Artificial intelligence can summarize a medical paper in seconds. Patients can read specialist guidelines before breakfast, compare treatment experiences over lunch, and arrive at an afternoon appointment carrying a proposed diagnosis, a laboratory wish list, and several articles saved on their phones.
Historically, the principal barrier to medical knowledge was access. Information lived in expensive books, universities, teaching hospitals, and the minds of trained practitioners. Acquiring it required money, geography, time, and access to teachers. Medical education was demanding and difficult to obtain.
A relatively small group of people—selected by opportunity, ambition, resources, and endurance—spent years acquiring knowledge and clinical experience. Everyone else depended on their judgment, honesty, and competence. The physician was not simply an adviser but the custodian of information. This arrangement helped produce the paternalistic tradition of medicine: the doctor knew, the patient complied.
No one anticipated the opposite problem: a world in which nearly every piece of information would become available to nearly everyone, almost instantly and at little cost.
This extraordinary access is changing the meaning of medical expertise.
However, the need for medical expertise remains greater than ever but the approach to problems presented and solutions suggested must evolve.
Patients no longer require a physician simply to obtain facts. They need help determining which facts are relevant, how reliable they are, and whether they apply to their unique situation. A search engine can produce a list of diseases associated with weight loss. It cannot easily determine which possibility deserves attention in a particular patient, in what order, and with what degree of urgency.
What distinguishes a plausible explanation from a probable one? When does an anecdote reveal an overlooked phenomenon, and when is it merely memorable? What can an observational study establish? What does a randomized trial leave unanswered? Which expert should be trusted when qualified people disagree?
The following conversations are now commonplace in medical offices.
A young adult—let us call him Ryan—comes to the clinic complaining of weight loss. He is otherwise well, and the change began after he started using cannabis regularly. The physician takes a history, performs an examination, and considers the range of possible explanations. Based on the chronology, cannabis use belongs on that list.
The physician recommends Ryan stop using cannabis for a short period while his symptoms and weight are monitored. Ryan refuses. He is convinced the cannabis is not the problem and is convinced he has cancer. He wants extensive testing to find the “real” cause.
Later, a medical technician also questions the physician’s reasoning.
“Cannabis doesn’t make people lose weight,” she says. “It gives them the munchies.”
“The relationship is more complicated than that,” the physician replies. “We still have to consider other causes, but the timing and overall history makes cannabis relevant.”
“Have you ever smoked it?” the technician asks. “Real cannabis makes you hungry.”
In a few sentences, three forms of knowledge collide.
The technician offers direct personal experience. Ryan contributes knowledge of his own body, symptoms, fears, and priorities. The physician brings formal training, an understanding of physiology and disease patterns, and experience evaluating many patients rather than one.
The technician’s observation is not absurd. THC can acutely stimulate appetite, and some cannabinoid medications have been used to address anorexia and wasting. But cannabis is not one uniform exposure, and “increased appetite” does not settle every question about its relationship to body weight.
Products differ in THC and cannabidiol content, potency, dose, route, frequency, and contaminants. Effects may also differ between occasional and sustained use. Research on cannabidiol, for example, has reported appetite reduction and weight loss in some studies, although the evidence is heterogeneous and does not justify a broad claim that cannabis reliably causes substantial weight loss. [Systematic review indexed by PubMed](https://pubmed.ncbi.nlm.nih.gov/36180814/) Chronic cannabis use may also be associated with nausea, vomiting, reduced intake, and weight loss in cannabinoid hyperemesis syndrome, but that diagnosis requires an appropriate symptom pattern; weight loss by itself is not enough. [Systematic review indexed by PubMed](https://pubmed.ncbi.nlm.nih.gov/34724666/)
The familiar “munchies” are therefore relevant, but they are not dispositive.
This is where anecdotal experience is frequently misunderstood. An anecdote can establish that something happened. It cannot, on its own, tell us how often it happens, what caused it, or whether it will happen to someone else. Observational research can identify patterns across populations, but it may be distorted by confounding variables. Randomized trials are better suited to answering certain causal questions, yet they may study narrowly selected participants under conditions that differ from ordinary practice.
Evidence is not a contest in which one category wins every time. Different forms of evidence answer different questions. Clinical reasoning involves understanding both their value and their limits.
Who is the expert?
The patient is the expert on lived experience: what symptoms feel like, what risks are tolerable, which outcomes matter, and what treatment burdens are acceptable. The clinician is the expert on differential diagnosis, probability, physiology, testing thresholds, and the consequences of acting too early—or too late. Other members of the healthcare team contribute knowledge derived from their training, observations, and sustained contact with patients.
Good care requires these forms to meet without pretending they are identical.
Personal experience with a substance is not a prerequisite for evaluating its medical effects. An oncologist need not experience chemotherapy to understand its toxicities. A psychiatrist need not have taken every medication prescribed. Clinical expertise is built through scientific study, supervised training, pattern recognition, and repeated encounters with outcomes across many patients.
At the same time, credentials do not make clinicians infallible. Training can produce blind spots. Evidence changes. Patients may notice patterns that formal studies have not yet captured. The proper response to those limitations is neither automatic deference nor automatic distrust. It is disciplined curiosity.
Managing uncertainty without abandoning judgment:
In Ryan’s case, the physician should neither dismiss his concern nor declare the cause proven. Unintentional weight loss can have many explanations, ranging from changes in diet and substance use to endocrine, gastrointestinal, infectious, psychiatric, medication-related, and malignant conditions. The appropriate evaluation depends on the amount and rate of weight loss, associated symptoms, examination findings, risk factors, and initial laboratory results.
A reasonable approach would combine several elements:
* Take the concern seriously and clarify how much weight has been lost, over what period, and whether there are warning signs.
* Review cannabis exposure precisely, including product, potency, frequency, route, timing, and associated nausea, vomiting, or appetite changes.
* Conduct a proportionate medical evaluation rather than assuming either cannabis or cancer is the explanation.
* Explain why a temporary reduction or cessation can function as a diagnostic intervention, while acknowledging that improvement would support—not conclusively prove—a causal relationship.
* Agree on a monitoring plan, including when further testing or more urgent assessment would be warranted.
The physician’s responsibility is to recommend what the evidence and clinical picture justify. The patient retains the right to accept or decline that recommendation. Both responsibilities remain intact.
The physician’s changing role:
The democratization of information should not be treated as a threat to medicine. Better-informed patients can ask sharper questions, recognize symptoms earlier, identify errors, and participate more meaningfully in decisions. But access to information is not the same as the capacity to appraise it.
The modern physician is therefore becoming less of a gatekeeper and more of a guide: someone who can separate signal from noise, translate population evidence into individual probabilities, disclose uncertainty without becoming paralyzed by it, and recommend a course of action without confusing authority with control.
That may be a more demanding role than simply possessing the facts.
Medicine’s next era will not be defined by whether patients or professionals “win” the contest over knowledge. It will be defined by whether they can construct a more mature partnership—one in which expertise remains meaningful and experience remains respected,