A grant totaling nearly $1.3 million has been awarded to Candace Walkington, Ph.D., the Annette and Harold Simmons Centennial Chair and Professor Gerald J. Ford Research Fellow.
Professor Walkington learned that she received the $1,296,683 grant from the National Science Foundation (NSF) Innovative Technology Experiences for Students and Teachers (ITEST) on July 11. Walkington, Principal Investigator for the project, says the research will involve working with teachers and using large language models to write math problems that are personalized to the interest areas of middle school students.
According to Walkington, the research will advance theories of interest development where there is a lack of intervention studies targeting motivation and guidance on how to support students at different interest development phases. “Interest in math has been shown to decline over adolescence and this research will explore how to make math meaningful to middle school students. This grant allows us to conduct research foundational to the future of personalized learning, capitalizing on very recent advances in AI that offer novel opportunities to bring these approaches to scale.” Walkington says the project will also advance theories of teacher problem-posing, examining teacher characteristics, knowledge, and attitudes, and the problems teachers encounter when teaming with AI.
Walkington says she is thrilled to begin working with her distinguished Co-PIs including Dr. Tiffini Pruitt-Britton, a Ph.D. graduate of SMU, currently at American Institutes for Research (AIR) as well as other national leaders in Generative AI and Large Language Models Ryan Baker, Andrew Lan, and Neil Heffernan.
The research has a start date of August 1, 2024, when Walkington and her team will begin recruiting and working with 7th grade teachers using ASSISTments. The team will conduct studies looking at the effect of deep authentic forms of personalization using LLMs versus other alternatives. They will test generating visual illustrations to accompany math problems using LLMs. And they will explore approaches teachers can use for prompt engineering to create the best personalized math problems. They will examine teacher and student outcomes in each of these studies.