AI-Supported Feedback and the Development of Critical Academic Writing Skills among English Language Learners: A Mixed-Methods Study

Abstract

There is a speed with which artificial intelligence (AI) integration into higher education is altering the teaching, practice, and editing of academic writing. AI-supported feedback is especially important for English language learners (ELLs) since it can rapidly offer tailored, individualised responses to content on language, organisation, argumentation and revision. However, the pedagogical value of this type of feedback cannot be assessed by grammatical accuracy alone. Learners must take an opposing standpoint, assess evidence pertinent to a concept, create logical ties between any claim, query intended facts and information based on theory choice, and retain mental ownership regarding the content. This mixed-methods study investigates how feedback powered by AI supports the learning of critical academic writing skills among ELLs in argumentation, critical analysis, coherence, integrating evidence and academic vocabulary, as well as feedback literacy and self-regulated revision. This is a convergent mixed-methods design analysing both pre- and post-intervention writing assessments with the results of semi-structured interviews and reflective journals. This study aims not only to find out if writing performance has any transformation after long-term interaction with AI feedback, but also how students read, interpret, accept or reject, and transform that feedback in revision. It also examines the dangers of offloading cognitive effort, formulaic prose writing, excessive reliance on auto-generated recommendations, and diminished authorial agency. Overall, the evidence points towards a human-in-the-loop model in which AI plays the role of a dialogic feedback tool rather than an autonomous writer or substitute for teacher judgement. This paper argues that the actual educational potential of AI-supported feedback relies less on the provision of a mighty model per se and more on other factors related to feedback literacy, critical judgement, task design, and pedagogical guidance in which it is embedded.