The rapid development of generative artificial intelligence calls for agent technology to deeply optimize graduate academic writing teaching. By studying and analyzing the basic concept of agent, an agent system is constructed through specialized equipment, and the chain reasoning mechanism is used to promote the deep collaboration between teachers, students and technology. Based on 104 valid questionnaires, the dimensions and item network structure and centrality indexes of the academic writing assistant scale were systematically analyzed by using the regularized Gaussian graph model, which revealed the core position of the agent in improving the efficiency and quality of writing, and the multi-node synergistic effect of interaction experience and use intention dimensions. The research results not only enriched the theoretical perspective of Al-enabled writing teaching in higher education, but also provided an operational path and strategy for the construction of intelligent writing support practice for graduate students.
K. Ren, Y. Wang and Y. Wang, "From Teaching Assistant to Co-Intelligence: Agent Technology Reshapes the Paradigm of Teaching Academic Writing to Graduate Students," 2025 5th International Conference on Educational Technology (ICET), Chongqing, China, 2025, pp. 173-177, doi: 10.1109/ICET67421.2025.11380589.