SCIENCE

Compelling signs of artificial life

Digital ‘creatures’ that clone themselves may really he alive

MARK NICHOLS June 7 1993
SCIENCE

Compelling signs of artificial life

Digital ‘creatures’ that clone themselves may really he alive

MARK NICHOLS June 7 1993

Compelling signs of artificial life

SCIENCE

Digital ‘creatures’ that clone themselves may really he alive

Three years ago, Thomas Ray, a biologist at the University of Delaware, began testing a model of evolutionary principles that he had created on a computer. To set the system in motion, Ray fashioned a digital creature made up of a string of computer instructions and injected it into the model. Within a few hours, the solitary creature had begun to proliferate, spawning a race of clones that lived, died, evolved and gave rise to new groups of mutants that competed with each other in a struggle to survive. Ray was amazed. When he designed the system, called Tierra, “nobody knew what was going to happen,” he remembers. “But it turned out that evolution worked just as well in a computer system as in the real world.” And Ray maintains that systems like Tierra do more than just mimic living things, they are living things. ‘To me, anything that lives and replicates is alive,” says

Ray. “It doesn’t have to be wet and squishy.” Ray is not alone in believing that some electronic creatures squirming through a digital world or darting across computer screens may share the spark of life with humans, animals and plants. During the past decade, the idea of creating artificial life has attracted a following among North American and European scientists. Working on computers, they have devised systems whose colorful displays show digital creations that resemble insects and plants flourishing in a silicon world. Even scientists who stop short of claiming that some computer inhabitants are really alive say that they sometimes seem eerily lifelike. Christopher Langton is director of the artificial life program at the Santa Fe Institute in Santa Fe, N.M. After setting up an artificial life system in 1981, recalls Langton, “I wondered if I had the right to turn it off. I had created a universe in which there existed something that resem-

bles life. I began to wonder what rights the creator has.”

So far, there is little agreement among scientists about what constitutes life, and whether ingenious squiggles on a computer screen meet any valid definition of it. But many researchers in the field share a burning ambition to help give birth to nonbiological creatures that will qualify as life forms. In the process, they expect to gain a greater understanding of the cosmic logic that underlies organic life. “Nobody yet has made life in a computer,” says Steen Rasmussen, a Danish physicist who has worked at the Santa Fe Institute since 1988. “But we are getting closer. And I think that within the next 10 years, somebody will make something that we will have to call a living process.”

The origins of computer-based artificial life systems can be partly traced to the work of John von Neumann, the brilliant Hungarian-born, American mathematician. During the 1950s, he devised a tool for investigating artificial life called a cellular automaton. Consisting of an array of squares on a vast checkerboard, the automaton behaves according to a set of rules governing each square; as well, each square is affected by the state of the squares bordering it. During the 1960s, mathematicians at England’s Cambridge University began playing an elaborate board game, called Life, which worked on the basis of a cellular automaton. Once set in motion, with each square in a giant grid being “alive” or “dead” (a dead square would be empty) depending on the state of the squares adjacent to it, the grid appeared to take on a life of its own. Patterns and shapes mysteriously appeared in the grid.

As journalist Steven Levy wrote in his 1992 book Artificial Life, “sometimes objects broke up only when other newborn cells tampered with the equilibrium; at other times they were temporary configurations, doomed to dispel into quiescence.”

ITie behavior of cellular automatons fascinated scientists. Within a few years, theoreticians at American universities had begun playing Life on computers, creating dazzling images as thousands of cells on their screens, obeying a few simple rules, winked on and off, forming complex and unexpected patterns. Part of the fascination was the idea that computer-generated systems might mirror nature itself. In the new field of study known as chaos theory, which developed during the 1980s, scientists had discovered that structures, or patterns, could be discerned even in systems that appear to be completely disorganized. One of the basic ■ questions that they wanted to answer, says Rasmussen, is “what is it in matter that enables it to have an incredible variety of forms, including life?”

During the past decade, a cadre of scientists have pursued the answer to that and other related questions. One of them was Langton who, after graduating with a science degree from the University of Arizona in Tucson, decided to see if he could create a selfreproducing artificial organism on his computer. He began by creating four-sided loops with a short tail extending from one side. The loops contained information that determined their behavior, instructing the tail to extend itself to create a new loop. The process

worked, and the loops began to multiply, forming a colony of identical loops. The experiment convinced Langton that biological processes could be reproduced in machines.

Meanwhile, Stuart Kauffman, a leading biologist and artificial life theorist who is affiliated with the University of Pennsylvania medical school in Philadelphia, had developed a theory to explain how organic life may have developed on earth as the result of a set of underlying rules working in a complex, primordial “soup” of chemicals. Rasmussen, during the late 1980s, created a computer model that imitated primordial conditions with an artificial chemistry consisting of millions of computer instructions. As the computer ran through thousands of generations, digital “proto-organisms” died and new ones were created. Rasmussen concluded, among other things, that the evolution of lifelike forms depended on symbiotic, or co-operative structures, which seemed to emerge spontaneously within the system. “I think this is a law of the universe,” says Rasmussen, “and it makes possible the jump from nonliving to living.”

In a dramatically different approach to artificial life, a Calgary scientist has developed a way of simulating plant growth on computer screens. In the system devised by Polishborn Przemyslaw Prusinkiewicz, strikingly lifelike computer graphics can show a lily-ofthe-valley growing from a seed to maturity in less than a minute. To make his digital plants grow, Prusinkiewicz, 41, uses a few lines of computer instructions to represent the genetic code contained in a real plant’s seed, then adds the digital equivalent of nutrients, sunlight and water. Since he began developing his system during the mid-1980s, Prusinkiewicz has created models for about 20 plants. He can even simulate a stand of pine trees, complete with millions of needles, on his computer screen. Prusinkiewicz’s blending of science and art is already being used by biologists to test

hypotheses about plant life. Prusinkiewicz does not claim that the plants he creates are alive, but that “they simulate processes that take place in real organisms.”

At the Santa Fe Institute, Langton is currently trying to create in a computer the digital equivalent of the conditions needed to support cellular life. To do this, Langton will provide in digital form the molecules needed to produce water, enzymes, nutrients and the other essential ingredients needed to sustain life. Then he will try to introduce into the chemical “soup” an artificial cell and watch to see if it survives—and reproduces. Langton, who is also a research scientist at the U.S. National Laboratory at Los Alamos, N.M., predicts that it will be several years before he can achieve his goal.

Even if Langton succeeds, he is unlikely to persuade skeptics that a synthetic cell is a genuine form of life. Doyne Farmer, a Santa Fe-based expert on artificial life, concedes that creating such a cell would be “a significant breakthrough.” But critics, he adds, might object that the cell does not really possess the attributes of life, because “Langton

has to some extent rigged up his environment to produce the results he wants to see.” In fact, some scientists maintain that the promise that artificial life studies once seemed to hold has already begun to fade. William Macready, a 30-year-old Ottawa physicist who is currently doing postgraduate studies on complex systems at the Santa Fe Institute, says that once he “found the idea of artificial life intriguing. But now I’m not so sure that people in the field are learning as much as they think—or whether the systems they build are anything more than computer simulations.” Artificial life scientists, including Langton and the University of Delaware’s Ray, have different ways of look-

ing at the issue. Life on earth, Langton argues, has developed along hierarchical lines, with single-celled organisms evolving into more complex creatures. Now, says Langton, after billions of years of evolution, the human race may be at the point of constructing “the basis for a new organization of life.” Ray is convinced that the digital creatures he has created on a computer can evolve into more complex forms of life. “And for such creatures,” he says, “intelligence is the next frontier. We are living proof that evolution is capable of creating intelligence out of virtually nothing.” Like other scientists in their field, Langton and Ray contend that life may not exist only in organic form—that silicon worlds can also teem with forms of life that human beings are helping to create.

MARK NICHOLS