I grew up in an anti-intellectual culture, and was persecuted for "trying to be white" when I focused on academics, proper english, and polite behavior.
I made it out. Many others I knew didn't.
If you can't understand how "urban" culture, with their thugs, gangstas, misogyny, violence and victimhood mentality, cause massive problems for those stuck in it, either you've never been there, or you're part of the problem.
If the wrong skin colors are coming into Google, look towards the *cultures* of the people who don't make it, rather than the skin color. Backwards urban cultures (where sadly most self-identified blacks and latinos live), are anti-intellectual and actively discourage those who try to make it out through education by shaming them as not being "real".
So, the question is, should Google be in charge of destroying thug gansgsta culture, and forcing urban youth to speak proper english, work hard in school, treat women with respect, and avoid violent destructive behavior?
As for men/women, they've got different brains, so you'll get different outcomes. There is no shame in being a man with less empathy than a woman, and no shame in being a woman with more empathy than a man.
Even beyond any possible quantum effects required, the fact of the matter is that you're challenged by the pure mathematics and scale of it.
1) simulation of individual molecules requires many orders of magnitude *more* molecules
2) while deterministic, complex stochastic systems are not computationally tractable
If you count artificial insemination as creating an "artificial brain", great, we've done it. But to assert that we can *computationally* reengineer an artificial brain is to misunderstand just how complex and poorly understood the brain is.
As for other natural phenomenon we've failed to reengineer or model, just take weather - while we've gotten upwards of maybe 5 days of fair local accuracy, nobody in their right mind assumes that the modeling of weather can be arbitrarily extended much further.
No offense, but, bullshit. Anything that supposedly "learns" in an algorithmic sense is *programmed* to "learn". It doesn't learn "just like a person", it may perform some bayesian filtering and a random number generator that makes its product seem non-deterministic (in cases of differing random seeds), but all the buzzwords in the world doesn't change the fact that at its core, it's an algorithmic computation that has been defined, at its core, by machine code instructions.
While certainly we exist, and it's very true that the laws of physics are deterministic, the problem we face is that it very well may be that we need to simulate things down to the molecular level to truly simulate intelligence. The hardware required to properly model a single molecule, of course, will be many, many, many times larger than the molecule being simulated. You can certainly grow natural neurons through artificial insemination and the eventual production of an "artificial" human being, but the idea that we can etch artificial neurons and create intelligence assumes that all we need are neurons - and that ignores the myelin, hormones, fluids, etc, etc, that interact, moderate, and importantly effect the neuron. By the time you get down to etching artificial myelin, hormones, fluid, etc, etc, you get to the "need millions of molecules to simulate a single molecule".
In the end, not only is it possible, but it is most probable that you cannot ignore the complexity at the molecular level that contributes to human level intelligence.
Behavior at the molecular level significantly affects the ultimate behavior of neurons. Systems can only be modeled at higher levels if the lower levels are not greatly impactful on outcome - however, this is clearly not the case with neurons.
I suggest the book "The Talent Code" which goes into significant depth on just how impactful myelin is. Here's a good cite from the book:
"For a good overview of what might soon be called the myelin revolution, see R. Douglas Fields's "White Matter Matters," Scientific American (March 2008), 54-61, as well as his "Myelination: An Overlooked Mechanism of Synaptic Plasticity?" Neuroscientist 11, no. 6 (2005), 528-31."
Now, whether or not it's possible to accurately model myelin activity, we *know* there are significant, impactful consequences from myelin on neurons.
Spacecraft trajectory calculations are of such a massively different *type* of calculation, it's hardly relevant to compare it to the modeling of biological intelligence. Yes, we can make models of simple things - nobody denies that. What is denied is that intelligence is a "simple" thing.
Single neurons, in fact, are *not* properly modeled when they ignore things like the activity of the myelin sheath (a cellular process, outside of the neuron, that dramatically affects how the neuron behaves). If a *massively* important component (myelin), cannot be properly modeled, how can we expect our simulated neuron to actually work the way a real one does?
We *know* that neurons are affected by the biological environment in which they reside, filled with fluids, hormones, cells, and all sorts of other factors that directly and *significantly* affect their behavior. It may very well be that in order to simulate these interactions with the cell membrane of the neuron, you'll need to get down to the molecular level.
I've heard the buzzwords change since the 70s, but every current approach I've ever seen, at its core, still asserts an algorithmic approach. Calling it a "neural network", when you're still doing 1s and 0s under the covers with the same machine code operations every other algorithm based approach uses, doesn't make it non-algorithmic.
I'm not sure if physics suggests that all physical processes are computable - especially with the observer effect.
My guess is this - the human brain is not an independent seat of intelligence. The biological matter surrounding all of our neural tissue is required, including all the various cells, fluids, myelin, etc. Until you get to the point where you can simulate that complex set of molecules, you'll never get to the point where you can simulate a human intelligence. And even when you get to that point, you'll be doing it in an incredibly inefficient way, since each modeling of each molecule will, in practice, require more than one molecule of computing device to simulate it.
When someone models a single celled bacteria, 100%, at the molecular level, I'll start believing that eventually it's possible. Thus far, I've heard of no such molecular simulation, much less one that could compute in anything close to real-time.
If something is sufficiently inefficient, given the realities of the physical limits of space and time, it can be practically physically impossible.
It's certainly physically possible to create a 1-to-1 scale map of the milky way galaxy, but it may be impossible to finish before the heat death of the universe.
Forget physics for a moment, let's talk mathematics:
Do you believe that there are some non-computable problems?
If human intelligence is indeed a non-computable problem, then assuming that an algorithmic design will ever be able to compute it is like insisting that the way we'll land on the moon is with a hot air balloon.
Put another way, it's quite possible that biological intelligence is the most efficient way of organizing intelligence, and that any digital simulation of it, even if it went down to the atomic level, would be more wasteful in application.
The "race-based medicine" trope is useless, as you already admitted in your initial comment regarding it *not* be a genetic construct. While you can make some narrow claims about some very narrow populations, those claims *don't* hold along the arbitrary racial lines of self or other identification, and are completely inapplicable to anyone of any significant mixture - which, if truth be told, is most of us. If someone is going to do *genetic* based medicine, great - identify the specific genes of a person before treatment. Asking them if their grandmother was Dutch is useless.
It sounds like you're against "bucketing", but for "bucketing" - I firmly disagree that there is any positive outcome of racial bucketing, *period*.
Why is it a concern? Can't it be possible that men and women make different choices naturally without some sort of nefarious, systematic discrimination against either one?
Should we also start explaining why some people are tall and some people are short? Why some people like roses and others like daffodils? Why some car colors are overrepresented?
Just because a subgroup of the population doesn't have the same proportions of some trait as the general population, doesn't mean that something evil has happened.
So, get rid of the label so that some bigot doesn't look at my paperwork and decide "oops, wrong race!"
Look, you've got two versions of racial identification - self identification, and other identification. The two don't match, and aren't even *consistent within themselves*!
Pretending that somehow we'll stop bigotry by labeling, categorizing, and dividing people into imaginary, socially constructed, arbitrary, contradictory and malleable buckets is silly.
Actually, the ones with the most to complain about in this whole mess are recent asian immigrants who didn't have a damn thing to do with slavery, but are now being quota-ed out of universities because they work hard, and end up over-represented in high test scores.
It's funny how the collateral damage of the victimhood mentality that demonizes whites ends up affecting a racial group that was legally discriminated against all the way up till 1952 (http://en.wikipedia.org/wiki/Immigration_and_Nationality_Act_of_1952)
Funny how if you're a dark Okinawan with good test scores, you get shafted by some light skinned mulatto because they want to stick it to some lily white guy in a picture from 150 years ago.
I actually fought for both of my daughters to be listed as "human" - but was told that it wasn't allowed because it was facetious.
I was never more sincere about something in my life.
In the end, I chose "decline to state", because multi-racial was limited to only 3 choices maximum - as if there was some upper limit on how many different countries your ancestors could come from.
Should hospitals and schools have to explain how the vast majority of their nurses and teachers are women?
Statistically significant differences between any given group, and the general population, shouldn't require some special explanation - that's guilty until proven innocent.
Frankly, if you have a workforce of 10,000 white people, no matter how they self-identify, I'll almost guarantee you that a significant portion of them have black ancestry. Our most recent common ancestor is quite possibly only a few thousand years ago (http://www.stat.yale.edu/~jtc5/papers/CommonAncestors/NatureAncestorsPressRelease.html) - to place some company under strict scrutiny because their employee base didn't self identify in the same proportions as the general population is silly when you realize that we're *all* simply cousins of one degree or another.
Slashdot Mirror
I grew up in an anti-intellectual culture, and was persecuted for "trying to be white" when I focused on academics, proper english, and polite behavior.
I made it out. Many others I knew didn't.
If you can't understand how "urban" culture, with their thugs, gangstas, misogyny, violence and victimhood mentality, cause massive problems for those stuck in it, either you've never been there, or you're part of the problem.
...culture does.
If the wrong skin colors are coming into Google, look towards the *cultures* of the people who don't make it, rather than the skin color. Backwards urban cultures (where sadly most self-identified blacks and latinos live), are anti-intellectual and actively discourage those who try to make it out through education by shaming them as not being "real".
So, the question is, should Google be in charge of destroying thug gansgsta culture, and forcing urban youth to speak proper english, work hard in school, treat women with respect, and avoid violent destructive behavior?
As for men/women, they've got different brains, so you'll get different outcomes. There is no shame in being a man with less empathy than a woman, and no shame in being a woman with more empathy than a man.
Even beyond any possible quantum effects required, the fact of the matter is that you're challenged by the pure mathematics and scale of it.
1) simulation of individual molecules requires many orders of magnitude *more* molecules
2) while deterministic, complex stochastic systems are not computationally tractable
If you count artificial insemination as creating an "artificial brain", great, we've done it. But to assert that we can *computationally* reengineer an artificial brain is to misunderstand just how complex and poorly understood the brain is.
As for other natural phenomenon we've failed to reengineer or model, just take weather - while we've gotten upwards of maybe 5 days of fair local accuracy, nobody in their right mind assumes that the modeling of weather can be arbitrarily extended much further.
No offense, but, bullshit. Anything that supposedly "learns" in an algorithmic sense is *programmed* to "learn". It doesn't learn "just like a person", it may perform some bayesian filtering and a random number generator that makes its product seem non-deterministic (in cases of differing random seeds), but all the buzzwords in the world doesn't change the fact that at its core, it's an algorithmic computation that has been defined, at its core, by machine code instructions.
While certainly we exist, and it's very true that the laws of physics are deterministic, the problem we face is that it very well may be that we need to simulate things down to the molecular level to truly simulate intelligence. The hardware required to properly model a single molecule, of course, will be many, many, many times larger than the molecule being simulated. You can certainly grow natural neurons through artificial insemination and the eventual production of an "artificial" human being, but the idea that we can etch artificial neurons and create intelligence assumes that all we need are neurons - and that ignores the myelin, hormones, fluids, etc, etc, that interact, moderate, and importantly effect the neuron. By the time you get down to etching artificial myelin, hormones, fluid, etc, etc, you get to the "need millions of molecules to simulate a single molecule".
In the end, not only is it possible, but it is most probable that you cannot ignore the complexity at the molecular level that contributes to human level intelligence.
Behavior at the molecular level significantly affects the ultimate behavior of neurons. Systems can only be modeled at higher levels if the lower levels are not greatly impactful on outcome - however, this is clearly not the case with neurons.
I suggest the book "The Talent Code" which goes into significant depth on just how impactful myelin is. Here's a good cite from the book:
"For a good overview of what might soon be called the myelin revolution, see R. Douglas Fields's "White Matter Matters," Scientific American (March 2008), 54-61, as well as his "Myelination: An Overlooked Mechanism of Synaptic Plasticity?" Neuroscientist 11, no. 6 (2005), 528-31."
Now, whether or not it's possible to accurately model myelin activity, we *know* there are significant, impactful consequences from myelin on neurons.
Spacecraft trajectory calculations are of such a massively different *type* of calculation, it's hardly relevant to compare it to the modeling of biological intelligence. Yes, we can make models of simple things - nobody denies that. What is denied is that intelligence is a "simple" thing.
Single neurons, in fact, are *not* properly modeled when they ignore things like the activity of the myelin sheath (a cellular process, outside of the neuron, that dramatically affects how the neuron behaves). If a *massively* important component (myelin), cannot be properly modeled, how can we expect our simulated neuron to actually work the way a real one does?
We *know* that neurons are affected by the biological environment in which they reside, filled with fluids, hormones, cells, and all sorts of other factors that directly and *significantly* affect their behavior. It may very well be that in order to simulate these interactions with the cell membrane of the neuron, you'll need to get down to the molecular level.
Okay, what about 100,000,000,000,000x wasteful? What if the entire planet earth needs to be scarified to digitally replicate your brain?
Is your brain really that important? :)
Okay, so we artificially inseminate a woman, biologically grow an artificial human, and claim victory on AI!
If AI *isn't* biological, then every time we manipulate breeding we're creating "artificial" intelligence, and we're done :)
I've heard the buzzwords change since the 70s, but every current approach I've ever seen, at its core, still asserts an algorithmic approach. Calling it a "neural network", when you're still doing 1s and 0s under the covers with the same machine code operations every other algorithm based approach uses, doesn't make it non-algorithmic.
Why not just run AI on biology? Just artificially inseminate someone, and grow an artificial human intelligence :)
Okay, so let's assert that with artificial insemination, we've created artificial intelligence. Done :)
I'm not sure if physics suggests that all physical processes are computable - especially with the observer effect.
My guess is this - the human brain is not an independent seat of intelligence. The biological matter surrounding all of our neural tissue is required, including all the various cells, fluids, myelin, etc. Until you get to the point where you can simulate that complex set of molecules, you'll never get to the point where you can simulate a human intelligence. And even when you get to that point, you'll be doing it in an incredibly inefficient way, since each modeling of each molecule will, in practice, require more than one molecule of computing device to simulate it.
When someone models a single celled bacteria, 100%, at the molecular level, I'll start believing that eventually it's possible. Thus far, I've heard of no such molecular simulation, much less one that could compute in anything close to real-time.
Something can exist without be computable.
http://en.wikipedia.org/wiki/C...
http://en.wikipedia.org/wiki/H...
If something is sufficiently inefficient, given the realities of the physical limits of space and time, it can be practically physically impossible.
It's certainly physically possible to create a 1-to-1 scale map of the milky way galaxy, but it may be impossible to finish before the heat death of the universe.
Forget physics for a moment, let's talk mathematics:
Do you believe that there are some non-computable problems?
If human intelligence is indeed a non-computable problem, then assuming that an algorithmic design will ever be able to compute it is like insisting that the way we'll land on the moon is with a hot air balloon.
Put another way, it's quite possible that biological intelligence is the most efficient way of organizing intelligence, and that any digital simulation of it, even if it went down to the atomic level, would be more wasteful in application.
The "race-based medicine" trope is useless, as you already admitted in your initial comment regarding it *not* be a genetic construct. While you can make some narrow claims about some very narrow populations, those claims *don't* hold along the arbitrary racial lines of self or other identification, and are completely inapplicable to anyone of any significant mixture - which, if truth be told, is most of us. If someone is going to do *genetic* based medicine, great - identify the specific genes of a person before treatment. Asking them if their grandmother was Dutch is useless.
It sounds like you're against "bucketing", but for "bucketing" - I firmly disagree that there is any positive outcome of racial bucketing, *period*.
So home pornos are now to be considered like engagement rings? :)
Dating sure is complicated in the digital age :)
Maybe your practice of masturbation differs from mine, but it usually involves some significant "dwelling", and certainly happens "all the time".
I suppose some people are better at loving themselves than others :)
Wait, so you're not allowed to masturbate while thinking of someone once you break up with them? It's now immoral to have sexual fantasies?
Does this mean a girl who masturbates while fondling a diamond ring she got from her ex needs to return the diamond ring?
Why is it a concern? Can't it be possible that men and women make different choices naturally without some sort of nefarious, systematic discrimination against either one?
Should we also start explaining why some people are tall and some people are short? Why some people like roses and others like daffodils? Why some car colors are overrepresented?
Just because a subgroup of the population doesn't have the same proportions of some trait as the general population, doesn't mean that something evil has happened.
Not true - a white woman can have a black child (Obama), but a black woman can't have a white child (Angela Howard).
Why do you think that is?
So, get rid of the label so that some bigot doesn't look at my paperwork and decide "oops, wrong race!"
Look, you've got two versions of racial identification - self identification, and other identification. The two don't match, and aren't even *consistent within themselves*!
Pretending that somehow we'll stop bigotry by labeling, categorizing, and dividing people into imaginary, socially constructed, arbitrary, contradictory and malleable buckets is silly.
Actually, the ones with the most to complain about in this whole mess are recent asian immigrants who didn't have a damn thing to do with slavery, but are now being quota-ed out of universities because they work hard, and end up over-represented in high test scores.
It's funny how the collateral damage of the victimhood mentality that demonizes whites ends up affecting a racial group that was legally discriminated against all the way up till 1952 (http://en.wikipedia.org/wiki/Immigration_and_Nationality_Act_of_1952)
Funny how if you're a dark Okinawan with good test scores, you get shafted by some light skinned mulatto because they want to stick it to some lily white guy in a picture from 150 years ago.
I actually fought for both of my daughters to be listed as "human" - but was told that it wasn't allowed because it was facetious.
I was never more sincere about something in my life.
In the end, I chose "decline to state", because multi-racial was limited to only 3 choices maximum - as if there was some upper limit on how many different countries your ancestors could come from.
Should hospitals and schools have to explain how the vast majority of their nurses and teachers are women?
Statistically significant differences between any given group, and the general population, shouldn't require some special explanation - that's guilty until proven innocent.
Frankly, if you have a workforce of 10,000 white people, no matter how they self-identify, I'll almost guarantee you that a significant portion of them have black ancestry. Our most recent common ancestor is quite possibly only a few thousand years ago (http://www.stat.yale.edu/~jtc5/papers/CommonAncestors/NatureAncestorsPressRelease.html) - to place some company under strict scrutiny because their employee base didn't self identify in the same proportions as the general population is silly when you realize that we're *all* simply cousins of one degree or another.