University of Florida Team Aims to Double Math Learning with AI Innovation

The University of Florida is thrilled to announce its participation in the Learning Engineering Virtual Institute (LEVI). LEVI selected a cohort of seven teams committed to harnessing the potential of AI and machine learning to enhance middle school math education. We are excited to be one of the seven!

LEVI teams are already making remarkable strides toward accomplishing this ambitious goal. Together we are developing innovative interventions to tackle this problem, ranging from increasing student motivation and engagement to AI-powered on-demand tutoring.

UF’s Wanli Xing, associate professor of educational technology, will lead the project titled: AI-augmented Learning by Teaching to Enhance and Renovate Math Learning (ALTER-Math). The ALTER-Math team aims to double math learning progress for middle school students by developing AI-powered teachable agents using large language models (LLMs) through UF’s HiPerGator supercomputer. These agents will be integrated into the UF Lastinger Center for Learning’s Math Nation, a core curriculum used by over one million students annually.

ALTER-Math will build on the success of Math Nation, which has been shown to significantly increase students’ state-required math exam passing rates. The platform currently features video-based recorded lectures and practice problems. Flipping roles, the students will engage with AI-powered teachable agents. This reversal has been demonstrated to significantly increase student engagement in several ways, including:

• Promoting the self-other effect and the protégé effect: Teachable agents can be programmed to create a sense of social connection with students.
• Integrating social psychological interventions: ALTER-Math will integrate social psychological interventions into the teachable agents at scale, helping to motivate students and enhance their mathematics attitudes and growth mindsets.
• Helping students identify their own difficulties and misconceptions: Students begin to identify their own difficulties and misconceptions by observing the problem-solving capabilities of their teachable agents, fostering conceptual change.

ALTER-Math is a promising new approach to math learning with the potential to significantly improve the outcomes of middle school students, especially those from low-income families. Specific examples of how ALTER-Math’s AI-powered teachable agents could be used to improve math learning, include:

• helping a student to learn a new math concept by providing step-by-step instructions and feedback.
• challenging a student with more difficult problems once they have mastered the basics.
• helping a student to identify and correct their mistakes.
• providing a student with encouragement and support when they are struggling.
• making math learning more fun and engaging by using games, simulations and other interactive activities.

ALTER-Math has the potential to revolutionize the way that math is taught and learned.

UF is one of seven teams, including: Carnegie Learning, Carnegie Mellon University, Eedi, Rising Academies, the University of Colorado Boulder and Khan Academy

We are incredibly proud of our progress thus far and are excited about the journey ahead. By doubling the rate of middle school math progress and ensuring the affordability of these interventions, we are on a mission to revolutionize math education globally – an endeavor that wouldn’t be possible without the commitment of our partner teams and the invaluable support of our community.