The AI agents can be used to change students’ understanding of technical concepts and perception of themselves, Du said.
“The system adopts the learning-by-teaching framework and incorporates learning science theories and instructional strategies, such as conceptual change and social psychological interventions, to engage and guide students toward deeper math understanding,” she said.
An example of these interventions is the ability for the AI to generate prompts that guide students through the problem-solving process. It can also offer real-time assessment and feedback to correct mistakes and address misconceptions to help deepen student understanding. Students’ anxiety and confusion are expected to be lessened through collaboration and discussion with the support of AI, Du said.
One barrier for AI-enhanced learning is concerns over the trustworthiness of information, and research has pointed out students do not sufficiently trust AI when learning new concepts, Du said. Through her research, Du hopes to find ways to empower educators on how to integrate AI into their classrooms and build student trust.
“[Large language models] do not take responsibility for students’ learning, but educators do,” she said. “I believe our work about how to integrate AI in support learning provides valuable insights for educators and researchers about approaching AI for education, as well as practical references about what to know, and potentially how to, when integrating AI for learning.”
The project is currently at the development and data collection stage, but Du hopes the study will allow researchers to utilize data-mining techniques and learning theories to identify meaningful learning behaviors that can be widely applied in K12 math education or even in a broader context for K12 STEM education, she said.
“By exploring the possibilities of applying AI for large-scale learning platforms like Math Nation, we hope to provide insights about how AI can be used to support formal education, identify the advantages and challenges of this method, develop a practice model for training teachers to teach with AI, and present a model of combining research-based pedagogies and advanced learning technologies to support learning,” Du said.
Mikayla King (‘17) covers the College of Science and Engineering and Woodring College of Education for the Office of University Communications. Reach out to her with story ideas at kingm24@wwu.edu.
