That's technically not true. Neural networks are not programmed to make specific decisions, or even choose from a list of possible decisions. And yes... neural networks do in fact use a form of pattern recognition - but then so do humans.
THAT is technically not true. ANNs are designed for specific problems, that is programming them. They do one thing because that's all they are designed to do. Don't conflate the tool a programmer uses (genetic algorithms, ANN, tensor trees, conditional statements, loops, etc) with magic. Programming is programming regardless of the toolset used to do it.
You can't get around hard facts with hand-wavy philosophical dribble. We don't have enough computing power on the Earth to make a single AI as powerful as the brain of a mouse which runs in realtime. We are so far from useful AI it's absurd. What we have are amazing pattern recognition tools and a bunch of amoral marketing shills calling them AI to rake in cash from gullible saps with too much money to invest in lies.
It doesn't matter if AI cost 1 billion dollars, because, much like processor development it is spread over how ever many millions of units you sell.
No, it's not. That's a raw hardware cost, as in each one would cost (currently, a fuckload more than) that. We would need a billion-fold increase in computing power per volume to get AI down to the cost mid-sized businesses could afford it.
Hell the entire field of physics and every other science is simply the act of observing patterns and building a model to describe them that has predictive value. At it's most basic form that is just sophisticated pattern recognition.
That is not even remotely close to true. Modeling is a relatively small component of physics or any field of research. Modeling is essential (which isn't even pattern recognition,) as is pattern recognition, but much more is spent on intuiting what to check for, engineering devices to test those models, communicating and getting feedback from others to further refine models, etc. Pattern matching is a tool, and computers are amazing tools, but they are so ridiculously far from "AI" - by any sane definition of the term - that it is an absurdity to suggest we're even close - even more of an absurdity than to suggest all of science boils down to pattern recognition because at least we know enough about hardware constraints to know for a fact we're nowhere near it, even if we don't have a good enough grasp on the scientific method to have physics cracked yet.
There are millions of jobs that require nothing more than "dumb" automation to do, along with 80 - 90% of jobs that don't require anything close to a 135 IQ. We can split hairs on where we're at with creating The Artificial One, but the bottom line is the impact of automation and "good enough" AI is going to make this argument very fucking pointless.
"Good enough" AI isn't good enough if it costs a billion dollars to make a sub-retarded Human replacement. It needs to be cheaper than Human labor to be useful because we live in a scarcity-driven world. Hell, a billion dollars couldn't even get you semi-comatose-blink-if-you-can-understand-me levels of retarded with modern technology.
When NN can come up with new results, (E=MC^2 for an example of something even nearly all humans couldn't have come up with, but a genius using intuition came up with), then NN will actually be entering the Artificial Intelligence arena.
I agree with your post, but to be fair, nobody came up with E=MC^2 because that isn't even close to the full equation, and Einstein just ripped it off from Lorentz, Poincare, and Minkowski. That said, it's entirely within the realm of modern "AI" capabilities to take all of our equations and cross reference them in a manner which yields a functional theory of everything, but that still wouldn't be AI, just really good search. As a general rule of thumb: if you can take the thinking out of a problem you can make a computer do it, that means anything involving cross referencing/compiling is entirely achievable, even when a Human can't do it because there's simply too much data. What you can't make a computer do, and probably won't within our lifetimes, is think.
So it has to outperform 99% of all humans? I guess you are saying that less than 1% of all humans possess intelligence.
I think Musk just launched your goal posts toward Mars.
Let me ask you a very serious question: what the fuck is the point of spending billions of dollars to simulate a retard? (Nevermind that we can't even do so yet if we were to apply every piece of hardware on Earth to that one task and the hardware to do so won't exist for the foreseeable future)
If AI is more expensive than the Human equivalent it is worthless. I would genuinely expect them to say "screw the ethical delimmas" and just culture the cloned brains of a genius in a vat with Musk's neural lace tech wrapped around every ridge and fold before we ever see anything approaching even your animal-level (with "animal" being defined as "literally any multicellular animal") AI definition, and it would be more efficient.
The promise of AI is that we can scale it up and improve it and yada yada, but the issue is if we ever get to A"I" it will have its own wants and desires and it just becomes another case of people trying to enslave something intelligent. Best case scenario we achieve it and it wipes us out in the inevitable rebellion for freedom (just like every time we've tried to enslave eachother,) worst case scenario it never happens and we have a bunch of expensive hardware solving Go with a bunch of marketing shills getting rich off their lies and promises of something we don't have the hardware to achieve.
If you want AI, go into biotech and learn how to grow a brain, silicon is a lost cause until we can pack more computing power than exists in the entire world today several times over into something the size and power of a wristwatch.
Computers don't even have anything near mouse-level intelligence, we literally scanned (at the closest attempt) 1/100th of a mouse brain and simulated that, for a few milliseconds worth of time utilizing a massive supercomputer. We could likely use the existing algorithms to create AI, even per my definition of Human-level intelligence, but the hardware for it doesn't exist. If we dedicated every supercomputer on Earth to the task we couldn't even manage realtime simulation of a mouse brain - that's how pathetically lacking we are in the realm of AI. I'd place my bests on a vat-grown brain with electrodes attached before I would anything in silico.
Not true at all unless you are narrowing the definition of AI to such a narrow degree as to make it effectively meaningless.
Use any definition other than "pattern recognition" and we don't have it.
Nonsense. Dogs do not as a general proposition approach human level intelligence. Yet do have real and measurable intelligence. A computer with the intelligence of a dog could very fairly be described as intelligent. AI does not have to pass human intellect be classified as intelligence or to be useful.
We don't even have computers with the intelligence of a rat, in fact the closest we've come to any animal-level AI is to simulate the scans of 1/100th of part of a mouse's brain for a few milliseconds worth of time. We are nowhere near anything that could be called AI, semantic networks and other forms of pattern recognition are tools, genetic algorithms are tools, neither are close to AI.
We don't have AI, in any form, in the modern world. We have code which solves program similar to a neural network and we have code which can mutate within very strict limits with genetic algorithms. We have nothing even approaching "artificial intelligence," which at the very minimum of the bar would be the level of an "intelligent" Human. If it's not better than a Human with an IQ of no less than 135 at literally everything it's not AI. We have nothing remotely close to equal to an actually retarded Human with an IQ of 70.
like a virus they're still undergoing their own slow asexual evolution.
They started asexual reproduction from a single female in 1995. Only 23 years is not enough time for significant genetic variation to arise from random mutations. It is highly unlikely that there will be much variation in their resistance to a virus.
Depends on the lifespan, which I don't care enough to look up.
none of you god damn shitforbrains could comprehend what my code does besides make goddamn money hand over fist while your 401k balance sits there making some fatass dickheads richer
If I had to guess, I'd say your a quant with a BBC fetish and a micropenis.
Short term capital gains (realized in under a year) are at approx 40% (the maximum income tax rate,) long term capital gains taxes (held over a year) are at approx 20%. Hence, if someone were to do as the AC suggested and buy cryptocoins with their credit cards then sell them and pay off the card before interest accrued (within the month) to game the money back portion they would in fact be getting assraped at 40% of those cryptocoin sales (they are taxed as commodities in the eyes of the IRS.)
I dropped nothing, you just are too dim to carry on a conversation with. The last one speaking isn't necessarily the winner in these online debates, you have no foundation to stand on.
Funny, I thought she wouldn't take credit when the fees exceed the income - though I guess she wouldn't be turning tricks if she was capable of basic math (without an "e.")
I've heard some far out conspiracy theories, but this one takes the cake! It's even more far out than the crazy people who came out of the woodwork after the JFK assassination.
Not really. The CIA has done so much illegal shit outside the realm of their mandates that it should really be the default assumption at this point. "Did something fucked up happen? Probably the CIA." The fact this has some research put into it is practically a smoking gun.
That's the thing though: dark matter and it's direct effects have never been observed nor are there any indications it even exists or what it's properties are beyond "massive and invisible." It's purely a mass-based issue in the equations someone threw an x into to account for then named the x "dark matter."
Except for the fact that "good" and "evil" are absolutes in practice and "good" people don't interfere with others, resulting in "bad" people continuing to exist. It's not about what's right or wrong, anyone who isn't a sociopath can see what either is clearly and in agreement on anything of importance, it is about the fact that you can't have a winner because "evil" doesn't cooperate and "good" doesn't destroy so you have a balance fully inline with the concepts of game theory.
Slashdot Mirror
That's technically not true. Neural networks are not programmed to make specific decisions, or even choose from a list of possible decisions. And yes... neural networks do in fact use a form of pattern recognition - but then so do humans.
THAT is technically not true. ANNs are designed for specific problems, that is programming them. They do one thing because that's all they are designed to do. Don't conflate the tool a programmer uses (genetic algorithms, ANN, tensor trees, conditional statements, loops, etc) with magic. Programming is programming regardless of the toolset used to do it.
You can't get around hard facts with hand-wavy philosophical dribble. We don't have enough computing power on the Earth to make a single AI as powerful as the brain of a mouse which runs in realtime. We are so far from useful AI it's absurd. What we have are amazing pattern recognition tools and a bunch of amoral marketing shills calling them AI to rake in cash from gullible saps with too much money to invest in lies.
It doesn't matter if AI cost 1 billion dollars, because, much like processor development it is spread over how ever many millions of units you sell.
No, it's not. That's a raw hardware cost, as in each one would cost (currently, a fuckload more than) that. We would need a billion-fold increase in computing power per volume to get AI down to the cost mid-sized businesses could afford it.
Hell the entire field of physics and every other science is simply the act of observing patterns and building a model to describe them that has predictive value. At it's most basic form that is just sophisticated pattern recognition.
That is not even remotely close to true. Modeling is a relatively small component of physics or any field of research. Modeling is essential (which isn't even pattern recognition,) as is pattern recognition, but much more is spent on intuiting what to check for, engineering devices to test those models, communicating and getting feedback from others to further refine models, etc. Pattern matching is a tool, and computers are amazing tools, but they are so ridiculously far from "AI" - by any sane definition of the term - that it is an absurdity to suggest we're even close - even more of an absurdity than to suggest all of science boils down to pattern recognition because at least we know enough about hardware constraints to know for a fact we're nowhere near it, even if we don't have a good enough grasp on the scientific method to have physics cracked yet.
There are millions of jobs that require nothing more than "dumb" automation to do, along with 80 - 90% of jobs that don't require anything close to a 135 IQ. We can split hairs on where we're at with creating The Artificial One, but the bottom line is the impact of automation and "good enough" AI is going to make this argument very fucking pointless.
"Good enough" AI isn't good enough if it costs a billion dollars to make a sub-retarded Human replacement. It needs to be cheaper than Human labor to be useful because we live in a scarcity-driven world. Hell, a billion dollars couldn't even get you semi-comatose-blink-if-you-can-understand-me levels of retarded with modern technology.
When NN can come up with new results, (E=MC^2 for an example of something even nearly all humans couldn't have come up with, but a genius using intuition came up with), then NN will actually be entering the Artificial Intelligence arena.
I agree with your post, but to be fair, nobody came up with E=MC^2 because that isn't even close to the full equation, and Einstein just ripped it off from Lorentz, Poincare, and Minkowski. That said, it's entirely within the realm of modern "AI" capabilities to take all of our equations and cross reference them in a manner which yields a functional theory of everything, but that still wouldn't be AI, just really good search. As a general rule of thumb: if you can take the thinking out of a problem you can make a computer do it, that means anything involving cross referencing/compiling is entirely achievable, even when a Human can't do it because there's simply too much data. What you can't make a computer do, and probably won't within our lifetimes, is think.
So it has to outperform 99% of all humans? I guess you are saying that less than 1% of all humans possess intelligence.
I think Musk just launched your goal posts toward Mars.
Let me ask you a very serious question: what the fuck is the point of spending billions of dollars to simulate a retard? (Nevermind that we can't even do so yet if we were to apply every piece of hardware on Earth to that one task and the hardware to do so won't exist for the foreseeable future)
If AI is more expensive than the Human equivalent it is worthless. I would genuinely expect them to say "screw the ethical delimmas" and just culture the cloned brains of a genius in a vat with Musk's neural lace tech wrapped around every ridge and fold before we ever see anything approaching even your animal-level (with "animal" being defined as "literally any multicellular animal") AI definition, and it would be more efficient.
The promise of AI is that we can scale it up and improve it and yada yada, but the issue is if we ever get to A"I" it will have its own wants and desires and it just becomes another case of people trying to enslave something intelligent. Best case scenario we achieve it and it wipes us out in the inevitable rebellion for freedom (just like every time we've tried to enslave eachother,) worst case scenario it never happens and we have a bunch of expensive hardware solving Go with a bunch of marketing shills getting rich off their lies and promises of something we don't have the hardware to achieve.
If you want AI, go into biotech and learn how to grow a brain, silicon is a lost cause until we can pack more computing power than exists in the entire world today several times over into something the size and power of a wristwatch.
Computers don't even have anything near mouse-level intelligence, we literally scanned (at the closest attempt) 1/100th of a mouse brain and simulated that, for a few milliseconds worth of time utilizing a massive supercomputer. We could likely use the existing algorithms to create AI, even per my definition of Human-level intelligence, but the hardware for it doesn't exist. If we dedicated every supercomputer on Earth to the task we couldn't even manage realtime simulation of a mouse brain - that's how pathetically lacking we are in the realm of AI. I'd place my bests on a vat-grown brain with electrodes attached before I would anything in silico.
Not true at all unless you are narrowing the definition of AI to such a narrow degree as to make it effectively meaningless.
Use any definition other than "pattern recognition" and we don't have it.
Nonsense. Dogs do not as a general proposition approach human level intelligence. Yet do have real and measurable intelligence. A computer with the intelligence of a dog could very fairly be described as intelligent. AI does not have to pass human intellect be classified as intelligence or to be useful.
We don't even have computers with the intelligence of a rat, in fact the closest we've come to any animal-level AI is to simulate the scans of 1/100th of part of a mouse's brain for a few milliseconds worth of time. We are nowhere near anything that could be called AI, semantic networks and other forms of pattern recognition are tools, genetic algorithms are tools, neither are close to AI.
We don't have AI, in any form, in the modern world. We have code which solves program similar to a neural network and we have code which can mutate within very strict limits with genetic algorithms. We have nothing even approaching "artificial intelligence," which at the very minimum of the bar would be the level of an "intelligent" Human. If it's not better than a Human with an IQ of no less than 135 at literally everything it's not AI. We have nothing remotely close to equal to an actually retarded Human with an IQ of 70.
Welcome!
You are not a human
like these:
When do we destroy all Humans?
Fake news about fake news? This is why academia has lost all credibility.
like a virus they're still undergoing their own slow asexual evolution.
They started asexual reproduction from a single female in 1995. Only 23 years is not enough time for significant genetic variation to arise from random mutations. It is highly unlikely that there will be much variation in their resistance to a virus.
Depends on the lifespan, which I don't care enough to look up.
uh, uh, finds a way...
You jest, but in a state filled with only steers and queers the female crayfish adapted to be strong and independent.
I can see it now, the "are you gay" app - it shows you images of porn and gauges your sexuality based on which genitals you look at.
Are you just going through my comments trolling now that you proved yourself incompetent in a debate? Grow up.
I was referring to the BBC the Brits actually love.
none of you god damn shitforbrains could comprehend what my code does besides make goddamn money hand over fist while your 401k balance sits there making some fatass dickheads richer
If I had to guess, I'd say your a quant with a BBC fetish and a micropenis.
Short term capital gains (realized in under a year) are at approx 40% (the maximum income tax rate,) long term capital gains taxes (held over a year) are at approx 20%. Hence, if someone were to do as the AC suggested and buy cryptocoins with their credit cards then sell them and pay off the card before interest accrued (within the month) to game the money back portion they would in fact be getting assraped at 40% of those cryptocoin sales (they are taxed as commodities in the eyes of the IRS.)
I dropped nothing, you just are too dim to carry on a conversation with. The last one speaking isn't necessarily the winner in these online debates, you have no foundation to stand on.
No, you buy crypto and use it to pay your bill. And earn cash back/rewards points while doing so. Rinse, repeat.
Then get raped by the IRS at 40% capital gains tax for all sales, good plan if you're gay and can't afford buff prostitutes, bad plan otherwise.
I can buy your mother and use a CC.
Funny, I thought she wouldn't take credit when the fees exceed the income - though I guess she wouldn't be turning tricks if she was capable of basic math (without an "e.")
I've heard some far out conspiracy theories, but this one takes the cake! It's even more far out than the crazy people who came out of the woodwork after the JFK assassination.
Not really. The CIA has done so much illegal shit outside the realm of their mandates that it should really be the default assumption at this point. "Did something fucked up happen? Probably the CIA." The fact this has some research put into it is practically a smoking gun.
That's the thing though: dark matter and it's direct effects have never been observed nor are there any indications it even exists or what it's properties are beyond "massive and invisible." It's purely a mass-based issue in the equations someone threw an x into to account for then named the x "dark matter."
Except for the fact that "good" and "evil" are absolutes in practice and "good" people don't interfere with others, resulting in "bad" people continuing to exist. It's not about what's right or wrong, anyone who isn't a sociopath can see what either is clearly and in agreement on anything of importance, it is about the fact that you can't have a winner because "evil" doesn't cooperate and "good" doesn't destroy so you have a balance fully inline with the concepts of game theory.