Effects of generative AI-supported feedback on academic writing quality in senior secondary education

Timely, specific feedback is central to writing development, yet teacher-provided feedback in senior secondary classrooms is often constrained by time and class size. This study examined whether generative AI-supported, rubric-aligned feedback with brief teacher mediation improves students' academic writing quality and writing self-efficacy. A quasi-experimental pretest-posttest design was implemented with 60 senior high school students assigned by intact class to either teacher feedback only or AI-supported feedback. Writing quality was evaluated with an analytic rubric covering content, organization, vocabulary, language use, and mechanics, and self-efficacy was measured at pretest and posttest. ANCOVA showed significantly higher adjusted posttest writing quality in the AI-supported condition, F(1, 57) = 8.77, p = 0.004, partial η² = 0.13, with the largest gains in organization, vocabulary, and mechanics. Writing self-efficacy was also higher in the AI-supported condition. These findings indicate that generative AI can strengthen classroom writing instruction when used as a supplement to teacher feedback. For practice, schools may adopt generative AI as a feedback aid when its use is guided by rubric criteria, teacher mediation, privacy safeguards, and responsible-use policies.