A doctor wrote in the New York Times that AI has become a valuable assistant in her daily diagnosis and treatment, effectively improving the quality of patient care. However, she is deeply concerned about the group of medical students who are currently undergoing training: will the readily available medical AI lead young doctors to lose their ability to reason and judge?
Recently, there have been many discussions about the disruptive risks that AI might pose to human society. In the medical field, another concern is also growing: will AI one day replace doctors?
This summer, she started using an AI platform exclusive to healthcare professionals. With this tool, she screened antibiotics for patients and interpreted abnormal blood test results. When dealing with a dizzy female patient, the AI suggested a possibility of migraine that was easily overlooked. She also discussed treatment options for elderly patients with lung cancer with large models. In private, she has also used this tool to consult with her family, friends, and about her own health issues.
As a general practitioner, she deals with a wide variety of diseases on a daily basis, and she encounters unfamiliar symptoms or syndromes almost every month. She is confident in collecting medical histories, making clinical logical judgments, and communicating with patients. However, medical knowledge is constantly evolving, and no one can remember all the details. Medical AI fills this gap.
However, the medical students and residents she teaches are also using this AI. She is worried that exposing them to such powerful decision-making tools too early could have negative effects. “The amount of information they need to remember is far below my expectations, and their thinking becomes almost passive when faced with these tools.”
Of course, it is not realistic or in line with the development direction of medicine to completely isolate AI from medical education in 2026. Practicing medicine is different from creative work; the essence of medicine is service. Any tool that can improve the quality of patient care should be accepted. She believes that cultivating doctors in the AI era means training them to think critically with the help of AI.
Twenty years ago, this doctor entered medical school. At that time, when dealing with difficult cases, it was common to consult literature, read journals, and discuss with colleagues. Her white coat always carried a classic manual in her pocket, which she could flip through at any time. But there is a prerequisite for using such manuals: you must clearly know what you want to find. Books cannot make judgments or formulate treatment plans for you.
Nowadays, she no longer carries paper manuals with her everywhere; instead, she always has a mobile phone with her. In her opinion, all the conclusions provided by AI are accompanied by references to literature. However, not all of the insights offered by AI are useful or accurate, as many of the medical literature cited by AI itself contains flaws. To make good use of this tool, one must remain skeptical when browsing information.
Currently, more than half of American doctors are using this AI tool. The platform plans to expand the free access to it, making it available to doctors in developing countries.
This doctor admitted that AI often suggests treatment approaches that she hadn’t thought of. Through such human-computer interaction, she has gained a lot. AI breaks through conventional thinking patterns, ultimately benefiting the patients.
The same tools can have completely different advantages and disadvantages for resident doctors. AI responds quickly and can predict problems, acting as a helper, but it can also become a “stumbling block” in thinking. Experienced doctors, when faced with treatment recommendations for back pain given by AI, can rely on their years of accumulated clinical thinking models to instinctively check whether the proposed solutions are reasonable. But conversely, if a novice is led to think with machines from the beginning, can they still develop this core judgment framework?
For this reason, she consulted Adam Rodeman, a doctor at Harvard Medical School and an AI researcher. Rodeman shares the same concerns: young medical students are no longer able to benefit from the 'effortful deep thinking training' that was practiced by their predecessors. Rodeman explains that experienced doctors may experience a slight decline in skills when using auxiliary tools, but this impact is limited. However, students who are still in school and interns who become dependent on these tools will develop cognitive inertia that hinders the development of core clinical skills.
This doctor does not advocate that medical students memorize every diagnostic and treatment plan, just as ordinary people do not need to remember all the phone numbers of their relatives and friends. However, she hopes that students will master the most fundamental skills of diagnosis: treating the patient’s condition as a complete story, identifying clues from it, and forming a chain of logical reasoning; developing the most important intuition for clinical judgment—recognizing the severity of the illness and determining the priority of different symptoms; and at the same time mastering communication skills to uncover key information that patients may not share voluntarily.
She believes that AI is just another technology that forces the medical field to rethink: in practicing medicine, there are certain abilities that cannot be replaced. When she was studying medicine, imaging techniques had already partially replaced the use of stethoscopes, but she still insisted on practicing the ability to identify heart murmurs. Today, the principle remains the same: even if machines will increasingly be used to remember medical knowledge, medical students still need plenty of opportunities to develop clinical thinking through hands-on experience.
