AI-Assisted Feedback in EFL Writing: Implications for Writing Accuracy and Learner Autonomy

Abstract

This structured thematic literature review synthesizes peer-reviewed empirical research published between 2020 and 2026 on AI-assisted feedback in English as a Foreign Language (EFL) writing, examining its dual implications for writing accuracy and learner autonomy. Drawing on foundational theories of written corrective feedback, self-regulated learning, and learner autonomy, the review analyzes studies comparing AI-generated, teacher-provided, and hybrid feedback models. Findings indicate that AI feedback appears to improve surface-level accuracy in the reviewed studies, although its effectiveness remains inconsistent regarding higher-order concerns, including coherence, argumentation, and organization. Furthermore, its influence on learner autonomy depends largely on pedagogical design, learners’ critical engagement, and teachers’ guidance. The review argues that accuracy and autonomy are interdependent outcomes shaped by instructional integration rather than technological tools alone. It identifies critical research gaps, particularly the lack of longitudinal evidence and limited research involving lower-proficiency learners, and proposes pedagogical recommendations for human-centered AI implementation in EFL writing classrooms.