Abstract
Generative artificial intelligence (GenAI) refers to technology that leverages large language models to generate content in response to human prompts and it is being increasingly used in school classrooms. A systematic literature review was conducted to identify empirical research which has investigated the application of GenAI for neurodiverse student populations at school. The study approaches and outcomes were analysed through the lens of Universal Design for Learning (UDL) and Strengths-Based Differentiated Instruction (SB-DI). The analysis suggested that GenAI can support some elements related to the three main principles of UDL. First, to create diverse formats of information (e.g., concept maps and summaries), thereby supporting multiple means of representation. Second, to provide feedback (e.g., on writing and language learning), thereby supporting multiple means of action and expression. Third, to personalise learning, provide novel challenges, and foster interest, thereby supporting multiple means of engagement. GenAI also showed evidence of being able to support some SB-DI strategies, identifying and leveraging strengths, suggesting individualised support, and opportunities to address challenges to learning. However, the analysis suggested that GenAI requires careful implementation for neurodivergent students to ensure that it is used safely and ethically, gives accurate information, and complements, rather than replaces, effective teaching practices. Further research is needed given the small number of studies conducted to date and this research needs to include more diverse populations and stronger study designs. Nevertheless, combining UDL's proactive design principles with strengths-based differentiation strategies has the potential for educators to be able to maximise GenAI's benefits for neurodivergent students.