Innovation Amnesia

We Call Yesterday’s Innovation Today’s Medicine

Lance Black, MD

Family and Flight Physician, Bioengineer and Innovator advancing human health through science and technology

Don’t Forget: You Were New Once

Innovation amnesia is our tendency to forget that every familiar technology was once unfamiliar. Each required clinicians to change their behavior, redesign their workflows, and reconsider what counted as competent care. Once a technology becomes embedded in practice, however, we stop seeing it as technology. We call it medicine. That forgetting leaves us chronically suspicious of whatever comes next, as though disruption were unprecedented rather than one of the profession’s oldest patterns.

Consider artificial intelligence. If I hear one more person say, “AI will not replace physicians, but physicians who use AI will replace those who do not,” I may lose my mind. The line presents itself as bold and forward-looking, but it is often a deeply reassuring compromise: AI will be transformative, just not transformative enough to threaten the role or value of the person repeating the quote. It is the Goldilocks position on technological change. AI matters, but only the right amount.

The less comforting truth is that technology replaces things all the time. It replaces tasks, workflows, and entire categories of expertise. It may not replace a physician as a human being, but it can absolutely replace enough of a physician’s work to transform, or even eliminate, the role that person currently occupies. More on that later. First, it is worth looking backward, because medicine has lived through this story before. Consider the history of ultrasound.

When Ultrasound Was Just Ultranoise

Ultrasound did not arrive in medicine as an obvious success recognized by your typical practicing clinician. One of its earliest pioneers was Karl Dussik, an Austrian neurologist who worked with his brother Friedrich, a physicist, to send ultrasonic waves through the human skull. In the 1940s, they published images that they believed depicted the brain’s ventricles. They got it wrong. Much of what they had recorded was artifact produced by the skull itself. One of medical ultrasound’s first celebrated applications was, by modern standards, largely a failure.

Imagine how easy it would have been to seize on that failure as proof that the technology itself was useless. How many physicians saw those crude images and concluded that ultrasound was an expensive distraction searching for a clinical purpose? That is the advantage enjoyed by every established technology: it is judged by what it has become. A new technology is judged by its earliest and least capable version.

The breakthrough came more than a decade later from a different partnership. In Glasgow, obstetrician Ian Donald joined physician John MacVicar and engineer Tom Brown to adapt equipment originally designed to detect flaws in industrial materials. Their experiments culminated in a landmark 1958 paper demonstrating that pulsed ultrasound could distinguish abdominal masses and produce images of pregnancy and the fetus. The images were crude, the machines cumbersome, and the technique difficult to master. Its future was anything but assured.

Obstetrics nevertheless offered ultrasound something every emerging technology needs: an opening. Existing imaging depended heavily on X-rays, which provided limited information about soft tissue and exposed the fetus to ionizing radiation. Ultrasound offered something fundamentally different: a way to look repeatedly inside the pregnant body without that exposure. It did not need to outperform every established imaging method in every specialty. It needed to solve one important problem that existing tools solved poorly.

That narrow opening became a widening breach. Through the 1960s and 1970s, ultrasound improved, accumulated evidence, and spread beyond obstetrics into cardiology, emergency medicine, critical care, anesthesia, and nearly every corner of clinical practice. Today, handheld ultrasound is routinely described as the second stethoscope. A technology once limited by questionable images, unwieldy equipment, and uncertain clinical value is now so deeply embedded in medicine that many physicians would feel naked without it.

This is innovation amnesia in miniature. We remember ultrasound as indispensable, but forget the decades during which it was experimental, technically limited, and unevenly trusted. We remember the destination and erase the journey.

AI Isn’t Coming. It’s Already Practicing.

That history should make us cautious about how confidently we judge artificial intelligence in its infancy. We are quick to find the artifact in the image, the hallucination in the answer, or the failure at the edge of a model’s competence and declare that we have discovered the limits of the technology itself. We compare an immature AI system with the fully developed version of modern medicine and congratulate ourselves when the established system wins. But that is precisely what innovation amnesia allows us to do: judge the new technology by what it cannot yet accomplish while forgetting how long our indispensable technologies took to become indispensable.

Anyone who understands the underlying capabilities of AI can see that it will not simply be added to healthcare. It will reconstruct healthcare from the ground up. Imagine a surgeon no longer manually performing one operation at a time, but supervising several AI-directed robotic systems capable of completing increasingly autonomous portions of surgery. The surgeon’s expertise would remain essential, but its expression would change, from continuous manual execution to planning, oversight, exception management, and accountability. We have seen this transformation elsewhere: skilled workers once assembled automobiles with their own hands; today, many supervise systems of mechanical arms that perform those same tasks with greater consistency and scale.

Now consider medical imaging. Radiology is built around representations designed for the human eye: shades of gray arranged so that a trained observer can distinguish normal anatomy from abnormality. But an AI system does not necessarily need to “see” a CT scan as a radiologist does. It may detect relationships within the underlying data that no human eye could perceive, regardless of training or experience. It could integrate those signals with prior imaging, laboratory results, genomics, physiology, and the longitudinal medical record, then construct a three-dimensional model or another representation designed not for diagnosis, but for human understanding and clinical action. The future of imaging may therefore involve more than machines reading our pictures. It may involve machines discovering that our pictures were only ever accommodations to the limits of human perception.

So, will AI replace physicians? The honest answer is yes: it will replace many physicians as their roles are presently constructed. It will replace tasks, reorganize workflows, eliminate some positions, and create others that we do not yet have names for. That does not mean medicine will have no physicians. It means the physician of the future may bear surprisingly little resemblance to the physician of today.

Our responsibility is not to preserve every task simply because it currently belongs to us. Nor is it to dismiss an emerging technology because its first images contain artifacts. Our responsibility is to determine what physicians should become when intelligence, perception, and technical execution are no longer exclusively human capabilities, and to shape that transformation around the interests of patients rather than the preservation of our own livelihoods. Innovation amnesia makes us forget that medicine has undergone this transformation before. This time, we should try to remember.