Artificial Intelligence in the Classroom: Implications for Pedagogical Design and Student Agency in Secondary Education

As artificial intelligence (AI) technologies become increasingly embedded in global education systems, their pedagogical implications remain underexplored particularly within secondary school contexts. This study investigates how AI tools are shaping teaching strategies and student agency in Chinese secondary classrooms, focusing on three AI pilot schools in Zhejiang Province. The study addresses three core research questions related to AI’s role in pedagogy, learner autonomy, and instructional challenges. In a qualitative case study design, data were gathered from semi-structured interviews with 15 teachers, non-participant classroom observations, and analysis of lesson plans and digital artefacts. Teachers viewed AI as not a substitute for teaching but as a collaborative co-planning partner with adaptive pedagogy, customized feedback, and real-time amendments. AI platforms enabled self-regulated learning but raised concerns regarding automation bias and over-reliance on algorithmic feedback. Student autonomy was not limited to cognitive freedom but also encompassed emotional comfort and rapport in class. Teachers noted that AI feedback caused less anxiety than human correction, which increased motivation. These affective-emotional factors are particularly applicable in high-stakes, performance-oriented systems such as China's. While provincial infrastructure and policy facilitation existed, issues such as few training opportunities, shared smart classrooms, and testing pressures limited meaningful innovation. Teachers frequently employed AI for performance-oriented activities instead of conceptual exploration. When aligned with class needs, though, AI facilitated differentiated instruction and communication with parents and students. This research underscores the importance of pedagogical fluency, institutional alignment, and ethical integration in realizing AI’s educational potential. The findings contribute to ongoing discourse on human-centered AI in education, emphasizing the need for guided agency, teacher empowerment, and context-aware strategies to ensure sustainable, equitable adoption in secondary schools.