I recently came across the idea that AI would turn software engineers from bricklayers into architects. But we look for different abilities in a great bricklayer and a great architect. If someone moves from one to the other, we would normally say they changed professions. Why assume this is the same job with better tools?
Try a thought experiment. Take a profession and write down the abilities that would have made someone exceptional at it five years ago. Then ask what happens to each as AI becomes more capable. Some might matter less, others more. New abilities might enter the picture. How much can that combination change before we are describing a different job?
In software, understanding a problem was never enough. You also needed to build the solution. Someone who knew Rails deeply could turn an idea into working software much more effectively than someone struggling through every implementation decision. If a tool supplies more of that knowledge, the difference could narrow. Choosing the right solution and recognizing whether it works might explain more of the difference between them.
My sister is an architect, and I remember her studying during the transition from physical architectural models to digital tools. Designing a good space and building a model of it require different abilities. Model-making can help develop an idea, but someone’s ability to express that idea can also limit how others perceive it. Better tools could reveal strengths we previously underestimated.
That possibility interests me most in medicine. Could AI change which people become exceptional physicians? And would we recognize them using the same signals we looked for in medical school?
My first instinct is to think of students who could absorb enormous amounts of information, understand difficult concepts, and produce the right answer under pressure. But being an excellent student, becoming a prestigious physician, and providing exceptional care are different accomplishments. We shouldn’t assume the same abilities explain all three.
For this experiment, I would start with five: retaining medical knowledge, reasoning through problems, obtaining reliable information from patients, turning decisions into effective care, and learning from experience. The balance would differ by specialty. Still, it gives us something more specific to examine than “good judgment.”
Suppose AI makes relevant medical information much easier to retrieve and apply. Rapid recall could become less distinguishing while understanding remains essential. A physician still needs to understand enough to recognize a bad recommendation. An ability can remain necessary without conferring the same advantage over one’s peers.
Reasoning belongs inside the experiment too. It is tempting to say machines will supply facts while physicians supply judgment. But what happens if tools become increasingly capable of connecting findings and comparing possible treatments? We cannot assume the boundary will stay where we find it comfortable.
Perhaps a physician’s advantage shifts toward recognizing misleading information, identifying which uncertainty matters, or understanding why a plausible recommendation does not fit this patient. Perhaps deep expertise becomes more valuable because it helps the physician use the tool well. Different tasks could move in different directions.
Obtaining the information deserves equal attention. Someone has to ask the useful follow-up question, perform the examination, or discover that a medication in the record is not actually being taken. If interpretation becomes easier, establishing what is happening could account for more of the physician’s contribution.
And a recommendation still has to become care. Someone must perform the procedure, explain the choices, and work out whether the patient can follow the plan. A treatment accomplishes little if the patient cannot obtain it. These abilities have always mattered. Their relative weight could increase.
New situations would require practice too: deciding what to delegate, detecting unsupported conclusions, and resolving disagreements with an AI system. Physicians would need to notice when the tool helps them think and when it encourages them to stop thinking too soon.
The result could be a different combination of strengths defining the exceptional physician. For someone already practicing, that raises useful questions. Which strengths should I deepen? Which advantages might weaken? What can I now accomplish despite a previous limitation? What should I learn that my training never required?
Then take the experiment up a level. If the abilities required to deliver good care change, why would healthcare remain organized in the same way?
Some activities could move between physicians, nurses, other professionals, and patients. Others might need more concentrated specialist attention. Training, supervision, and accountability would shape those changes. We could keep the existing titles while substantially changing how work is distributed among them.
The definition of an exceptional clinic could change too. If many clinics have access to similarly capable AI, their differences might depend more on how well they gather information, handle uncertainty, and follow through. Imagine one clinic producing an excellent plan that nobody implements, while another reliably notices when a patient never obtains the recommended test. Generating a better answer addresses only part of the problem.
My hypothesis is that some of healthcare’s largest changes will come from reorganizing care around this new distribution of abilities. As certain capabilities become more available, the constraints elsewhere become more consequential. Cheaper clinical interpretation does not automatically provide access to examinations, procedures, or treatment. The opportunity is to understand what remains difficult and build around it.
That brings me back to the question about names. A profession carries assumptions about what someone knows, what they can do, and how we identify the best people. Those assumptions influence education, hiring, and the way entire organizations operate. They deserve another look when the work changes.
We may keep calling people physicians, engineers, and architects. But if we keep looking for excellence in exactly the same places, we could miss both the people and the organizations capable of delivering it next.