Researchers at Dublin City University are busy constructing a completely new life form, one based on computer software rather than cells. This "living technology" will grow, reproduce and evolve just like the real thing however, and in time may help to producing computers and robots that can repair themselves, writes Dick Ahlstrom.
A form of "life, but not as we know it" is the simplest way to understand the work under way in DCU's Research Institute for Networks Communications Engineering (RINCE). Dr Barry McMullin leads DCU's €8.5 million contribution to a wider EU research initiative called PACE, Programmable Artificial Cell Evolution. Its goal is to build artificial life forms.
"It is taking an engineer's view of the problem of evolution," explains McMullin. "How do we build robust, self-replicating computer systems." The closest thing to this goal at the moment is that bane of computer-users everywhere, the software worm. These self-perpetuate and bore into computer files, damaging information as they go.
The aim of PACE, funded by the EU's 6th Framework Programme, is to produce self-sustaining "software agents" that can change and evolve as the computer environment in which they live changes, says McMullin.
"They are self-sufficient little computer programs that can grow, replicate, diversify within the computer system," he explains. "The field is known as artificial life," and the "holy grail", he adds, would be the development of the "spontaneous, evolutionary growth of complexity in a computer system".
The engineers want to model such systems on biological life, which evolves as its environment changes. Some changes end a biological life form, others increase its chances of survival, but these changes also increase the level of complexity within it.
A computer is more complex than a stone but a single bacterium is more complex than a computer, he says. "The biological process leads to a spontaneous growth in complexity. We would like to achieve something similar in computing."
The approach being taken does not rely on applying artificial intelligence to help a software agent to "think" its way out of a problem in order to survive. Rather, like a biological agent that uses whatever resources happen to be available to it, the software agent spontaneously incorporates small programme elements into its overall makeup and sheds others in response to changes in its environment.
"This is a fundamentally new way to engineer software systems, by growing them rather than designing them," says McMullin. "The most complex machines we have don't do this."
One benefit would be a computer system or programme that can repair itself. By extension this might lead to a robot that can correct its own errors or learn from its mistakes.
McMullin also takes a wider view, suggesting that if we could build machines that can evolve then we should gain new insights into evolution itself. "If you could synthesise living organisms then you would get a better understanding of how they are organised and how life originated."
PACE is a four-year programme involving 13 research partners and two co-operating groups from eight European countries.
The consortium includes chemists who will help create the microscopic chemical elements that should allow the artificial cells to assemble themselves automatically from non-living material.
DCU's involvement focuses on trying to do all this in software simulations, models that will help to explain what can be achieved in the long term.