Enhancing physical activity through a relational artificial intelligence chatbot: A feasibility and usability study

Area of Science:

• Digital Health
• Behavioral Science
• Artificial Intelligence

Background:

• Mobile app-based interventions are increasingly used for behavior change.
• The role of relational capacity in AI chatbots for health interventions is under-explored.
• Physical activity is crucial for public health, yet adherence remains a challenge.

Purpose of the Study:

• To assess the usability, feasibility, and initial efficacy of a mobile app-based relational AI chatbot (Exerbot) for increasing physical activity.
• To compare a relational AI chatbot with a baseline non-relational AI chatbot in a randomized controlled trial.

Main Methods:

• A 1-week pilot randomized controlled trial involving 36 participants.
• Participants were assigned to interact with either a relational AI chatbot or a control (non-relational) AI chatbot.
• Physical activity was objectively measured using smartphone pedometers.

Main Results:

• The study demonstrated feasibility with high enrollment (36 participants) and retention (94%).
• Daily engagement with the AI chatbot exceeded 88% across both groups.
• Participants using the relational AI chatbot maintained step counts, while the control group showed a decrease; stronger social bonds were reported with the relational chatbot.

Conclusions:

• Relational AI chatbots show promise for cost-effective, accessible, and sustainable behavior change interventions.
• This AI-driven approach may offer benefits for individuals with limited access to traditional interventions.
• Further research into AI's relationship-building capabilities for health behaviors is warranted.