Medical council with rifles

Basically, it all comes down to how much you believe not in statistics but - statistics. If she tells you that, according to available data, a self-driving car is 70 to 90 percent safer than one with a human behind the wheel, will you let it drive you through the Morača canyon while you take a nap? With a note that the driver does not drink, does not exceed the speed and knows the road by heart even in low visibility, which are the most common causes of traffic accidents. 

If we take this to the next level, literally, would you get into an airplane that has no pilot but is driven by autopilot? And this is, for the most part, already a reality because in most of today's passenger planes, the pilots monitor the operation of the computer, except during landing and takeoff when they are more directly involved. Although that is also possible without human participation, it is just not allowed.

Finally, we come to the key question of today's column: would you let artificial intelligence make your medical diagnosis and prescribe treatment? Although many are already doing it with the help of the famous "Dr. Google", we are still far from the doctor's council being replaced by a computer. How far? Microsoft company says - five to ten years away.

Microsoft presented its "path to medical superintelligence" which is traced by their team formed for this purpose, headed by Mustafa Suleiman, a British technology expert. The system functions as a simulation of a panel of specialist doctors who solve extremely demanding and diagnostically complex cases.

In cooperation with the advanced model o3 by OpenAI, Microsoft's System VI successfully solved eight out of ten challenging clinical studies. In comparison, when the same tasks were given to medical practitioners without access to colleagues, literature or digital tools, accuracy was two out of ten.

In addition to greater accuracy, the VI system is, according to Microsoft, a cheaper option because it more efficiently determines the necessary analyzes, thus reducing unnecessary costs in healthcare.

Suleiman told the "Guardian" that he expects the system to work almost without error in the next five to ten years, that is, that in the next decade the systems will become practically faultless, which, in his opinion, will be a relief for the health systems of the world.

This statement is quite bold, and if any (already frowning) doctor is reading this, they will surely agree that technology has already dramatically raised the quality of medicine. Starting with cameras and laser operations, scanners, not to mention diagnostic devices and laboratories. So the point is not to replace the living doctor, who will be necessary as long as we treat people and not robots. The bottom line is that that doctor has an assistant at all times who has all the medical knowledge in the world and the ability to use it when faced with the patient's symptoms. But the doctor still establishes the necessary relationship with the patient and gives the final word by evaluating what he sees and recognizes. It just means that medical studies and specialization will probably take even longer than now in order to get an expert who is a useful assistant to YOU, and not someone who will blindly believe in these findings.

Although they say this will reduce healthcare costs, it is hard to believe. Diagnosis VI will be faster and more accurate, but eternal learning of the system will be expensive, and consequently its use, i.e. rental. Probably lawsuits when it fails, obviously those two misdiagnoses are too expensive for Microsoft and the system won't be active until it's down to one in a thousand. I was going to say one in a million.