AI-Assisted Feedback versus Teacher Feedback in Improving EFL Learners’ Writing Accuracy

Abstract

: This mixed-methods study compares AI-assisted feedback and teacher feedback in improving EFL learners' writing accuracy. Sixty intermediate Chinese university EFL learners were assigned to an AI group and a Teacher group. Data came from pre-/post-tests, error analysis, interviews, and journals. Both feedback types significantly improved overall accuracy, with no significant group difference. AI feedback was stronger for surface-level errors, whereas teacher feedback was stronger for deep-level errors such as coherence and word choice. Learners valued AI for immediacy but preferred teachers for contextual relevance, emotional support, and complex explanation. The study supports a blended feedback model guided by AI literacy and teacher mediation.