Multiplier Effect: Drexel Joins Gates Foundation Initiative to Develop AI Tools for Math Teachers

- Researchers Develop Effective Predictor of Severe COVID-19
- Novel Enzyme Activator Discovered as Potential Alzheimer’s Disease Therapy
- New Research Reimagines Primary Care Visits for People Living with Obesity
- Pearlstein Gallery Exhibitions Showcase Digital Rendering Talents of Drexel Community, Explore the Nature of Consciousness

Researchers from Drexel University’s School of Education will join peers from Ursinus College and the 21st Century Partnership for STEM Education in an effort to strengthen artificial intelligence tools that will enhance math teaching and learning. This work is part of a Gates Foundation initiative to improve math education by better supporting students. Drawing on mentoring data and insight from math educators, the team aims to improve how AI tools respond to student work. The project will also explore how AI can assist teachers in providing focused feedback that equips students to build their conceptual understanding of math over time.
“Timely, personalized feedback is essential for helping students make sense of mathematical ideas, reflect on their thinking and stay engaged in the learning process,” said Jason Silverman, PhD, a professor of mathematics education and associate dean for Faculty Affairs and Research in the School of Education, who is helping to lead the project. “Engaging with each student’s thinking is a central goal of teaching, but one that’s often difficult to achieve consistently. Large class sizes and limited prep time make it hard for teachers to provide feedback to students regularly.”
The collaboration will use a $500,000 grant from the Foundation over the next year to lay out a proof-of-concept for a system that uses a large language model trained on data generated by the Problems of the Week Service, an initiative of The Math Forum, an online mathematics education community, formerly housed at Drexel, and most recently part of the National Council of Teachers of Mathematics. Key to the project’s success is the collaboration between educators and students to refine the large data set and train the model to focus on supporting sustained interest and engagement — and not be limited to remediation of common mathematical errors.
“By drawing on an extensive database of student work and expert mentor responses — and combining that with current teacher expertise — a tool like this can help teachers move toward their goal of providing each student with timely, meaningful, individualized support,” Silverman said. “Even if the AI-generated feedback isn’t perfect, starting with a high-quality draft makes it far easier and faster to revise than creating responses from scratch, an important advantage when the time required to reach every student individually can be overwhelming.”
While this project’s proof-of-concept system, called MENTIR-AI (short for Mathematics Education Network for Teaching and Instructional Responses), focuses on coaching teachers’ feedback, the same model could be extended to provide live, in-class feedback to students, out-of-class tutoring systems and support in many other educational contexts.
“Many students come away from math classes believing that their thinking doesn’t matter. In the best cases these students try to follow and mimic the thinking of others; in the worst cases, they give up,” said Stephen Weimar, director of Online Math Programs at the 21st Century Partnership for STEM Education and former director of the Math Forum. “Feedback that recognizes and builds on the thinking of the student can change a student’s sense of capacity for mathematics and cultivate their mathematical identity.”
Over the next year, the team will work with teachers in Baltimore, Washington, D.C. and California to design, test and refine the system in classroom settings. In addition, the team will create a feedback framework to evaluate the instructional purpose, tone and relational quality of the guidance provided by the tool. The team will make its data and tools available to research and practitioner communities through a series of workshops, conference presentations and publications.
This project builds on previous research, funded by the National Science Foundation, that brought together a community of teachers and students to collaborate on designing more effective feedback for students. These efforts began with The Math Forum, an award-winning online math learning and teaching community.
For more information about the Gates Foundation’s AI-powered innovations in mathematics teaching & learning initiative, visit: https://usprogram.gatesfoundation.org/news-and-insights/articles/ai-powered-innovations-in-mathematics-teaching-and-learning-rfi.
Drexel News is produced by
University Marketing and Communications.