Have you ever wondered why your doctor spends most of his time looking at a screen and not at your eyes? Because he's become an overpaid secretary. But Microsoft and Google have just entered the office with tools that promise to change that - or send doctors to the unemployment office. Is this the solution to healthcare or the beginning of the end of the white coat? Is the profession of a GP dead?
Imagine this global phenomenon: you sit in a sterile waiting room, surrounded by magazines from 2018 and people coughing at you with the enthusiasm of an opera singer. When it's finally your turn, the doctor greets you with a mellow "good day," then the following Dedicate ten minutes to a passionate relationship with your keyboardYou're talking about chest pains, and he's typing like he's writing the next big literary hit. This is the reality of modern medicine everywhere in the world. – a bureaucratic hell where diagnosis has become a secondary activity of the administration. But, dear readers, get ready. It is coming. artificial intelligence (AI) that doesn't need coffee breaks, doesn't go on strike, and remembers every symptom you've mentioned in the last twenty years. Is the profession of a GP dead?
When Microsoft enters the operating room
Let's get one thing straight first. When we talk about this revolution, we're talking about Microsoft's the actual weapon of mass destruction of bureaucracy – Nuance DAX Copilot (Dragon Ambient eXperience). Microsoft paid a whopping $19.7 billion (approximately €18.5 billion) for Nuance. Why? Because they knew that healthcare was ripe for a disruptive revolution and that the biggest problem facing doctors today was paper, not viruses.
DAX Copilot is not just a dictaphone you would find in a dusty 90s lawyer's office. It is a system that uses ambient intelligence and the power of the GPT-4 model to "listen" to the conversation between a doctor and a patient. It doesn't need commands. It's just there. While the doctor checks your vital signs and asks you about your lifestyle, DAX, in the background, writes a complete clinical record in seconds, structured according to all the highest medical standards.
The results? The numbers are so good that any engineer would love to stick them on the side of a supercar:
- 7 minutes saved per patient visit.
- 70 % reducing burnout in doctors.
- 50% reduction in time spent on documentation.
- And most importantly: 93 % patients say that the doctor is more “human” using this system.
Ironic, right? We needed the coldest technology in the world to bring warmth back to doctors. The system not only records, but also understands context, cuts out irrelevant chatter about the weather, and focuses on clinical facts. The doctor ultimately just scans the record, confirms it in his EHR (Electronic Health Record) system, and the case is closed.
Google Med-PaLM 2: Genius in a box
If Microsoft the one who organizes the papers and acts as a top administrative assistant, Google is the one who wants to diagnose. Their model Med-PaLM 2 is a medical encyclopedia on steroids. This is not your average chatbot that will prescribe emergency surgery for a mild headache. Med-PaLM 2 was the first AI model to sit on the US medical licensing exam. (USMLE) scored above 85 %. This is the “expert level”.
This thing can read an X-ray, analyze your genomics, and connect symptoms that the average GP would miss because they were thinking about where they parked their car. It's multimodal, meaning it understands text, images, and sensor data all at once. Imagine if your car mechanic knew everything about every bolt ever made—that's Med-PaLM 2 for the human body.
The Cold War of Numbers: Silicon vs. Meat
If we strip away the romanticism of “doctor’s instinct” and look at the raw data, the picture becomes quite painful for the human ego.
1. Diagnostic accuracy (General knowledge)
Here is the benchmark USMLE (United States Medical Licensing Examination) – the “holy grail” of medical exams.
-
Man (doctor): It takes approximately 60 %The average doctor achieves solid results, but is limited by memory.
-
AI (Google Med-PaLM 2): This model scored in the same exam 86,5 %.
-
Result: The AI is a “nerd” who has read and memorized every medical book, research paper, and note written in the last 100 years. A human simply doesn’t have enough “RAM” in their head to match that database.
2. Radiology: Eyes that never get tired
Radiology is the first battlefield where man began to lose.
-
Study (Google Health, Breast Cancer): When analyzing mammograms (breast cancer), the AI system reduced the number false positive results (when the doctor says it's cancer, but it's not) for 5,7 % and false negatives (when the doctor misses the cancer) for 9,4 %.
-
Human deficit: Perceptual fatigue. A radiologist experiences a drop in concentration after reviewing 100 X-rays. His eyes physically see the anomaly, but his brain doesn't register it. The AI sees the 10,000th image with the same sharpness as the first.
3. Dermatology: Melanoma
-
Study (Annals of Oncology): Comparison between 58 dermatologists and a convolutional neural network (CNN) in melanoma recognition.
-
Man: Dermatologists correctly identified melanoma in 86,6 % cases.
-
AI: The algorithm achieved 95 % accuracy.
-
Result: AI rarely mistakes a dangerous sign for a benign mole.
4. Where are doctors in the greatest “deficit”? (Human Achilles’ heel)
Human error statistics are what drives the introduction of AI. Humans are not designed to process huge amounts of data.
-
Misdiagnosis rate: It is estimated that the rate of diagnostic errors in primary care (personal doctors) is somewhere between 5 % and 15 %In the US, this means that approximately 12 million adults receive a misdiagnosis each year.
-
Cognitive bias (Anchoring Bias): Doctors often “anchor” themselves to the first diagnosis that comes to mind and ignore data that contradicts it. AI has no ego and no “favorite disease.” It processes every piece of information with equal weight.
-
Rare diseases: The average GP may see a rare disease once in their career (or never). As a result, they often overlook it and attribute the symptoms to something common (“it’s probably just stress”). The AI has in its database all 7,000+ known rare diseases and immediately connects symptoms that a person would otherwise miss.
So is the doctor dead?
This is where things get interesting. If I have a system that is better at diagnosing (Google) and a system that is better at writing reports (Microsoft), why exactly do I need a tired, error-prone human in the middle?
The cynic in me says: for responsibility. When AI It gets confusing – and sometimes it will – who are you going to sue? An algorithm in the cloud? A server room in Arizona? Tough. We need a human with a signature to carry the moral and legal weight of the decision.
But the optimistic part of me sees something else. I see the end of the era of “typing zombies.” The medical profession is not dead, but the medical administrator profession is. Those doctors who refuse to AI, They will be run over faster than a pedestrian on a racetrack. Those who accept it will become “super-doctors.”
Imagine your doctor having all the medical knowledge in the world at their disposal in real time (via Google Med-PaLM) and that he doesn't have to write a single letter (thanks, Microsoft DAX). What's left for him? He can actually pay attention to you. He can look you in the eye. He can use what AI doesn't (yet) have and probably never will: empathy, intuition, and the ability to calm you down when you're scared.
Conclusion: Is the profession of a traditional GP dead?
We are at a turning point. The technology they are introducing Microsoft and Google, is not just a “software update.” It is a complete transformation of the system, comparable to the transition from horse-drawn carriages to internal combustion engines. DAX Copilot is not cheap, but the cost of inefficiency in healthcare is measured in lost lives and billions of dollars.
Personally, I think that in the near future, a visit to the doctor will look like this: you will walk in, sensors will discreetly measure your vitals (pressure, pulse, saturation), AI will listen to your conversation and suggest a diagnosis, and the doctor will be there like a pilot in a modern airplane - monitoring the systems and making sure you land safely.
Is the profession of a personal physician dead? No. But if your doctor in the year 2025 still typing with two fingers and is looking at the screen instead of at you, then it might be time to replace it. Because the future waits for no one.