Developing students’ agency and voice by using generative AI in an online EAP module

Using a three-stage small-scale classroom action research methodology, this article presents how the emergence of AI systems underpinned by the computational architecture of large language models (LLMs) has both empowered and challenged language teachers and learners. The university's RAG (‘traffic-light’) approach for using AI chatbots in assignments and its ‘amber’ policy were recently implemented in an online English for Academic Purposes module taught over the past three years. The study explores how structured AI chatbot use helped address student challenges from the previous iteration of the module, where unstructured AI use had led to increased student anxiety and poor learning outcomes. With structured AI assistance, students demonstrated much better understanding of the research proposal genre and showed more meaningful engagement with the tasks. The integration of AI in learning has empowered students to become more confident in academic communications, potentially increasing their output capabilities.HIGHLIGHTSWhen adequately scaffolded, AI use in EAP can enhance student agency.Introducing AI tools early in the academic writing process, particularly during the research conceptualisation phase, can help students engage more confidently with research, reduce anxiety, and enhance learning.