Exploring the interactions among instructor, prospective teachers and AI in facilitating mathematics learning

Abstract

Mathematics teacher educators (MTEs) have an important obligation to prepare prospective elementary mathematics teachers (PEMTs) to use artificial intelligence (AI) in their classrooms. To better understand how AI can mediate the interactions between students, teachers, and mathematics, we examined PEMTs' perceptions of AI-integrated learning sessions in a foundational content course. We employed inductive and deductive thematic analysis to categorize these perceptions along three edges of a didactic tetrahedron for teaching and learning (AI–PEMT interactions, PEMT–instructor interactions, and AI–instructor interactions). The instructor's use of Khanmigo as an AI tutor to facilitate small-group problem-solving yielded several experiences categorized along the AI–PEMT and instructor–PEMT edges of the didactic tetrahedron. AI–PEMT interactions created robust opportunities to communicate and construct mathematical knowledge with Khanmigo. Instructor–PEMT interactions provided a mathematical and technological “safety net”, mitigating challenges PEMTs faced in communicating with AI. Limited evidence of PEMT descriptions of AI–instructor interactions suggests MTEs should intentionally model their instructional decision-making and partner with AI to facilitate their lessons. PEMTs can benefit from discussing how teachers foster meaningful student engagement with AI beyond getting answers to problems. Implications for recognizing challenges and realizing the affordances of using AI to deepen mathematics learning in content courses for PEMTs are offered.