Digging the Doctor Out

Here is the thought I keep returning to. What if AI's real job is not to be the doctor, but to dig the doctor out from under the pile?

The JAMA authors have a word for this. Excavation — digging carefully through layers to find what is buried underneath. They say it plainly:

AI cannot replace the physician, but it can remove the burden of what replaced the physician's clinical time.

That is from Martinelli et al., 2026. Read it twice. AI is not the next pile of work added on top. It might be the first thing strong enough to take a pile away.

Why this is different

Until now, every fix to doctor burnout has done the same thing: take the paperwork away from the doctor and give it to somebody else. A nurse. A scribe. An assistant. The pile did not get smaller. It moved sideways. The other person got tired.

AI is the first tool that can actually absorb the pile instead of just shuffling it. A computer can write the visit note from the conversation. A computer can fill out the insurance form. A computer can sort the inbox. None of this is replacing the doctor. It is clearing the desk so the doctor can stand up and look at the person in the room again.

Think of it like a dishwasher in a kitchen. The dishwasher does not become the cook. It does the dishes so the cook can cook. Without it, the cook is washing plates instead of making dinner.

Or think of a calculator and your math homework. A calculator does not do your homework for you. It frees up your brain so you can think about the harder parts of the problem.

That is what AI looks like when it is doing its job.

The careful part

Here is the line in the paper that I think is the most important. Not the most quoted. Not the most exciting. The most important.

AI does not replace physician judgment, it preserves it.

Read it slowly (Martinelli et al., 2026). The doctor still has to be in the room. The doctor still has to sign off on what the AI suggests. The doctor still has to notice when the AI is wrong — and the AI will be wrong sometimes, because every tool gets things wrong sometimes.

A doctor who is not in the room cannot check the AI's work. A doctor who is in the room can. AI without a present doctor is like a self-driving car with no driver paying attention. It works until it does not, and there is no one there to catch the mistake.

The danger here is real. If a hospital uses AI to remove the doctor instead of free the doctor, the same problem comes back, just faster. A buried doctor was bad. A doctor sitting somewhere else, not in the room, signing off on whatever the AI generated — that is worse.

Two kinds of work

Reading the paper, I started sorting the doctor's day into two piles in my head.

The first pile is work that can be handed off to a tool. Filling out forms. Looking things up. Sorting messages. Spotting patterns in test results. Writing the first draft of a note. None of this is what the doctor came for. All of it can be helped by a good tool.

The second pile is work that cannot be handed off. A hand on the shoulder. Sitting with someone who is scared. Noticing the thing the patient did not say out loud. Holding someone's gaze when the news is hard. This is the work that needs a human in the chair across from another human.

AI gets the first pile. The doctor gets the second pile. That is the whole idea, and it is a good one.

Why this matters outside medicine

The shape of this question is not just for doctors. Almost every job has these two piles. The work that can be handed off, and the work that needs a human present. AI is about to test every profession on whether it can tell the difference.

The next chapter is about one place where I think we have to be the most careful — and the most honest about what AI is, and what it is not.

References

Martinelli, C., Carnevale, V., Ercoli, A., et al. (2026). Artificial Intelligence Is Not the End of the Physician. JAMA. doi:10.1001/jama.2026.4356