An ECNU Review of Education study explores how Large Language Models can transform teaching by acting as personalized assistants, automating routine tasks

As personalized education gains recognition as crucial for student achievement, the traditional one-size-fits-all approach to teaching faces significant challenges. A study by Jiayi Liu, Bo Jiang, and Yu'ang Wei from East China Normal University investigates how Large Language Models (LLMs) like ChatGPT can help educators overcome these challenges by automating teaching material generation, assessment, and feedback provision. This article was made available online on January 02 2025, in ECNU Review of Education.

The study identifies two key areas where LLMs can significantly enhance personalized teaching. First, LLMs excel at generating customized educational materials across diverse subjects, creating resources such as language learning quizzes, programming exercises, and curriculum-aligned learning objectives. Second, LLMs can streamline assessment processes by helping design evaluations, providing automated scoring, and generating targeted feedback for students, thereby substantially reducing the time educators traditionally spend on these tasks.

"The potential of LLMs to handle routine educational tasks allows teachers to focus on what they do best—mentoring students and creating meaningful learning experiences," says Jiayi Liu, lead author of the study. "This human-AI collaboration represents a promising direction for the future of education."

The study acknowledges that while LLM-generated content requires supervision and adjustment, it substantially reduces teachers' workload during pedagogical preparation. As these models continue to evolve, they may soon offer sophisticated, ready-to-use teaching materials across various subjects.

LLMs have demonstrated their ability to assist in various educational domains by generating quizzes, programming exercises, and curriculum-aligned learning objectives. These AI-driven tools help educators design structured and engaging course content while automating student assessments with accuracy comparable to human evaluation. By creating physics curriculum tasks, scoring essays, and providing personalized feedback, LLMs reduce teachers' administrative workload, allowing them to focus on student engagement. However, despite these benefits, AI-generated content often requires supervision and refinement to ensure accuracy and relevance. Researchers emphasize that the success of LLM-driven education depends on a balanced human-AI collaboration, where teachers curate and adjust AI-generated materials based on students' unique needs.

The study highlights the ideal teacher-LLM collaboration model, where educators act as orchestrators, integrating AI-generated materials into lesson plans while LLMs function as assistants providing structured content and automated assessments. "LLMs are not here to replace teachers but to enhance their capabilities," states Jiayi Liu, emphasizing that thoughtful AI integration can create dynamic and personalized learning environments. The research calls for further empirical studies to optimize AI-generated content, improve adaptability in feedback mechanisms, and explore integration strategies across diverse educational settings. With careful implementation, LLMs have the potential to revolutionize education by making personalized learning more scalable and accessible worldwide.