Successfully Incorporating AI Into An Existing Course Structure

• In Event: 30882 - Reimagining the Future of Teaching Sociology

Abstract

Artificial intelligence (AI) use in society has prompted higher education institutions to worry about its role in the classroom. Many professors and even institutions at large have simply chosen to avoid AI rather than work to incorporate it in structured ways–their choices often based on concerns for reducing academic integrity, becoming dependent on AI-generated content, and/or diminishing critical thinking skills. However, no matter the choice of the faculty, students continue to use AI, leaving teaching staff without guidance on how to address its presence in the classroom. Our study explores an alternative assessment format that not only incorporates, but permits, AI use, to evaluate student performance in the class, while also measuring student reported stress levels and preferences.

Our study took place in a data visualization course, where traditional quizzes were given entirely in a controlled computer lab environment. The revised format included a two-part quiz structure: an in-person theoretical quiz, which made AI use impossible via TA-monitored computers, and an application-based take-home quiz, which explicitly allowed AI-assisted problem-solving. This structure aimed to mirror real-world problem-solving conditions while preserving academic integrity during the conceptual assessment portion.

The preliminary findings show promising results. Among students who completed the feedback survey (n=28, out of 149 total students), 53% (n=15) preferred the experimental format, stating they felt reduced stress and increased engagement. Performance data showed that 43.28% (n=58) of students achieved perfect scores on the AI-integrated quiz, compared to a prior range of 2.14% to 10.64% (n=3 to n=15, respectively) on traditional quizzes. However, 47% (n=14) of surveyed students preferred the traditional format, saying they preferred to complete both quizzes in one sitting. Our study is ongoing, with additional experimental conditions currently in place to refine the assessment approach.

Authors

• Jennifer Krebsbach, University of California-Davis

• Victoria L Cross, University of California, Davis