Spyke

This the same AI that was able to suss out people's race from metadata or whatever and used that to make poor diagnostic decisions?

18

You almost have to admire the balls on any healthcare exec still willing to be so brazen with their enshitification...

6

I'd agree if I'd had more useful experiences with radiologists.

1
lemmy.dbzer0.com

if the test comes back negative, it’s wrong only about 3 times out of 10,000,”

This is a hard one. While this could lead to an increase in the frequency of body scans, it's being said by a healthcare ceo so I kind of need him to play a game a Luigi's Mansion before I can listen to him.

7

if the test comes back negative, it’s wrong only about 3 times out of 10,000,”

So there-in lies the problem, which is that things like cancer, are already rare. Say there are 6 real instances of cancer in 10,000. Say the system says 20 are cancer, 3 actually are, and it missed 3. That means it was only positively correct at a rate of 50%, even though its overall accuracy is well over 99%

14

and "AI" is NOT something that learns from its mistakes. ever.

This is from someone who could program together a radiologist "AI" within a week that could beat the average radiologist.

The problem is not the hit %. It's the accountability chain and the economy.

Regulate "AI" properly, and have a healthy safety net that would leave a radiologist comfortable after a layoff, and we could start to talk.

Before then? Fuck you.

3

I had three radiologists look at CTs of my lungs and say they were (more or less) normal. I showed those same images to a pulmonologist and she immediately saw the problem. The serious problem. And after she showed me how she identified it, I started to be and to recognize the pattern as well.

The radiologists couldn't.

🤷👋

1

You reached the end