Encouragement as Pedagogy: Using AI-Mediated Mathematical Participation to Address Math Anxiety in Education and Outreach

Math anxiety, a persistent apprehension about mathematical tasks that measurably degrades working memory and performance, affects a large share of students and adults and leads many to avoid mathematics entirely. We argue that the same division of labor that allows a non-mathematician to contribute to frontier mathematical research, supplying framing and encouragement while an AI system supplies technical execution, offers a promising template for math education and outreach aimed at anxious learners. Rather than asking an anxious student to close a knowledge gap before engaging with mathematics, this model lets them engage first, in a low-stakes register (asking questions, expressing uncertainty, requesting another attempt) while an AI tutor absorbs the technical load and a human retains responsibility for pacing, tone, and correctness. We connect this argument to two independent evidence bases: research on math anxiety and avoidance, and recent randomized evidence that supervised generative AI tutoring can match or exceed human one-to-one tutoring on mathematics learning outcomes. We argue that the encouragement-only mode of participation documented in AI-assisted mathematical research is not a curiosity restricted to elite problems, but a design pattern with direct application to classrooms and public outreach, and we outline the safeguards this pattern requires, particularly continued human oversight of tone, pacing, and factual accuracy.