Integrating Large Language Models into Academic Writing Courses: Application and Practice

Abstract

As large language models (LLMs) are increasingly integrated into higher education, this study explores their application in thesis writing instruction. Based on an English academic writing course in a medical college, we designed and implemented a blended “AI +teacher+student” teaching model. This framework embeds an LLM throughout the entire learning process, including pre-class preparation, in-class practice, and post-class enhancement. The results demonstrate that the LLM provides intelligent writing assistance, personalized learning pathways, and real-time feedback, significantly improving students’ writing efficiency and quality while stimulating their interest in learning. Consequently, the teacher’s role shifts from a traditional knowledge transmitter to a facilitator and enabler of learning. While affirming these pedagogical benefits, the study also analyzes challenges observed during implementation, such as student over-reliance on the model, inaccuracies in AI-generated content, and issues with adherence to academic norms. To address these challenges, we propose strategies including strengthening academic integrity education, establishing clear AI usage guidelines, and enhancing teachers’information literacy. The findings suggest that integrating LLMs can effectively overcome traditional instructional bottlenecks, provided there is careful guidance to ensure proper use and uphold academic standards. This study offers empirical insights for innovating academic writing pedagogy, highlighting AI’s potential to empower education while stressing the critical importance of balancing technological advantages with educational ethics.