We already know that the human brain changes substantially and structurally over time (and that we can change it further by meddling).
Critically, the structure is spatiotemporally contiguous throughout these changes - which is totally unlike the transfer hypotheticals.
the human brain with no obvious connection to what materials the underlying machinery is composed of
Again, this is just asserting the conclusion that the physical structure of the brain is unimportant, and then reasoning backwards from that conclusion.
It's like claiming that a car won't drive, if we make it out of aluminum instead of out of steel or the wheels of wood not rubber.
I think what you're saying is akin to claiming that something without wheels, differentials or a steering column is still a "car" which "drives." It may be a highly efficient vehicle, but it's not going to "feel the same."
That's the point here - the mind isn't a homunculus inhabiting your head, which can simply get a new job managing a different theater. All evidence to date supports the materialist proposition that to radically alter the physical structure of the mind/brain would be to radically alter its subjective character as well.
After all, there are many other problems you run into when you try that game, such as whether a mind is the "same" ten minutes later...a mind can change over time to some degree without changing its categorization.
It's a non sequitor - we're talking about hypotheticals which feature entirely different physical structures, or similar physical structures composed of physically distinct sets of atoms, not single spatiotemporally connected sets of atoms. We are talking about instance identity (the "same" mind), not categorization.
No, because definition by definition does not mean that.
You are begging the question, by simply assuming that human mental processes are exactly representable in entirely different physical structures.
And who knows, maybe it's impossible for me to sleep suspended from my ankles. After all, I haven't tried that either.
The main reason AI might kill us all is that it is not anthropomorphic...You only get one wish, you have to make that wish in machine language...
Isn't this just as true of any programmatically automated solution? Control software bugs/misfeatures have already caused various unintended consequences without any learning/AI components. E.g. that big Northeastern blackout a while back.
a large proportion of wishes that could be made would result in a planet covered in solar panels and computer factories.
And isn't this just a general argument for not hooking up any shiny new control software (again, regardless of learning/AI capacity) to the nukes until after it's been thoroughly tested?
The issue of internal biology in neurons is a big unknown though. We know that cell biology has a big affect on cognitive ability. The real question though is how much of an affect it has on the actual processing capabilities.
Very true, but it's common to see sensational explanations of the ANN, so it's a limitation I like to point out.
There is also the issue of fixed-size inputs (not just for ANN but for essentially all current statistical learning methods). There are some workarounds (like tiling with convolutional networks), but in general it's a pretty severe limitation that will require big advances in data reduction algorithms to overcome. Active Appearance Models have an ingenious solution, the "shape free patch," which could maybe be a template for other approaches. We'll also need to develop robust methods for handling real semantics, which is a prerequisite for most tasks we consider truly "intelligent."
it is reasonable to expect that AI can overlap with the category of intelligences that have such motivations.
Fair enough, but we aren't dealing with the belief that AI can in principle have such motivations, but the belief that any intelligence will have such motivations.
I wasn't aware that saying something is "antimaterialist", especially when it's not, was somehow an argument that anyone would take seriously.
That wasn't supposed to be an argument, in and of itself. I think this "vessel" viewpoint is a kind of closet dualism often exhibited by self-proclaimed materialists when pop psychological notions aren't closely examined.
Then the model of body (and also, the organ of the brain) as vessel for mind is demonstrated by actually being able to move the mind to a new and demonstrably different body.
But this seems to rest on an assertion that it would be the same mind. Set aside whether or not it's possible in principle to "transfer" the intelligence to a new type of "vessel" and just consider the old teleportation problem: is the new copy "you?"
Say, an alien transforms you to a silicon-based machine while preserving your mental processes
Again, that this is possible in principle is just an assertion. It is also possible that one's mental processes can by definition not be preserved in a silicon-based machine, so long as direct simulation is excluded. (I do think that if you simulate every atom in a brain using a physically correct simulation, the simulated brain would feel like a physical one "on the inside," all though this still leaves us with the "is it you" problem.)
Especially if those with domain-specific expertise need to save their careers.
The problem with this argument is that it always rules out the opinions of those most likely to have correct opinions on any subject. You need evidence for specific instances of corruption involving the specific individuals whose statements are being evaluated in order for this to-the-wallet argument to have any weight.
I disagree that domain-specific expertise permits objectivity in that domain.
(i also agree with you)
Good point - though I didn't say that exactly, just that IMHO fame ought to be a lesser factor than domain knowledge in estimating the truth value of statements.
IMHO, all of the fear mongering is based on anthropomorphizing silicon. It implicitly imputes biological ends and emotionally motivated reasoning to so-called AI.
I think that folks who don't have hands on experience with machine learning just don't get how limited the field is right now, and what special conditions are needed to get good results. Similarly, descriptions of machine learning techniques like ANNs as being inspired by actual nervous systems seems to ignore 1) that they are linear combinations of transfer functions (rather than simulated neurons) and 2) even viewed as simplified simulations, ANNs carry the very strong assumption that nothing happening inside a neuron is of any importance.
One man, Harry Daghlian, working alone at night, let slip one cube too many, frantically grabbed at the mound to halt the chain reaction, saw the shimmering blue aura of ionization in the air, and died two weeks later of radiation poisoning. Later Louis Slotin used a screwdriver to prop up a radioactive block and lost his life when the screwdriver slipped. Like so many of these worldly scientists he had performed a faulty kind of risk assessment, unconsciously mis-multiplying a low probability of accident (one in a hundred? one in twenty?) by a high cost (nearly infinite).
(Emphasis mine). That quote is from Genius, James Gleick's biography of Richard Feynman - the author of TFA apparently makes the same mistake.
Well there is bound to be some shit floating around any gravity well. As long as we are talking spherical cows in a vacuum just make the massless parachute larger? How do you actually define "atmosphere" anyways?
Say we want to land on Mars. The atmosphere is 100 times less dense than on Earth, so we'll need 100 times more parachute. I'm sure that's still a lot less mass and complexity than the balancing rockets...but what if we want to make a round trip, and need to land on Mars and Earth?
We're going to need parallel parachute systems of different sizes, or a parachute that can be re-packaged combined with variable unfurling, and suddenly the parachute option is looking a lot more complicated. On the other hand, if you get the balancing rockets to work, you can use them anywhere, over and over again.
I see you've decided to respond with the predicted "it's not tangible, so therefore it doesn't count" excuse.
That's simply false - as several posts below have noted.
This completely ignores the fact that things do not have to be tangible to be considered to have a measurable value.
No, it doesn't, and further you are conflating scarcity and tangibility. Regardless of the measurable value, or tangibility, of a thing, it must be scarce to be stolen.
Referring to copyright violations as "theft" is nothing more than a lobbying tactic based on perverting our language into a simplified form in which nuance cannot be expressed, but I'm not surprised you've fallen for it given your demonstrated lack of legal and linguistic knowledge.
Unfortunately for your position, "that one" is the entire argument. Our civilization possesses codified laws in which the words used to define the laws actually matter.
In the laws that we actually have, as opposed to the ones you wish we had, the taking of scarce goods is theft, while making unauthorized reproductions of copyrighted material is copyright violation.
Slashdot Mirror
Why is there an assumption that intelligence implies will? Why would an AI have goals?
Exactly! I tried to make the same point in this comment, which has apparently sparked a lively discussion.
We already know that the human brain changes substantially and structurally over time (and that we can change it further by meddling).
Critically, the structure is spatiotemporally contiguous throughout these changes - which is totally unlike the transfer hypotheticals.
the human brain with no obvious connection to what materials the underlying machinery is composed of
Again, this is just asserting the conclusion that the physical structure of the brain is unimportant, and then reasoning backwards from that conclusion.
It's like claiming that a car won't drive, if we make it out of aluminum instead of out of steel or the wheels of wood not rubber.
I think what you're saying is akin to claiming that something without wheels, differentials or a steering column is still a "car" which "drives." It may be a highly efficient vehicle, but it's not going to "feel the same."
That's the point here - the mind isn't a homunculus inhabiting your head, which can simply get a new job managing a different theater. All evidence to date supports the materialist proposition that to radically alter the physical structure of the mind/brain would be to radically alter its subjective character as well.
I didn't say they would help, I said "we would find out" because the nuclear powers would undoubtedly try it.
Still, some asteroids are relatively loose agglomerations of rocks, so maybe in that case it would make a difference.
After all, there are many other problems you run into when you try that game, such as whether a mind is the "same" ten minutes later...a mind can change over time to some degree without changing its categorization.
It's a non sequitor - we're talking about hypotheticals which feature entirely different physical structures, or similar physical structures composed of physically distinct sets of atoms, not single spatiotemporally connected sets of atoms. We are talking about instance identity (the "same" mind), not categorization.
No, because definition by definition does not mean that.
You are begging the question, by simply assuming that human mental processes are exactly representable in entirely different physical structures.
And who knows, maybe it's impossible for me to sleep suspended from my ankles. After all, I haven't tried that either.
Right - you're just guessing.
The main reason AI might kill us all is that it is not anthropomorphic...You only get one wish, you have to make that wish in machine language...
Isn't this just as true of any programmatically automated solution? Control software bugs/misfeatures have already caused various unintended consequences without any learning/AI components. E.g. that big Northeastern blackout a while back.
a large proportion of wishes that could be made would result in a planet covered in solar panels and computer factories.
And isn't this just a general argument for not hooking up any shiny new control software (again, regardless of learning/AI capacity) to the nukes until after it's been thoroughly tested?
The issue of internal biology in neurons is a big unknown though. We know that cell biology has a big affect on cognitive ability. The real question though is how much of an affect it has on the actual processing capabilities.
Very true, but it's common to see sensational explanations of the ANN, so it's a limitation I like to point out.
There is also the issue of fixed-size inputs (not just for ANN but for essentially all current statistical learning methods). There are some workarounds (like tiling with convolutional networks), but in general it's a pretty severe limitation that will require big advances in data reduction algorithms to overcome. Active Appearance Models have an ingenious solution, the "shape free patch," which could maybe be a template for other approaches. We'll also need to develop robust methods for handling real semantics, which is a prerequisite for most tasks we consider truly "intelligent."
it is reasonable to expect that AI can overlap with the category of intelligences that have such motivations.
Fair enough, but we aren't dealing with the belief that AI can in principle have such motivations, but the belief that any intelligence will have such motivations.
I wasn't aware that saying something is "antimaterialist", especially when it's not, was somehow an argument that anyone would take seriously.
That wasn't supposed to be an argument, in and of itself. I think this "vessel" viewpoint is a kind of closet dualism often exhibited by self-proclaimed materialists when pop psychological notions aren't closely examined.
Then the model of body (and also, the organ of the brain) as vessel for mind is demonstrated by actually being able to move the mind to a new and demonstrably different body.
But this seems to rest on an assertion that it would be the same mind. Set aside whether or not it's possible in principle to "transfer" the intelligence to a new type of "vessel" and just consider the old teleportation problem: is the new copy "you?"
Say, an alien transforms you to a silicon-based machine while preserving your mental processes
Again, that this is possible in principle is just an assertion. It is also possible that one's mental processes can by definition not be preserved in a silicon-based machine, so long as direct simulation is excluded. (I do think that if you simulate every atom in a brain using a physically correct simulation, the simulated brain would feel like a physical one "on the inside," all though this still leaves us with the "is it you" problem.)
That's a very short term view.
With regards to my second paragraph, it may indeed be only a matter of (a very long) time. However:
when we create something at least as clever as us, that may very well be the end of our era.
What I tried to critique in my first paragraph is exactly this implicit imputation of human-like motivations to the supposed AI.
If general AI is made, there's no reason to think it will have a motivation structure that agrees with humans or that we can even easily model.
My point, which might have been more clearly made, is that there's no reason to think it will have any motivation structure whatsoever.
Especially if those with domain-specific expertise need to save their careers.
The problem with this argument is that it always rules out the opinions of those most likely to have correct opinions on any subject. You need evidence for specific instances of corruption involving the specific individuals whose statements are being evaluated in order for this to-the-wallet argument to have any weight.
I disagree that domain-specific expertise permits objectivity in that domain. (i also agree with you)
Good point - though I didn't say that exactly, just that IMHO fame ought to be a lesser factor than domain knowledge in estimating the truth value of statements.
-- quest for power / energy / food, survival, and maybe even reproduction
But where do these come from? I submit that each one of these is only suggested here because we already have these motivations.
we're a biological vessel for intelligence
I consider this antimaterialist. Our bodies aren't vessels (except in that they're literally full of fluids) we inhabit, they are us.
Well, the google car has been rear ended 7+ times
I believe the quoted number is for their entire fleet of automated cars, which AFAIK is of unknown size.
IMHO, all of the fear mongering is based on anthropomorphizing silicon. It implicitly imputes biological ends and emotionally motivated reasoning to so-called AI.
I think that folks who don't have hands on experience with machine learning just don't get how limited the field is right now, and what special conditions are needed to get good results. Similarly, descriptions of machine learning techniques like ANNs as being inspired by actual nervous systems seems to ignore 1) that they are linear combinations of transfer functions (rather than simulated neurons) and 2) even viewed as simplified simulations, ANNs carry the very strong assumption that nothing happening inside a neuron is of any importance.
the opinion of people like Stephen Hawking, Bill Gates and Elon Musk
I disagree with the premise, that fame is more important than domain-specific expertise.
One man, Harry Daghlian, working alone at night, let slip one cube too many, frantically grabbed at the mound to halt the chain reaction, saw the shimmering blue aura of ionization in the air, and died two weeks later of radiation poisoning. Later Louis Slotin used a screwdriver to prop up a radioactive block and lost his life when the screwdriver slipped. Like so many of these worldly scientists he had performed a faulty kind of risk assessment, unconsciously mis-multiplying a low probability of accident (one in a hundred? one in twenty?) by a high cost (nearly infinite).
(Emphasis mine). That quote is from Genius, James Gleick's biography of Richard Feynman - the author of TFA apparently makes the same mistake.
do we have the technology to alter the course of a mile wide asteroid in 2 months?
My guess is we'd soon find out if ICBMs work.
zero times infinity is zero, but a number very slightly greater than zero times infinity is...infinity.
How about Obelix?
Well there is bound to be some shit floating around any gravity well. As long as we are talking spherical cows in a vacuum just make the massless parachute larger? How do you actually define "atmosphere" anyways?
Say we want to land on Mars. The atmosphere is 100 times less dense than on Earth, so we'll need 100 times more parachute. I'm sure that's still a lot less mass and complexity than the balancing rockets...but what if we want to make a round trip, and need to land on Mars and Earth?
We're going to need parallel parachute systems of different sizes, or a parachute that can be re-packaged combined with variable unfurling, and suddenly the parachute option is looking a lot more complicated. On the other hand, if you get the balancing rockets to work, you can use them anywhere, over and over again.
*there
Maybe they want the system to work whether or not their is an atmosphere.
I see you've decided to respond with the predicted "it's not tangible, so therefore it doesn't count" excuse.
That's simply false - as several posts below have noted.
This completely ignores the fact that things do not have to be tangible to be considered to have a measurable value.
No, it doesn't, and further you are conflating scarcity and tangibility. Regardless of the measurable value, or tangibility, of a thing, it must be scarce to be stolen.
Referring to copyright violations as "theft" is nothing more than a lobbying tactic based on perverting our language into a simplified form in which nuance cannot be expressed, but I'm not surprised you've fallen for it given your demonstrated lack of legal and linguistic knowledge.
I'll have to give you that one.
Unfortunately for your position, "that one" is the entire argument. Our civilization possesses codified laws in which the words used to define the laws actually matter.
In the laws that we actually have, as opposed to the ones you wish we had, the taking of scarce goods is theft, while making unauthorized reproductions of copyrighted material is copyright violation.
+1 Contains economics.