Effects of Generative AI on Engagement and Performance Within Asynchronous Online Courses

This study examines the role of a generative AI course assistant (Spark) in relation to student engagement and performance in asynchronous online courses. As online learning expands, sustaining active engagement–particularly in cognitive, affective, and behavioral dimensions–remains a persistent challenge. By providing on-demand clarification, formative feedback, and just-in-time guidance, Spark is designed to reduce cognitive friction, bolster feelings of support, and prompt more frequent task-oriented interaction with course materials—aligning with cognitive, affective, and behavioral dimensions of engagement.Using a mixed methods, quasi experimental design, we analyzed survey data, sentiment scores, GPA expectations and outcomes, and usage metrics from over 2,000 student course instances at Los Angeles Pacific University. Students could optionally interact with Spark, an AI assistant embedded into each course for clarification, feedback, and guidance. In the cognitive domain, Pearson product moment correlations indicated that expected GPA and actual course GPA were strongly aligned for both Spark users and non-users, suggesting that use of the tool did not substantially change students’ accuracy in predicting their performance. When Spark users were grouped into low, medium, and high usage tiers based on message counts, exploratory analyses showed a modest upward trend in mean GPA across tiers. These differences, however, were not statistically significant, indicating that heavier use did not yield clear additional gains in performance. Differences in survey based engagement between users and non-users were generally small or insignificant; however, longitudinal analyses suggested modest increases in self reported motivation, support, and engagement among Spark users in later terms as implementation matured. Sentiment analysis indicated a statistically significant but small difference in polarity scores, with Spark users describing their course experiences in slightly more positive emotional terms. Taken together, these findings suggest that the key distinction in this context is whether students use the AI course assistant at all, rather than how intensively they use it, and that Spark functions as a complementary support with modest, context dependent associations with academic and affective outcomes rather than as a primary driver of achievement or engagement.