Big Blue will partner with leading US universities to build a computer (ed: a loosely used term) capable of mimicking the human brain as IBM tries to break the "conventional programmable machine paradigm".
The firm has enlisted the help of IBM and some of the brightest minds at Stanford, Cornell and the University of California, Los Angeles to develop this computing initiative. Unsurprisingly, the US government is financing the research by pumping in nearly $5 million through the Defense Advanced Research Projects Agency or DARPA.
An IBM statement said that the end goal would be "ubiquitously deployed computers imbued with a new intelligence that can integrate information from a variety of sensors and sources, deal with ambiguity, respond in a context-dependent way, learn over time and carry out pattern recognition to solve difficult problems based on perception, action and cognition in complex, real world environments"
The system will be designed to fit into a small volume and be power efficient, capable of simulating and reproducing the brain's fundamental abilities : sensation, perception, action, interaction and cognition.
The human brain has an estimated 100 billion neurons and 100 trillion synapses and IBM pins its hope of using Artificial Intelligence to emulate the brain of a small mammal.
The current fastest supercomputer system, the Roadrunner, has been built by IBM and has a peak speed of 1.11 PFLOPS. According to Raymond Kurzweil, the renowned expert in the field of artificial intelligence, as little as 10^16 floating point operations are needed to simulate human brain fucntions, with 10^19 (1000 times 10^16) being the strict minimum to make human brain Neural simulation for uploading a reality.
In comparison, ATI's Radeon HD4870 has a peak floating point operation performance of 1.2 TeraFLOPS (10^12).
IBM plans to build virtual human brain