Slashdot Mirror
Why We Should Build a Supercomputer Replica of the Human Brain
An anonymous reader sends this excerpt from Wired:
"[Henry] Markram was proposing a project that has bedeviled AI researchers for decades, that most had presumed was impossible. He wanted to build a working mind from the ground up. ... The self-assured scientist claims that the only thing preventing scientists from understanding the human brain in its entirety — from the molecular level all the way to the mystery of consciousness — is a lack of ambition. If only neuroscience would follow his lead, he insists, his Human Brain Project could simulate the functions of all 86 billion neurons in the human brain, and the 100 trillion connections that link them. And once that's done, once you've built a plug-and-play brain, anything is possible. You could take it apart to figure out the causes of brain diseases. You could rig it to robotics and develop a whole new range of intelligent technologies. You could strap on a pair of virtual reality glasses and experience a brain other than your own."
Comment removed based on user account deletion
Simulating how the neurons and connections function won't be enough. You also need an initial state for each of them. Get even a tiny precentage of them wrong, and the result would probably be a virtual seizure.
sudo cat /dev/me > /dev/you
You are not in the sudoers file. This incident will be reported to God.
I doubt Kim Jong Un would volunteer to help the project anyway.
It doesn't need to be a mirror image, but it needs to "develop" in the same manner.
The brain is plastic.
For large sets, this will be our guide even unto death, for the LORD will work for each type of data it is applied to...
This is why I regularly thank my toaster.
An enigma, wrapped in a riddle, shrouded in bacon and cheese
I think the stake holders need to think about that simple question. The last thing we need is some sentient silicon running around like a pestilent child lobbing nukes between hemispheres for fun.
Pestilent children are the worst, with all their plagues and their boils and their oozing pustules.
Sentience? I bet it won't be capable of meaningful phrases!
C'mon. You can model every circuit in the brain - and assuming it's really just like a big, deterministic watch works, you could still get a Jerry Falwell or Ryan Seacrest instead of a sentient being.
"Flyin' in just a sweet place,
Never been known to fail..."
Robots will be so good at complex tasks that they will find it overkill to use one for simple tasks. They'll simply say, why waste a robot on this task when we have all of these stupid humans who are willing to do it for basically nothing. Half the quality at an eighth the price. Can't beat that.
We've long established that the source of the human "soul" is in the brain. Those interconnections give rise to consciousness and self-awareness -- and sentience. If you build something that precisely models the brain, you will be creating sentience. I have to question how we can create a sentient creature simply to experiment upon it and still claim to have a shred of humanity to us.
I know that this is not as dazzling and interesting as building the device to geeks like us, but we cannot simply ignore the ethical consequences of our actions. All vocations, all manner of human endeavor, must move forward with an eye towards a respect for life. This may not be human life we're creating, or even organic life, but it is no less deserving.
Someday we're going to have cybernetic life walking about. And I have to wonder -- how well will they treat us, when they find out how ethical we were in creating it?
#fuckbeta #iamslashdot #dicemustdie
The brain "develops" in humans for a very long time though, to work around /with that the mechanical brain would either need to be able to develop itself or start off in an adult state.
I have my doubts about the success of this project, but we've got to start somewhere & we'd learn a lot with this project, not like we don't spend our country's money on wars, or policing / giving aid to people who hate us instead.
Say this actually works. We create a brain and start down the long path of "teaching" it just like with new-born humans.
What happens when we detect that the brain is "experiencing pain" (we already know that pain has a detectable neurological basis right?)
What happens when we detect the brain is experiencing depression?
What are our responsibilities then? Is this thing a human, a lab-rat, or a machine?
What exactly are "the functions of all 86 billion neurons"? I sense massive oversimplification here. Neurons have lots and lots of functions we have no idea how to simulate exactly, such as all the details of the thousands of networked internal metabolic mechanisms of any large mammalian cell, which most neural network simulations simply neglect.
Furthermore, we have plenty of evidence that the non-neuronal components of the brain (glia and oligodendroglia) massively influence brain functioning, and may be required for adequate cognition. Furthermore we have no way of knowing if a brain-in-a-vat will work the way a brain in the body, with all its connections, works. The above issues are just a start to the limitations of the scheme.
To put it in perspective, that 86 billion neurons would be 86 "giga-neurons"; huh, conceptually not too overwhelming. Then we have the 100 trillion connections between them, or 100 "tera-connections"? Forget it.
Not to even mention (as someone already did) the initial state, then the learning process. To even form this structure in RAM would require, what? 40-50 more Moores Law iterations? Which I doubt is even physically possible.
I think this is the wrong approach, and even if possible, not in our lifetimes....
Global Thermonuclear Chess? I'm in!
As a neuroscientist, this seems absurd. Not all neurons perform the same functions, some are very different in terms of structure and connections (pyramidal cell vs interneuron for example). We don't have a good sense for all the multitude of ways they can connect (via axon projections, or through retrograde signals at a given synapse). And we're just starting to appreciate the role that non neuron brain cells play in cellular communication - astrocytes release signaling molecules that modulate neuronal function (caffeine interferes with these) and they also regulate the amount of ions around neurons - in essence they enable neurons to change states.
i don't know karate, but i know ca-razy
How do you know you're not trapped inside a computer right now?
2 reasons:
1 - no respawn
2 - my cat won't respond to regular expressions
An enigma, wrapped in a riddle, shrouded in bacon and cheese
If it's on the internet, maybe it will post to Slashdot as an A/C.
you insensitive clod!
some of us simulations have registered accounts!
... The self-assured scientist claims that the only thing preventing scientists from understanding the human brain in its entirety — from the molecular level all the way to the mystery of consciousness — is a lack of ambition.
This.
Also, the lack of any sort of a roadmap as to how to do this.
Also, the lack of any sort of definition for "consciousness", or any indication that it is an emergent property, or any way to measure when you've succeeded in making consciousness, or any theoretical evidence at all that it would arise from any specific plan.
We could model as many neurons as we like and it *still* wouldn't be a human brain unless we figure out how those neurons connect with each other. With no detailed plan, it's like trying to build a house by tacking boards together.
The "self-assured scientist" could start by telling us how a Cortical Column is wired up, how the feedback and feed-forward between columns works, and why artificial neural nets have inputs on one side and outputs on the other, when the brain apparently has both inputs and outputs on one side (in the sense of a functional diagram; ie - the efferent and afferent neurons connect to the same level of layer), and what the distinction is between these models.
If he can't solve basic issues, how can he hope to succeed in such a complex and ambitions project?
Indeed, this seems to be something these sorts of projects forever overlook - the point. If you create a conscious model of the human brain, then you have all the same ethical problems experimenting on it as you would on an actual human, all you've done is drastically increase the potential benefits of doing so, and I for one do not particularly want to live in a world where it's accepted that you can experiment on someone's brain just because "the benefits are worth it".
You could possibly learn something new by just being able to watch it in action in excrutiating detail, but all the parts at least are only going to work in the manner you programmed them to, so really it comes down to a test case to see if our understanding of the component mechnisms of the brain has captured the "secret sauce" of consciousness. Even that though has major ethical considerations - it's unlikely to work right the first time, and all the intermediate attempts are rather analogous to intentionally creating children with severe brain damage.
And that's not to mention the fact that we may well need completely new technology to simulate a brain effectively - all existing computers are clocked, and any simulation is going to by necessity work in discrete time slices, which is completely unlike the totally asynchronous, continuous operation of an organic brain. Even if we can somehow manage the simulation by, for example, using extremely fine time slices and running it at a tiny fraction of real-time, it will still likely require several orders of magnitude more processing power than the human brain itself possesses. I mean the architectual differences mean it took a decent Pentium-class machine in order to be able to simulate an ancient AtariXT in real time, and those two systems are practically identical compared to a brain.
--- Most topics have many sides worth arguing, allow me to take one opposite you.
From the article:
"There are too many things we don't yet know," says Caltech professor Christof Koch, chief scientific officer at one of neuroscience's biggest data producers, the Allen Institute for Brain Science in Seattle. "The roundworm has exactly 302 neurons, and we still have no frigging idea how this animal works."
That's the problem. Just because we can extract the wiring diagram doesn't mean the components are well understood yet. Also, if we understood the components and how to wire them up, it would be cheaper to just build hardware. Simulating neurons is slow. It's like running SPICE instead of building circuits. Works, but there's about a 1000x or worse speed, power, and cost penalty. GPUs are often simulated at the gate level before making an IC; NVidia uses twenty or thirty racks of servers to simulate one GPU during development.
What bothers me about claims of strong AI is that I've heard it before. Ed Feigenbaum, the "expert systems" guy at Stanford, was running around in the 1980s, promising Strong AI Real Soon Now if only he could funding for a giant national AI lab headed by him. He even testified before Congress on that. Expert systems were a dead end.
Rod Brooks from MIT went down this road too. His COG project had a robotic head and some arms, some facial expressions, and a lot of hype. Work ceased on that embarrassment in 2003. He'd done good artificial insect work, but the jump to human level was way too big.
This is the hubris problem in AI. Too many people have approached this claiming their One Big Idea would lead to strong AI. So far, not even close.
All the mammals have similar DNA and brain architecture. A mouse brain is about 1g; a human brain is about 1000g. So build a simulated mouse brain and demonstrate it works, or STFU.
And naturally a petulant pestilent child is that much worse.
1 - Strict mode.
2 - Cats don't seem to respond predictably to any input. I posit they're our perceptual interpretation of random number generators.
Blank until
The last thing we need is some sentient silicon running around like a pestilent child lobbing nukes between hemispheres for fun.
If scientists persist in trying to play God with projects like this, they are going to unleash the Four Horsemen of the Apocalypse:
War, Famine, Death, and Petulance.
-
- - You can't take something off the Internet! That's like trying to take pee out of a swimming pool.
$ csh
$ Why rub two sticks together?
Why: No match.
Truth is stranger than fiction, but it is because Fiction is obliged to stick to possibilities; Truth isn't. Mark Twain.