Research on Teaching Guidance and Personalized Feedback of Generative AI in Academic Writing

Abstract

Generative AI is already part of academic writing classes. But it comes with problems. It weakens the power of critical thinking, it gives generic feedback, and it provides few teaching hints. All these problems harm students’ depth of writing. Existing research on generative AI is messy too, particularly on mixing teaching guidance with personalised feedback. This review evaluates 58 papers (published 2023 to 2026) from the Dimensions database. This study organises these into four themes: quality of generative tools, what teachers and students think about them, current teaching models, ethics These findings reveal four things. The first is that AI helps with language but does not have much impact on creativity or judgment. The second is that most teachers and students like AI but are concerned about cheating, misusing AI, and poor feedback. The third is that schools now go beyond treating AI simply as a tool to become a teacher-AI-student approach to teaching. Fourth is that discussions about ethics all emphasise originality, bias and fairness. Little research exists on teaching effective AI use. Future research will need to include guided feedback, customized teaching, and longer-term studies with more explicit integration of technology in academic ethics.