In Focus delivers deeper coverage of the political, cultural, and ideological issues shaping America. Published daily by senior writers and experts, these in-depth pieces go beyond the headlines to give readers the full picture. You can find our full list of In Focus pieces here.
As with many developments here in the early stages of the artificial intelligence revolution, the idea of a non-human delivering our healthcare is something that, in some ways, we’ve been preparing for as a culture for some time. Star Wars: Revenge of the Sith famously depicted medical robots delivering Luke and Leia as newborns and giving Darth Vader his life-saving cybernetic implants and suit. Star Trek Voyager’s “EMH” (emergency medical hologram) was a full member of the crew, spending seven years as the ship’s only full-time physician.
Those popular stories were told three or four decades ago, but even way back in 1928, a sci-fi story titled The Psychophonic Nurse described how an Eastinghouse Electric Company developed a mechanical entity who “doesn’t eat anything, is on duty twenty-four hours a day, and draws no salary.” Said a buyer: “She’s cheap at the price I paid.”
Fast forward to today. AI can already refill your medicine without any human oversight. Just days ago, the FDA approved a device that can draw your blood “without hands-on operator intervention.” Anthropic, maker of the ‘Claude’ AI, signed a three-year MOU with the Government of Rwanda, committing to deploy AI toward eliminating cervical cancer, reducing malaria, and lowering maternal mortality.
Just two weeks ago, Ezekiel Emanuel co-published an article in the prestigious Journal of the American Medical Association titled, “Will Autonomous AI Exceed AI-Aided Physicians as the Best Medical Care?” The central claim is that the data now show that AI alone consistently outperforms both human beings alone and humans working with AI in five key tasks:
- Reviewing systems, eliciting complaints, taking family health history, getting past and current medications, etc.
- Choosing the correct tests to order
- Ranking the true diagnosis correctly
- Offering treatments in line with commonly accepted guidelines and practices
- Management of chronic diseases like diabetes (better patient adherence and less distress)
And with AI models improving so fast, they predict that in less than four years, autonomous AI will be ready to deliver health care in “some, maybe many” clinical workflows.
Are we, in fact, ready for where this seems to be headed?
A Pew survey released last week found that 1 in 3 Americans now use AI chatbots for at least one health reason: Everything from diagnosing symptoms to understanding test results. Significantly, 18% use chatbots for matters they are uncomfortable disclosing to a human physician. Nearly all users find chatbots in this context at least somewhat helpful.
This data tracks with, if not our own experience, stories we have likely heard over the last couple of years. One of these anecdotes offered by Dario Amodei (CEO of Anthropic, one of the most influential figures in the AI revolution) shared some of the details of his sister’s struggle with a pregnancy. She “wasn’t getting the answers or support she needed from her care providers,” but “she found Claude to have a better bedside manner” and even that it was “better at diagnosing the problem.” Amodei wonders if “a human touch” may not be that significant for medicine (and other practices) after all.
In my opening piece for this column, I made the case that the AI revolution has forced our culture’s weightiest questions wide open. What is a person? What is war? What is work? What is education? What is friendship? These questions, previously marginalized by the managers of our public discourse to philosophy classes, religious texts, and late-night dorm room conversations, are now front and center.
In my second installment, I dove into the question of what makes acts of war human and (therefore) moral or immoral. I argued that consequentialist logic, focused on which practices will win the most wars and result in fewer innocent deaths, though significant, cannot settle the debate. The advent of autonomous AI-powered weapons put the very idea of central human activity at risk, and I argue that this means we should find a way to keep war human. Even if it produces worse consequences.
CHARLES CAMOSY: SHOULD A ROBOT DECIDE WHO DIES ON THE BATTLEFIELD?
It is, of course, natural for Amodei or any of us to push for new practices that are likely to produce better consequences for our sick family members. And, like autonomous AI-directed weapons, autonomous AI-powered health care providers are likely to produce some significantly better outcomes.
But that is not the only consideration here. As with war, the AI revolution is asking us fundamental questions about the nature of healthcare itself.
On the one hand, one might claim that ‘health care’ is whatever we want it to be. The “provider” simply gives us what we want. In medical ethics, many of us call it the “Burger King” model of medicine: The customer gets it “their way.” Even the term “provider” itself suggests a passive, market-determined process. If you want a Whopper without a bun, then who is a Burger King cashier to tell you different? You want sex reassignment surgery because all your friends are doing it, then who is your doctor to tell you different? He or she is merely a vendor with a technical skill (or credential) you pay to give you what you want.
But we must be absolutely clear: A physician practicing healthcare is not merely a vendor with technical skill. A physician is practicing an ancient profession and art that must be directed toward objective ends: practicing a human relationship of care in seeking health of the whole person — body, mind, and spirit.
There are no oaths or codes of ethics that guide and limit someone according to the nature of a sandwich when making a customer a Whopper. But there are still oaths and codes of ethics that guide and limit a physician in the practice of healthcare when helping a sick person get better.
Significantly, Emanuel got exactly this kind of pushback online in response to his Journal of the American Medical Association piece. Because healthcare is essentially about human beings caring for other vulnerable human beings, his critics argued, AI cannot and must not be allowed to do what his article argued for. If permitted, it would undermine and destroy the very nature of healthcare practice itself.
But Emanuel’s devastating rejoinder to such critics (including to the CEO of the AMA) is something everyone should fully absorb:
For the last 75 years, medicine has emphasized science rather than the art of medicine. We do not screen patients for admission to medical school on the basis of their emotional intelligence and human sensitivity. What are the prerequisites? We ask organic, physics, calculus, not the humanities. When we are in medical school, what are the first two years of training? They’re all pathophysiology, this and that. Yes, the physician-patient relationship is in there, a tiny, tiny bit. OK. How do you evaluate patients, medical students in the clinic, the length of their differential diagnosis, did they get it right, the workup, etc. Not their interaction with the patient. OK. This idea that art is so important to medicine, we don’t behave like that. And the claim that that’s what medicine is going to do requires a very different training, a very different emphasis, and it’s not where medicine is today.
AI-delivered “healthcare” can only be plausible as an idea if we operate with a deeply impoverished way of thinking about healthcare. Unfortunately, it is one we have been using for many decades, and now AI looks like it can perform many of these tasks at far lower cost, with virtually infinite time and “patience,” sharper cognitive and technical skills, and better outcomes. There are also serious arguments being made that AI “providers” will be necessary to solve our health care labor shortage.
But the AI revolution (as Emanuel’s argument, its critiques, and his rejoinder indicate) has now forced open the question, “What is healthcare?” and pushes for an answer that goes beyond, “Technocratic vendors giving customers what they want.”
CHARLES CAMOSY: INFANTICIDE IS BACK
Especially in an era in which cutting off healthy genitals, killing prenatal human beings because of their disability, and optimizing embryos for intelligence and good looks (while discarding those which don’t measure up) are all fiercely defended as “healthcare,” there is a golden opportunity enter the fray on a question that, once marginalized as esoteric and academic, is now fully up for cultural grabs.
My next column will attempt to do just that, by offering a vision of health care that honors the ancient profession, incorporates AI as a tool, and remains firmly committed to staying human.