Mastery Without a Master: Clinical Deskilling and the Post-Hegelian Paradox of Artificial Intelligence

In the modern hospital, a doctor's work has always been a conversation between human judgment and accumulated knowledge. For centuries, this judgment was built by doing: a physician learns to diagnose by seeing many patients, weighing uncertain clues, and practicing the mental steps of framing a problem, considering alternatives, and explaining a choice. This process is not just about getting the right answer today; it is about keeping the mind sharp enough to get the right answer tomorrow. We know from experience that skills, whether in surgery or in reasoning, require regular exercise to remain strong. If a person stops practicing a complex task, even if they still hold the title of an expert, their ability to perform that task without help can quietly fade. This is the background against which a new question has emerged: what happens when the very tools designed to help doctors think also change the way they think?

A recent paper by Andrea Mangiameli explores this question by looking at how artificial intelligence is reshaping clinical medicine. The author argues that while AI can handle many tasks efficiently, relying on it too heavily for the core work of reasoning creates a hidden danger. This danger is not that the machine will make a mistake, but that the human doctor will slowly lose the ability to think through a problem on their own. The paper suggests that when a doctor repeatedly lets an AI system frame a case, generate a list of possibilities, or draft an explanation, the doctor stops practicing those specific mental muscles. Over time, this leads to a state the author calls "abdicated mastery." In this state, the doctor still holds the official title and the final responsibility for the patient's care, but their actual capacity to exercise that judgment independently has eroded. The system remains under human control in name, but the human capacity required to truly control it has diminished.

The paper builds this argument by examining how skills are formed and maintained. It draws on the idea that professional expertise is not a static thing you possess, but a living ability that depends on continuous practice. When a doctor uses an AI to summarize a patient's history or suggest a diagnosis, they are often engaging in a process of evaluation rather than creation. They are checking a box rather than building the box. The author points out that this shift seems harmless in the moment because the assisted care often looks good. The AI provides a correct-sounding answer, the paperwork is complete, and the patient is treated. However, the paper notes that this smooth performance hides a long-term cost. The doctor is no longer practicing the difficult work of starting from scratch, of wrestling with uncertainty, or of constructing a fresh explanation from the ground up.

This process is described as a "post-Hegelian paradox," a reference to an old philosophical idea about masters and servants, but applied here to doctors and machines. In the original idea, a master who relies entirely on a servant to interact with the world eventually loses their own connection to that world. In the medical version, the doctor is the master and the AI is the servant. The paradox is that the doctor can remain the formal authority while the AI does the heavy lifting of reasoning, yet the doctor's own ability to reason independently slowly disappears. The paper suggests that this is not a sudden failure but a gradual drift. A doctor might use the AI for a problem list, then for a differential diagnosis, then for a treatment plan. Each step seems like a small, sensible shortcut. But after months or years of these small shortcuts, the doctor may find they can no longer generate a problem list or a treatment plan without the machine. The authority remains, but the substance of the skill has been hollowed out.

The research highlights that this problem is different for experienced doctors and for those still in training. For an experienced doctor, the risk is "deskilling," where a capability they once had is lost because they no longer use it. For a trainee, the risk is "foreclosed mastery," where they never get the chance to build the skill in the first place. If a student doctor always starts with an AI-generated summary, they may never learn how to take a vague patient story and turn it into a structured medical problem. They might reach the correct conclusion, but they will have skipped the mental struggle that builds the deep understanding needed to handle unusual or complex cases. The paper argues that this creates a dangerous gap: the doctor can follow the AI's path, but if the path disappears or leads to a dead end, the doctor may not have the resources to find a new way.

The author emphasizes that this is not about blaming doctors for using technology. The paper acknowledges that doctors often use AI because they are busy, because the hospital expects it, or because it helps them focus on other parts of the patient's care. The issue is systemic. The way hospitals are organized, the way electronic records are designed, and the way training is conducted all encourage this kind of delegation. The paper notes that current measures of success in medicine often look only at the immediate result: did the patient get better? Did the AI help? These metrics miss the long-term change in the doctor's ability. A doctor might perform well for years with AI assistance, only to face a situation where the AI fails or the case is too unique for the system, and the doctor finds they lack the independent skill to step in.

The concept of "meaningful human control" is central to the paper's conclusion. The author argues that for a doctor to truly be in control, they must have the actual ability to question, change, or reject the AI's suggestion. This ability is not just a legal right to sign a form; it is a practical skill that must be kept alive through practice. If a doctor can only repeat what the AI says, they are not exercising meaningful control. They are merely relaying a message. True control requires the capacity to understand why a recommendation fits a specific patient, to spot when it does not, and to construct a different plan when the standard one fails. The paper suggests that this capacity is fragile. It can be preserved, but only if the medical system intentionally creates space for doctors to practice these difficult, unassisted tasks.

The paper does not claim that AI is bad or that it should be banned. Instead, it warns that the path of total reliance is a trap. It suggests that the future of medical judgment depends on how we manage the balance between assistance and independence today. If we let the AI do all the thinking, we risk creating a generation of doctors who are formally in charge but substantively unable to think for themselves. The solution proposed is not to stop using the tools, but to ensure that the tools do not replace the practice that keeps the human mind sharp. Doctors must continue to frame cases, generate alternatives, and justify their choices, even when the AI offers to do it for them. This is the only way to ensure that when the machine is silent, the doctor is still capable of speaking with authority.

Ultimately, the paper presents a quiet but urgent warning about the nature of expertise in the age of artificial intelligence. It suggests that the most valuable thing a doctor has is not their access to information, but their ability to synthesize that information into a judgment that fits a unique human life. This ability is built through the slow, often difficult work of practice. If that work is delegated away, the ability fades. The paper concludes that preserving the human capacity to judge is not just a technical detail; it is an ethical necessity. A system can look like it is under human control, with doctors signing off on every decision, but if the doctors have lost the capacity to make those decisions independently, the control is an illusion. The future of medicine, the author implies, depends on recognizing that the way we use our tools today will shape the kind of doctors we have tomorrow.