Shifting student AI literacy through customised feedback to support writing assessment

As generative AI tools become increasingly embedded in students’ academic writing practices, English for Academic Purposes (EAP) teachers face renewed challenges in supporting meaningful feedback uptake. This article reports on a classroom-based action research project conducted in a high-stakes Direct Entry Course at the Centre for English Teaching, University of Sydney. The project explored the use of a customised, rubric-aligned AI chatbot designed to provide timely, on-demand writing feedback while promoting student feedback literacy. Rather than replacing teacher feedback, the tool was introduced through explicit scaffolding, modelling, and classroom activities that supported critical engagement with assessment criteria. Findings from this action research suggest that AI-mediated feedback is most effective when students are taught how to interpret, question, and apply feedback purposefully. The article concludes that the educational value of AI lies less in the sophistication of the technology itself and more in how teachers purposefully design, scaffold, and incorporate its use to support learning.