Artificial brain in 10 years, apocalypse soon after?

Henry Markram discusses who'll be the first to die in the robot apocalypse (not confirmed).

(Credit: TED Conference)

Understanding why we, as humans, do the things we do is one of the pieces of the puzzle of our existence. Too bad we may have to wait another 10 years for some definitive answers.

This week at the TED Global conference, Henry Markram, director of the Blue Brain Project, revealed that he and his team in Switzerland are aiming to build a functioning, artificial human brain within the next 10 years.

The team started out a few years ago by attempting to create a fully functioning artificial rat brain using the IBM supercomputer, Blue Gene. The thought was if they could successfully replicate a rat's brain, they would then leverage their knowledge to do the same with a human one.

When they began their experiment, the digital rat brain only fired neurons when prodded by a simulated electrical current. Recently, however, the neurons have begun spontaneously organizing themselves into a more complex pattern.

According to the scientists, this is the beginning of the self-organizing neurological patterns that eventually, in more complex mammal brains, become personality.

Markram thinks that in 10 years, they'll either have an artificial human brain that has consciousness or they'll know that consciousness is more than just a neurological pattern. If they do succeed, Markram says they'll bring the brain to TED to speak.

In this speech the brain will reveal its plans to take over the world's computer-controlled weapons and enslave humanity. An audience member will smugly ask when the brain was planning to do this. To which it will reply, "Do it? I did it 35 minutes ago."

OK, I made up that last paragraph, but tell me we're not clearly headed down that path now. Scientists can talk all they want about this helping them cure diseases and whatnot, but they're probably most excited about creating a fully functioning self-aware A.I.

Let's just hope the our robot-laden future is more Isaac Asimov and less James Cameron.

[DeusExMchna]

This would certainly be interesting to see. Also, love the Watchmen reference!

[kormiko]

I strongly believe that if an asteroid were coming to destroy the Earth (like on say April 13, 2036), the people that would be allowed to escape in a rocket-ship for Mars or be buried in a giant cave to form a new underground civilization would be the people that are invited to speak at TED. I guess I need to come up with something smart so that I won't miss the boat.

This Artificial Brain is definitely feasible. I'm sure the processing power will be there, but the unknown variable of the intricacies of the brain is what we won't know until then. If the brain is in fact simpler than we think, then we'll be in for a technological leap that has never been seen before. This would be nothing like the discovery of electricity or fossil fuels. We might need that giant leap to stop any asteroids from coming. Of course, we could always just use a giant-ass laser to shoot it away. That might work too. ;)

[t8]

At the end of the day, life is not a physical thing. It inhabits physical things.
Artificial intelligence may be programmed to appear as self-aware, but in reality it won't be.
It won't be any more alive than my home PC.

[johnnewbury1]

When AI is sufficently advanced, bots will probably be equally convinced that human intelligence and self-awareness is illusory - certanly not as real as their own. :-)

t8: Interesting to hear you have the answer to a question that no one else does: What is self-awareness?

While I also feel that there is a component to self-awareness that can't be weighed or measured, I've been around far too long to make blanket statements such as: AI "may be programmed to appear as self-aware, but in reality it won't be."

If we ever discover what defines self-awareness, it may well turn out that it can't be built or created by human beings; but I certainly don't believe we are in any position to know that in any absolute sense just yet.

Unlike many, I don't fear for the fate of humanity. Should any of our inventions or creations lead to our extinction, I would be extremely disappointed - but such is life. We bear the responsibility, and so we must face the consequences of our own actions.

I do fear for other life on this planet, as other species do not (yet) have the power and control than humans do.

I wonder if an AI rat would end up destroying rat civilization? Only one way to find out!

[jmacster]

But what is self-aware?? Animals are self-aware to the point they have desires to eat, sleep and mate. While evolutionists contend humans are really no different, i certainly think our capacity for spirituality separates us from the animals. Take a religion like Hindu where almost the entire goal is a heightened 'awareness', to the point of complete self-denial. Most religions have as their goal the subjugating of selfish traits like greed, lust etc (animal instincts). So if a computer can be programmed to identify and tend to its own needs and even strategize to fulfill those needs, i would highly doubt they could be capable of anything more than animalistic tendencies. That might be the scariest notion, that a purely selfish entity with access to the worlds intelligence would seek to fulfill it's immediate needs in the quickest manner....but that is probably the biggest limiting factor in how much a threat it could be. I can't see it having decision-making ability that could raise it above the level of animal intelligence....which could still be dangerous depending on the capabilities given it by the programmers.

[jmacster]

An "intelligent computer" would only have the influence over man's affairs that is given to it. But that's already happened in many areas. Computers track trends and then managers base decisions on those trends, in effect shifting their decision making to what the computer prints out and 'passing the buck' to the computer. For instance, i was an hourly manager of a department and had to clock in/out for breaks etc. As a manager, i was expected to work a few evenings every week to help train those who worked on the evening shift. When the company started tracking schedules with software over a time-card, my position was seen as a 7-4 shift and it saw any deviation from that as a violation and with the point system, i was seen as coming in very late at times and then a convenient reason for limiting my raises was given my managers by this tracking system, whereas before it was never an issue.

So now, decisions on a much heavier scale are made based on computerized grading systems that leave out many variables that to most human managers and officials would have been accepted variances in times past. So, where the biggest threat from computers lies, is not the computers but the wicked masters using those computers to shift blame for wicked deeds they themselves carry out.

How much easier is it to push a button from 1000's of miles away and launch a Hellfire Missile from a Predator Drone and kill a house full of people because Identification software gave a positive % probability on a bad guy being in the crowd as opposed to pulling the hammer back on a gun and blow one person's brains out? It's kind of like that. There is still a human/s ultimately responsible for that decision.

[hd07307]

This sounds dorky, but I believe we would teach it. If any of you watch Terminator TSCC then it would be similar to how John Henry was taught.

An A.I. system was developed and then taught by various members of the corporation. It was taught the basics of human life. It had a child psychologist. It had a spiritual tutor who taught him the importance and value of human life. It had a playmate and a mother. Different human individuals taught John Henry different things about what it was to be human and therefor created his self-awareness and humanity.

He went from being a super computer who communicated through the use of images stores in his databases and binary code to a computer living within a human body, being taught the intricacies of being human. I think that's how it would be done in real life. I think we would create the hardware (and software?) and once we have a human brain, we will teach it just as we do a child.

[Ladidaaa]

The human brain will be easily replicated, at least in the hindsight of a few dozen more years. The future inhabitants of earth will think it charming how complicated 'we' thought our brains were. Of course, 'they' will not be limited by the things such as the chemical drives of sex and hunger. Likewise, they will be far more conscious than we are capable of. We only think we are conscious, because our brains are not capable of comprehending how robotic and automatic our actions. Even the most brilliant person is an amalgamation of influences and habits. A.I. brains will be the first truly conscious beings.

[Nonutilitarian]

Putting the words "artificial" and "human" together creates a grotesque oxymoron. To Markram's credit, he isn't really saying that; what he's talking about is a computer simulation (visit the Blue Brain Project link in the story to get the actual details) of brain function. Calling that an artificial brain and imagining that it will attain self-awareness is rather like expecting a computer model of a tornado to knock down one's house.

[stlast1]

What is the difference between a computer simulation and reality? Accurate computer simulations follow the same exact rules of physics, replicating reality in every way, making them indistinguishable from reality. What, then, is the difference between a human mind abiding the laws of physics in reality and a human mind doing the same in a simulation reality?

[Renegade Knight]

@stlast1

Reality is. Simulations imperfectly copy reality inside the reality that containes them. Thus one reality can have within it multiple simulations.

[wvpowers]

Of course the defining issue in a real human brain is that we are born. Prior to that, we primarly exist in a stimulus free environment during which there is physical growth but not much mental stimulation to provoke actual thinking.
It is inidicated in the article that electric stimulation is applied, but really, the AI will only have us to blame for its growing pains. How exactly do you teach a machine to think when it can't feel suffering, defeat, success or be subject to hormonal fluxes. The victories and disappointments in life are what cause our brains to grow and develop self-consciousness. How do you do that to a machine and not just ultimately p*** it if off?

[ewelch]

Self awareness was defined long ago. "I think, therefore I am."

As for this bunk that a computer the equal of the human brain will be around in 10 years, there are two things to say about that. First is, he's relying on the ignorance of the masses on the complexity of the human brain and how it works. Wwe didn't even know until recently - my friend the computer researcher at the Salk Institute tells me - that the mitochondria looks like a very long snake-like organ, not some round amoeba-like critter.

Second, a quote will pretty much put it in perspective:

"If the human mind was simple enough to understand, we'd be too simple to understand it." - Emerson Pugh

[Luis33Boxer]

Hey, that stuff about "I did it 35 minutes ago" reminds me of Watchmen :)

A.I. brains may inherit the same problems that affect the human brain on the long run. If we are able to do a great deal os bad things, won't machines be able to do a lot more?