Study: AI Chatbots Can Aid Healthy Lifestyle Changes but Struggle with Early Motivation
Krissy Vann | Host, All Things Fitness and Wellness
Researchers from the ACTION Lab at the University of Illinois Urbana-Champaign have discovered that while large language model-based chatbots hold promise for promoting healthy behavior changes, they struggle to accurately recognize users' motivational states. The study, led by doctoral student Michelle Bak and information sciences professor Jessie Chin, was published in the Journal of the American Medical Informatics Association.
Large language model-based chatbots, also known as generative conversational agents, have seen increasing use in healthcare for patient education, assessment, and management. Bak and Chin aimed to determine whether these AI tools could effectively support behavior change. They tested models from ChatGPT, Google Bard, and Llama 2 across 25 health-related scenarios, including low physical activity, diet and nutrition concerns, mental health challenges, cancer screening, sexually transmitted diseases, and substance dependency.
The scenarios were designed to represent the five stages of behavior change: resistance to change, awareness of problem behavior, intention to take action, initiation of behavior change, and maintenance of behavior change for six months. The study found that while the chatbots provided relevant information for users with established goals and a commitment to take action, they struggled to assist users in the early stages of change.
The study revealed that when users were resistant to change, the chatbots failed to provide information that would help them evaluate their problem behavior and its consequences or assess environmental influences. For example, individuals resistant to increasing their physical activity might benefit from understanding the negative consequences of a sedentary lifestyle. Without such tailored information, the chatbots were unable to generate the emotional engagement necessary to motivate users.
Conversely, for users ready to take action, the chatbots effectively provided information to help them move toward their goals, including advice on replacing problem behaviors and seeking social support. However, they lacked guidance on maintaining motivation through reward systems or reducing environmental stimuli that could trigger relapses.
The researchers concluded that while large language models are not yet adept at recognizing motivational states from natural language conversations, they can support behavior change for individuals with strong motivations and readiness to act. Providing chatbots with more specific knowledge could enhance their effectiveness in helping people change their behaviors.
Elevate Your Expertise: Tune in to Our YouTube Channel for Exclusive Podcasts with fitness and wellness Industry Titans and Game-Changers
