Posted on 05/25/2022 2:36:23 PM PDT by Roman_War_Criminal
According to new research, deep learning models based on artificial intelligence can identify someone’s race merely by looking at their X-rays, which would be impossible for a human doctor looking at the same photos.
The findings raise several serious concerns concerning AI’s role in medical diagnosis, assessment, and treatment: Could computer algorithms mistakenly apply racial bias when analyzing photographs like these?
An international team of health researchers from the United States, Canada, and Taiwan tested their AI on X-ray images that the computer program had never seen before after training it with hundreds of thousands of existing X-ray images annotated with specifics of the patient’s race.
Even when the scans were taken from people of the same age and sex, the AI was able to predict the patient’s claimed racial identification on these photos with amazing accuracy. With some groups of photos, the system achieved 90% accuracy.
(Excerpt) Read more at strangesounds.org ...
Someone trained a model (taught the AI) with X-rays and normal pictures of a person with this interesting result. Not sure what they wanted to find. Maybe trying to find a way to diagnose without X-rays? This is not racist and if the prediction has a very high probability, it is just a fact. This is called knowledge that would normally be added to our scientific base that might be useful at some point. However, now it is verboten, and will be expunged.
Evidently some people spend every waking moment looking for nonsense to get hysterical about.
Could computer algorithms mistakenly apply racial bias
Just mayyyybe there are moral issues with the use of some technology?? Who woulda thunk it?
Damn!!! Racist robots.
If doctors can’t tell the difference, how did they program robots to detect the difference?
I heard a rumor that AI Bigots can tell the difference between a biological Male and a biological Female.
Shocking new evidence emerges that AI does not believe belief Trumps Biology.
dogmatists, dogmatizers, partisans, sectarians
doctrinaires, fanatics, purists
jingoists, nationalists
racialists, racists, supremacists
chauvinists, sexists
BAN THEM ALL
They are not predicting anything. If you have the physical attributes of ‘pick your type of human’ , that is simple categorization. If you want the AI to be racist, that will take additional programming.
I find this concern bizarre. Me = retired clinical pharmacist. The race of the person is important as different drug regimes are superior for some races and inferior for other races, in particular drugs for hypertension. Most of the drugs are indifferent to race but some are.
AI better ignore the crime rates—believing their lying eyes is about as racist as it gets!
I learned from anthropology class 50 years ago which races had certain bone features. Any Medical examiner can tell from a fresh skeleton.
There are significant differences in bone density between races (blacks have higher density), something that would readily be apparent in an x-ray.
https://academic.oup.com/jcem/article/82/2/429/2823249
But can that AI tell the preferred pronouns from those X-Rays?
Probably.
Just check the density of the skull.
When it is low, or zero, change the initial predicted sex.
People from different races have numerous minor differences. Why would anyone think that differences in X-rays would not also show differences?
X-raycism
Just apply radio-opaque tape to select areas to change the x-ray image obtained. It would subtract data. Find out the most predictive areas and alter the resultant image. Or add tape which can add data to the image.
I can predict the race of people only by knowing the race of their parents, without equipment of any kind.
Just wait til the AI starts programming the AI
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.
