A CASE STUDY EXPLORING THE RELATIONSHIP BETWEEN INSTRUCTION TYPES AND PROMPT LEVELS IN PRESERVICE MATHEMATICS TEACHERS' CHATGPT-BASED LESSON DESIGN

This study examines how preservice mathematics teachers integrate artificial intelligence into lesson design and how the cognitive levels of the student prompts they propose vary according to their modes of AI utilization. Mock lesson materials created by seven groups enrolled in a mathematics education course were analyzed using open coding, which yielded four distinct types of AI-supported instructional design: using AI as a tool for answer verification (Type A), for delivering content (Type B), for guiding modeling processes (Type C), and for facilitating teacher-student interaction (Type D). The prompts that preservice teachers envisioned students would use were further classified according to the six cognitive levels of the Revised Bloom's Taxonomy. The analysis revealed a systematic alignment between instructional types and prompt levels: Type A was associated with the Remember level; Type B extended to Understand; Type C included Apply; and Type D generated higher-order prompts corresponding to Analyze, Evaluate, and Create. These findings indicate that the ways in which teachers position AI within the instructional process shape the cognitive demand placed on students. In particular, higher-order thinking tended to emerge when AI functioned as a partner within teacher-student-AI interactions rather than as a mere answer-providing tool. This study highlights the importance of preparing teachers to design AI-supported interactions that promote deeper mathematical reasoning and provides implications for preservice mathematics teacher education in the era of generative AI.