I recently read the article by N. Shah (2019), "Health care in 2030: will artificial intelligence replace physicians?", and I do not think that artificial intelligence (AI) is likely to replace physicians anytime soon. But with exponential technological advancement in this area, it is not implausible, either. As Shah (2019) clearly affirmed, we do not exactly know how the algorithms that drive AI work, therefore our trust in them is limited. Trust and empathy, which would be difficult for AI to mimic, are essential to satisfactory, successful patient-provider relationships. I would be willing to bet, at least at this current point in time, many people would not trust an algorithm with a serious medical decision, and may have distinct opinions about receiving life-changing health news from a robot.
Another reason I do not think AI will cause physicians to become obsolete is that, while they certainly rely on health information technologies to perform their duties, most physicians use a non-linear approach to figuring out medical conundrums. They ask a series of questions that lead to a hierarchy of trial-and-error scenarios, and often ultimately rely on creativity, problem-solving, and investigative skills to uncover the root cause, and subsequent treatment, of the disease. Unlike a game of chess, the complexities involved in the myriad different reasons why a patient has fallen ill are too vast and nuanced for even the smartest algorithm.
Additionally, there are tasks that physicians perform that would not be feasible for AI to accomplish, including emergency procedures when a patient's status could change rapidly and other hands-on treatments. I believe that AI, such as the "digital assistants" mentioned in the Shah article (2019), could be useful as a tool for physicians to use to help guide decision-making and perform tedious duties, but that the need for a physician's interpretation or approval of AI suggestions continues to be critical.
More inclusively, clinical informatics is emerging as a medical subspecialty. Read more here if interested: https://jamanetwork-com.proxy2.library.illinois.edu/article.aspx?articleid=1871416
References
Detmer, D.E., Shortliffe, E.H. (2014). Clinical informatics: Prospects for a new medical subspecialty. Journal of the American Medical Association, 311(20), 2067-2068.
Shah, N. (2019). Health care in 2030: will artificial intelligence replace physicians? Annals of Internal Medicine, 170, 407-408
------------------------------
Kate Moore
Program Coordinator
SIU School of Medicine
------------------------------
Original Message:
Sent: 11-23-2020 06:58
From: Stephanie Weiner
Subject: Question of the Week: Will artificial intelligence replace doctors?
Questions about the potential of artificial intelligence (AI) to advance medicine and health continue to provoke discussion in our community. AAMCNews asked this question-and provided some evidence toward an answer-in an article last year. In a recent Nature commentary, several academics warned that AI studies are often not transparent and reproducible, and are published without complete evidence of code, models, and methodology; this makes AI research vulnerable to "hype."
What are your thoughts about the potential of AI? Will it replace doctors? Will it advance healthcare delivery?
------------------------------
Stephanie Weiner
Director, Digital Strategy & Engagement
Association of American Medical Colleges
------------------------------