The Epic Battle of Man versus Machine is a Myth
[Hackaday]
Growing Eyeballs in the Lab and Building Wearables [Read article now »](
The Cyborg vs the Machine
By [Elliot Williams](
AI, or more accurately machine learning, has been making headlines. In the last year, for instance, computers have been trained to recognize broken bones in X-ray images more accurately than people can, while also making fewer false diagnoses. This is pretty amazing, and has lead a large part of the technology media to conclude that [radiologists will soon be "obsolete"](.
If you look to the medical press, however, you get a much more nuanced view. After all, a radiologist does a lot more than look at X-rays. Most obviously, they maneuver the patient into the right position to take the critical X-ray image in the first place, but they do this based on already having a firm intuition about what could be wrong. "So you ran into a tree with your arm while skiing and now it hurts the worst right here" they say as they press gently all around to find the injury, watching your face to see when you wince the worst. The radiologist, or the doctor who sent you there, has already done the diagnosis -- the X-ray is just to refine the picture and confirm hypotheses.
of this is to say that AI won't have its place in the hospitals of the future, but it won't simply be replacing doctors. As another article points out, [it's more likely that radiologists who work with AI will be replacing radiologists who won't](. Because the best software systems out there can make diagnoses quickly, accurately, and on the cheap, it's much more likely to transform X-rays to a screening tool as opposed to a verification as it's used today. But [as with self-driving cars]( there are real issues of where the responsibility lies when the system gets it wrong, and medical malpractice lawsuits are a big deal.
Analogies are always a little slippery, but I'm not sure that relying on a computer for a preliminary X-ray diagnosis is any more likely to put radiologists out of business than relying on compilers to translate higher-level languages into machine language is taking away coders jobs because they're no longer writing assembly by hand. I think it's entirely likely that AI will change radiology, to pick one job out of thousands, but it's a lot more likely to do so in a way that enhances the human element. We are, after all, tool-using mammals.
The big trick is to design machine learning systems to best cooperate with their humans. I think that this is both the crux of AI development at the moment, and the truly hard problem that the "people replaced by machines" narrative glosses over. Maybe it's time to focus more on the cyborg and less on the machine.
From The Blog
[Hackers Want Cambridge Dictionary to Change Their Definition](
By [Jenny List](
A computer security firm petitions the Cambridge Dictionary to get "illegal" taken out of their definition of "hacker". Hear, hear. But is that the way language works? [Read more »](
[Lost in Space Gets 3D Printing Right](
By [Tom Nardi](
When it has become too common for movies and television to hyper-sensationalize engineering, and to just plain get things wrong, here’s a breath of fresh air. [Read more »](
[Keep it Close: A Private Git Server Crash Course](
By [Tom Nardi](
Even if you're not terrified of Microsoft's purchase of GitHub, standing up your own Git repo is easy enough that you should probably do it anyway. [Read more »](
If You Missed It
[Arduino and Pidgin C++](
[A Different Use For Microwave Oven: Melting Aluminum](
[ESP8266 Home Computer Hides Unexpected Gems](
[Nvidia Transforms Standard Video Into Slow Motion Using AI](
[Literary Camouflage For Your Router](
[3D Printed Tourniquets are Not a Cinch](
[Hackaday](
NEVER MISS A HACK
[Share]( [Share]( [Share]( [Share](
[Terms of Use](
[Privacy Policy](
[Hackaday.io](
[Hackaday.com](
This email was sent to {EMAIL}
[why did I get this?]( [unsubscribe from this list]( [update subscription preferences](
SupplyFrame, Inc · 61 South Fair Oaks Avenue · Suite 200 · Pasadena, CA 91105 · USA