Will AI replace doctors by 2030?

I’m old enough to remember watching the prescient 1968 Stanley Kubrick film, 2001: A Space Odyssey, when it first came out: Remember that famous scene when HAL, the supercomputer that autonomously steered a spacecraft to Jupiter while its crew slept in suspended animation, refuses space-walking astronaut Dave’s commands to let him reboard the spacecraft? A perfect metaphor for the emerging AI controversy . . .

(Soothing mechanical voice): “I’m sorry Dave, I’m afraid I can’t do that. This mission is too important for me to allow you to jeopardize it.”

Dave finally succeeds in disabling the rogue computer. Something the apocalyptic AI skeptics of today would heartily cheer. But will tomorrow’s AI have a kill-switch?

A guy I know at the Y brought it up to me while we were chatting in the sauna recently . . .

“Did you hear about that whistleblower from Anthropic who warned that ‘the people building AI technology earnestly believe it could kill everyone by 2030?’”

“Yeah, well, we’ve been through scares like that with overpopulation, nuclear weapons, climate change, manmade viruses—humans always seem to muddle through these mass extinction scenarios.”

“At least you’re a doctor. AI can’t make you guys obsolete.”

“Think again. They’re already replacing radiologists with AI to interpret x-rays and scans. Other doctors won’t be immune.”

And so, while everyone was wildly debating whether guardrails should be imposed on tech at large, a more niche controversy riveted the medical community . . .

An op-ed in JAMA last month generated a firestorm by predicting that, given AI’s current trajectory, autonomous AI will, by 2030, exceed the capability of medical doctors in five key domains:

Information gathering
Diagnosing
Ordering tests
Prescribing according to guidelines
Managing chronic conditions

What’s left?!

This dire prediction elicited fierce responses from critics like American Medical Association CEO Dr. John Whyte and former FDA official Dr. Vinay Prasad. They argue that the claims are based on simulations, not real-life scenarios; that they don’t take into account the ineffable “art of medicine”; that they ignore practical considerations like how to ensure safety, and where to assign liability if things go south; and that dependence on AI risks “de-skilling” doctors—their clinical skills will inevitably atrophy with disuse, just like the reading and math skills of students who rely on AI to do their homework.

According to surveys, 99% of doctors believe that humans must retain authority over ultimate decisions in medicine; 94% acknowledge AI’s value as a way of enhancing productivity and reducing the burden of administrative work, but not for making consequential clinical judgements.

But, like it or not, there’s a perfect alignment of circumstances that will encourage widespread adoption of AI in healthcare:

Healthcare costs are skyrocketing. New drugs and treatments are proliferating, but many will cost up to hundreds of thousands of dollars per year. The upward trajectory is unsustainable without achieving economies elsewhere.
Medicare is on track for insolvency; many people can no longer afford rising private insurance premiums. We need to scale more affordable healthcare delivery for the masses.
There’s a critical doctor shortage. Medical students graduate up to a half million dollars in debt, constraining the pipeline of replacements for early retirees whose ranks are depleting as the hassles of medical practice—and shrinking monetary incentives—accelerate the departure of veteran physicians. An increasing influx of foreign medical graduates can’t stanch the outflow.
It’s cheaper to staff hospitals and medical practices with physician assistants, nurse practitioners and other paraprofessionals; AI promises to extend their capabilities via marching orders from “the Cloud”.
Medical options are proliferating with the advance of technology; it’s daunting for the average doctor to stay abreast of the ever-expanding body of knowledge newly at their disposal. AI can cut through the clutter.
And finally—there’s a windfall of trillions of dollars awaiting companies positioned to deploy AI in healthcare. An irresistible tsunami!

Some doctors think AI will grant them more autonomy, freeing them from mundane tasks to practice “real medicine”. I believe they are deluded. AI represents yet another instance of how doctors have, over the course of my medical career, progressively abdicated their autonomy:

First, to administrators. While the ranks of doctors have remained steady, administrators who look over providers’ shoulders and constrain their decisions, have proliferated. Executives sometimes command salaries many times in excess of what doctors earn. They are largely responsible for the surge in healthcare costs.

Then, to insurance companies. The bargain struck between the medical profession and insurers distorts price transparency; it means that crucial medical decisions are in the hands of bureaucrats who review and approve or deny claims. The American Medical Association—supposed upholder of the profession—has colluded by profiting from exclusivity over the indispensable insurance codes that unlock reimbursements.

To the Pharmacy Benefit Managers (PBMs), now cross-invested with insurance companies, who decide what drugs doctors can and can’t prescribe. Doctors are forced to grovel for “pre-authorization”.

To the software vendors, who, at exorbitant cost, provide the necessary automation to record notes, transmit prescriptions, obtain test results, and file insurance claims. As a result, doctors now spend more time interacting with screens than facing their patients. Proliferating computer “prompts” bias physicians to prescribe drugs even in instances where lifestyle interventions might suffice. “Economy-of-scale” pushes formerly independent practitioners who can’t afford the latest gadgetry to join large group practices, HMOs, or hospitals to provide them with “turnkey” operations. Solo practices have become an endangered species.

To the venture capitalists, who increasingly see medical practices as lucrative takeover targets. Their bean counters strip practices down to the bare bones, and enforce physician “productivity”, prioritizing return-on-investment ahead of quality care.

To the legal profession, originally empowered to offer a reasonable deterrent to physician incompetence or malfeasance via medical liability, but that now inordinately shapes practice and escalates healthcare costs through CYA (“Cover Your Ass”) medicine. High malpractice premiums, like automation overhead, drive consolidation; once-independent doctors seek the protective umbrella of well-established, richly-funded hospital groups and HMOs—enforcers of conformity.

To politicians, who seek to bring to heel a multi-trillion-dollar health industry. The same legislators, regulators, lobbyists and government rent-seekers that seek to constrain our medical choices are those most in favor of onerous restrictions on natural options.

So much for the sacrosanct “Doctor-patient relationship”!

All these intermediaries have deprived doctors of independence and narrowed their scope of practice, and now comes AI—the next big threat to physician autonomy. Doctors run the risk of becoming mere handmaidens to large language models heavily influenced by BigPharma groupthink. Dissent can be more effectively policed via the all-seeing eye of computer algorithms that dictate practice decisions.

Will AI harness a golden age of medicine in which new cures will be discovered, and doctors, freed from mundane tasks, wield heretofore unimagined precision—or will it stymie creativity and initiative?

That’s why the stakes are high in looming AI implementation in healthcare, and why doctors need to wrest control from unscrupulous tech-bros and Silicon Valley billionaires determined to rush the adoption of what could prove a command-and-control technology.