Microsoft says its AI diagnoses patients more accurately than doctors
00:00 Speaker A
The use of artificial intelligence, technologies such as Chat GPT is a part of our daily dictionary and thinks that Microsoft AI may be in your doctor’s office. The Microsoft AI team is investigating how AI can help diagnose medical cases. Today, the new work of the team shows that the diagnostic orchestrator using the company’s open AI technology can accurately identify 85% of the case procedures in the New England Magazine Medical Magazine. This is four times higher than a group of experienced doctor and is more costly than the findings. Now join me, Mustafa Suleman, Microsoft AI CEO. It’s nice to see you, Mustafa. The follower of your business for a long time. So let’s start the news today. How will this breakthrough be, how will this health, UH, the progress of the health system will help?
01:39 Mustafa Suleman
The incredible thing is that we get 50 million health -related interrogation in Microsoft AI a day from Copilot and Bing. And our aim is to provide and provide the cheapest health advice and support for the highest quality, most accurate and ultimately to our consumers and more in the future. Although this is only an early research, we hope that as it turns it into production, we hope that it will provide access to very high quality health information.
02:26 Speaker A
The first thing I think is Mustafa, why do I need my doctor when I see this? And does the doctor disappear over time because of these advances in artificial intelligence?
02:47 Mustafa Suleman
No, you definitely still need your doctor. Your doctor will be there to call for the right decision, to plan the treatment immediately after the diagnosis, and to maintain and organize and keep responsible for AI’s self -control. Essentially, what these models do is to look at a wide series of medical information, synthesize and present back to people at the right time. This is an incredible interactive experience. AIS now asks questions about the data themselves, suggests which tests to be obtained, and then improves the accuracy of the diagnosis as you learn more information. And to do it jointly with doctors, you know, they will have themselves clearly.
04:05 Speaker A
How are physicians trained in AI?
04:11 Mustafa Suleman
I mean, today, that is, you know, you need to pass the Usmle, Medical License Exam and you know, you need to get about 60% in multiple choice questions. These models get 95% accuracy on these questions. Um, but it’s really unique about this model, not just using an artificial intelligence. In fact, Big Major AI also uses four of the developers and actually goes to the right thing at any time. And this is the function of an orchestrator. When you decide when you will use a large model or a small model, you will use it too much to think too much or a little, you can get a much more accurate answer by taking the best possible, UH, UH, models.
05:34 Speaker A
UH, in many cases health decisions are interested in life or death, Mustafa. How do you trust this technology?
05:45 Mustafa Suleman
What is great about this method is that you can really see why AI has reached a certain conclusion. And as you add more data, it updates the best forecast for what diagnosis may be. And this gives very transparency about how it thinks, which data it takes into account. Therefore, this is, of course, an unprecedented transparency that is the key to trust, because you can actually return and look at many speeches that are coordinated with each other.
