1 Introduction

This section introduces the application of generative artificial intelligence (GAI) in foreign language education, particularly its role in providing feedback in writing instruction. The advancement of GAI technology offers new solutions to the challenge of delivering personalized writing feedback within limited time constraints. Research indicates that both GAI-generated feedback and teacher feedback can enhance students' writing quality, with GAI demonstrating advantages in speed, personalization, and cost-effectiveness. However, the quality of GAI feedback depends on prompts, carries the risk of erroneous feedback, and overreliance may weaken students' independent thinking skills. This study aims to explore the added value of GAI feedback in improving the quality of university students' English essays when combined with teacher feedback.

2 Literature Review

This section discusses the application of generative artificial intelligence (GAI) in foreign language writing instruction and its current status in writing feedback research. Compared to traditional AI, GAI emphasizes creativity over prediction, processes a broader range of data types, and has facilitated a shift from automated writing evaluation (AWE) to GAI feedback in foreign language writing instruction. Applications of GAI in this field include curriculum reform, academic writing instruction, grammar correction, and essay assessment. Studies show that GAI can provide timely and effective feedback, helping students grasp research trends, enrich writing ideas, and improve writing efficiency. Research on GAI feedback indicates that effective writing feedback enhances students' writing skills, with GAI offering new approaches and tools. Existing studies fall into two categories: those focusing on teacher perspectives by comparing GAI and teacher feedback, and those analyzing student responses to GAI feedback. However, limitations include an overemphasis on feedback content comparison rather than writing quality improvement, a lack of research allowing students to freely generate GAI feedback, and insufficient exploration of the effects of multiple rounds of GAI and teacher feedback. This study aims to address these gaps by comparing changes in writing quality before and after teacher feedback versus teacher+GAI feedback to assess the added value of GAI feedback.

3 Research Methods

This section describes the research methodology, including participants, data collection, and data analysis. Participants were first-year non-English majors from a coastal 985 university, divided into experimental and control groups with comparable English proficiency. Data collection occurred during regular coursework, with students completing four essays. Teaching assistants and another rater scored the essays using CET-4 grading criteria to ensure reliability. The experimental group revised essays using GAI alongside teacher feedback, while the control group relied solely on teacher feedback. Data analysis was conducted using SPSS 27, including normality tests, within-group paired-sample t-tests, and between-group independent-sample t-tests to evaluate the effects of the two feedback approaches on writing quality improvement.

4 Research Results

This section analyzes the performance of both groups across four essays, revealing that both groups showed improvement in draft and revised essay scores, with revised versions consistently outperforming initial drafts. The teacher+GAI feedback group demonstrated greater score improvements than the teacher-only group. Within-group paired-sample t-tests confirmed that post-feedback scores were significantly higher than pre-feedback scores, with large effect sizes. Independent-sample t-tests indicated significant differences in score improvement between the two feedback approaches in the latter two essays, with the added value of GAI feedback becoming more pronounced as feedback iterations accumulated.

5 Discussion

This chapter examines the impact of teacher+GAI feedback versus teacher feedback on Chinese university students' English writing quality. Findings suggest that both approaches significantly enhance writing proficiency, with GAI feedback exhibiting a potential cumulative effect—its added value becomes increasingly evident with repeated use. GAI feedback matches teacher feedback in quality and aids writing improvement. Students' initial draft scores steadily increased under both feedback conditions, indicating the positive role of corrective feedback in skill development. GAI feedback can generate high-quality responses quickly, saving teachers time and improving feedback efficiency. However, GAI cannot fully replace teachers; educators should maintain a collaborative relationship with GAI, treating it as a supplementary tool. Early-stage GAI users may lack experience in effectively utilizing feedback, limiting its potential benefits. Over time, however, students adapt and refine their GAI feedback usage, allowing its added value to emerge. Teachers should guide students in crafting appropriate prompts and applying GAI-generated feedback effectively to enhance their GAI feedback literacy and maximize its utility.

6 Conclusion

This chapter concludes that both GAI and teacher feedback improve university students' English writing quality and recommends combining the two in writing instruction. GAI cannot fully replace teachers, who must oversee feedback quality and guide students in its proper use. Study limitations include a small sample size, a single essay type, potential ceiling effects in scoring, and variability in students' GAI feedback literacy. Future research should further investigate the efficacy of GAI feedback.