Future Teachers Overconfident in AI-Generated Math Lessons, JMU Study Finds

JMU Professors Earn National Fellowship for Research on AI in Math Education

Amanda Gantt Sawyer and Marcus Wolfe, professors in James Madison University’s College of Education, received the 2025 National Technology Leadership Initiative (NTLI) Fellowship from the Association of Mathematics Teacher Educators. Their awarded paper, "Exploring Preservice Teachers’ Adaptations of ChatGPT Created Mathematics Curricula: Discovering Overconfidence," examined how preservice teachers use AI tools like ChatGPT to develop and modify math lessons.

Their findings shed light on how future educators interact with AI-generated teaching materials and what this means for both students and teachers.

Why Study AI in Education?

The rise of online resources such as Teachers Pay Teachers and AI tools like ChatGPT has made it easier for educators to access ready-made lesson plans. However, Sawyer and Wolfe noticed a growing reliance on these resources without critical evaluation. They saw that lessons created by AI might not always encourage deep thinking or meet educational standards.

Because critical thinking is essential in teaching, they decided to investigate how reliable AI-generated lessons really are and how preservice teachers engage with them.

Overview of the Research

The professors developed a unit aimed at helping preservice teachers critically assess AI-generated math curricula. Their study focused on how these future teachers adapted lessons created by ChatGPT.

• They found that 64% of adaptations involved only superficial changes, such as improving the visual design.
• Many preservice teachers did not critically analyze the content or question its suitability for elementary students.

This confirmed concerns about overconfidence in AI-generated material and highlighted the importance of teaching future educators how to evaluate AI tools thoughtfully.

Surprising Insights

Despite receiving training on evaluating educational resources, many preservice teachers made only surface-level edits to AI-created lessons. They trusted ChatGPT’s responses even when the content was not fully appropriate.

This reveals a gap between knowing evaluation strategies and applying them effectively when working with AI. It also emphasizes the need to prepare educators to question and verify AI outputs rather than accepting them at face value.

Practical Tips for Integrating AI in Teaching

Teachers can make AI a helpful part of their toolkit by:

• Teaching students about AI’s limitations, including potential biases and inaccuracies.
• Encouraging critical thinking and careful review of AI-generated content.
• Using AI as a supplement rather than a replacement for strong teaching skills.

Proper guidance ensures AI tools support rather than hinder the learning process.

Impact of Receiving the Award

The fellowship holds special meaning for both educators. For Dr. Wolfe, early-career recognition motivates continued growth and contribution. For Dr. Sawyer, a faculty member since 2014, it honors her dedication to the field of mathematics education.

Such recognition also helps highlight the importance of research focused on AI’s role in education.

For educators interested in understanding more about AI tools and how to apply them effectively, resources like Complete AI Training’s ChatGPT courses offer practical guidance.