There's a lot of interesting things going on at the intersection of medicine and AI. That said, it's important to understand some of the limitations: 1.) Medic

There's a lot of interesting things going on at the intersection of medicine and AI. That said, it's important to understand some of the limitations: 1.) Medical data is often quite sparse and quite poor quality You may only get a few years here and there for a patient, and a lot of the things mentioned in the article ("cough is raspy", "I have a feeling it might be pneumonia") aren't always in the medical record, and even if they are they aren't in a form that's easily accessible to a computer. 2.) Interaction matters Seeking medical care is an extremely vulnerable state to be in. A good doctor isn't doing just diagnosis, but teasing out the right bits of information. It's unclear how a computer will handle, "I feel I'm not getting the full story from this patient" situations. A good doctor (not all of them are) will have the interpersonal skills to get the full story. Finally, even if you solve diagnosis, then what? You have to take action. For a lot of expensive chronic conditions, it's not like the answer isn't obvious or even particularly needs diagnosis. If you're overweight, you probably already know you should eat less and maybe be more active. Many times even people with certifiable diabetes diagnoses do not change their lifestyle appropriately. How you handle putting the diagnosis into action is a tough problem, and it's not entirely clear how things like AI will fix it. Convincing people to change habits is damn hard. Disclosure: I work as a data scientist at a medical AI startup (www.lumiata.com)