AI-Chabots as Digital Mental Health Interventions: Assessing their Effectiveness in Reducing Depression, Anxiety and Stress among University Students

Abstract

This study explored the effectiveness of AI-based mental health interventions—specifically Wysa and Serenity—in reducing psychological distress among Pakistani undergraduate students. Using a quasi-experimental pre-test-post-test design, 297 students exhibiting mild to moderate symptoms of depression, anxiety, and stress were selected from four public universities in Khyber Pakhtunkhwa. Participants were randomly assigned to either an experimental group, which engaged with AI chatbots over an eight-week period, or a control group that received no intervention. The chatbot sessions were autonomous, flexible, and ethically safeguarded, allowing students to interact privately with digital agents trained in evidence-based therapeutic techniques. Quantitative analyses revealed significant reductions in psychological distress within the experimental group, with large effect sizes ranging from d = 0.72 to 0.93, while the control group showed negligible change. Gender-specific findings indicated that female students benefited more across all domains of distress. Structural equation modeling further confirmed that chatbot engagement significantly predicted lower levels of depression, anxiety, and stress, with the strongest effect observed for stress (β = −.47). These results underscore the potential of AI chatbots as scalable, low-cost, and culturally sensitive tools for mental health support in resource-constrained university settings. The findings contribute to the growing body of digital mental health literature by empirically validating chatbot efficacy in a collectivist cultural context. The study also highlights the importance of integrating AI-based interventions into campus mental health frameworks to overcome barriers related to stigma, accessibility, and limited professional resources.