The short answer is no. Machines aren't beating doctors at human tasks but that's a good thing. There's just a crazy buzz these days about machines outperforming surgeons and pathologists. Though we should know that's completely not the case, according to Dr. Luke Oakden-Rayner's blog post properly explaining why.
However through his interesting post, it also in turn explains what AI can practically tackle. In short, the message is we should use these intelligent machines to do what we can't do, or what we aren't efficient at doing. Machines shouldn't replace human medics, or at least for now. Building AI should help doctors, rather than threaten their employment. Think about it, with AI doing things that doctors can't and doctors don't things that AI can't, we have an all-star team today.
In summary AI today is a tool more than anything and that's where I see the mid-term future of medicine. AI shouldn't be considered a doctor itself. Machine learning should work conjointly with human intelligence to promote medical precision beyond levels we see today.
I remain convinced that we have yet to see a machine outperform a doctor in any task that is relevant to actual medical practice. The slowly building wealth of preliminary research suggests that won’t last forever, but for now I haven’t seen a case where the robots win.