AI-Generated vs. Peer Feedback in EFL Writing: Effects on Writing Skills, Self-Efficacy, and Enjoyment by Proficiency Levels

Feedback is essential in writing instruction because it helps learners identify strengths, weaknesses, and areas for improvement. This study examined the effects of peer feedback and AI-generated feedback (iWrite and ChatGPT) on Chinese EFL college students' writing skills, writing self-efficacy, and enjoyment. Participants were 242 non-English-major undergraduates enrolled in a 16 -week university writing course in China and assigned to one of three conditions: peer feedback, iWrite, or ChatGPT. Students completed multiple writing tasks, received feedback, and revised their drafts across the semester. Pretest-posttest assessments showed no significant between-group differences on writing performance, self-efficacy, or enjoyment after controlling for pretest scores. English proficiency also did not significantly moderate these effects. These findings suggest that, in a rubric-aligned course context, AI-generated feedback can yield outcomes similar to peer feedback and may serve as a complementary support for EFL writing instruction.