Should we be training disabled students how to use Generative AI effectively?

About me

Rhona is employed at the University of Dundee as the Assistive Technology & Alternative Formats Officer. While working, she completed a part-time MSc in Educational Assistive Technology, graduating in 2025. With a keen interest in Generative Artificial Intelligence and the use of mainstream technology for assistive purposes, she designed her MSc dissertation to investigate the potential of GenAI as an assistive technology.

Why research this?

Although assistive technology can prove highly beneficial for disabled students, many encounter delays or barriers in accessing these tools. Common challenges include lengthy waiting lists for ADHD and autism diagnostic assessments, Disabled Students’ Allowance delays and the stigma associated with seeking support from Disability Services or in using assistive technologies. As such, the project aimed to explore the potential of a highly popularised, free mainstream technology (Microsoft CoPilot) when used for assistive purposes. It hoped to identify if GenAI could prove an effective additional tool to help disabled students succeed in Higher Education.

Methodology

The research project aimed to identify disabled student perspectives on the use of Microsoft CoPilot as a tool for exam revision. Participants were invited to attend a 45-minute group training session to develop an understanding of key terminology, learn a prompt framework, discover example prompts relating to revision and discuss the ethical concerns. Following the session, participants completed an online survey to investigate their use of GenAI before the training, the impact of the training on their self-reported confidence to use GenAI for study purposes, and their perceptions of CoPilot as a tool for revision.

Following institutional data security concerns about the proposed student use of Microsoft CoPilot in the training, an Initial Data Risk Assessment was completed before the sessions took place. This identified concerns regarding ‘data grounding’, defined as, ‘the ability for GenAI technology to utilise other information sources such as web content or user data to help it respond to a prompt’ (Google Cloud, 2024). As a result, students were not permitted to use CoPilot during the sessions and key security measures within CoPilot were highlighted.

Findings

29 disabled students participated in one of eleven training sessions, 24 female, 5 male. Specific learning disability or mental health conditions were the most common disabilities.

• There was a wide-ranging reported use of GenAI pre-training, with 41% reporting no prior use, while 31% used it at least once a week.
• 90% reported that they were more or much more confident in using GenAI post-training.
• 96% probably or definitely would use GenAI to support their revision after the training.
• A fear of academic misconduct was the most common concern regarding GenAI use at university.
• Hallucinations (inaccuracies) was reported as the most common concern regarding GenAI use for revision.

Recommendations

The project proved highly successful, with students appreciating the opportunity to increase their knowledge of GenAI, learn how to interpret the university’s GenAI guidance and discuss the associated ethical concerns. The key recommendations for other Higher Education Institutions are:

• Disability Services may wish to consider the potential benefits of delivering prompt training.
• Universities must provide clear guidance outlining the extent to which GenAI use is permitted.
• A Data Protection Impact Assessment should be undertaken for GenAI tools before they are used in educational institutions.

Get involved

Please get in touch with Rhona (RPayne001@dundee.ac.uk) if you would like further information about the project or wish to discuss similar research or professional work that you have undertaken.

We’ll also be exploring more around current research in assistive technology with an online event in the new academic year. Join the Assistive Technology Network Jiscmail community for updates.

Want to learn more?

Turing lectures on AI

Alshaigy, B. and Grande, V. (2024) October. Forgotten Again: Addressing Accessibility Challenges of Generative AI Tools for People with Disabilities. In Adjunct Proceedings of the 2024 Nordic Conference on Human-Computer Interaction (pp. 1-6).

Fletcher, R. et al. (2024) How is GenAI used by Dyslexic, Neurodivergent and Disabled students in Higher Education?. Available at: 10.13140/RG.2.2.16852.82563

Hadar Souval, D. (2025) ‘Transforming perceptions: exploring the multifaceted potential of generative AI for people with cognitive disabilities. JMIR Neurotechnology, 4, p.e64182.

Zhao, X. et al. (2024) A Report on the Use & Attitudes Towards Generative AI Among Disabled Students at the University of Sheffield. Available at: 10.15131/shef.data.25669323.vi

References

Google Cloud (2024) Grounding Overview, Available at: https://cloud.google.com/vertex-ai/generative-ai/docs/grounding/overview