In Search of Artificial Life

By Philip Elmer-DeWitt

They are born. They live their brief life. The fittest of them survive long enough to produce offspring. Over time their descendants evolve, adapting to changes in their environment. Or they fail to adapt and become extinct. They behave, in short, just like living things -- except that they are not flesh and blood but programs that inhabit the memory of a computer.

Can something that "lives" inside a computer really be alive? That is the bizarre question at the heart of artificial-life research, a fast-growing scientific field that seeks to illuminate the nature of life by recreating lifelike behavior in nonliving systems. In laboratories around the world, scientists tapping at computer keyboards are creating electronic versions of biological entities -- proteins, microbes, ants -- that bear a striking resemblance to their living counterparts. In the process, the researchers are raising questions that touch on some of biology's most enduring mysteries: How does nature create order from chaos? How did life emerge from nonlife? What does it mean to be alive?

The most notorious computer life-forms are the electronic viruses that have been injected, inadvertently or maliciously, into computer networks. Like real viruses, these programs are strings of instructional code that have the ability to infect a host computer and reproduce without restraint, sometimes causing considerable damage. But computer viruses are not really alive. They do not evolve or metabolize. And they are created, fully formed, by human programmers. The proponents of artificial life want their life-forms to create themselves, to emerge from nonliving components just as life on earth arose from the primordial ooze.

Nobody claims to have created true artificial life -- yet. But some have come intriguingly close. Christopher Langton, a researcher at New Mexico's Los Alamos National Laboratory, gets credit for coining the term artificial life. $ He was fiddling in the mid-'80s with programs known as cellular automata when he stumbled on a loop-shaped figure that could spontaneously reproduce itself. "That was a watershed," he says. "If you could capture self- reproduction, what else could you do?"

Today hundreds of people are exploring that question. At Bellcore, the research affiliate of the Bell telephone companies, David Ackley makes little creatures with humanoid faces that roam around a computer-simulated world consuming resources, evading predators and multiplying like rabbits. At UCLA David Jefferson and Robert Collins have created colonies of randomly generated "ants" that over many generations evolve the ability to navigate electronic mazes and search for symbols representing food.

Not all artificial life-forms are confined to a computer screen. At M.I.T.'s mobile robot lab (also known as the "artificial insect lab"), Rodney Brooks is building tiny six-legged creatures that are controlled by interconnected computer chips and that display behavior (scurrying for cover, stalking prey) that seems quite purposeful.

Are these things alive? That depends on how the term is defined. Surprisingly, there is no clear definition of "life." Most of the criteria put forward in the past are anthropocentric. Life on earth is carbon-based and built around the nucleic acids RNA and DNA, but that may be a historical accident. Most living things metabolize and multiply, but not all. Viruses have no metabolisms of their own; mules cannot reproduce. Many living things grow, but so do clouds and garbage dumps.

Still, most people have an intuitive sense of what it means to be alive. They know life when they see it. That is what is so disturbing about a good computer simulation. Take Craig Reynolds' flocking birds. By specifying a couple of simple rules -- keep a few wings' distance from your neighbors, try to fly as fast as they do -- Reynolds, a computer scientist at Symbolics, Inc., got bird-shaped objects on a screen to exhibit a flocking behavior that is absolutely convincing. The birds are artificial, but the flocking is real.

It is the same with life in general. Contends Langton: "Artificial life will be genuine life. It will simply be made of different stuff." This is the leap of faith made by a growing number of scientists, many of whom are associated with the Sante Fe Institute, a research facility that is the center of the artificial-life movement. "They feel like they are taking the first step into taboo territory," says Steven Levy, a New York City-based author who is writing a book on artificial life. "It's almost a religion."

Like religion, artificial life has evolved certain tenets. One is that lifelike behavior cannot be imposed from the top down. Rather, it emerges from the bottom up, like flocking among birds, when large numbers of parts obey a few simple rules. Another principle, derived from recent advances in the theory of chaos, is that when a system is sufficiently complex -- like the mix of chemicals in the primordial sea -- a lifelike order will spontaneously emerge.

Scientists have begun to think of possible uses for adaptive, self- replicating machines -- cleaning up toxic wastes, perhaps, or exploring outer space. There is a danger, though, that such machines could multiply uncontrollably, like the viruses that have disrupted computer networks. Doyne Farmer, a physicist at the Los Alamos lab, points to a cautionary science- fiction tale by Stanislaw Lem. In Lem's Fiasco, space explorers discover a Saturn-like planet with a ring around it. On closer inspection, the ring turns out to be a swarm of attack satellites and killer robots, part of a "star wars" defense shield that had reproduced itself over and over again. Artificial life, says Farmer, could turn out to be man's most beautiful creation. Or, like Lem's swarming robots, it could be a nightmare.