The medical AIs are incredible. Mayo Clinic has one that is guaranteed to put hundreds of medical informatics people out of work right now. Do you want to hire me to explain skin cancer in the immunocompromised patient, or do you want to ask an AI and get an answer in a minute? Well, do you want a right answer or a wrong answer? Because …
The medical AIs are incredible. Mayo Clinic has one that is guaranteed to put hundreds of medical informatics people out of work right now. Do you want to hire me to explain skin cancer in the immunocompromised patient, or do you want to ask an AI and get an answer in a minute? Well, do you want a right answer or a wrong answer? Because that matters to some people. Unfortunately, it doesn't matter as much as you might expect.
I think of AI like Family Feud, not Jeopardy. You don't get the correct answer, you get the most popular answer. For a lot of questions, that's the same thing. I know the difference, but the people who would hire me don't know, and I'm not sure they care. Until someone gets sued for recommending the wrong treatment, but frankly I don't know how that would happen. One thing I know is that medical uses have to be blessed by the FDA. Mayo Clinic will clear that hurdle but small companies won't. That's bad for small companies and competition, but at least it prevents awful people from telling us all to stop getting vaccinations.
Outside of medical applications? There's no FDA watching your back.
Medical imaging and analysis is a great example of where AI is probably better than humans at catching things. Initially it will be AI+human to validate findings. But just like the computer eventually beating the chess grandmaster, once they can do it better than a human there’a no going back.
The medical AIs are incredible. Mayo Clinic has one that is guaranteed to put hundreds of medical informatics people out of work right now. Do you want to hire me to explain skin cancer in the immunocompromised patient, or do you want to ask an AI and get an answer in a minute? Well, do you want a right answer or a wrong answer? Because that matters to some people. Unfortunately, it doesn't matter as much as you might expect.
I think of AI like Family Feud, not Jeopardy. You don't get the correct answer, you get the most popular answer. For a lot of questions, that's the same thing. I know the difference, but the people who would hire me don't know, and I'm not sure they care. Until someone gets sued for recommending the wrong treatment, but frankly I don't know how that would happen. One thing I know is that medical uses have to be blessed by the FDA. Mayo Clinic will clear that hurdle but small companies won't. That's bad for small companies and competition, but at least it prevents awful people from telling us all to stop getting vaccinations.
Outside of medical applications? There's no FDA watching your back.
Medical imaging and analysis is a great example of where AI is probably better than humans at catching things. Initially it will be AI+human to validate findings. But just like the computer eventually beating the chess grandmaster, once they can do it better than a human there’a no going back.