The article argues that artificial intelligence (AI) has so far failed to fundamentally change how math is taught in classrooms — and there’s little reason to think it will do so soon. Despite widespread excitement about AI’s potential in education, experienced educators and specialists say the technology has mostly produced sketchy, incomplete lesson materials rather than high-quality, usable instruction. Typical AI-generated resources often lack depth, omit key steps, and require teachers to rework them extensively before they can actually be used in class, limiting their practical value for everyday teaching.
A major reason AI hasn’t taken off is that effective math instruction depends on social, interactive learning between teachers and students — something that AI tools struggle to replicate. Good math teaching involves conversation, exploration and building shared understanding, and teachers play a central role in guiding those discussions, responding to student ideas, and creating high-trust learning environments. Simple automation or chatbot responses, by contrast, can’t foster the kinds of teacher-student interactions that drive deep understanding.
Surveys show that very few math teachers are actively using AI tools in their instruction, with only a marginal increase in adoption compared with previous years, and most usage confined to peripheral tasks like writing emails, generating ideas or preparing supplemental resources. Teachers also report that much of the current AI output needs heavy editing or additional context before it’s classroom-ready, and they don’t feel that AI yet helps them address the core challenges of engaging diverse learners in rigorous mathematical reasoning.
The article also cautions against the hype that AI will replace traditional instructional challenges such as building problem-solving, argumentation and communication skills. Evidence from existing research suggests that promising improvements often fade when studies control for other factors like tutoring time or parental support, and in some cases AI use has even been linked with weaker learning outcomes because students outsource thinking rather than reason through problems themselves. For now, educators emphasize that foundational pedagogy and human interaction remain central to math learning, with AI potentially blending into background tools only when it truly supports, rather than distracts from, instruction.
