Communications of the ACM
Riders on a Swarm
Russell Cobb
One of the bugaboos that authors of science fiction sometimes use to scare their human readers is the idea that ants may develop intelligence and take over the Earth. The purposeful collective activity of ants and other social insects does, indeed, look intelligent on the surface. An illusion, presumably. But it might be a good enough illusion for computer scientists to exploit. The search for artificial intelligence modelled on human brains has been a dismal failure. AI based on ant behaviour, though, is having some success.
Ants first captured the attention of software engineers in the early 1990s. A single ant cannot do much on its own, but the colony as a whole solves complex problems, such as building a sophisticated nest, maintaining it, and filling it with food. That rang a bell with people like Marco Dorigo, who is now a researcher at the Free University of Brussels and was one of the founders of a field that has become known as swarm intelligence.
In particular, Dr Dorigo was interested to learn that ants are good at choosing the shortest possible route between a food source and their nest. This is reminiscent of a classic computational conundrum, the travelling-salesman problem. Given a list of cities and their distances apart, the salesman must find the shortest route needed to visit each city once. As the number of cities grows, the problem gets more complicated. A computer trying to solve it will take longer and longer, and suck in more and more processing power. The reason the travelling-salesman problem is so interesting is that many other complex problems, including designing silicon chips and assembling DNA sequences, ultimately come down to a modified version of it.
From The Economist
View Full Article
No entries found