenges integrating AI into their established routines, with those who have anxiety-related conditions like OCD particularly resistant to disrupting tried-and-tested workflows. Limited AI literacy support meant students only used basic features, unaware of customization options that could better serve their needs. The trial-and-error nature of prompt engineering added cognitive burden rather than reducing it, especially problematic for students already struggling with time management and procrastination. A striking finding was the "flattening of personality" — students with autism and language-processing conditions found that AI suggestions erased their distinctive voice, producing text that felt "dry," "robotic," or "corporate." Students also described a persistent guilt about using these tools, feeling it was "like cheating" even when they recognized the tools as accessibility aids. When envisioning the future, students wanted GenAI to function as a dedicated assistant capable of managing schedules, organizing tasks, monitoring energy levels, and prioritizing work — essentially supporting executive function rather than just generating content.
Relevance
This paper offers an important counterpoint to the dominant narrative that GenAI tools are straightforwardly beneficial for neurodivergent users. By applying crip time as a critical lens, the authors show how these tools can inadvertently reinforce neuronormative standards of productivity — speed, efficiency, linear output — rather than supporting the nonlinear, fluctuating ways neurodivergent students actually work. The three proposed design values of flexibility, adaptability, and self-authenticity provide concrete guidance for developers building accessible AI tools. Practical recommendations include pacing controls for output speed, context-aware nudges based on energy patterns, lightweight mood check-ins, and mode switching between focused and exploratory states. For accessibility practitioners, this research underscores that making AI tools accessible goes beyond interface compliance; it requires fundamentally rethinking what productivity means and ensuring tools adapt to users rather than pushing users toward neurotypical norms.
Tags: neurodiversity · generative AI · cognitive accessibility · higher education · disability studies · qualitative research · crip time · executive function