M14: Scaling up active learning activities via LLMs in an intermediate statistics class

Information

We look at transforming an intermediate statistics class (<20 students who have had an introductory statistics non-calculus based course with a mix of majors) at a private SLAC in central PA to be more interactive and hands-on by using a large language model (LLM) to quickly build class activities (via applets) with actual datasets. Pre-LLMs, my class would be lecture, where I cover output from one dataset together with theory, and labs where students write code to do data analysis. With an LLM, my class is now an active learning class where students do pre-readings assessed partially in homework, and then do worksheets / group activities in non-lab days using these applets. Students now get to visualize (many of the) plots, statistical software output with many datasets, even before starting a lab. I assess the effectiveness by the students' participation (asking interesting questions) in class, as well as their exam grades. I envision LLMs being used to design many interactive applets by StatEd instructors, in a short fraction of the time it would normally take, leading to more engaging and interactive classes for students.