OVER the past half-century, our efforts to create machines in our own image have been stymied by three things: life, consciousness and intelligence.
We may labour under the comforting illusion that we understand these concepts. In reality, no one can agree on precisely what they mean.
It is no surprise, then, that the 鈥渢op down鈥 approach to creating artificial intelligence, life and consciousness has failed. If we can鈥檛 define what intelligence really is, attempts at AI are doomed. Ditto the rest.
Advertisement
But there is an alternative 鈥 evolving digital life from the 鈥渂ottom up鈥. At long last, this approach seems to be living up to its extraordinary potential.
As we report this week (see 鈥淭he genesis machine鈥), digital evolution has advanced in leaps and bounds since the idea first inspired computing pioneers, such as John von Neumann in the 1940s and then John Holland, the father of genetic algorithms.
Two decades ago, Tom Ray did some of the earliest work with self-replicating programs. Like DNA, his 鈥淭ierra鈥 programming language could mutate. He also added natural selection, enabling more successful organisms to monopolise resources.
Then Chris Adami of Caltech pushed for software that could be used for the systematic study of evolutionary processes. In this way the computer world Avida was born. Now, crawling around inside Charles Ofria鈥檚 computers at Michigan State University, East Lansing, are digital microbes, known as Avidians. They consume computing time as 鈥渇ood鈥, have equations for genetic codes, self-replicate, mutate and evolve into new, more intelligent artificial life forms.
In other experiments, digital pets have evolved to have rudimentary memory, and symmetrical brains. Ofria鈥檚 team has also created organisms with altruistic traits, who will sacrifice themselves for their kin.
鈥淒igital pets have evolved with rudimentary memory and symmetrically organised brains like ours鈥
Evolving life inside a computer lays the groundwork for understanding the selective pressures that led to life, consciousness and intelligence the first time around. A new chapter in our understanding will open when we get evolution to do it all over again.