And weather is understood pretty well, unlike the brain.
Many would argue that mapping and simulating the human brain is a much easier problem than mapping and simulating the Earth's atmosphere, simply as an issue of scale. There're a lot more atoms in the air than there are in one's head, amusingly enough.
Maybe we won't need store clerks any more but we will need somebody doing something new and unanticipated at this time.
Why? Why would we need somebody doing something new and unanticipated at this time, when a robot could do it for cheaper?
Or are you one of those people that believes that there are jobs that it will never be possible to automate? Do you believe that it will necessarily be humans that build and/or repair robots?
So, why exactly do you believe that things like law, software development, media, etc., can not be automated?
A large portion of the service economy is being automated as we speak. Really, most service jobs where a person stands behind a counter and interacts with a computer or cash register are on the verge of being gone entirely.
Rather than having a room full of accountants entering journal entries by hand on a paper ledger we have one accountant keeping the books in some software and everyone else does something more productive.
There are those among us that would question the productive value of flipping burgers at McDonalds and greeting people as they enter Walmart.
We're talking about automation eliminating jobs. If some people lose their jobs as what they do can be done more cheaply by new tech, why would the jobs created elsewhere go to people? The very premise is that robots can do the job better. Surely the newly created jobs would go to robots as well, no? I mean, what business owner would pay more for human workers when they could pay less for robots?
Or are you one of those people that believes that there are jobs that it will never be possible to automate? Do you believe that it will necessarily be humans that build and/or repair robots?
Whalers, buggy whip manufacturers, and people whos jobs can be trivially replaced by robots are all going to be displaced when technology improves.
So, in an era of ubiquitous automation, when all people's jobs can be trivially replaced by robots, this won't be a problem because... because we'll just do other jobs? Why wouldn't those jobs be done by robots?
Or are you one of those people that believes that there are jobs that it will never be possible to automate? Do you believe that it will necessarily be humans that build and/or repair robots?
I think you vastly underestimate how big it would have to be.
Okay, so let's say I'm off by a factor of 1000. That's roughly 10 iterations of Moore's Law, or 15 years. Or even off by a factor of 1000000, or six orders of magnitude. That's still only a 30 year wait, well within my expected lifetime.
As I said earlier (not too sure who to), modeling a brain would help us understand it, but the model won't think any more than the model of a nuclear explosion produces radiation.
A model of a nuclear explosion does produce radiation, in the model. I'd imagine any simulation of a nuclear explosion that doesn't include radiation in the calculations wouldn't be very useful to anyone, primarily because a nuclear explosion is driven by said radiation.
but we'll never create an electronic computer that can actually think.
Everyone's entitled to their own predictions of what the future will bring, but your argument isn't very convincing. We can simulate a collection of atoms being acted upon by the laws of physics. We can determine the atomic structure of real physical objects. It seems self-evident to me that simulating a human brain is just a matter of scaling up existing technologies. It seems self-evident to me that if Moore's Law (or the Law of Accelerating Returns, the general form of Moore's Law) holds up, simulations of human brains are inevitable. No need to resort to understanding the nature of consciousness or any other philosophical puzzles.
So you're saying that other countries have better outcomes at lower costs because those governments decide to not cover expensive treatments, which is functionally no different than having lifetime caps? How does this account for the better outcomes? Are you suggesting that treatments that are ruled out by lifetime caps don't significantly impact outcomes? Or are you saying that the profit generated by "health insurance" as compared against "health care" sufficiently explains this discrepancy?
Isn't that the point of simulating complex systems? To gain understanding of them? Or are you suggesting that we only simulate systems that are well understood, not to understand them, but just for the fuck of it?
Why do we need theories to deal with that level of complexity? Do the laws of physics break down when there's a lot of particles to simulate? How is this not simply an issue of scaling up computing infrastructure to support simulations of larger data sets?
if you think we know essentially nothing about most of the cells in the brain, I find it odd that IBM has been simulating everything from neocortical columns to cat brains in-house. Perhaps you should drop them a line explaining to them that they don't know what they're doing.
Regarding "experimental data", I don't understand why we'd need anything more than an accurate map of the brain's structure, and an understanding of the laws of physics that govern the interaction between the brain's constituent particles. Perhaps you can explain to me why the laws of physics don't apply to the human brain?
It would take a big computer. Good thing Moore's Law suggests that we'll have a big computer soon enough. Or rather, a small computer that's got sufficient transistor density at a reasonable cost, to anyone itching to make pedantic remarks. In any case, yes, we already simulate the folding of proteins. It would seem that scaling that up would have us simulating brains before long.
Anyway, I don't think anyone's talking about designing a brain. That's why I'm confused by half the comments here, with people complaining that we don't know how the brain works. That's the whole point of running simulations. That's why we run aerodynamic simulations before we do wind-tunnel testing. That's why we run nuclear simulations instead of detonating warheads. When we don't know how a complex system works, we resort to the brute-force method of simulation.
If you believe that there's more to the human brain than the laws of physics acting on a bunch of matter, the onus is on you to demonstrate that. I, personally, don't ascribe any supernatural properties to what seems like a glob of cells.
Why base simulations on experimental data? Wouldn't it be easier to simply map the structure of these glial cells (say, using an atomic force microscope), and then run a simply physics simulation on that collection of atoms?
Isn't the whole point of this project to use simulations instead of experimental data, due to the ethical concerns that stand in the way of getting sufficient experimental data?
A $6,000,000,000,000 program based on the hunch that Saddam had weapons of mass destruction? There's a reason I wish my government would sponsor comparatively affordable research into brains with my tax dollars.
Are you saying that the human brain doesn't abide by the laws of physics? Or that we simply can't figure out what it's made of?
Or are you saying (much like half the other commenters here) that there's some magical voodoo in the brain. That there's some arcane mystery that we'd need to solve before we could simulate the effect of the known laws of physics on the known structure of the brain.
We simulate fly brains, neuron for neuron. We simulate cat brains, neuron for neuron. You suggest that we can't scale this up to human brains because... because... because why, exactly?
Yea, that happened already. Also, I feel like less of a person for linking to Popular Mechanics. I hope you're happy.
We already did the fly brain thing. Apparently now we're at the cat brain scale. Obviously, this isn't the first effort to simulate a brain in silicon. Google is your friend.
Slashdot Mirror
ODB = Ol' Dirty Bastard
OBD = On-Board Diagnostics
In my experience, in addition to the 18% gratuity, there is also a 20% service charge. And of course sales tax.
Make that trillions
Okay, so in 45 years? That's not quite "never".
And weather is understood pretty well, unlike the brain.
Many would argue that mapping and simulating the human brain is a much easier problem than mapping and simulating the Earth's atmosphere, simply as an issue of scale. There're a lot more atoms in the air than there are in one's head, amusingly enough.
To me the suggestion we're living in a simulation serves no other purpose that throwing out something wacky to stump people at parties
... whoa... </neo>
Maybe we won't need store clerks any more but we will need somebody doing something new and unanticipated at this time.
Why? Why would we need somebody doing something new and unanticipated at this time, when a robot could do it for cheaper?
Or are you one of those people that believes that there are jobs that it will never be possible to automate? Do you believe that it will necessarily be humans that build and/or repair robots?
You're overlooking the fact that the jobs that are created can be automated too, so they don't really help with employment.
That's the thing with automation. It kills jobs. Not just some jobs, creating others. All jobs, eventually.
So, why exactly do you believe that things like law, software development, media, etc., can not be automated?
A large portion of the service economy is being automated as we speak. Really, most service jobs where a person stands behind a counter and interacts with a computer or cash register are on the verge of being gone entirely.
Rather than having a room full of accountants entering journal entries by hand on a paper ledger we have one accountant keeping the books in some software and everyone else does something more productive.
There are those among us that would question the productive value of flipping burgers at McDonalds and greeting people as they enter Walmart.
No, even if new industries come along, there's still nothing else for these people to do.
Why would you think new industries won't be subject to the same trend of automation that has been sweeping across all sectors of the economy?
We're talking about automation eliminating jobs. If some people lose their jobs as what they do can be done more cheaply by new tech, why would the jobs created elsewhere go to people? The very premise is that robots can do the job better. Surely the newly created jobs would go to robots as well, no? I mean, what business owner would pay more for human workers when they could pay less for robots?
Or are you one of those people that believes that there are jobs that it will never be possible to automate? Do you believe that it will necessarily be humans that build and/or repair robots?
Whalers, buggy whip manufacturers, and people whos jobs can be trivially replaced by robots are all going to be displaced when technology improves.
So, in an era of ubiquitous automation, when all people's jobs can be trivially replaced by robots, this won't be a problem because... because we'll just do other jobs? Why wouldn't those jobs be done by robots?
Or are you one of those people that believes that there are jobs that it will never be possible to automate? Do you believe that it will necessarily be humans that build and/or repair robots?
So you're saying that Walmart could easily afford to increase their employees' wages to $12/hr without even raising prices on their products?
I think you vastly underestimate how big it would have to be.
Okay, so let's say I'm off by a factor of 1000. That's roughly 10 iterations of Moore's Law, or 15 years. Or even off by a factor of 1000000, or six orders of magnitude. That's still only a 30 year wait, well within my expected lifetime.
As I said earlier (not too sure who to), modeling a brain would help us understand it, but the model won't think any more than the model of a nuclear explosion produces radiation.
A model of a nuclear explosion does produce radiation, in the model. I'd imagine any simulation of a nuclear explosion that doesn't include radiation in the calculations wouldn't be very useful to anyone, primarily because a nuclear explosion is driven by said radiation.
but we'll never create an electronic computer that can actually think.
Everyone's entitled to their own predictions of what the future will bring, but your argument isn't very convincing. We can simulate a collection of atoms being acted upon by the laws of physics. We can determine the atomic structure of real physical objects. It seems self-evident to me that simulating a human brain is just a matter of scaling up existing technologies. It seems self-evident to me that if Moore's Law (or the Law of Accelerating Returns, the general form of Moore's Law) holds up, simulations of human brains are inevitable. No need to resort to understanding the nature of consciousness or any other philosophical puzzles.
So you're saying that other countries have better outcomes at lower costs because those governments decide to not cover expensive treatments, which is functionally no different than having lifetime caps? How does this account for the better outcomes? Are you suggesting that treatments that are ruled out by lifetime caps don't significantly impact outcomes? Or are you saying that the profit generated by "health insurance" as compared against "health care" sufficiently explains this discrepancy?
Do other developed Western nations' socialized healthcare programs have lifetime caps?
If not, isn't it safe to say that lack of lifetime caps isn't the reason our healthcare is so expensive?
Why not?
In other news, there's a beekeeper on slashdot.
Sometimes I just don't understand...
But one of them costs 0.167% as much as the other.
The expression is penny wise, pound foolish.
Again with the "we don't understand" argument.
Isn't that the point of simulating complex systems? To gain understanding of them? Or are you suggesting that we only simulate systems that are well understood, not to understand them, but just for the fuck of it?
Why do we need theories to deal with that level of complexity? Do the laws of physics break down when there's a lot of particles to simulate? How is this not simply an issue of scaling up computing infrastructure to support simulations of larger data sets?
if you think we know essentially nothing about most of the cells in the brain, I find it odd that IBM has been simulating everything from neocortical columns to cat brains in-house. Perhaps you should drop them a line explaining to them that they don't know what they're doing.
Regarding "experimental data", I don't understand why we'd need anything more than an accurate map of the brain's structure, and an understanding of the laws of physics that govern the interaction between the brain's constituent particles. Perhaps you can explain to me why the laws of physics don't apply to the human brain?
It would take a big computer. Good thing Moore's Law suggests that we'll have a big computer soon enough. Or rather, a small computer that's got sufficient transistor density at a reasonable cost, to anyone itching to make pedantic remarks. In any case, yes, we already simulate the folding of proteins. It would seem that scaling that up would have us simulating brains before long.
Anyway, I don't think anyone's talking about designing a brain. That's why I'm confused by half the comments here, with people complaining that we don't know how the brain works. That's the whole point of running simulations. That's why we run aerodynamic simulations before we do wind-tunnel testing. That's why we run nuclear simulations instead of detonating warheads. When we don't know how a complex system works, we resort to the brute-force method of simulation.
If you believe that there's more to the human brain than the laws of physics acting on a bunch of matter, the onus is on you to demonstrate that. I, personally, don't ascribe any supernatural properties to what seems like a glob of cells.
Why base simulations on experimental data? Wouldn't it be easier to simply map the structure of these glial cells (say, using an atomic force microscope), and then run a simply physics simulation on that collection of atoms?
Isn't the whole point of this project to use simulations instead of experimental data, due to the ethical concerns that stand in the way of getting sufficient experimental data?
Actually, scientists understand how the brain works very well.
It's made of matter.
The known laws of physics act on that matter.
The end. Fire up the simulation.
A $6,000,000,000,000 program based on the hunch that Saddam had weapons of mass destruction? There's a reason I wish my government would sponsor comparatively affordable research into brains with my tax dollars.
Are you saying that the human brain doesn't abide by the laws of physics? Or that we simply can't figure out what it's made of?
Or are you saying (much like half the other commenters here) that there's some magical voodoo in the brain. That there's some arcane mystery that we'd need to solve before we could simulate the effect of the known laws of physics on the known structure of the brain.
We simulate fly brains, neuron for neuron. We simulate cat brains, neuron for neuron. You suggest that we can't scale this up to human brains because... because... because why, exactly?
Yea, that happened already. Also, I feel like less of a person for linking to Popular Mechanics. I hope you're happy.
We already did the fly brain thing. Apparently now we're at the cat brain scale. Obviously, this isn't the first effort to simulate a brain in silicon. Google is your friend.