IBM working on a computer with a mind of its own
By John Lister
IBM is teaming with five universities to develop a computer with similar reasoning to the human brain. The ambitious project is getting part of its funding from a $4.9 million government grant.
The scheme, titled Cognitive Computing via Synaptronics and Supercomputing (C2S2), aims to use nanoscale components to produce a computer which not only mirrors human thinking, but has similar size and power requirements as a brain.
The New York Times quotes a professor involved in the project as explaining that computers perform well only in clearly defined tasks; they struggle with open-ended problems that require a variety of skills and approaches, such as a person trying to remember where their car is parked in a busy lot. Computers can’t currently recreate the human ability to process information from a variety of senses.
Meanwhile the senior IBM scientist on the project, Dharmendra Modha, says the process will be equivalent to “reverse-engineering the brain”. Researchers have long wanted to produce a computer which can think, but it’s only now become a practical possibility thanks to advances in both computing and understanding of how a brain operates.
The project is backed by a $4.9 million grant from DARPA, an agency of the Department of Defense which works on technology for military use. Those involved haven’t made any public statement on whether or how the findings will translate into a military setting.
You may be wondering what the point is of creating a computer to equal a human brain’s capability when six billion such ‘devices’ already exist. The answer is that doing so would allow a combination of human ‘rational’ thought and a computer’s processing power.
For example, IBM argues, computers are much more capable of processing data in situations such as financial markets or monitoring water supplies. However, it takes humans to understand what the data actually means and make decisions based upon the information. To bring the two skills together, it should be a lot simpler to do the difficult task of making computers think in the same way as humans than to do the impossible task of making humans able to cope with the same amount of simultaneous data as a computer can.
Related:





Stumble It!
