Slashdot Mirror
The Human Brain Project Kicks Off
Velcroman1 writes "What if you could build a computer that works just like the human brain? You could invent new forms of industrial machinery, create fully autonomous thinking cars, devise new kinds of home appliances. And a new project in Europe hopes to create a computer brain just that powerful in the next ten years — and it's incredibly well-funded. The Human Brain Project kicks off Oct. 7 at a conference in Switzerland. Over the next 10 years, about 80 science institutions and at least 20 government entities in Europe will figure out how to make that computer brain. The project will cost about 1.2 billion euros — or about $1.6B in U.S. dollars. The research hinges on creating a super-powerful computer that's 1,000 times faster than those in use today."
I think that conversion ratio is wrong. $13.57 USD
$1.3 Billion and they forget to install a kill switch.
Edit that original post before someone notices your euro to dollar conversion mistake and the dollar sign when mentioning euros.
I like to think the editors at /. would understand that the $ hasn't just rocketed in value.
Also, this was copied verbatim from the Fox News website. Over-valuing of the $ might be normal there but lets keep it off tech sites.
It will just be replaced by a human in 5 years. They take less power and will work for less than 10 billion euros.
"We have only bits and pieces of information but what we know for certain is that at some point in the early twenty-first century all of mankind was united in celebration. We marveled at our own magnificence as we gave birth to AI.”
RETURN without GOSUB in line 1050
If it works just like a human brain, at what point should it be considered to have the same rights as a human?
Technoli
Ouinnnnnn,
and the "parents" decide that the power bill is too high,
so who gets to kill the new sentient being ?
And who goes to jail ?
Since noone posting is actually visiting the Human Brain Project's website....
The goal of the Human Brain Project, in a nutshell (skullshell?) is to create new neuroscience informatics and modeling software, and new computers powerful enough to run them. This will, in theory, allow "in silico" experiments to test various hypotheses about brain organization, diseases, etc. The proposed "Brain Simulation Platform" supercomputer is just one component of the overall project.
So no...they are not trying to make artificial brains to drive an autonomous car, terminator robot, or flying toaster.
source: https://www.humanbrainproject.eu/documents/10180/17646/Vision+Document/8bb75845-8b1d-41e0-bcb9-d4de69eb6603
I understand that we have far more invested interest in modelling the human brain for medical purposes than any other type of brain. However, if you're going to try to create a model of something vastly complex you should probably start with something easy (and by easy I mean less vastly complex). A short list of neuron amounts in various animals is here, an aplysia(sea slug) or fly brain, I would expect to be a much more reasonable starting point and one with the obvious advantage that you can experiment on, breed whole lines of defective forms to study, just generally have far more control and face no ethical issues with.
Oh and whatever differences may be present in moveing from fly to rat to monkey to human it isn't in the neuron itself those, from what I understand, are almost indistinguishable across species.
This project will not, and I suspect will make no meaningful attempt at, creating a thinking human brain simulation and is really just about better medicine for various mental diseases, which we do sorely need. If it was attempting to take a stab at hard AI "The research hinges on creating a super-powerful computer that's 1,000 times faster than those in use today" is most certainly a false statement: my smartphone is no more creative than the computers of yore that it is 1,000 times faster than.
I suspect they went the thinking machine angle just for the attention... Is it just me or is there a chill in the air?
Only 20% of the cells in the cortex are neurons. We have very little idea what the other cells are doing.
Exactly. I don't even think we quite understand how the brain does what it does enough to build a computer that does what it does. If we really understood how the brain worked, we wouldn't have people battling drug addiction or mental illnesses, because we would be able to fix their problems. Building a computer that operates even close to the capabilities of the human brain doesn't just require a faster computer. It requires algorithms that don't even exist yet. If they could actually build this computer, they would already have a working prototype that worked, but at a slower speed than the human brain.
Anthropic principle: We see the universe the way it is because if it were different we would not be here to see it.
... simulate the complete human brain on supercomputers to better understand how it functions. The end hopes of the HBP include being able to mimic the human brain and being able to better diagnose human brain diseases and mental problems.
The confusion seems to have come from the Fox News article, the author mentions that the computer to simulate the human brain must be much more powerful than we currently have. But it's not supposed to be powerful because it's based on the human brain, it's supposed to be powerful to SIMULATE the brain.
He says a computer brain will consume gigawatts of power, require new forms of memory, and force scientists to look at cutting edge storage techniques. But the immense technical hurdles will be worth the effort. The first phases will help us understand how the brain functions. In later phases, we’ll find out how we learn, how we see and hear, and why the brain sometimes doesn’t process information correctly.
TLDR: they're building a supercomputer to model the human brain, not building a computer modeled on the human brain to be super.
According to this Computerworld article from 2008, a lot of that "steaming pile of 1980s-style AI" is in use every day.
I, for one, am looking forward to the payoff of this new, basic research 30 years from now.
[Sir Garlon] is the marvellest knight that is now living, for he destroyeth many good knights, for he goeth invisible.
It'll take at least 3 decades before their artificial human brain is cost competitive with a human brain.
Except that the billion euros is the development cost, not the unit production cost. The development of the human brain took 4.5 billion years, and the resources of an entire planetary system, although there were some inefficiencies in the process.
The goal (or "vision" as they put it):" ...a global collaborative effort to understand the human brain and its diseases and ultimately to emulate its computational capabilities." This sounds more like a finite element model of the chemistry of the brain, with the main goal of modeling diseases and basic switching functions.
The beauty of an artificial brain is that you don't have to put up with it behaving like an average lazy human. You can torture it until it does your bidding, with no legal repercussions.
But seriously, if you have thinking minds working for you, and you do not allow them self-determination and pay them for their work, you're a slavemaster.
This is not what you want to automate your factory, or run your car. The idea of a slave in my garage is disgusting.
Machine learning goes back to the 1950's, and it has been a part of AI ever since. The techniques used in speech recognition are standard machine learning techniques (hidden Markov models, Gaussian mixtures, neural networks, Bayesian networks). What you call "1980's style AI" may be symbolic, non-probabilistic AI: rule-based systems, inference engines, logic, etc.. And even that is in day-to-day use, in everything from databases to compilers, graphics programs, and games.
(In different words, you have no idea what you're talking about.)
It's a pipe dream. Before you're going to build a computer that works like a human brain you're going to have to figure out how the human brain actually works. Neuroscientists aren't clueless, but they don't have very many clues. The science is in its infancy, and thinking you can replicate something you don't understand is the height of ignorant hubris.
Yes, you can easily program a computer to fool a human into thinking it thinks like a human. Trivially easy, humans are easy to fool. Just ask the Amazing Randi or David Copperfield; that's how IBM's Watson "thinks". Smoke and mirrors. A logic gate has no resemblance whatever to a neuron or axion, and an electronic bit has no analog to serotonin or other brain chemicals.
These folks are fools or charlatans or both.
Free Martian Whores!
Dennett is a populist hack. He is to philosophy what Deepak Chopra is to physics.
Required reading for internet skeptics
Do you have any idea how big a computer that would take? You would have to model every subatomic particle in the entire nervous system.
And tell me, why do physicists need engineers? Thinking that physics is the only key to the human brain is a mistake; I can know everything about how transistors and capacitors and resisters and coils work, and understand the physics behind electricity, but that doesn't mean I can design an amplifier -- I have to know how an amplifier works first.
Free Martian Whores!
A simulation of a brain produces thought like a simulation of an atomic explosion produces radiation. Of course computers will be helpful in understanding how a brain works, but brains are chemical-analog, not binary-electrical. You're not going to produce true thought with a Turing machine. The best you'll get is a simulation.
Free Martian Whores!