A Use-Case for Implementing ChatGPT to Augment Teaching an Introductory Statistics Course

Abstract

This study investigates the efficacy of incorporating ChatGPT (Radford et al., 2021), a cutting-edge natural language processing (NLP) tool, into higher education, specifically for an introductory statistics course. Drawing from the pedagogical framework of constructivist learning theory (Piaget, 1954; Vygotsky, 1978), the author, a statistics professor, explores the potential benefits and drawbacks of utilizing ChatGPT in the learning process. The study consists of four stages: (1) employing ChatGPT to perform preliminary data analyses and determine the appropriate statistical tests, (2) executing both parametric and nonparametric tests with ChatGPT’s assistance, (3) reviewing the output and analysis for accuracy, and (4) evaluating the potential impact on students’ understanding of statistical concepts and procedures. The findings suggest that ChatGPT offers valuable support in data analysis and interpretation, by providing students with models to compare their results and understanding against the AI-generated output in a similar manner as described by Bishop, 2006. However, potential drawbacks include the risk of students becoming overly reliant on ChatGPT, leading to diminished understanding of statistical concepts (Kirschner et al., 2006), or possessing insufficient knowledge to identify errors introduced by the AI (Hattie & Donoghue, 2016). The study concludes that, while ChatGPT can enhance the learning experience in statistics education, careful implementation and guidance are necessary to prevent potential hindrances to students’ development of a comprehensive understanding of the subject matter due to incorrect results and incorrect interpretations and citations.