The UCF Mathematics Colloquium, held every Monday from 3:30-4:30 p.m. in MSB 318, offers a diverse platform for research scholars, faculty, students, and industry experts to share and exchange ideas, fostering discussion and networking across various areas of mathematics.
Dr. Seshashayee Murthy, from Halıcıoğlu Data Science Institute at the University of California San Diego, will speak at this week's colloquium on Research on Using AI to Enhance STEM Learning, Course Materials and Engagement at All Levels: The Aspire Project.
Abstract: This talk will present the open-source ASPIRE project, which leverages Large Language Models (LLMs) to enhance personalized learning in STEM. Large Language Models (LLMs) represent a significant development in facilitating natural language interactions with students, and in creating course materials. There is a possibility that they can help bridge the Bloom 2 sigma gap. LLMs however currently exhibit an error rate of 20% to 40% on tasks related to precalculus. Initial findings also suggest that when LLMs simply solve problems for students, there is a risk of reducing student engagement and performance. Our research attempts to address these issues by having SMEs supervise LLMs in creating course content and in answering questions.
ASPIRE uses LLMs, directed by subject matter experts (SMEs), to construct detailed domain models: directed acyclic graphs that capture detailed pre-requisite information for concepts. These models assist LLMs, instructors, and students in evaluating students' mastery of prerequisites for specific modules and in creating personalized learning pathways. SMEs supervise LLMs to generate questions to assess students' knowledge of the concepts in the domain model. We use these to assess students' knowledge of prerequisites and to bridge knowledge gaps when necessary. Additionally, students receive preview questions to prime them for key concepts before lectures and review questions after lectures to reinforce their understanding and retention.
We also use LLMs to rapidly answer relevant questions from learners in a safe group environment (like Piazza) with SME supervision. The group setting helps ensure that LLM mistakes do not affect the student. SMEs correct LLM mistakes. Students evolve critical thinking skills from critiquing LLM responses in a group.
ASPIRE has been piloted over the last year with over 500 students across four courses: differential calculus, a bridge course for differential calculus, introduction to probability and statistics for data scientists, and introduction to Python programming. This presentation will focus on the work conducted in these courses and outline future plans for expanding personalization to improve student outcome.
Read More