From Prompting to Pedagogical Design: Secondary Mathematics Teachers’ Experiences with a Mixed-Model Prompt Engineering Framework for AI-Assisted Exemplar Development

Abstract

Generative artificial intelligence is increasingly used in mathematics lesson planning, yet teachers often lack a structured process for designing, refining, and validating prompts. This study examined the experiences of 39 secondary mathematics teachers who participated in an AI-assisted exemplar-development training using the Mixed-Model Prompt Engineering framework. The model guided teachers through persona, task, and context definition; output structuring; validation and iterative refinement; curriculum alignment; and final human checking. A convergent mixed-methods program-evaluation design with qualitative priority was employed. Quantitative data were obtained from an 11-item post-training evaluation, while written responses from 36 participants were analyzed thematically. The program received a high overall rating (M = 4.85, SD = 0.38), with speaker effectiveness and content relevance obtaining the highest ratings (M = 4.95). Five themes emerged: prompt engineering as a professional practice, AI-assisted instructional design, learner-responsive mathematics teaching, teacher empowerment, and conditions for sustainable implementation. Participants valued structured prompting, lesson-exemplar development, and reduced preparation demands, but identified the need for additional practice, reliable connectivity, continued training, and human verification. The study contributes to mathematics education by proposing pedagogical prompt literacy as an emerging teacher competence that integrates mathematical knowledge, instructional judgment, contextual awareness, and critical evaluation of AI-generated materials.