In my second year of residency in 1997, I was introduced to mainstream evidence-based medicine, and I was all in. It felt like the answer to so many problems in medicine. We should not practice based on habit, personality, or hierarchy. We should practice based on data. It felt not only intellectually satisfying, but morally right.
What I did not understand then was how much market forces would shape what counted as evidence in the first place. It became obvious to me that the people and organizations most able to produce evidence at scale were not individual physicians, or even small groups of clinicians who were close to patients and asking important practical questions. It was large pharmaceutical companies, device manufacturers, and other powerful organizations with deep pockets because someone has to fund the trial, design it, recruit the patients, analyze the results, publish the findings, and then promote them widely enough that they become part of the mainstream conversation. That takes huge amounts of money, generally millions of dollars.
Meanwhile, the everyday physician is expected to absorb an ever-growing body of literature, guidelines, updates, performance metrics, and protocols. The information load is enormous. Many physicians feel as if they are permanently behind, no matter how hard they work. And the more medicine becomes framed as “follow the evidence,” the easier it is for physicians to start doubting themselves, as if their own training, experience, and judgment are somehow less legitimate unless validated by the latest publication or guideline update.
That, to me, has been one of the quieter losses in modern medicine.
I do not think anybody set out intending for that to happen. Evidence-based medicine was supposed to improve judgment, not replace it. But in practice, it has often been operationalized in ways that make care look more standardized, more protocol-driven, and more interchangeable. Once that happens, the system starts asking a dangerous question: If medicine is mostly following guidelines, why does it require so much physician training?
Real physician voices, twice a week
Free, and one click to unsubscribe.
And that question has consequences. If a less-trained clinician can follow the same algorithm, check the same boxes, and prescribe according to the same protocol, then from the perspective of private equity or a cost-cutting health system, the physician starts to look expensive rather than essential.
I still want us to be data-driven. Absolutely. But I also think we need to expand who gets to contribute to the data and how that data are used.
This is where I think AI may actually help in a meaningful way. For most of modern medicine, individual physicians have had very little ability to systematically generate evidence from their own practices unless they were part of large research infrastructure. But AI is starting to change that. It may become possible to analyze patterns across a physician’s own patient population, identify which subgroups respond differently, detect signals that would otherwise be missed, and learn from real-world care in a more continuous way.
That excites me, because it suggests a future where physicians are not just passive consumers of evidence created elsewhere. We can become contributors again. We can help build knowledge that is more clinically grounded, more patient-specific, and more reflective of real life than many traditional trials.
To me, that is the direction we should be heading. Physicians need to be part of the data revolution. Many physicians are “allergic” to new technology, thinking it’s beyond us. There is nothing more complex than a human body. If we can learn medicine, we can learn anything.
Vaishali Popat is an endocrinologist.
