You had one boob job: Three retrospective studies compared AI systems with the clinical decisions of the original radiologist, including 79,910 women, of whom 1878 had screen detected cancer or interval cancer within 12 months of screening. Thirty four (94%) of 36 AI systems evaluated in these studies were less accurate than a single radiologist, and all were less accurate than consensus of two or more radiologists.
From 2020 – Artificial Intelligence Makes Bad Medicine Even Worse
Wait, am I correct in understanding that AI systems couldn’t manage breast exams properly?
Were there no teenage boys available?
(drat, can’t find the YouTube link to the “let us check for you!” ad that made the rounds a couple of years ago with teenaged boys dazzled at the chance to approach real breasts.)
And here we are in Canada unable to talk to a doctor except by phone unless we are taken to hospital by ambulance. How can a doctor diagnose anything with out actually being physically present with a patient. I wonder how many will die because of the push for AI?
Can’t disagree with you, vowg. One of our daughters was just diagnosed with bronchitis, over the phone. Back in May, I was diagnosed with shingles, over the phone.
I think its Bell currently running ads about doctors now making house calls, virtually, of course. What a joke!
Then again, I don’t really want to visit my doctor and be bullied and chastised for not getting jabbed.
“Then again, I don’t really want to visit my doctor and be bullied and chastised for not getting jabbed.”
You too, eh?
My GP lectured me on a phone call.
Our lives and ability to live them as we wish are rapidly shrinking.
We haven’t received a phone call from our doctor, yet. We did receive emails stating she will NOT provide a letter of exemption or fill out exemption forms for COVID-19 vaccines unless an allergist/immunologist-confirmed severe allergy or anaphylactic reaction to a previous dose of a COVID-19 vaccine or to any of its components that cannot be mitigated; a diagnosed episode of myocarditis/pericarditis after receipt of an mRNA vaccine).
Not in Canada, but my unselfishly noble offers to assist female friends in this regards have all sadly been declined…
Darn…
If they only had a procedure to raise men’s balls – we wouldn’t have so many knee injuries.
+++Steve
I’m always thinking of you 🙂
I am of that small percentage of people who have been saying for decades that computers will be our downfall ( or one big factor in it )
We rely way too much on them
the following is not directly about AI but related,
did you know that the data ( family photos, videos, screen shots of bills you paid, the book you wrote etc ) you save on CDs, USB drives, internal or external hard drives start to erase themselves after only 10 years?
Don’t believe me? google it.
if you are lucky it may last 15 years, but eventually the data will be lost, unreadable by any device.
We are putting the entire knowledge of humanity on something that starts erasing itself after only 10 years…
the problems are numerous, from hackers stealing your information/identity to hackers asking million of dollars with viruses like ransomware ( and getting the millions of dollars as easily as getting a coffee at starbucks!)
One day China or some rogue nation will hack/take control of the Pentagon’s computers and the USA will surrender without one shot being fired.
We rely W-A-Y too much on computers!!! They will be our downfall !!!
interesting fact,
the shelf life of a sheet of paper is 500 years
In my best HAL 9000 voice.
” I’m sorry Dave, I can’t allow you to do that “.
As somebody who wrote his first computer program fifty-three years ago and has made his living as a software developer, I have found it wise always to attach the word “yet” to any statement that a computer can’t do something. Fifty-three years ago, a computer couldn’t beat a chess master, much less a grand master. Twenty-nine years later, a computer beat Garry Kasparov, the reigning world chess champion, in a six-game match. It wouldn’t surprise me if an AI application does better than radiologists within five years.
Yet … it may kill a few thousand women while it gets improved. And then when it is “perfected” … there will be no recourse for women killed by it as computers are “perfect” and there can be no malpractice committed by machine. Right? Machines are inerrant.
All I know is this: my wife’s life was saved by a highly trained, dedicated, competent, caring (that’s the important one) and experienced Radiologist who brought all his knowledge and human nuance to the analysis of my wife’s mammograms … and saved her life by catching her breast cancer at the earliest stages. God Bless his humanity. God bless his dedication and competency (along with my wife’s oncologist and surgeon).
All of whom work in America’s … ewwwwwww … FOR PROFIT medical system. Ewwwwww … they all made PROFFFFFITTTT (said with spittle) off my wife’s illness. How awful that they saved her life
In our mastery of fire, millions have been killed. In our mastery of nuclear power, many thousands. In our mastery of transportation tech, millions. Are you suggesting we not try to master new techniques because people may die?
You seem to be infected with modern society’s obsession with safety.
https://www.youtube.com/watch?v=e3EsCIjvrSw
Thank you for your courteous comment (that’s not a snark – the comment is quite courteous).
I don’t know whether it would be the usual thing to do, but there’s nothing preventing the simultaneous use of the AI application and flesh-and-blood radiologists. In fact, if the AI application is using a neural network approach, the application learns from the verdicts of experts.
I, too, am a fan of the for-profit medical system that we have here in the USA. My appreciation of it grew greatly when I reported to a hospital in London with a temperature in excess of 102 F / 39 C, and was told that I should just keep taking soluble aspirin.
Kate is legit funny.
Twenty-nine years later, a computer beat Garry Kasparov, the reigning world chess champion, in a six-game match. It wouldn’t surprise me if an AI application does better than radiologists within five years.
Yes, there is no living person who can beat Stockfish, which is an open source program you can download for free,(Mate in 27 loser! I can’t believe you stumbled into that obvious trap!) and AlphaZero can beat Stockfish. They get this good by playing millions of games, within chess’s narrowly defined rules, against themselves and learning from the results of these matches. That’s still a long way from ‘intelligence.’ There is no way to apply this kind of ‘learning’ to self-driving without letting the cars on the roads and potentially kill huge numbers of people until they finally learn how to not do it very often. People don’t follow rules 100% of the time, unlike computers in chess.
Maybe you can roughly describe how one would go about programming a computer to ‘think’ rather than calculate.
BTW, evolution had no problem with killing billions, even trillions of individual organisms, to get us to where we are in terms of ability to think our way through situations.
We have machines that can “out-think” many insects.
It took evolution almost 4 billion years to evolve muti-cellular critters with brains.
Humans have made devices that are capable of similar levels of abstraction in less than 200 years.
That’s about 50 000 000 times faster than evolution did it.
“Maybe you can roughly describe how one would go about programming a computer to ‘think’ rather than calculate.”
Are you suggesting that ants and mayflies “think”?
Insects operate in a space of environmental options much larger than that which vehicles operate in…”…within chess’s narrowly defined rules…” the rules of traffic are “narrowly defined…” for a given value of narrow, and our machines are slowly learning optimized behavior within this space of environmental options…and so far, even in its infancy, there has not been a calamity that has killed huge numbers of people, and to suggest that it has potential to do so, well, so did the mastery of fire.
First, look into deep neural networks.
Then look into genetically evolved algorithms and hardware.
Then consider the fact that the programmers who made these programs could not play chess, or go, or what have you, anywhere near as well as the programs they made.
You seem really invested in self-driving cars. Nobody is going to accept the level of destruction that went along with “the mastery of fire” for your self-driving cars.
Insects don’t “think,” humans think. Insect ‘programming’ is based on the survival of the species, not individual insects, a standard that won’t be applied to self driving cars. Show me the program that can out-think a species of insects, which has its ability to evolve. We have our hands full out-thinking a virus.
As for teaching computers to think, you have suggested that “genetically evolved” algorithms will hit on an answer by a determined effort to achieve a lucky miracle. Maybe, maybe not. It’s certainly no sure thing. Quantum computers can solve tiny classes of problems really fast, but you could not balance your checkbook with one.
Neural nets have the same issues. Our subconscious has evolved over hundreds of millions of years, since the first mammal, to get anything really recognizable; yet obviously, that mammal didn’t evolve from scratch. Trillions of experiments later, somehow we have hit on some kind of magic group of weightings for the neurons, after trillions of mistakes that lead to the death of trillions of individual organisms.
Ah, so where does mere calculation leave off, and “thinking” begin?
What size and architecture of brain?
Do cats and dogs think? Chimps? Lizards?
What about a man with an IQ of 50 who is too dumb to drive?
What about a man with an IQ of 75 who just scrapes by his test while alert and pumped? Should we ban him from driving while tired?
Soon enough, we will have software that can pass a driver’s exam, both written and road tests.
“Nobody is going to accept the level of destruction that went along with “the mastery of fire” for your self-driving cars.”<-correct, because society is now a bubble-wrapped bunch of cowards who are scared of their own shadow, and are not willing to "share the road" with anyone or anything that does not conform to their level of cowardice.
Just another example of the mentality that seeks to freeze the world into some sickened configuration made up in the minds of sickened men.
“Ah, so where does mere calculation leave off, and “thinking” begin?”
Calculation is the application of pre-determined rules to solve a problem. Thinking solves a problem, even if there are no pre-determined rules.
What about a man with an IQ of 50 who is too dumb to drive?
That’s true. After all, when was the last time we saw Prinz Dummkopf behind the wheel of a car?
Tim, according to your definition, neural networks “think.”
Human brains also work with a set of pre-determined rules, we call them the laws of physics.
Neural nets can detect pattens that rise to a certain level within the bounds of their training, which is constrained by the imagination and the resources of those training them. A human can look at a one-off case and say “Hmm….” and then investigate it from other angles and maybe learn/discover something new. A neural net can only modify its weights a tiny bit when it sees a rare case.
“Human brains also work with a set of pre-determined rules, we call them the laws of physics.”
LOL! So you think that at some point computers are going to be able to perfectly model human brains by complete understanding of the laws of physics? Well, you have a measurement problem there, with Plank’s Constant and uncertainty. It is theoretically impossible to measure the states of all of the neurons in the human brain accurately, even if you could attempt it without killing the subject.
The human brain does not need to understand the laws of physics, only to obey them. It’s a huge difference, it’s the difference between intelligent design and evolution. Evolution is far more powerful, even if it is incredibly wasteful. It only needs to succeed in a tiny number of experiments out of massively uncountable experiments performed, to achieve things that a designer could likely not even conceive of as a goal.
Not to mention the limitations of sensors. Can a computerized camera really tell the difference between a rabbit and a dog? A squirrel and a crow? A small child? A bouncing ball when there is a game in a vacant lot? Can it tell if a pedestrian is blindly walking looking at a cell phone or carefully walking to wait at the curb for a safe time to cross? Someday cheap cameras will improve and computers may improve enough to rapidly apply their algorithms to the orders of magnitude larger numbers of pixels, but there is no guarantee of Moore’s Law, as past performance is no guarantee of future results.
It’s still a pipe dream. It could be that fusion is about to happen, and maybe it will, but optimistic people who follow the field closely put the odds at 50-50.
And no, I don’t want to accept the risk of self-driving cars just so that we can give computers and those who hold all of the wealth and power one more avenue of control over us.
“Joe Schmoe wants to drive across the border.”
“Wait a minute, Mr Schmoe has some problematic comments on some Canadian right-wing website, no can do.”
You cheer on your brave new world. It’s not brave, and it’s not new. Totalitarians have always used these kinds of tools.
Instead of:
“I, For One, Welcome Our New Self-Driving Overlords”,
the headline should read:
“Fear-mongering luddite post of the day.”
Indeed, if women ran the world, we would still be living in caves…without fire.
Lastly, the lesson here seems to be “don’t try this new tech, because people will die!!!”
This sort of thinking is endemic on modern society, with its sickened obsession with safety.
“Caveat Emptor”??? WTF does that even mean??? Latin?? That’s too hard!
“Let the buyer beware”??? NO! I’m too lazy, and wish to outsource my due diligence to institutions that have proven time and again that they are flawed, and yet I still insist on outsourcing my risk assessment and whining about it!
How an obsession with safety creates sick minds and a sick society:
https://www.youtube.com/watch?v=e3EsCIjvrSw
When they invent a machine that can make Spock’s head explode (using logic) then I’ll be impressed.
I think.
Therefore, I am.
I am, therefore I think. But I’m a man.