Personalized learning, tailored to the needs of individual students, is widely recognized as one of the most effective approaches to improving educational outcomes. However, its implementation in public school classrooms has been impractical due to the one-to-many teaching model, while private education can afford it, further widening educational inequality in Korean English education. In this context, AI is emerging as an alternative pedagogical tool to facilitate personalized learning, prompting the Korean Ministry of Education to announce the implementation of AI digital textbooks (AIDT) for English education in primary and secondary schools starting in 2025. AIDT, powered by AI-based learning analytics, adaptive learning, and human-centered systems, provides personalized learning experiences. However, AIDT alone cannot guarantee successful learning outcomes and requires evolving roles for both students and teachers. Students will need to develop greater agency and self-directed learning skills to engage effectively with AI systems, while teachers will need to transition to roles as facilitators, curators, and social-emotional supporters—aspects that AIDT cannot provide. The chapter concludes by exploring the challenges of AIDT, including data management, privacy concerns, the AI divide, and educational inequities resulting from data and algorithmic biases, and proposes viable solutions.