Are they? Because the journalist writes some prose about Terminator? Machine intelligence is already running large portions of the western economy, deciding how best to sell you stuff you don't need, and figuring out who needs to be assassinated.
Anyway, they're looking ahead for the next couple of centuries. It's not unrealistic that there will be real live, general AIs by whatever non-mystical definition you choose around by then. What exactly do you want as demonstrating general AI "to a degree?"
Okay, I should have been more specific. An AI wouldn't have human society evolved into it. If we're willing to wait a few million years we could evolve an AI that's symbiotic with our species. If we want one sooner we might conceivably evolve one in fast forward, although not necessarily, there may be limits on how fast you can make AI-type architecture go. Just because computers switch really fast doesn't mean you can make an AI do the same. But even if you could, the best it's going to have from it's evolutionary process is a familiarity with simulated humans, which don't have much in common with the real live, excruciatingly slow living ones.
Not if it's tap water. That's the point. Your resources are going to be wasted regardless of the "treatment" you prescribe. The only difference is what prescription you give. Here are the options, and the likely consequences:
a) prescribe nothing. This is the logical prescription. Unfortunately it doesn't satisfy the patient and he goes and wastes some other doctor's time until he gets one of the other two results.
b) prescribe something active but (somewhat) harmless like antibiotics for a cold. The patient quits bugging you but there is a nonzero chance that he will be harmed by a useless treatment, likely harm to society via things like resistance, and the cost of any drug is greater than that of tap water.
c) prescribe a placebo. It used to be sugar pills, now it could be water. Treatment is free. No harm to patient or society, unless you believe the patient could somehow have been argued out of believing in quackery.
The only issue is that what the NHS decided to do was pay quacks actual money to prescribe tap water. If they'd chosen instead to pay real doctors 0 dollars to prescribe tap water, you'd have situation (c).
I would have replied to you and we could have discussed an interesting question, but since you chose to be an ass in your last sentence it's almost certainly not worth the effort. You must be a delight at parties.
Do you understand the details of the processor in the computer you typed that on? Probably not. The designer of that processor likely could not have designed it without the help of other, less sophisticated processors. And the designers of those processors probably couldn't have made them without the help of still less sophisticated processors.
We already build AIs that are focused on killing humans. They're not terribly smart yet, but they're getting smarter and we seem to be ever willing to give them bigger weapons.
Somewhat. Except that with humans the basic wiring was not determined by other humans but by evolution. Being social animals we've evolved the instinct that having others around us, and happy with us, is usually a pretty good idea. Screwed up parents or the occasional mental illness can blunt that instinct, but in general it's fairly prevalent. An AI wouldn't have that social evolutionary heritage built in.
Let's see. I wouldn't smile if the AI shot someone. I also wouldn't smile if it didn't shoot someone. A psychopath might smile if the AI shot someone, particularly if he knew the AI liked people who smile.
AI doesn't have to mean sci fi self aware AI. A robot that can do image processing, identify targets and attack those targets on its own is AI. Drones exist that can attack things autonomously. Suppose someone put a similar AI system in charge of a nuclear arsenal. It's a great idea right? If a first strike is detected your counterattack is launched faster than any human would be able to react. But you made a boo boo in your pattern recognition code, or your training failed to include a crucial example, and the robots launch on each other when they shouldn't have. Hello Skynet.
We've already got the technology for doomsday AI and we're already using it, on smaller scales.
There are military drones around that have the capability to decide to kill you, all on their own. Welcome to the real world. You're just as dead whether the robot decided to kill you because it thought you were a valid target or because it made some philosophical decision that you should die.
There was a discussion about the national health service in the UK covering homeopathic treatments a while ago on Slashdot. The average Slashdotter (sorry, average of the ones who are not idiots) thought it was a horrible idea. I thought it was a great idea. Provided the "homeopathic treatments" were tap water, prescribed by a real physician. The physician could prescribe water to patients who wanted it for a variety of common ailments such as the common cold, instead of things like antibiotics, but if anyone with something treatable came in he could give them real treatments. You'd have to tell the patient it was an expensive homeopathic remedy of course.
The criticism of that idea is that you'd be reinforcing the superstitious beliefs of people who already believe in homeopathy. My counter argument is that people who really believe things like that tend not to be swayed by telling them it doesn't work anyway.
I see you don't understand what a blinded trial or the placebo effect is either. You're welcome to broadcast your ignorance but please, PLEASE stop trying to convince people to make decisions regarding their health based on it.
"First of all lots of chinese medicine is: standardized, purified and most of all validated."
No, it's not. The first thing you learn when you look into Chinese medicine is that everything is done a little bit differently by each practitioner. When an actual clinical trial of some technique or concoction fails to show an effect the first criticism from believers is usually "oh, you didn't do it right. You have to do it the way THIS school/practice/group/individual does it!" If you do a clinical trial of Advil, Adex, Actron, Anadin or any of the other Ibuprofen brands, you get the same results, because each of them is exactly the same thing.
Regarding Homeopathy, if it would rely on the placebo effect only a small number of patients would be "cured"
No. The placebo effect is quite strong. It can be measured and quantified, although it does depend on the circumstances and the effect in an individual depends very much on that individual's psychology and how they view the treatment they're getting. Modern clinical trials peg the placebo effect at around 30%. Homeopathic remedies HAVE been run through randomized clinical trials and they do not perform better than a placebo.
I don't think you know what homeopathy is. Homeopathy specifically involves diluting substances (ranging from herbal extracts to things like arsenic) until there it is very unlikely there is even a single molecule of the active substance remaining. That is, homeopathic remedies are water. The "theory" underlying homeopathy is that water molecules have a memory of other molecules they've been near and somehow this memory effect turns the water itself into an active drug.
The camomile example you give sounds like an herbal remedy. Many people confuse the two. Many "homeopathic" practitioners probably hope people confuse the two since herbal remedies have a LOT better chance of actually working (i.e. greater than zero). Herbs do indeed contain active ingredients that could potentially be purified, standardized, and validated. However, studies to date (and the US government has invested billions in doing these studies) have resulted in finding one traditional herbal remedy (that hasn't already been turned into a drug) that performed better than placebo: ginger for nausea. Other plant extracts are already used extensively. Aspirin (from willow bark) is the standard example. Currently drug companies are "mining" tropical rain forests looking for drugs.
I'm afraid you don't know what you're talking about, regarding homeopathy or anything else you've mentioned. Please stop trying to educate other people. And no, I'm not being harsh. People like you and the quacks that practice homeopathy are screwing around with people's lives. I have one friend who lost her mother because by the time they realized the alternative "medicines" weren't working it was too late. I remember the day she asked me if I could recommend a good oncologist.
There are bits of Chinese medicine that might potentially be real, if they could be standardized, purified and most of all validated. Homeopathy on the other hand, relies purely on the placebo effect. You don't need expensive water for that.
The captcha's now are harder than they used to be but I have no doubt that if you run a few hundred through a breaker you'd get a few hits. Not quite human level, but impressively close considering where we were five years ago. Someone with some serious computer power to put behind it could probably do significantly better.
AI got a bad name because of the promises it made in the 60s and 80s, and there are lots of mystics who are critical of any AI, but practical things that have come out of AI research are in use every day by Google, Apple, Microsoft and millions of regular people.
Imagine what one of those 60s AI researchers (or even one from the 80s) would think if they saw the translator app I've got on my phone.
There have been several stories about captchas being broken, to the point where secure ones today have to be barely decipherable by humans. That suggests the character recognition algorithms are performing very similarly to humans.
Actually, it seems your post is the vague one. "normal human "learning" is much more fluid." What does that mean?
Learning: (dictionary.com) 1. knowledge acquired by systematic study in any field of scholarly application. 2. the act or process of acquiring knowledge or skill. 3. Psychology. the modification of behavior through practice, training, or experience.
Many machine learning algorithms "learn" exactly the way you'd teach a child. They see examples, you tell them what the object, word, etc. is, and they remember that answer imperfectly. Repetition improves their accuracy and a breadth of examples improves their generality. After not seeing something for a while, they may forget it.
As the other poster pointed out, "deep" describes algorithms that are better able to teach multi-level systems. The changes associated with learning are better propagated to deeper levels, better utilizing all the capacity of the system.
No, it's not just you. There are a lot of people who see the brain as the last bastion of their identity as some kind of special and privileged creature, therefore it must be magical and any attempts to explain how it works are misguided, childish and silly. Whether that's your actual belief or not, that's what your post sounds like. Modern computational neuroscience has actually come a long way. We're even capable of producing chips that can be implanted and replace some parts of the brain. It's not magic.
No, "deep learning" refers mostly to new training algorithms. More computer power helps of course, but the problem previously was that your training became less efficient the bigger your system got. If that doesn't happen, you can scale things up indefinitely.
It's just a sailing outrigger canoe with hydrofoils. It looks different, but it's not really that radical. You can actually buy hydrofoil trimarans for not much more than standard monohulls that also don't look exactly like regular sailboats.
He's right, you clearly don't understand how one time pads work.
With a properly used one time pad, ANY message (of the same length) is equally valid. Typically you salt the message with some nonsense or whitespaces too, so any message of length = the length of the encrypted message is possible.
So you can make up any message you want, gibberish or real words, and you have no idea if it's the real message or not. You cannot use frequency analysis, dictionary attacks, content hints, or anything else against a properly used one time pad.
You're thinking of simpler encryption algorithms that DON'T use completely random pads. Things like Enigma. If you know something of the content of the message that can help immensely in decrypting those messages, but again, prior knowledge, guesses or whatever have no effect on the security of a properly used OTP.
I know reading comprehension isn't among the best of a typical Slashdotter's abilities, but really? It's not a one time pad if it's used more than once, is it?
Slashdot Mirror
Are they? Because the journalist writes some prose about Terminator? Machine intelligence is already running large portions of the western economy, deciding how best to sell you stuff you don't need, and figuring out who needs to be assassinated.
Anyway, they're looking ahead for the next couple of centuries. It's not unrealistic that there will be real live, general AIs by whatever non-mystical definition you choose around by then. What exactly do you want as demonstrating general AI "to a degree?"
Okay, I should have been more specific. An AI wouldn't have human society evolved into it. If we're willing to wait a few million years we could evolve an AI that's symbiotic with our species. If we want one sooner we might conceivably evolve one in fast forward, although not necessarily, there may be limits on how fast you can make AI-type architecture go. Just because computers switch really fast doesn't mean you can make an AI do the same. But even if you could, the best it's going to have from it's evolutionary process is a familiarity with simulated humans, which don't have much in common with the real live, excruciatingly slow living ones.
Not if it's tap water. That's the point. Your resources are going to be wasted regardless of the "treatment" you prescribe. The only difference is what prescription you give. Here are the options, and the likely consequences:
a) prescribe nothing. This is the logical prescription. Unfortunately it doesn't satisfy the patient and he goes and wastes some other doctor's time until he gets one of the other two results.
b) prescribe something active but (somewhat) harmless like antibiotics for a cold. The patient quits bugging you but there is a nonzero chance that he will be harmed by a useless treatment, likely harm to society via things like resistance, and the cost of any drug is greater than that of tap water.
c) prescribe a placebo. It used to be sugar pills, now it could be water. Treatment is free. No harm to patient or society, unless you believe the patient could somehow have been argued out of believing in quackery.
The only issue is that what the NHS decided to do was pay quacks actual money to prescribe tap water. If they'd chosen instead to pay real doctors 0 dollars to prescribe tap water, you'd have situation (c).
I would have replied to you and we could have discussed an interesting question, but since you chose to be an ass in your last sentence it's almost certainly not worth the effort. You must be a delight at parties.
Do you understand the details of the processor in the computer you typed that on? Probably not. The designer of that processor likely could not have designed it without the help of other, less sophisticated processors. And the designers of those processors probably couldn't have made them without the help of still less sophisticated processors.
Who do you suppose will buy the vast amounts of crap that the machines are making? Who supplies these "bucks" to the companies?
We already build AIs that are focused on killing humans. They're not terribly smart yet, but they're getting smarter and we seem to be ever willing to give them bigger weapons.
Somewhat. Except that with humans the basic wiring was not determined by other humans but by evolution. Being social animals we've evolved the instinct that having others around us, and happy with us, is usually a pretty good idea. Screwed up parents or the occasional mental illness can blunt that instinct, but in general it's fairly prevalent. An AI wouldn't have that social evolutionary heritage built in.
Let's see. I wouldn't smile if the AI shot someone. I also wouldn't smile if it didn't shoot someone. A psychopath might smile if the AI shot someone, particularly if he knew the AI liked people who smile.
AI doesn't have to mean sci fi self aware AI. A robot that can do image processing, identify targets and attack those targets on its own is AI. Drones exist that can attack things autonomously. Suppose someone put a similar AI system in charge of a nuclear arsenal. It's a great idea right? If a first strike is detected your counterattack is launched faster than any human would be able to react. But you made a boo boo in your pattern recognition code, or your training failed to include a crucial example, and the robots launch on each other when they shouldn't have. Hello Skynet.
We've already got the technology for doomsday AI and we're already using it, on smaller scales.
There are military drones around that have the capability to decide to kill you, all on their own. Welcome to the real world. You're just as dead whether the robot decided to kill you because it thought you were a valid target or because it made some philosophical decision that you should die.
But Samsung didn't add more devices in a fit of pique, or a tantrum.
I miss the real Slashdot. Circa 1998.
There was a discussion about the national health service in the UK covering homeopathic treatments a while ago on Slashdot. The average Slashdotter (sorry, average of the ones who are not idiots) thought it was a horrible idea. I thought it was a great idea. Provided the "homeopathic treatments" were tap water, prescribed by a real physician. The physician could prescribe water to patients who wanted it for a variety of common ailments such as the common cold, instead of things like antibiotics, but if anyone with something treatable came in he could give them real treatments. You'd have to tell the patient it was an expensive homeopathic remedy of course.
The criticism of that idea is that you'd be reinforcing the superstitious beliefs of people who already believe in homeopathy. My counter argument is that people who really believe things like that tend not to be swayed by telling them it doesn't work anyway.
I see you don't understand what a blinded trial or the placebo effect is either. You're welcome to broadcast your ignorance but please, PLEASE stop trying to convince people to make decisions regarding their health based on it.
No, it's not. The first thing you learn when you look into Chinese medicine is that everything is done a little bit differently by each practitioner. When an actual clinical trial of some technique or concoction fails to show an effect the first criticism from believers is usually "oh, you didn't do it right. You have to do it the way THIS school/practice/group/individual does it!" If you do a clinical trial of Advil, Adex, Actron, Anadin or any of the other Ibuprofen brands, you get the same results, because each of them is exactly the same thing.
No. The placebo effect is quite strong. It can be measured and quantified, although it does depend on the circumstances and the effect in an individual depends very much on that individual's psychology and how they view the treatment they're getting. Modern clinical trials peg the placebo effect at around 30%. Homeopathic remedies HAVE been run through randomized clinical trials and they do not perform better than a placebo.
I don't think you know what homeopathy is. Homeopathy specifically involves diluting substances (ranging from herbal extracts to things like arsenic) until there it is very unlikely there is even a single molecule of the active substance remaining. That is, homeopathic remedies are water. The "theory" underlying homeopathy is that water molecules have a memory of other molecules they've been near and somehow this memory effect turns the water itself into an active drug.
The camomile example you give sounds like an herbal remedy. Many people confuse the two. Many "homeopathic" practitioners probably hope people confuse the two since herbal remedies have a LOT better chance of actually working (i.e. greater than zero). Herbs do indeed contain active ingredients that could potentially be purified, standardized, and validated. However, studies to date (and the US government has invested billions in doing these studies) have resulted in finding one traditional herbal remedy (that hasn't already been turned into a drug) that performed better than placebo: ginger for nausea. Other plant extracts are already used extensively. Aspirin (from willow bark) is the standard example. Currently drug companies are "mining" tropical rain forests looking for drugs.
I'm afraid you don't know what you're talking about, regarding homeopathy or anything else you've mentioned. Please stop trying to educate other people. And no, I'm not being harsh. People like you and the quacks that practice homeopathy are screwing around with people's lives. I have one friend who lost her mother because by the time they realized the alternative "medicines" weren't working it was too late. I remember the day she asked me if I could recommend a good oncologist.
There are bits of Chinese medicine that might potentially be real, if they could be standardized, purified and most of all validated. Homeopathy on the other hand, relies purely on the placebo effect. You don't need expensive water for that.
And you successfully demonstrated that jumping out of a plane with a working parachute does indeed improve your chances of survival.
What was the point of that silly post anyway?
Here's one you can try out yourself: http://code.google.com/p/captchacker/
The captcha's now are harder than they used to be but I have no doubt that if you run a few hundred through a breaker you'd get a few hits. Not quite human level, but impressively close considering where we were five years ago. Someone with some serious computer power to put behind it could probably do significantly better.
AI got a bad name because of the promises it made in the 60s and 80s, and there are lots of mystics who are critical of any AI, but practical things that have come out of AI research are in use every day by Google, Apple, Microsoft and millions of regular people.
Imagine what one of those 60s AI researchers (or even one from the 80s) would think if they saw the translator app I've got on my phone.
There have been several stories about captchas being broken, to the point where secure ones today have to be barely decipherable by humans. That suggests the character recognition algorithms are performing very similarly to humans.
Actually, it seems your post is the vague one. "normal human "learning" is much more fluid." What does that mean?
Many machine learning algorithms "learn" exactly the way you'd teach a child. They see examples, you tell them what the object, word, etc. is, and they remember that answer imperfectly. Repetition improves their accuracy and a breadth of examples improves their generality. After not seeing something for a while, they may forget it.
As the other poster pointed out, "deep" describes algorithms that are better able to teach multi-level systems. The changes associated with learning are better propagated to deeper levels, better utilizing all the capacity of the system.
No, it's not just you. There are a lot of people who see the brain as the last bastion of their identity as some kind of special and privileged creature, therefore it must be magical and any attempts to explain how it works are misguided, childish and silly. Whether that's your actual belief or not, that's what your post sounds like. Modern computational neuroscience has actually come a long way. We're even capable of producing chips that can be implanted and replace some parts of the brain. It's not magic.
No, "deep learning" refers mostly to new training algorithms. More computer power helps of course, but the problem previously was that your training became less efficient the bigger your system got. If that doesn't happen, you can scale things up indefinitely.
It's just a sailing outrigger canoe with hydrofoils. It looks different, but it's not really that radical. You can actually buy hydrofoil trimarans for not much more than standard monohulls that also don't look exactly like regular sailboats.
Sure. Just not for very long and it has to go across the wind for a bit first.
He's right, you clearly don't understand how one time pads work.
With a properly used one time pad, ANY message (of the same length) is equally valid. Typically you salt the message with some nonsense or whitespaces too, so any message of length = the length of the encrypted message is possible.
So you can make up any message you want, gibberish or real words, and you have no idea if it's the real message or not. You cannot use frequency analysis, dictionary attacks, content hints, or anything else against a properly used one time pad.
You're thinking of simpler encryption algorithms that DON'T use completely random pads. Things like Enigma. If you know something of the content of the message that can help immensely in decrypting those messages, but again, prior knowledge, guesses or whatever have no effect on the security of a properly used OTP.
From the GP's quote:
"never reused in whole or part"
I know reading comprehension isn't among the best of a typical Slashdotter's abilities, but really? It's not a one time pad if it's used more than once, is it?