An AI-Powered Positive Psychology App for Student Well-Being: A Pilot Study (Preprint)

BACKGROUND Higher education institutions in the U.S. are facing an unprecedented mental health crisis, yet the development and validation of effective mental health promotion and prevention resources have been limited so far. OBJECTIVE This two-week pilot study evaluated the effectiveness of a mobile app (“Flourish”) that integrates interactive, evidence-based positive psychology activities with an AI-powered chatbot, designed to promote everyday well-being practices and provide in-the-moment emotional support at scale. METHODS Students from eight classrooms at a two-year community college in Northern California were randomly assigned at the classroom level to either a treatment condition (n = 40), where they completed at least two well-being activities from the Flourish app weekly, or a control condition (n = 19), where they did not receive the mobile app. We assessed positive and negative affect, emotional challenges, and perceptions of school support, both at baseline and two weeks later. RESULTS A 2 (Condition: Treatment vs. Control) × 2 (Time: Before vs. After) repeated-measures ANOVA revealed a significant Condition x Time interaction effect for positive affect, F(1, 39) = 4.56, p = .039, η2 = 0.10, ε2 = 0.08; specifically, those in the treatment condition exhibited a marginally significant increase in positive affect after using the app for two weeks (p = .080, 95% CI -0.39, 0.02, d = 0.18), whereas those in the control condition showed no changes (p = .182). Similarly, there was a significant Condition x Time interaction for emotional challenges, F(1, 39) = 4.50, p = .040, η2 = 0.10, ε2 = 0.08; specifically, those in the treatment condition reported a decrease in emotional challenges after using the app for two weeks (p = .062, 95% CI -0.01, 0.49, d = 0.19), whereas those in the control condition showed no changes (p = .217). Finally, across both timepoints, those in the treatment condition reported marginally higher perceptions of school support compared to those in the control condition, F(1, 39) = 3.68, p = .063, η2 = 0.09, ε2 = 0.06. CONCLUSIONS This pilot study suggests that a mobile app integrating evidence-based well-being activities and an AI-powered chatbot can improve positive affect, reduce emotional challenges, and enhance perceptions of school support among college students. These findings highlight the potential of scalable, evidence-based interventions to address student mental health challenges and provide a foundation for further research on long-term impacts and broader applications.