AI-powered mental health communication: Examining the effects of affection expectations on health behavioral intentions

Objectives: This study proposed and examined users' expectations of chatbots in mental health communication and compared them with human doctors. Focusing on the affective expectation, we aimed to unveil the impacts of changes in affection on individuals' expectations and behavioral intentions. Methods: A 2 (chatbot vs. human doctor) x 2 (affective vs. neutral) online experiment (N = 364) was conducted. Participants were asked about their expectations towards communicating health with chatbots and human doctors, watched pre-recorded conversations, and answered questions about expectancy violation toward affection and willingness to further health actions. Results: Individuals have higher expectations from human doctors in almost all aspects. Individuals' expectation of online chat with doctors is more likely to be violated compared with chatbots. Significant results are found on individuals' behavioral intentions, with a greater likelihood of switching to online chatbots when chatbots use an affective approach or doctors use a neutral tone. Individual's past experiences with chatbots play a key role in expectancy violations. Conclusion: Individuals have higher expectations of human doctors, but using an affective approach may better meet individuals' expectations toward chatbots and switch their behavioral intentions. Practice Implications: Chatbots that meet expectations can be designed and applied to address mental health concerns.