Bhuvan Kala (University of Illinois), Sanjana Gongati (University of Illinois), Laura Bandi (Toronto Metropolitan University), V.N. Vimal Rao (University of Illinois)
Abstract
Background. The emergence of generative AI tools has almost overnight changed the practice of data science. Within the field of data science education, recent literature has primarily focused on how teachers and students can use generative AI for knowledge building. For example, Ellis and Slade (2023) consider how ChatGPT can be used by instructors to generate course content or by students to answer content related questions such as “What is a p-value?”.
We instead focus on the potential of generative AI technologies to remove barriers to participation in data analysis, especially to develop a basic understanding of data science for all undergraduate students regardless of their major. Specifically, we consider the potential of a new software application called Rtutor.AI. Rtutor.AI combines the power of ChatGPT with the statistical software program R.
With Rtutor.AI, students can conduct advanced statistical analyses or make complex multivariable data visualization with simple prompts such as “make a scatterplot of students’ exam grades based on how much they studied, with different colors based on their year in school”.
Here, we focus on whether these experiences help to promote students’ interest in STEM and the statistical and data sciences.
Methods. Using Social Cognitive Career Theory (SCCT) as an orienting framework, we recruited participants from an introductory statistics course that incorporates Rtutor.AI into instruction and weekly assignments. Survey responses related to relevant constructs per SCCT as well as students’ attitudes towards Rtutor.AI were collected before the semester started and once more towards the end of the semester. Additionally, 8 students were recruited to participate in a follow-up interview to further probe the potential impact of using Rtutor.AI. Survey responses were analyzed with a longitudinal structural equation model (LSEM) and interview transcripts were analyzed with a thematic analysis.
Findings. We are currently in the process of collecting data. We have survey responses from 783 students for the first of the two time-points. We have not yet conducted interviews nor collected survey responses from the second time point. These will be done, and analysis will be completed, by the time of presentation.
Implications For Teaching and For Research. In a changing technological landscape, it is essential for educators to develop new methods of assessment and instruction that capitalize on the strengths of AI technologies to promote students’ learning. Understanding the impact that AI has on students, both cognitively as well as socio-emotionally, will inform our curricular development efforts. The results of this study will help establish a foundation for understanding the broader impacts AI can have on all of our students, and will help us achieve the mission of data science for all.
Additional Information. Funding for this project is supported by the University of Illinois Center for the Social and Behavioral Sciences.