Chatbots or Apps? Design Tradeoffs and the Opportunity for Hybrid Health Interfaces

Abstract

Large language model (LLM)–based chatbots are emerging as persuasive interfaces in digital health, yet their advantages relative to established mobile apps remain uncertain. We present a comparative field study (N = 167) evaluating two smoking-cessation digital interventions: QuitBot, an LLM-driven conversational coach, and quitSTART, a structured mobile app. Participants selected one of these, used it for a week, and reported on usability, engagement, and perceived effectiveness. Quantitative findings showed that QuitBot was experienced as easier and more natural to use, while quitSTART prompted more frequent daily engagement and a greater proportion of users to plan a quit date. Qualitative analysis revealed parallel strengths: QuitBot fostered empathy, motivation, and social presence, whereas quitSTART reinforced predictability, goal-setting, and progress tracking. Together, the findings highlight a potential trade-off between relational adaptivity and structural scaffolding in persuasive health interfaces, suggesting that hybrid intelligent systems integrating both dimensions may better sustain engagement and support long-term behavior change.

Ähnliche Arbeiten