Generative AI–Driven Personalized Learning: Construction and Application of a New Teaching Model

Abstract

Generative artificial intelligence (GenAI) is rapidly reshaping how learners interact with content, teachers, and peers. Instead of merely automating question answering, large language models and related tools can be embedded into a structured pedagogy that adapts goals, resources, and feedback to individual learners. This paper proposes an integrated framework for GenAI-driven personalized learning that combines data-informed learner modeling, generative content orchestration, and teacher-in-the-loop regulation. First, we review recent work on AI-enabled personalization, learning analytics, and ethics, and summarize the theoretical basis for combining mastery learning, self-regulated learning, and cognitive apprenticeship with GenAI. Second, we construct a path framework that describes how learner data, generative services, and human decision-making are connected in practice. Third, we present a teaching implementation case in a university “Programming Fundamentals” course where GenAI supports diagnostic questioning, micro-scaffolding, and reflective writing. Finally, we report empirical results from a quasi-experimental study (N = 192) comparing a GenAI-personalized class with a conventional blended class. Results show higher post-test scores, improved self-regulated learning indicators, and reduced help-seeking inequality, while also revealing tension around data privacy and over-reliance. We conclude with design implications and cautions for institutions planning to deploy GenAI-driven personalized learning at scale.