Proposal Description

In the past few years, the advent of large language models (LLMs) has challenged how instructors teach and how students learn. With LLMs such as ChatGPT, Claude, and Copilot at students' fingertips, and even embedded in data analysis software such as Excel and Posit, educators are compelled to reconsider the role these technologies play in the classroom.

In this session, we will focus on several aspects of how large language models have impacted statistics and data science education: how instructors assess, how instructors design course materials, and how students create prompts. This IPS session includes four presenters from different continents (Africa, Australia, Europe, and North America). These presenters include educators and researchers, someone who has worked on national guidelines for teaching statistics, and instructors in both small- and large- enrollment courses. Presenters have a range of doctoral backgrounds from economics to statistics to education. This IPS will provide insights into how students are using AI and how to teach effectively for meaningful learning, with or without LLMs.

One paper examines the redesign of an introductory statistics course for economics students, guided by three key priorities: developing course materials with LLMs in mind, emphasizing student use of verification strategies, and evaluating the impact of technology on student attitudes. After implementation, student feedback suggests that technologies, including AI, can make statistics more accessible and relevant. However, instructors report more nuanced experiences, noting that while AI tools and supplementary resources such as educational videos can enhance engagement, they may also, in some cases, come at the expense of developing careful, methodologically sound statistical reasoning.

The second paper will explore how digital portfolios and interviews, integrated with LLM models, can foster significant learning aligned with Fink’s taxonomy. Students create a portfolio during their Mathematical Statistics course that they must be able to defend in an end-of-semester interview. These assessments help to establish significant learning through communication and reflection.

A third contribution addresses the challenges of large-enrollment courses, in which individualized assessments, such as interviews, are often difficult to implement. This paper will discuss an assignment in which students conduct an exploratory data analysis to create a poster. The study investigates how using AI to complete this assignment influences students’ perceptions of learning, their attitudes toward statistics, and their confidence in interpreting results.

The final paper will examine students' prompting in relation to their self-regulated learning. Drawing on Self-regulation Theory, this work will highlight how students' ability to plan, perform, and evaluate their prompting process affects their interactions with the LLMs. Since the specificity of prompts in LLMs determines the applicability of the generated material, understanding the prompt styles they use and their self-regulation, including motivation and learning strategies, can inform instructional design. This paper will examine results from two large-enrollment courses in which over 1,000 students completed the Motivation and Learning Styles Questionnaire with a prompting exercise.

Overall, this session aims to spark discussion and generate actionable ideas for fostering meaningful learning in an educational landscape where LLMs are increasingly ubiquitous.