ERNIE-assisted feedback and its impact on EFL learners’ revision strategies and writing performance

Generative AI tools have been increasingly adopted in higher education to facilitate English as a foreign language (EFL) academic writing. This mixed methods study examined EFL students’ revision strategies and performance when assisted by ERNIE, an AI-feedback system. We also examined feedback-uptake challenges and how personal resources and learning motivation influenced the seeking, receiving, and implementing of ERNIE feedback. Participants were 117 first-year Chinese undergraduates. Data included screen recordings of revision processes, revised scripts, and semi-structured interviews. A series of ANCOVAs showed significant improvements in writing quality across all five measures, with the effect sizes ranging from small to large. Feedback engagement varied notably depending on students’ motivation, proficiency levels, feedback efficacy, and accountability. The study provided new insights into research on feedback engagement, urging the research community to redefine the core concepts of originality and plagiarism in relation to ethical and beneficial use of AI tools in higher education.