Blending Human and Artificial Intelligence: An Empirical Study of Student Perceptions of AI-Generated Feedback in EFL writing Instruction

ABSTRACT: This study examines Korean university students’ perceptions of AI-generated feedback compared to traditional instructor feedback in an EFL writing course. Thirty-one undergraduate students received blended feedback from both their instructor and GPT-4 on essay assignments. Survey data assessed students’ ability to distinguish between feedback sources, perceived helpfulness, comprehension levels, and preferences. Results indicate that 83.9% of students found AI feedback helpful or very helpful, while 41.9% could not distinguish between AI-generated and teacher feedback. Notably, 67.7% preferred receiving both AI and teacher feedback rather than a single source. Students particularly valued grammar corrections, organizational suggestions, and concrete examples. These findings suggest that AI-generated feedback can effectively complement traditional instructor feedback when thoughtfully integrated into writing pedagogy.

Keywords : AI feedback, EFL writing, student perceptions, automated writing evaluation, generative AI, English language teaching