Will AI Make Physicians Obsolete?

Marcia Frellick

November 08, 2018

The hype and promise of artificial intelligence (AI) and machine learning in medicine is sky high. Individual physicians, patients, and experts have very different views on whether computing and machines can really take the place of physicians.

In a debate published online November 7 in the BMJ, Jörg Goldhahn, MD, says they will. "Doctors as we now know them will become obsolete eventually," writes Goldhahn, deputy head of the Institute for Translational Medicine at ETH Zurich in Switzerland.

He explains that the technology has already shown the potential to be more accurate than physicians at making diagnoses in certain specialties, producing prognostic models, and performing some surgical interventions.

"And in 2017 a robot passed China's national medical exam, exceeding the minimum required by 96 points," he writes.

As to the counterargument that patients will always need a physician's empathy and to feel trust, Goldhahn agrees but says that machines and computers may be able to fulfill those needs too.

"[M]achines and systems can be more trustworthy if they can be regarded as unbiased and without conflicts of interest," he writes. He adds that in some very personal cases that may involve shame, a robot's services may be preferred.

AI machines have language processing abilities that enable them to process volumes of scientific literature, and they can teach themselves, for instance, about drug interactions. With deep learning, they gain knowledge at a speed no human can match, Goldhahns continues.

The machines also address escalating healthcare costs and can monitor patients remotely, he added.

Physicians Will Always Be Vital to Healthcare

Others, though, say there are aspects of care that machines can never provide.

"In particular, physicians will remain better at dealing with the patient as a whole person, which involves knowledge of social relationships and normativity," write Vanessa Rampton, PhD, from the McGill Institute for Health and Social Policy, Montreal, Québec, Canada, and Giatgen A. Spinas, MD, professor emeritus in the Division of Endocrinology, Diabetes and Clinical Nutrition at University Hospital in Zürich, Switzerland.

They note that machines will augment physicians' work but "will never replace them entirely."

Only humans can "relate to the patient as a fellow person and can gain holistic knowledge of the patient's illness as related to his or her life. Such knowledge involves ideals such as trust, respect, courage, and responsibility that are not easily accessible to machines," they write.

In an accompanying commentary, three patient advocates agree and say there are some things machines just can't do.

Michael Mittelman, executive director of the American Living Organ Donor Fund in Philadelphia, Pennsylvania, Sarah Markham, PhD, visiting researcher with the Institute of Psychiatry, Psychology and Neuroscience at King's College London, United Kingdom, and Mark Taylor, head of impact at the National Institute for Health Research Central Commissioning Facility in London, say that knowing that a physician truly cares about what they are going through can make all the difference in managing their own care.

A machine just can't show true comfort, they say.

"Ultimately, no one wants to be told he or she is dying by an entity that can have no understanding of what that means. We see AI as the servant rather than the director of our medical care," they write.

AI Will Change Physicians, Not Replace Them

Jonathan H. Chen, MD, PhD, assistant professor with the Center for Biomedical Informatics Research in the Department of Medicine, Stanford Medicine, California, is writing a chapter for National Academy of Medicine guidance, which will be coming out next year, on the incorporation of AI in medicine.

On his website, he says that people who think AI can replace physicians either don't know medicine or don't know AI.

As for Goldhahn's contention that "[d]octors as we now know them will become obsolete eventually," Chen told Medscape Medical News that the same could and should be said for almost any profession.

He sees that statement as predicting major change, but he sees the change not as involving the elimination of physician jobs but as changing the clinician role.

"Computers are very good at information retrieval — they can answer the 'what' questions. They don't really tell you 'how.' They don't really tell you 'why.' They're not good at context," he said.

"One of the hardest things computers have to deal with is common sense," Chen said.

He said Goldhahn is right in that it's impossible for any human to keep up with the evolving complexity of medicine.

"Trying to do it alone doesn't make sense, so we should be using support systems," he said.

He urged physicians to recognize the strengths of AI in taking over parts of physicians' jobs that keep them from patient interaction, for instance, in automating data entry.

Trust Is a Complicated Thing

Jack Stockert, MD, was formerly in an internal medicine practice and is now managing director of Health2047 Inc, in Menlo Park, California, a nonprofit healthcare integrated innovation company founded by the American Medical Association.

He told Medscape Medical News that he agrees that roles of physicians will certainly change as AI advances. For example, physicians may now need to audit the rule sets that are written in the AI system.

But he said any statement that we won't need physicians anymore "is like saying we won't need to have Thanksgiving dinner in person with people again. It runs afoul of what healthcare is, which is a doctor and a patient — an individual living their healthiest life. There are many ways where technology and AI will be able to help improve that and there are many ways it will fall short."

He said he's not convinced by Goldhahn's argument that some patients will trust a machine more than a physician.

Trust is complicated and often comes over time, he notes.

"If it's not transparent and it's not conforming to standards of reproducibility and it's not identifying what steps we should take to address biases in these systems, then I struggle to understand how I could individually trust it," he said.

"Current surveys suggest that physicians are the most trusted entity in the marketplace," he said. "To think that that goes completely away is asking too much of technology at the moment."

Dr Goldhahn, Dr Rampton, Dr Spinas, Dr Chen, Michael Mittelman, Dr Markham, and Mark Taylor have disclosed no relevant financial relationships. Dr Stockert is managing director of Health2047 Inc in Menlo Park, California, which was founded by the American Medical Association.

BMJ. Published online November 7, 2018. Full text, Commentary

For more news, join us on Facebook and Twitter


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.