Vietnamese English as a Foreign Language Instructors’ Perceptions of AI-Generated Feedback on Students’ Writing: Benefits, Limitations, and Recommendations

Writing is an important language skill in English as a foreign language (EFL) context, yet many students encounter difficulties with both surface-level errors and higher-order writing skills. With the development of artificial intelligence (AI), AI-generated feedback has emerged as a potential tool for writing instruction. This qualitative case study explored Vietnamese EFL instructors’ perceptions of the benefits and limitations of AI-generated feedback on students’ writing regarding grammar, vocabulary, content, and organization, and their recommendations for its integration in classrooms. Semi-structured interviews were conducted via Zoom with 10 instructors at a university in An Giang province, Vietnam, who were selected through criterion-based purposive sampling. The interview data were analyzed using thematic analysis following Braun and Clarke’s (2006) six-phase framework. The findings revealed that, although instructors valued AI feedback as a multidimensional scaffold, their insights exposed critical pedagogical trade-offs such as cognitive passivity, formulaic writing structures, loss of personal voice, and reduced learner agency. To address these concerns, they recommended human–AI feedback integration, learner agency, and context-dependent integration. The findings suggest that AI-generated feedback should complement, rather than replace, teacher-generated feedback, highlighting the importance of balancing automated feedback with human mediation in higher education.