Effects of Static and Conversational AI-Generated Messages on Colorectal Cancer Screening Intent: a Randomized Clinical Trial
Large language model chatbots show increasing promise in persuasive communication, but their clinical utility remains uncertain, particularly in settings where sustained conversations are difficult to scale. In a randomized clinical trial, 915 U.S. adults (ages 45-75) who had never completed colorectal cancer (CRC) screening were randomized to no-message control, expert-written patient materials, single AI-generated message, or motivational interviewing (MI) chatbot. Both AI interventions significantly increased stool-test intentions by over 12 points (12.9-13.8/100), compared to a 7.5 gain for expert materials (p<.001). While AI interventions outperformed the no-message control for colonoscopy intent, neither showed improvement over expert materials. Notably, we did not detect a statistically significant difference between the longer, interactive chatbot and the simpler single AI message, despite participants spending 3 minutes more on average engaging with it. These findings suggest concise, demographically tailored AI messages may offer a scalable approach to increasing CRC screening intentions, although effects on completed screening remain unknown.