Canadian high-tech startup Intelligence Realm is constructing a distributed virtual brain, one computer at a time. Utilizing a computational model we’ve seen in such projects as SETI@Home, the system will harness the computing power of thousands of machines throughout the world to achieve its goal.

Is this a different kind of intelligence?

No. Intelligence is implemented in the same way across all biological species, from insects to humans. This system will be no different in this respect; it will use the same information processing mechanisms found in all species. Understanding the mechanisms of intelligence allows us to control and increase machine knowledge and understanding beyond human levels. Nonetheless if a person will have the same information in one field as a machine, that person will be able to reach similar conclusions with the machine.

Can we implement consciousness in a computer?

Yes. Consciousness is a state of mind, inside the brain. By simulating on a computer the same brain processes that raise consciousness we will obtain the same result, that is a conscious machine.

Is this a distributed computing project like Seti@Home or Folding@Home?

Yes, but it has some notable differences, mainly that it will not be used for intensive computations only but also as a large neuronal simulator. ‘

What is the future system capacity?

We intend to build a system that has the brain capacity of all people living in the world. This way we can ensure that no one will be left behind. Its capacity will go far beyond the petabyte scale.

If someone would have the program source code would they be able to find out how knowledge is represented in the brain?

Probably not. They won't be able to understand the knowledge representation mechanisms even if they have the entire source code. They will be in the same situtation that neuroscientists are now, where they have access to a lot of biological information, yet they are unable to understand how knowledge is represented in the brain. The fact that the software program is a neural simulator doesn't change the nature of the problem. It is only translated from one system (i.e. the brain) to another (i.e. the computer); having a computer that simulates brain processes doesn't solve this problem.

Can I donate time for this project while running in parallel another distributed computing project?

Yes. You can either run the program as a standalone application or as a screensaver. It is your choice. The only interaction with other projects is in the amount of computing power allocated for each of the projects.

How you can get Involved

Question

Overall this is a very ambitious project. Some say this is achievable within eight years, while others think that is wildly optimistic. What do you think?

Comments

Search

James Richardson  - Distributed is the way to go   |70.190.238.xxx |2009-05-30 14:50:57 I think they are on the right track relying on a distributed system instead of a giant centralized supercomputer. Additionally, they are getting FREE computing resources! Although 8 years does seem a bit aggressive, I think this project has the potential to do some really interesting things in that time period.

Jason Chaotic  - lol   |24.148.86.xxx |2009-05-31 09:08:52 8 years? yeah right, thats what AI peeps said 8 years ago already. Get real, stop watching too many startrek reruns while smoking cheap weed...try building skynet within 25 year timeframe and I may believe ya.

Frank Eriksson  - Does I get paid?   |81.228.155.xxx |2009-06-02 08:56:28 If I could do that, does you pay me for 25 years of work? ( 50% now, 50% afterwards. Rigth? ) Scientist may have overestimated, and sometime they have underestimated the power of Autonomus Machine Learning. I belive that we could have more complex systems already today ( I'm working on tomorrows solution already ). But the increasing processing power of computers are rather stable ( doubled each 18th month, see Moore's law ) wich means that a standard computer will be 278 times higher in 25 years! And using techicues as BOINC/GirdComputing we can achive extremly high computionrates by using quantity over [quality]. So yes, SkyNet is very posible in few years. But SkyNet is a system that strives for an other goal than it's creators intended for (by poorly thoughtout goal defenition). So the system is faulty. So, I'm just pondering do you want a SkyNet as it became in the terminator movies, or do you want one that is "correct".

Jeff  - Math and Minds   |66.68.64.xxx |2009-05-31 23:26:52 "Currently there are over 6000 individuals contributing the processing power of 14,000 computers for a total of 791 billion neurons. If you’d like increase the neuron count by 500,000 by adding your machine to the virtual mind:" If each of 14,000 computers are each simulating 500,000 neurons, how does that get you to 791 billion neurons? Also --- the various neural structures of the human brain don't have the characteristic disparities of latency that you'll see between e.g. neurons running on the same host vs. on different hosts, much less the kind of batch-oriented processing typical of BOINC. How do you account for this sort of thing? Finally, assuming a Hawkins-like hypothesis that the neocortex is all that matters, 14x10^9 neurons might be sufficient for human equivalency. Even at 14k * 500k = 7B, you're halfway there. But I'm assuming that the topology of your simulation isn't particularly based on e.g. a neocortical structure, but rather is more likely a standard "neural network" with some number of hidden nodes / layers, etc. What is the topology, and how / why did you choose it?

Robert Riley  - re: Math and Minds   |70.190.238.xxx |2009-06-01 00:16:02 Jeff wrote: If each of 14,000 computers are each simulating 500,000 neurons, how does that get you to 791 billion neurons? Good catch. Based on the totals Intelligence Realm is reporting along with wording in their results section my interpretation is: If a particular machine processes a work unit modelling 500,0000 neurons, finishes the unit and then starts on another unit modelling another 500,000 neurons they'll count the total number of neurons as 1,000,000. This does seem a bit odd considering at any one time you only had 500,000 neurons in operation.

Matt  - Isn't some sort of sensory input required?   |162.39.213.xxx |2009-06-01 00:45:37 Whats the point of running a neural net simulation when the "brain" has no visual or other sensory input to help it develop, how would a child develop awareness if it had no vision, hearing, touch, taste, etc. were building a Helen Keller -1 bot

anon   |68.109.174.xxx |2009-06-03 20:57:20 this isnt being used to create a sentient being. its used for science. no other simple way to put it.

John  - Digital computers cant be conscious   |86.4.194.xxx |2009-06-11 23:31:37 Digital computers might perform intelligent actions but they can't become conscious. They have defined states at the end of each clock pulse so believing a digital computer can be conscious is like believing a photo is conscious. One day we may have conscious machines but they won't be digital computers! See Materialists should read this first.

Only registered users can write comments!

3.26 Copyright (C) 2008 Compojoom.com / Copyright (C) 2007 Alain Georgette / Copyright (C) 2006 Frantisek Hliva. All rights reserved."