My Doctor is a Robot!: Public Trust in AI Robot Physicians

Theoretical Background and research questions/hypothesis:

Over several decades, Artificial Intelligence (AI) technology has continued to evolve and become more observable in people’s everyday lives. The adoption of AI within medical field is no longer ignorable as AI is already capable of replacing humans in diverse areas of works. Soon, we may see human-looking AI robots working in physicians’ offices and hospitals, since scientists have already invented human-looking AI robots, such as Sophia. A number of medical practitioners and scientists anticipate the introduction of AI robot physicians in the near future. Based on this anticipation of active adoption of AI technology in the medical field, this study aimed to measure and identify how much the public would trust AI robot physicians in comparison to human physicians if AI robot physicians are actually introduced to patients.

Methods:

This study measured the public trust in AI robot physicians by employing two well-refined and widely used scales in measuring patients’ trust in their physicians: 1) Trust in Physician scale with 11 questions designed by Anderson and Dedrick (1990) and 2) 10 item scale designed by Hall et al. (2002). This study employed two scales to increase validity and reliability of our study by comparing outcomes from two scales measuring the level of public’s trust in physicians. The modification of original questions for this study was minimal, 1) changing the subject term, doctor, to AI robot doctor, and 2) changing general verbs to probable verbs (e.g., from I trust to I would trust). Questions about demographic information and the experience of severe illness, selected by Kim’s (2018) study of factors affecting trust in physicians, were added for more specific analyses that revealed different levels of trusts in AI MDs among different groups of people.

In terms of data collection, this study used a social media-based data collection. The study asked university students at a southern university to post an invitational message to their social media with a link to the survey questions. Although the data collection was initiated from college students’ social media, their friends in social media were not necessarily students only. This study conducted a t-test to see if there is any difference between students’ answers and non-students’ answers. The test resulted in no difference. Therefore, this study combined all data together, which increased generalizability of the result of data analysis.

Results:

For the items asking accuracy of diagnosis, AI physicians received a higher level of trust, while human physicians were trusted more on items related to interpersonal cares.

Conclusions: Overall, the level of trust in AI physicians was notable.

Implications for research and/or practice:

Discussion and implications will include 1) current level of public trust in AI robot physicians without their direct experience, 2) attributes on the diffusion of AI technology in medical and health care fields, 3) social readiness for adopting AI robot physicians, and 4) necessary preparation for implementing AI technology in the medical field. This study will contribute not only to theoretical development of human-AI (machine) communication, but also effective adoption of the AI technology in medical and healthcare fields.