Trust in Health Chatbots.

A health chatbot is an artificial intelligence program that can conduct an intelligent conversation via auditory or textual methods regarding healthcare issues. An estimate by Market research firm Grand View Research is that the global chatbot market will reach $1.23 billion by 2025. Millions of chatbots have been sold to hospitals. Many health chatbots can be downloaded for free. These health chatbots provide a variety of services. For instance, Babylon Health allows users to report symptoms of their illnesses, checks them against a database of diseases, and offers appropriate courses of action. If needed, the chatbots will offer the users a live video consultation with a real doctor. Florence, a health chatbot, can remind a patient to take his/her pills, track the user's health, and find the nearest pharmacy and doctor's office if needed. The capabilities of health chatbots have been growing steadily. With the aging population and rising healthcare costs, health chatbots will likely become the first point of contact for primary care. Health chatbots are not making decisions per se but offer users rational options. Physicians must trust health chatbots to confidently use them as supporting tools. Patients must trust health chatbots if they are to have faith in the chatbots' diagnoses and recommendations. In the context of health chatbots, trust can be defined as the willingness of users to provide confidential information, accept the recommendations, and follow the suggestions. Reliability, transparency, and explainability affect the trust-building process (Siau and Wang 2018). The chatbots need to be consistent in their recommendations and suggestions. To facilitate trust, the chatbot's "black box" must be opened up. In other words, users need to understand how the recommendations were derived and what information contributed to the recommendations. In addition, the chatbots' algorithms rely heavily on data. The integrity, accuracy, privacy, and security of the data are essential. Otherwise, users will not provide sensitive healthcare information to use the health chatbots. Governance and ethical standards are still in the budding stage for artificial intelligence as a whole (Wang and Siau 2018) and these fields need to be further developed to facilitate the development of trust and adoption of health chatbots. Trust is dynamic -- involving initial trust building and continuous trust development. Many factors affect trust building. Previous literature has shown that personality, institution, cognition, knowledge, and calculative factors have an impact on initial and continuous trust (Wang and Siau 2018). Positive experience affects continuous trust. Are these the key factors that affect trust building in health chatbots? Are there other factors that are important in trust building in health chatbots because of the involvement of artificial intelligence and the health risks with wrong diagnoses? How to improve users' trust and acceptance of health chatbots? The research studies these research questions and develops a trust model that depicts the trust-building process between users and health chatbots. To develop a trust model, research models by Gefen et al. (2003), Bansal et al. (2015), and Siau andWang (2018) will be referenced. A case study involving experts fromthe healthcare industry, patients, and academicswill be conducted. [ABSTRACT FROM AUTHOR]