Post

Conversation

GODFATHER OF AI ADMITS HE WAS WRONG ABOUT RADIOLOGISTS geoffrey hinton predicted in 2016 that within about 5 years radiologists wouldnt be reading scans anymore 10 years later he gives two reasons he got it wrong first, making scans cheaper and faster didnt mean hospitals suddenly needed fewer radiologists. it meant they could just do more scans and second, he admits he didnt really understand the job itself his picture of a radiologist came mostly from one former student who spent his time reading scans and barely talking to people the AI part actually went pretty close to plan. machines got really good at reading scans but making one part of a job cheaper doesnt automatically make the whole job disappear AI can shrink a task without shrinking the job
The media could not be played.
David Watson 🥑
Post your reply

The last line nails it. Automating a task isn’t the same as automating a job.
Straight line projections as a class rarely predict the future. There are always retarding and amplifying forces that make confident predictions very risky.
thats probably the lesson here. u can predict the task pretty well and still completely miss what happens to the job around it
This is complete BS. We have radiologists still because medicine everywhere works on a closed guild system. Most systematic waste of IQ in human society.
Great demonstration of how inventors often get the implications of their intentions very wrong careful...We should be very careful about believing all the doomsday predictions we hear about the future.
Im sure it'll change soon, but AI is not increasing radiologist productivity on any large scale at this point. Radiologist demand is sky high. None of these things can function without a rad overseeing it, so why would you pay extra for it if you still need a rad to do the work?
Hinton is an actual communist and understands very little about the world outside his small niche.
Quote
Perry E. Metzger
@perrymetzger
Geoffrey Hinton on how to deal with AI: His answer is literally the word "Socialism". Just because someone helped start a field doesn't mean they should be listened to outside their area of expertise. (He has always had terrible views about politics and economics.)
The media could not be played.
The stats shows people distrust Ai Perhaps, this is the reason? They were told to fear a future by people who didn't fully understand the job being predicted away
and this is what all who work in AI and make predictions are like they will some parts of it right but the extrapolate on that to thinking they get the entire picture right and can make silly predictions don't listen to them
He’s really close to getting it!!
Quote
T Ay.
@ayedtay
100% agree with this and have been saying it for some time Even if you buy the ASI-in-X-years scenario it takes a looong time to actually replace the last 1% of human input and that 1% gets exponentially more valuable during this transition period which can very well last x.com/StockMKTNewz/s…
prof hinton is not aware he wrote another key paper on distallation which along with your work zeta law may help a lot in the technology side, non technical aspect probably will outweigh
> increased capability did not result in fewer radiologists, it resulted in more scans Damn if only there was an economic theory that could have predicted this
It's because people are afraid of giving jobs to computers, when we know computers can do it better, but they already fired me, so why not take their jobs as well. It's better for society.
And yet does he learn this lesson, the second part. All of the AI leaders think they know how our economy works across all sectors. In reality they do not.
You get charged 1,500$ for an MRI. The radiologist gets 40-50$. A radiologist will get 7 or 8 bucks for reading an Xray. Getting rid of radiologists will not reduce the cost in any appreciable way.
Yes, this is called demand expansion. When atm machines came, bank tellers didn't disappear. Just banking volume increased and we still need humans. We can see this in many areas. You will see this in many, if not all areas 10 years from now
This is a good example of how someone can be extremely smart in one narrow field and a complete idiot about everything else. Keep that in mind next time you see predictions from Musk, Altman, or any other "father of" whatever.
Translated from French
In medicine, doctors will be replaced (probably many in the long term), but not all dimensions of the profession can be replaced by AI... Simply because most patients like having a human in front of them. That AI verifies and optimizes our prescriptions, or interprets imaging,
What Hinton may not have first hand knowledge is the experience of radiologists in the AI era. While "radiologists" are the final interpreters; consulting AI before signing off the report is commonly prevalent. AI has not taken over the radiologist job, it has made it easier.
Geoffrey Hinton’s original premise was not wrong. He was wrong about the timeline. Technology (AI medical models) are not as of yet at the level to take over radiology or any other medical field but those models are getting there. Ask yourselves whom would you trust to diagnose
This is a well-worn pattern. Make something cheaper, and you get more of it. Straight from the book.
Why hasn't the US government appointed this guy to oversee AI development? He seems like one of the good ones.
He didn't realy understand the job. And thats true of most jobs the AI bros claim AI will replace. they even know they don't understand and just lie about it anyway for the money. Thats it. Theranos x a billion.
Hinton wasn't wrong. He just didn't understand how hospital billing works. If there is no reimbursement, there is no reason to change the billing practice.
Image