Mental health applications offer accessible alternatives to traditional care, with many now integrating AI features like chatbots and personalized recommendations. However, little is known about how AI is actually implemented or how users experience these features. Our study examines both developer positioning and user perceptions of AI in mental health applications. We systematically analyzed 244 mental health apps from the Apple App Store, identifying 12 distinct AI roles (e.g., coach, tracker, companion) and four interface types. We then conducted thematic and sentiment analysis of 996 user reviews from 27 AI-enabled apps to understand user experiences. Our analysis revealed recurring tensions around AI replacing human roles, trust, and augmentation. Our findings contribute a structured understanding of AI’s current roles in digital mental health and offer design recommendations for more effective and empathetic implementation.
ACM CHI Conference on Human Factors in Computing Systems