No matter how powerful generative AI becomes, physicians will still have jobs. But will those jobs … More
Will AI replace doctors? A year ago, most physicians would’ve confidently answered “no.” Medicine, they’d argue, is too complex, too personal, too human to be handled by machines, no matter how advanced.
Now, that confidence is starting to waver. Physicians, like other highly educated workers, are watching what’s happening in another once-secure, intellectually demanding profession: computer programming. Not long ago, coding was considered one of the most prestigious and future-proof careers in the modern economy. The brightest students pursued software engineering, drawn by high salaries, strong demand and the appeal of solving complex problems. Programmers were irreplaceable. Until they weren’t.
From Amazon to Meta to Salesforce, tech companies are laying off engineers by the thousands. At Microsoft, generative AI already writes a third of the company’s code, and some experts predict it could eliminate two-thirds of programming jobs by decade’s end. Companies like Shopify and IBM have gone even further, requiring managers to justify hiring humans over AI or freezing new hires for roles they believe GenAI tools will soon replace.
In medicine, large language models already outperform physicians at diagnosing complex cases and answering patient questions. But that doesn’t mean clinicians are at risk of losing their jobs.
Here are three reasons GenAI won’t replace doctors — followed by one major caveat.
1. Too Few Doctors, Too Much Work
Across hospitals and clinics, American healthcare is already stretched beyond capacity. The American Medical Association projects a shortfall of up to 124,000 physicians by 2036, including 48,000 in primary care alone.
Three major forces are driving this shortage:
- The population is aging and getting sicker. The number of Americans aged 65+ is expected to grow 45% by 2033. And nearly 95% of adults over 60 live with at least one chronic disease (hypertension, diabetes, heart failure, etc.). In recent decades, as medicine has shifted from treating acute illness to managing lifelong diseases, the number of patient needs (and physician responsibilities) has grown exponentially.
- The workforce is actually shrinking. One in five active doctors are already over 65, with another 22% between 55 and 64. Workplace stress is pushing many to leave the profession earlier than planned. A recent Medscape poll reports that nearly half of all physicians want to retire before age 65, with more than a quarter of hospitalists hoping to retire even sooner.
- There’s a serious bottleneck in training new doctors. Medical school enrollment has climbed nearly 40% since 2002, but students can’t practice without completing residency. These training positions are funded by Medicare, which capped its financial support in 1997. A 2021 relief bill added 1,000 new slots, but that’s a drop in the bucket. As a result, many top U.S. graduates go unmatched each year due to the shortage. And with physician training taking nearly a decade to complete, the system remains years behind growing demand.
Bottom line: The physician shortage is real and getting worse. GenAI can help fill the gaps, but it won’t eliminate the demand for human clinicians.
2. Cutting Doctors Is A Poor Way To Cut Costs
In most industries, replacing high-salaried workers with technology is the fastest path to profitability. But in healthcare, that approach misses the point entirely.
Take primary care as an example. It’s the backbone of the U.S. medical system, yet it accounts for less than 5% of the nation’s $4.9 trillion in healthcare spending. Only half of that percentage goes to salaries. So, even if we eliminated half of all primary care physicians (an unthinkable move), total costs would drop by just 1.25%.
In healthcare, the greatest opportunity for cost savings is in preventing and better managing chronic diseases like diabetes, hypertension and long-term heart failure. According to the CDC, improving prevention and chronic disease management could prevent 30–50% of their complications (heart attack, stroke, cancer and kidney failure). Avoiding these catastrophic medical events would save an estimated $1.5 trillion annually.
Value-based care models have already demonstrated what’s possible. Studies from leading health systems show that investing in proactive, team-based primary care reduces hospitalizations, improves outcomes and lowers annual per-patient costs by up to 23%.
That’s where the right combination of clinicians and generative AI offers the greatest value. Between visits, GenAI can track symptoms, alert patients to necessary medication changes and identify complications before they turn into crises. Paired with 24/7 telemedicine, GenAI can provide patients with real-time expertise and care for routine concerns, flagging serious problems when doctors aren’t normally available.
Bottom line: Controlling chronic disease offers 20 times the savings of cutting primary care jobs. Ultimately, the greatest cost reductions will come from better health, and the best way to achieve that is pairing generative AI with skilled clinicians and empowered patients.
3. AI Can’t Replace Human Connection
Generative AI is becoming more skilled with every update. In a recent dual study, ChatGPT provided answers to routine patient questions originally fielded by doctors. When clinicians and patients reviewed the anonymized responses, the AI was rated better than physicians in both quality and empathy.
But when the stakes are high — life-altering cancer diagnoses or complex treatment decisions —patients want a trusted human at their side.
Abraham Verghese, my colleague at Stanford and a bestselling author, notes: “Medicine at its heart is a human endeavor … the physician‑patient relationship is key; all else follows from it.”
He emphasizes the ritual of the physical exam as transformative, a structured encounter that seals trust, communicates care and calms fear. Studies show that the doctor’s touch reduces anxiety, boosts patient satisfaction, and even improves clinical outcomes.
Bottom line: Even when data show that generative AI is more accurate, patients still want to talk with a human when facing complex or life-threatening decisions.
A Caution Against Complacency
No matter how powerful generative AI becomes, physicians will still have jobs. But will those jobs be fulfilling or soul-crushing? That depends on what doctors do next.
If private equity firms or for-profit health insurers determine how GenAI is integrated into medicine, the technology will be used primarily to increase productivity: faster diagnoses, shorter visits, less support staff. Yes, technology can streamline tasks. But unless clinicians shape its deployment, GenAI will be used primarily to drive productivity, making today’s problems worse for both clinicians and patients.
By contrast, if physicians take the lead, they can harness generative AI to improve patient health, reduce burnout and lower costs by preventing complications like heart attacks, strokes, cancer and kidney failure. But that success will require more than technological tools. Doctors must organize into high-performing medical groups, integrate GenAI into all aspects of clinical care and negotiate payment models that reward improved outcomes — not just higher volume.
Bottom line: Bottom line: GenAI can cut corners or improve care, but not both. It can boost profits or improve lives, but not both. The path we take will depend on who takes the lead.
