Generative AI and the Transformation of EFL Writing: Feedback, Revision, and Learner Agency

Abstract

The rapid development of Generative Artificial Intelligence (GenAI) has significantly transformed the landscape of English as a Foreign Language (EFL) writing, particularly in relation to feedback, revision, and learner participation in the writing process. Generative AI tools have enabled learners to obtain immediate feedback, explore alternative linguistic expressions, revise their texts iteratively, and engage in more flexible forms of writing support. These developments have created new opportunities for improving writing practices while simultaneously raising concerns about the extent to which learners remain actively involved in evaluating feedback and making decisions about their own texts. This study aims to critically examine and synthesize existing scholarly literature on the role of GenAI in EFL writing, with particular attention to AI mediated feedback, writing revision, and learner agency. The study employs a library research approach by examining relevant academic publications addressing Generative AI, EFL writing, automated feedback, writing revision, learner autonomy, and learner agency. The selected literature is analyzed thematically to identify recurring findings, emerging patterns, pedagogical opportunities, limitations, and unresolved issues. The synthesis indicates that GenAI can provide accessible, immediate, and iterative feedback while supporting learners in engaging with multiple possibilities for textual revision. However, the educational value of AI generated feedback depends substantially on learners’ ability to evaluate its relevance, accuracy, and appropriateness rather than accepting its recommendations uncritically. Generative AI can strengthen learner agency when learners retain control over writing decisions, but excessive dependence on AI may reduce cognitive engagement and ownership of the writing process. Therefore, the transformation of EFL writing through GenAI should be understood not merely as the automation of feedback and revision, but as a broader transformation of how learners interact with feedback, exercise judgment, and participate in the construction of their written texts