How AI Is Flipping Power Dynamics in Modern Healthcare

A patient walks into the office, sits down, and before the doctor can say a word, opens with: "I think I have this — I checked with AI." A situation that was rare three years ago is now routine. Anyone who has led real technology rollouts in healthcare — the kind with physicians, clinical committees, and a budget on the line — recognizes right away that this is much bigger than an anecdote: it's the entire system shifting under our feet.

The data backs it up. In Mexico, 73% of patients already use artificial intelligence or digital tools to interpret their lab results and test findings before they ever set foot in a doctor's office. And that number is climbing toward self-diagnosis. Meanwhile, barely 9% of physicians use AI in their practice. The patient got there first. That's not a cute detail: it's the map of where power in healthcare is moving.

To understand why this changes everything, you have to name what's breaking. For centuries, medicine ran on one thing: information asymmetry. The doctor knew, the patient didn't. That asymmetry was the axis around which the entire system organized itself — the physician, the pharma company, the insurer, the lab. Each earned its place by being, in one way or another, the front door to medical knowledge. Today, that door is an AI the patient already consulted from their phone, for free, at 11 at night. When the entry point to knowledge moves into the patient's pocket, every player's position is up for renegotiation. And from the field, that renegotiation has already started.

The first to feel it is the consultation itself. The patient no longer arrives asking, "What do I have?" but, "is what the AI told me true?" The doctor stops being the oracle and becomes the interpreter: the one who contextualizes, corrects, and guides. That's frightening — it feels like losing authority — but misread, it becomes a trap. The physician who flatly dismisses what the patient brings from AI loses the patient. The one who sits down to review it with them becomes more valuable, not less. And this is exactly where technology, used well, plays in the human's favor: ambient documentation assistants that listen to the visit and draft the clinical note are already reducing physician burnout, giving the doctor back roughly half an hour a day that used to disappear into paperwork. This isn't AI replacing clinical judgment; it's AI pulling the keyboard out from between doctor and patient so the physician can do what now matters most: look the patient in the eye and interpret. Because once the patient already has a hypothesis, the scarce asset stops being the data and becomes trustworthy judgment.

The wave rolls on to the pharmaceutical industry, and in Mexico the blow is deeper than it looks, because it lands on a model that was already cracking. The old playbook — influencing the prescription through the physician and the rep — assumes the patient passes through the formal consultation. But in Mexico that stopped being the rule a long time ago: 41% of people respond to an ailment with home remedies or self-medication before seeking professional care, and of those who buy medication without a prescription, 22% do so because they "already recognize their symptoms." On top of that sits a phenomenon as Mexican as it is massive: pharmacy-adjacent doctor's offices, which today generate more than 10 million consultations a month, fusing diagnosis and sale at the same counter. The traditional sales visit was already competing against all of that. AI is the accelerant. Now, that patient who "already recognizes their symptoms" also shows up with the name of a molecule the machine suggested.

The instinct would be to push the product into that AI layer. But with prescription-drug advertising banned to the general public, the shortcut doesn't exist: the way through isn't to advertise better, it's to inform better. The only legal entry point into the conversation is scientific information — unbranded disease awareness, clinical evidence aimed at professionals — built with enough rigor that an AI finds it and cites it. That layer, which used to live in a forgotten corner subordinate to the commercial team, becomes the company's most valuable asset: it's the only thing that puts the brand in the conversation without crossing the line. The restriction that looked like a brake is precisely what forces pharma into the only game AI rewards: being a trusted source, not an advertiser.

And this is where pharma and medical devices end up playing exactly the same match. When the first "verdict" no longer happens in front of a regulated machine in a clinic but on a phone screen, the manufacturer faces the same challenge as the lab: you don't win by pushing, you win by earning trust. The clinician is asked to believe a reading they didn't produce and can't always explain; the patient, to tell signals from the noise they already carry in from AI. Neither of those problems is solved with a better spec sheet or more commercial pressure. They're solved with evidence, validation, and explainability — with being the name the machine, the doctor, and the patient trust when they have to decide.

And then there are the insurers, where the reshuffle turns almost poetic. For years, the information asymmetry made it possible to deny in bulk; in the United States, several insurers face class-action suits over algorithms that rejected care en masse, with human reviews of barely over a second per claim. But that same patient armed with AI is now using it to fight and appeal the denials they once accepted without a word. The insurer that used AI to shield itself against the patient suddenly finds a patient using AI to defend themselves — an arms race that, on top of new regulation, is lost on reputation. The lesson applies across the whole chain: technology deployed against the empowered patient ends up turning against whoever deployed it.

Four different effects, one single cause. The one gaining control in healthcare is the patient, which is why patient-centricity stops being a brochure line and becomes the structural logic of who wins and who loses. This is where leadership comes in, and where field experience teaches the same thing over and over: this isn't a technology problem, it's a problem of trust and position. One finding sums it up: executives' confidence in AI tends to run far higher than the front line's. The boss is already convinced; the person seeing the patient isn't. The leader who pushes from their own enthusiasm and ignores that gap ends up talking to themselves, with an investment that dissolves into teams pretending to use the tool. What works is less flashy: bring people into the decision before it's made, run quiet pilots so trust is built on evidence rather than a memo, be honest about the limits, and measure value in something that matters to the person using it — not on the committee's dashboard.

The digital transformation of healthcare won't be won by whoever has the best algorithm, but by whoever rebuilds their operation around a patient who no longer arrives blank. The patient has already changed. The question is who, across the whole system, is going to change with them. And that was never a technical decision.

Sources: 

Doctoralia/Funsalud (AI use by patients and physicians in Mexico); Research Land / ENSANUT and Anafarmex (self-medication and pharmacy-adjacent offices); JAMA Network Open / Mass General Brigham studies (ambient documentation assistants); ProPublica and U.S. litigation (insurer algorithms); COFEPRIS / RLGSMP framework (drug advertising).