US scientists build brainy leech computer

Forget artificial intelligence and neural networks, computers of the future could be based on little blood suckers.

A US team at the Georgia Institute of Technology, has developed a machine based on the neurons of leeches, in a pioneering project described by team leader Professor Bill Ditto as a "manipulation of the building blocks of the brain to do our bidding".

As researchers harness nature to solve the computation problems that conventional machines struggle with, computers are going organic. And the thinking behind the biological computer is relatively simple.

"Conventional computers are really very dumb," said Ditto. "Often a five year old can figure out what I am going to say next whereas a computer would have no idea. With conventional computers we have to micro-manage how they solve a problem. What we want is a computer that can come to a solution with just partial information," he said.

The neurons that control the thought process in living things can do just that according to Ditto. "We are trying to do it the way nature does it," he explained. "The brain can't go out and buy a faster processor, it is stuck with living neurons and computes through the connections between neurones."

The team is attempting to control the way neurons interact. The neurons are stimulated by electrodes and their interaction is controlled by a PC, a job which will eventually be done by a chip. The neurons are currently brushing up their maths skills. "Right now they are doing addition," Ditto said. The next stage will be to embed the neurones on silicon and a working model will be complete within a year and a half.

Despite being age-old fodder for Sci-Fi writers, computers that think for themselves is within our reach, said Ditto. "They will be the next, next generation of computers. We are talking in twenty years time."

The enormous capacity and availability of organic computing -- "it's relatively easy to grow cells," explained Ditto -- means biological machines will eventually replace silicon ones, Ditto believes. "It may not be practical on the desktop but there are many problems beyond the scope of conventional computers, like finding out how societies interact or economic problems where there are millions of variables."

Perhaps the most interesting application for the living chip will be a medical one. "We will be able to interface silicon with the nervous system and create a robotic arm to replace a missing limb or correct problems with the nervous system. This is our first glimpse of how to control the human nervous system," Ditto explained.