The Positive Power of Generative AI: Using AI for “Good” In Statistics Classrooms

1443 

Invited Paper Session 

Jacqueline Hicks (1), Rebecca Andridge (2), Jacqueline Hicks (1), Amanda Ellis (3), Andrea Lane (4), Justin Post (5)

(1) Boston University, N/A, (2) The Ohio State University, N/A, (3) University of Kentucky, N/A, (4) Duke University, N/A, (5) North Carolina State University, N/A

Jacqueline Hicks  
Boston University

Rebecca Andridge  
The Ohio State University

Jacqueline Hicks  
Boston University

Amanda Ellis  
University of Kentucky

Andrea Lane  
Duke University

Justin Post  
North Carolina State University

Artificial intelligence (AI) is everywhere and is constantly evolving. With the sudden deployment of and proliferation of generative AI – e.g., ChatGPT – educators are realizing a need to both assess how AI can be used to support student learning and to consider how it impacts academic integrity and students' ability to think independently. Throughout history, statistics educators have constantly adapted instruction to include new tools and techniques, but adapting to the presence of generative AI may feel especially daunting given its newness, potential "nefarious" use (e.g., concerns about academic integrity) and that AI generated content can be biased and inaccurate. This session will focus on the potential positive power of generative AI in a statistics and data science curriculum and how we may best leverage its use in the classroom. Through these talks we will illustrate how AI has the potential to enhance the teaching and learning experience by providing personalized learning, automating administrative tasks, and improving student outcomes in statistics and data science classrooms.

Speaker: Amanda Ellis, University of Kentucky
Presentation Title: Monitoring Students' Engagement with AI Features

Surveying and monitoring how students are utilizing AI tools in their coursework is a critical step in effectively integrating these tools into the classroom and enhancing the learning experience. By gathering data on students' interactions and preferences, educators can gain valuable insights into which AI features are most beneficial and how they impact student performance. This information can inform curriculum design and help tailor AI-driven content to meet the specific needs and learning styles of the students. Additionally, monitoring student engagement with AI tools allows for the identification of potential challenges or areas where additional support may be required, ensuring a more seamless and productive integration of AI into the educational environment.


Speaker: Andrea Lane, Duke University
Presentation Title: Using an AI learning management system to personalize student learning

Instructors often face the challenge of catering to varying technical backgrounds and learning paces of students. Classwise is a learning management system developed by Dr. Jon Reifschneider that incorporates AI to personalize student learning. Classwise offers a range of AI features to benefit both instructors and students, including a chatbot to summarize materials and check students' conceptual understanding. Instructors can then monitor student progress and use AI to develop assessments and other teaching materials.

Speaker: Justin Post, North Carolina State
Presentation Title: Using Generative AI to Supplement Learning in Statistical Programming Courses

This talk will discuss how large language models (LLMs) like ChatGPT have been integrated into statistical programming courses to help students get started on tasks, come up with ideas for descriptive analyses, and debug programs. The inclusion of LLMs required the rethinking of traditional assessments used in the course which has led to assignments that require more critical thinking and reflection. Successes and mistakes from these changes will be discussed.

No Additional Sponsor ³

Section on Statistics and Data Science Education ²

Section on Teaching of Statistics in the Health Sciences ¹