Generative AI for Inclusive Learning

About the Project

People with learning difficulties often face barriers in traditional education systems because standard teaching tools are not sufficiently adaptive to individual cognitive differences. Generative AI (such as large language models) offers a new frontier: the ability to dynamically personalise explanations, scaffold support, rephrase content, and respond interactively to learners. However, this promise has significant challenges: hallucinations and misleading responses, lack of long-term reliability, embedded biases, learner overreliance on AI, and weak modelling of individual learning profiles.

This PhD project will investigate how to design safe, robust, and adaptive generative AI systems that can support learners with learning difficulties (for instance, dyslexia, attention deficits, processing variances) without introducing new harms. The project may require you to combine software development, user interaction design, and empirical evaluation in real learning environments.

Key strands of the work may include one or several of the followings:

1. Mitigating hallucinations and misleading content: One of the biggest risks with generative models is that they sometimes produce outputs that appear convincingly correct but are actually incorrect or misleading answers. In educational settings, especially with learners who may struggle to detect errors, such outputs can reinforce misconceptions or confusion. You will explore mechanisms to detect or filter hallucinated content before it reaches learners — for example, using fact-checking modules, secondary verification agents, or confidence scoring.
2. Ensuring robustness and longitudinal reliability: Many AI-augmented learning tools are pilot systems evaluated over short periods. But learners with difficulties benefit most from sustained, consistent support over months or years. This strand investigates how to maintain stable performance, adapt to concept drift, and resist degradation over time — for example via continual learning, fail-safe fallback strategies, or monitoring of drift.
3. Addressing bias, representation, and inclusion: Models trained predominantly on “average learner” data risk embedding ableist assumptions or failing to represent neurodiverse profiles. Your research will examine how bias manifests (e.g. giving overly terse instructions, under-scaffolding, mischaracterising errors) and will propose mitigation strategies: inclusive training data, fairness constraints, adjustable output styles, and stakeholder review (e.g. with educators and learners) of generated materials.
4. Preventing overreliance and preserving learner agency: While AI can assist, it should not replace critical thinking or effort. There is a danger that learners might defer too readily to AI suggestions and stop trying themselves. You will explore interaction designs that maintain learner engagement (e.g. prompting learners to justify or check AI suggestions), graduated assistance (less help over time), or hybrid modes where the AI nudges but the learner leads.
5. Personalising via cognitive/learning profiles: General adaptation is not enough — you will model individual differences such as working memory capacity, processing speed, error patterns, attention fluctuations, and prior knowledge. The system will dynamically tailor explanation length, scaffolding depth, pacing, and feedback style to the learner’s profile, testing whether such personalised support improves learning and satisfaction.

You will develop prototypes of generative tutoring systems incorporating some of these features, and deploy them in small-scale, real settings. Evaluation will include controlled experimental measures (learning gain, error correction, retention), longitudinal monitoring (stability, drift), and qualitative user feedback (acceptance, trust, perceived usefulness).

Academic qualifications

Have, or expect to achieve by the time of start of the studentship a first-class honours degree, or a distinction at master level, ideally in Computer Science, Software Engineering or Artificial Intelligence, with a good fundamental knowledge of Machine Learning, Data Science / Informatics, Electrical Engineering, Cybersecurity, Mathematics / Applied Mathematics / Statistics, Cognitive Science / Computational Cognition (especially if the project bridges AI and learning theory), Education Technologies

English language requirement

IELTS score must be at least 6.5 (with not less than 6.0 in each of the four components). Other, equivalent qualifications will be accepted. Full details of the University’s policy are available online.

Essential attributes:

• Programming (Python)
• Software engineering fundamentals (data structures, algorithms, modular design)
• Strong written and oral English communication
• Self-organisation, initiative, and discipline

APPLICATION CHECKLIST

• Completed application form
• CV
• 2 academic references, using the Postgraduate Educational Reference Form (download)
• Research project outline of 2 pages (list of references excluded). The outline may provide details about:
  1. Background and motivation of the project. The motivation, explaining the importance of the project, should be supported also by relevant literature. You can also discuss the applications you expect for the project results.
  2. Research questions or objectives.
  3. Methodology: types of data to be used, approach to data collection, and data analysis methods.
  4. List of references.

The outline must be created solely by the applicant. Supervisors can only offer general discussions about the project idea without providing any additional support.

• Statement no longer than 1 page describing your motivations and fit with the project.
• Evidence of proficiency in English (if appropriate)

To be considered, the application must use

• the advertised title as project title

For informal enquiries about this PhD project, please contact Prof Ashkan Sami - A.Sami@napier.ac.uk