If AI is just regurgitating the prevailing bias that got us into this health predicament, I don't see how that helps get us out of it.
Junk (training data) in : Junk (answers) out
The Good
One year ago I slipped on ice and hurt my knee.
A couple of weeks ago, I finally got an MRI. The radiologist said it was negative.
Perplexed how I could have knee pain for a year without structural damage, I asked my brother, an MD, and he sent the MRI to another radiologist for a second
opinion.
This second opinion revealed a tear in my quad tendon.
Turns out misdiagnoses like this happen ALL THE TIME.
The initial radiologist read who-knows-how-many MRI's that day, was probably tired, and just missed it.
Would AI make this mistake?
A big part of my MRI bill is paying the radiologist for the (mis)diagnosis.
How much cheaper and more accurate would it be to have an AI read it?
Further, based on the diagnosis, what if the AI could then recommend the
list of potential treatment options, risks and benefits, while evaluating my insurance coverage, and recommending local providers?
So much potential time and money savings — all with technology we have today.
Takeaway:
I believe AI is going to provide a boon to healthcare.
But, at least in the short term, I worry about poor nutrition training data leading us down a continued sick path, because as Mark Twain said:
It ain't what you don't know that gets you into trouble. It's what you know for sure that just ain't so.