By: Emily LaRosa & David Danks
This paper discusses the potential consequences that can arise surrounding patient-doctor trust, since AI has become increasingly prominent in the healthcare and medical industry. First, it is mentioned that AI technology is well on its way to playing a significant role in healthcare, from performing particular tasks to conducting simple monitoring. With these still experimental advancements in mind, while analyzing human-machine interactions is important, the authors choose to focus their exploration on AI’s impact on human-human interactions instead, specifically the relationship of trust between doctor and patient. In the paper, two different types of trust are identified: behavioral trust and understanding trust. Behavioral trust is rooted in reliability, in the sense that the trustor can predict the trustee’s behavior in certain situations based on past experiences with similar contexts. The trustors do not have knowledge about how the trustee’s behavior is generated, but just that their behaviors are probable based on the circumstances. This type of trust is not extremely effective for novel situations since it heavily relies on situations that have been faced in the past. The second type of trust is understanding trust, which is rooted in the understanding of the trustee’s actions because of the trustor’s comprehension of the trustee’s values, beliefs, desires, and intentions. They may not be able to perfectly anticipate what the trustee will do, but can rather base their prediction on their understanding of the trustee’s personality, habits, morals, and other defining characteristics. Contrary to behavioral trust, understanding trust is much better suited for novel situations, as no prior experience is needed. LaRosa and Danks then argue that the trust patients use with doctors is considered role-based trust, a third type of trust that is not separate from behavioral or understanding trust but helps establish one or the other. A doctor’s role is largely defined by their skills and knowledge, so patients trust that the doctor will have specific beliefs, values, and intentions (because of the professional role they occupy). Three routes to trust are then outlined as ways in which patients develop understanding trust in their doctors. Firstly, the licensure and certification of doctors to practice medicine ensures individuals that their doctors satisfy widely-known, objective criteria, helping patients’ justify their expectations about the decisions that their doctors make. Secondly, doctors have a special social role, as they are responsible for ensuring care that aligns with their patients’ values. Thirdly, in order to develop patient-doctor understanding trust, they must share experiences and interact consistently so that the patient can gain increased insight into their doctors’ capabilities, expertise, and values. Given these roles, LaRosa and Danks propose three main regulatory principles in order to establish some boundaries for AI and robotic systems when they begin to integrate into the field of healthcare. Number one — doctors using AI systems and/or results must go through educational training that is overseen, measured, and approved by an independent party, verifying that they are credible and not just simply puppets of the AI technology. Number two — educated consent of the patient or caregiver is necessary when implementing AI into patient care, involving more active conversations about protocols and procedures, unlike plain informed consent. Number three — until healthcare AI is widely accepted as “standard of care”, doctors are obligated to provide the alternative of a human performing the assigned task, emphasizing improvement of this patient-doctor understanding trust, and less encouragement towards “blind faith” with doctors. These regulations guide the start to a more balanced and healthy relationship between stakeholders in the medical community, and can help make certain that AI’s entrance into healthcare is evolving in an ethical manner.
This topic of Medicine and AI is one of my favorites so far, and I really think that this article does a sound job of explaining some challenges and points of curiosity within this realm. I enjoyed the authors’ recommendations at the end for proposals that could boost and maintain trust within patient-doctor relations, and I think that this was a wonderful start to exploring the role of AI in the modern-day medical industry!