AI Is Not Going to Replace Doctors. It Is Going to Replace What Never Required One.

AI Is Not Going to Replace Doctors. It Is Going to Replace What Never Required One.

📝 This Week's Real-Life Lesson

I want to tell you about a conversation I had with a patient, or more accurately, a conversation I almost didn't have, because of all the things that had to happen before I could get to it.

This was a few years into my career. A woman came in with a sore throat. Straightforward enough. But before I walked into that room, I had already spent time navigating a prior authorization on an unrelated patient, responding to three inbox messages that should have been handled by someone else, documenting a visit from two hours earlier, and trying to find a lab result that had been filed in the wrong chart. By the time I actually sat with her, I had maybe eight minutes. She had questions. I had a schedule.

That's the problem nobody talks about when they argue about whether AI is going to replace doctors. The argument assumes the thing doctors spend most of their time doing is practicing medicine. It isn't. Study after study finds that for every hour of direct patient contact, physicians spend nearly two hours on administrative tasks. The most expensive, most trained professional in the clinical ecosystem is spending the majority of their day on work that does not require a medical degree. That is befuddling and infuriating at the same time!

That's not an intelligence problem. That's a misallocation problem. And that's the problem I've spent the last several years trying to solve, first as a practicing family physician, and now as the founder of ChatRx, an AI-assisted on-demand urgent care platform built on a principle I believe deeply: physician-governed AI, not AI replacing physicians.

I want to share what I've learned from being inside this problem, not as a commentator, but as a physician entrepreneur who has actually tried to build the solution.

What I Got Wrong at the Beginning

When I first started building ChatRx, I thought the goal was to make AI as smart as possible. Get the diagnostic accuracy high enough, compress the clinical logic tightly enough, and you'd have something that could handle urgent care at scale. I was thinking like a technologist, not like a physician.

Here's what the technology taught me: AI is extraordinarily good at a very specific kind of problem. Pattern recognition in defined domains. Symptom triage against a known differential. Checking for red flags in structured data. Routing the right question to the right protocol. In those domains, well-designed AI doesn't just help, it outperforms. A January 2026 Stanford-Harvard State of Clinical AI report confirmed what we'd been seeing in practice: AI systems can match or exceed physician performance on fixed, well-framed diagnostic cases. These findings were confirmed in a similar March 2026 article here.

But the moment you introduce uncertainty, incomplete information, a patient who's describing something that doesn't fit the textbook, a clinical presentation that requires judgment about what matters most given who this specific person is — performance drops significantly. The same Stanford-Harvard report found that under those conditions, AI systems performed closer to medical students than to experienced physicians. They also tended to commit strongly to an answer even when ambiguity was high.

That last part is important. Experienced physicians know what they don't know. AI, at its current state, often doesn't.

That's not a knock on the technology. That's a design insight. And it fundamentally changed how I built ChatRx.

The Real Question Is Not Replacement. It's Allocation.

The former president of the American Medical Association said something I think about often: "AI is not going to replace doctors. But doctors using AI will replace doctors who aren't using AI." That's the right frame. Not replacement. Competitive advantage through augmentation.

The clinical tasks that AI should own are the ones that never required a physician's judgment in the first place. Symptom intake. Protocol matching. Prior authorization documentation. Appointment scheduling. Routine follow-up messaging. Lab result routing. Refill management for stable, chronic patients on established regimens. None of these tasks require twelve years of medical training. All of them currently consume physician time.

When you peel those tasks away, what's left? The actual practice of medicine. The judgment calls. The conversation with the patient who is technically responding to treatment but clearly isn't okay. The decision about whether this chest pain is the one you need to take seriously right now. The conversation with a family that's scared and needs someone to explain what's happening and what comes next in plain language. The moment when the data says one thing and your clinical instinct says another, and you've learned, over years, to trust what that instinct is telling you.

That is physician work. AI cannot do it. Not because it isn't smart enough. Because it cannot be accountable. It cannot own a decision. It cannot bear responsibility for being wrong. The physician can, and does, every time we walk into a room.

What Building ChatRx Actually Taught Me

The architecture of ChatRx is built around a principle I call physician-governed AI. The AI handles the front end: structured symptom capture, red flag screening, differential triage, protocol matching, pharmacy routing. It does this faster and more consistently than any human intake process could. But every case that results in a prescription, a referral, or a clinical recommendation goes through a physician — not as a rubber stamp, but as the decision-maker who reviews the AI's work and owns the outcome.

This isn't a regulatory workaround. It's the right model. Because what I found is that when you give a physician an AI-assisted intake, structured, complete, and with red flags already screened; the quality of the physician's decision goes up, and the time required goes down. The doctor isn't starting from scratch. They're making a judgment call on a well-organized clinical picture. That's what physicians are trained for. Give them the friction-free version of that, and they're faster, sharper, and more satisfied.

A 2025 randomized controlled trial published in Nature Medicine found that physicians using AI assistance improved clinical decision accuracy from 47% to 65% in one patient group and from 63% to 80% in another, without introducing bias into care decisions. That's not AI replacing physicians. That's AI making physicians measurably better.

That's the future I'm building toward. Not a world where doctors are obsolete. A world where the parts of medicine that drain physicians and add no clinical value are handled by infrastructure, and physicians are freed to do the work that only they can do.

What This Means for You as a Physician Entrepreneur

If you're reading this as a physician who is thinking about entrepreneurship, here's what I want you to take from this:

The AI moment in medicine is not a threat to your profession. It is an opening. The physicians who understand how AI works, its actual capabilities and its real limitations, are positioned to lead the next decade of healthcare innovation in a way that no technologist without a clinical background can replicate.

You understand what a patient actually needs. You understand what happens when triage goes wrong. You understand the difference between a protocol that looks right on paper and a decision that is right for this patient, right now. That knowledge is not programmable. It took a decade to build. And it is the thing that makes physician-governed AI coherent, not just as an ethical standard, but as a clinical and business advantage.

The micro-business opportunity here is real. Telemedicine. AI-assisted DPC. Physician consulting for health tech companies building clinical AI. Medical advisory roles for startups that need someone who actually understands what happens inside an exam room. These opportunities exist right now, and they are growing. The physician who has a business structure to capture them, a micro-corporation, a professional brand, multiple income streams , is in a position to participate in the most significant transformation in healthcare since the invention of the EHR.

And unlike the EHR, this one is actually going to make your professional life better. If we build it right!

Monday Case Study: Dr. Oh Builds the Bridge

Dr. Oh (name changed) is an internal medicine physician who spent eight years in a large health system before approaching me about consulting on a health tech startup that was building an AI triage tool. She had no formal business training. What she had was something the startup's engineering team desperately needed: the ability to look at their clinical logic and tell them exactly where it would fail and why.

She set up her micro-corporation specifically to take on consulting engagements like this one. Within 18 months she was consulting for three different health tech companies simultaneously, reviewing clinical AI decision trees, flagging protocol gaps, and serving as the physician voice in product development meetings. She billed at $350/hour through her micro-corp. She kept her part-time clinical position because she wanted to. Not because she had to.

She told me: "The tech companies I work with are brilliant. They can build anything. But they kept building things that would never survive contact with an actual patient. They needed a physician who could translate. That's not something you can hire a developer to do."

That's the gap. And physician entrepreneurs are the ones who can fill it.

Read more about how physicians are building careers at the intersection of medicine and tech: Every Doctor Is A Business and The E-Myth for Doctors: Don't Build Another Job — Build a System.

Is This Deductible?

Scenario: I spent $2,400 this quarter on AI software tools, a clinical documentation assistant, a scheduling automation platform, and a medical knowledge database subscription, all used in my independent practice. Deductible?

If you're incorporated: Yes — 100% deductible as business software and technology expenses through your micro-corporation. These are ordinary and necessary business expenses under IRC Section 162. Document the business purpose for each tool and run them through your accountable plan or directly as corporate expenses.

If you're a W2 employee using these tools: Almost certainly not. Since the 2017 Tax Cuts and Jobs Act, unreimbursed employee business expenses are no longer deductible at the federal level. Your employer may reimburse you, but if they don't, that $2,400 is gone with no tax benefit.

Bonus insight: If you're a physician consultant advising a health tech company, any AI tools you use specifically for that consulting work are fully deductible business expenses through your micro-corporation. The consulting economy in medical AI is real, and the tax treatment favors those who are structured to participate in it.

→ Free Download: Tax Deduction Guide for Micro-Business Owners

→ Free Download: The Ultimate List of Business Deductions for Professional Micro-Corporations

Join the Movement

"AI will not replace the physician's judgment. It will make that judgment the most valuable thing in the room, because everything else will have already been handled." — Dr. Tod Stillson

The physicians who thrive in the AI era won't be the ones who fought the technology or surrendered to it. They'll be the ones who understood it well enough to govern it, build with it, and profit from it, on their own terms.

If you want to be in that group, the first step is building the structure that lets you participate. A micro-corporation. A professional brand. The business literacy to operate as an owner, not just a clinician.

→ Join PEA Explorer Membership — build with physicians who are already making this move

→ Book a Micro-Business Strategy Consultation — let's map your specific path

→ Free Download: The Physician's Guide to Entrepreneurship

→ Free Download: The Physician Enterprise Roadmap

→ Free Download: Job Options for Independent Physicians — Breaking Free from Corporate Medicine

→ Free Download: Every Doctor Is A Brand — Distinguishing Yourself as an Independent Doctor

→ Enroll: My top selling course: Creating a Practice Without Walls

The future of medicine is not AI versus physicians. It is AI as infrastructure that makes the physician more effective, more autonomous, and more valuable than ever. That future is already being built. The question is whether you're building it, or watching someone else do it.