The algorithm sees one thing. The doctor sees the whole picture.
A dermatologist speaking on a Portuguese health program this week offered a measured reckoning with a transformation already underway: artificial intelligence has entered the clinic not as a distant promise but as a present instrument, capable of reading disease into the body before the body itself knows to speak. João Maia Silva welcomed this shift toward anticipatory medicine while cautioning that the same tools, placed in untrained hands or trusted without question, can mislead as readily as they illuminate. The deeper question he raises is not whether AI belongs in medicine, but whether we have yet learned the wisdom to use it well.
- AI is already reshaping diagnosis in real time — detecting subclinical disease markers through image analysis before patients experience a single symptom.
- A growing number of people consult algorithmic health apps before seeing a doctor, creating a quiet epidemic of misplaced confidence in machine-generated assessments.
- The danger is not the technology itself but the blind trust it can inspire — patients who believe an app has solved their case may delay or entirely avoid proper clinical care.
- Silva insists the algorithm can flag, but only the trained clinician can judge — years of expertise and direct examination cannot be compressed into a digital output.
- The healthcare sector now faces the hard work of integrating AI innovation within structures of specialist oversight, so that earlier detection translates into better outcomes rather than diagnostic shortcuts.
João Maia Silva, a dermatologist, appeared this week on Efeito Placebo on Canal S+ to argue that artificial intelligence has not merely arrived at medicine's door — it has already walked in and begun rearranging the furniture. The shift he describes is profound: where medicine once waited for symptoms to surface before responding, AI-assisted image analysis can now detect disease markers at subclinical stages, visible to algorithms long before the body raises any alarm. Silva calls this evolution "wonderful" — a genuine opening toward medicine that prevents rather than merely reacts.
Yet his enthusiasm comes paired with a clear-eyed warning. Patients increasingly turn to digital apps and online platforms to interpret their symptoms before booking any appointment, and the convenience of an instant, plausible-sounding assessment can breed dangerous overconfidence. When someone trusts an algorithm's reading as final, they risk not only misdiagnosis but the subtler harm of avoiding proper care altogether — convinced they already have their answer.
Silva is unambiguous: AI can support clinical decision-making, but it cannot replace it. A dermatologist's judgment — built from training, experience, and direct examination of a patient in full context — remains something no app can replicate. The lesion an algorithm flags still requires a doctor to weigh it against a patient's complete history and determine what follows. The challenge ahead, as Silva frames it, is not a choice between AI and traditional medicine, but the harder discipline of learning to use both together, with honest clarity about what each can and cannot do.
João Maia Silva, a dermatologist, appeared this week on Efeito Placebo, a weekly opinion program on Canal S+ that airs every Tuesday at 9 p.m. on channel 129, to make a case that artificial intelligence has already arrived in medicine—it is not something we are waiting for, but something reshaping how doctors work right now.
The transformation is already visible in how diseases are diagnosed and how health problems are anticipated before a patient ever feels sick. Silva describes this shift as a "wonderful" evolution for medical practice. The technology is opening a path toward medicine that is less reactive and more preventive, less about treating what has already emerged and more about catching signals of disease while they are still subclinical—visible only to algorithms analyzing images and data, not yet to the human body's own warning systems.
What makes this possible is the ability to identify disease markers at a stage before symptoms appear. A dermatologist or any specialist can now use AI-assisted image analysis to spot patterns that suggest trouble ahead. This is not diagnosis in the traditional sense. It is anticipation. It is medicine learning to read the body's future.
But Silva also sounds a note of caution, and it is one that deserves attention. Many people now turn to digital applications and online platforms to search for information about their symptoms before they ever book an appointment with a doctor. The convenience is obvious. The danger is subtler. When someone reads an AI-generated assessment of their condition and finds it plausible, they may trust it too much. They may delay seeking proper medical evaluation. They may convince themselves that the algorithm has solved the puzzle.
This is where the technology becomes risky. Silva is clear: these tools can support medical decision-making, but they cannot replace it. A dermatologist's judgment, built on years of training and experience, remains irreplaceable. An algorithm can flag a suspicious lesion or pattern, but only a doctor can weigh that finding against the full context of a patient's history, examine the skin directly, and decide what comes next. Blind faith in what an app says—treating its interpretation as final—can lead to wrong diagnoses or, just as dangerously, to a patient avoiding proper care because they think they already know what is wrong.
Silva's comments, made on a program designed to air competing perspectives on health and medicine, reflect a broader tension now playing out across the healthcare sector. Innovation in artificial intelligence is real and valuable. The potential to catch disease earlier, to prevent suffering, to make medicine more precise—all of this is genuine. But that potential can only be realized if the technology is used as a tool in the hands of trained clinicians, not as a replacement for clinical judgment. The challenge ahead is not choosing between AI and traditional medicine. It is learning to use both together, with clear eyes about what each can and cannot do.
Notable Quotes
Silva describes AI's impact on medical practice as a 'wonderful' evolution, enabling more preventive and anticipatory medicine— João Maia Silva, dermatologist
Silva cautions that while AI tools can be useful as support, they should not substitute for medical evaluation by a trained specialist— João Maia Silva, dermatologist
The Hearth Conversation Another angle on the story
When Silva says AI is already here, not in the future, what does he mean exactly? Is he talking about tools doctors are using in hospitals, or something broader?
Both. In hospitals and clinics, dermatologists are already using AI to analyze skin images and flag potential problems. But he's also talking about the fact that the technology has already changed how people think about their health—they're consulting apps before they see a doctor. The future isn't coming. It's already arrived and people are living inside it.
The preventive angle interests me. He mentions catching disease at a "subclinical" stage. What does that actually mean in practice?
It means before you feel anything wrong. Before symptoms. An AI system analyzing your skin images might detect patterns that suggest melanoma risk or another condition is developing, even though you have no marks, no pain, nothing you'd notice yourself. That's the shift—from waiting for the body to announce a problem to having the technology read ahead.
But then he warns about people trusting apps too much instead of seeing doctors. Isn't that a contradiction? If AI is so good at catching early disease, why not trust it?
Because it's not that simple. An algorithm can be excellent at spotting a pattern in an image, but it can't ask you questions about your family history, can't feel the texture of your skin, can't know that you had a similar mark five years ago that turned out to be nothing. A doctor does all that. The app sees one thing. The doctor sees the whole picture.
So the real risk isn't that AI gets it wrong. It's that people think they understand their own diagnosis when they only have part of the story.
Exactly. Someone reads an app's assessment, it sounds authoritative, and they either panic unnecessarily or they delay seeing a real doctor because they think they've already solved the problem. Both outcomes are dangerous.
What does Silva think the healthcare system should actually do with this technology?
Use it as support, not replacement. Let AI flag what it's good at flagging. Let doctors do what only doctors can do—integrate that information into actual care. The sector has to figure out how to balance innovation with safety, which means keeping specialists in the loop, not pushing them out.