Site Presented By
Sunday, Nov 8, 2009
Steve Grand, designer of the artificial life program Creatures, talks about the stupidity of computers, the role of desire in intelligence and the coming revolution in what it means to be "alive."
Jan 2, 2002 | When Steve Grand developed his artificial-life computer game Creatures nine years ago, he never dreamed that 1 million people would play it and come to care deeply about the lives of their virtual pets. Creatures allowed players to design these pets, or norns, and observe how they interacted with their environment and with other norns. The norns have computer-simulated hormones and DNA. They eat and breed. They fall in love. According to Grand's book "Creation: Life and How to Make it," "Creatures was probably the closest thing there has been to a new form of life on this planet in four billion years."
That's a pretty startling claim, but as Grand explains in his strangely accessible and consistently surprising book, whether or not you believe it depends on your definition of what's alive. Grand -- now two years into building a 4-month-old robot orangutan named Lucy -- argues that our traditional notion of life is just now beginning to change.
Creation: Life and How to Make It
By Steve Grand
Harvard Univ. Press
230 pages
Nonfiction
Grand spoke to Salon from his home outside of Bristol, England (where he works out of his garage), about what artificial life says about the soul, why emotions are so important to intelligence and why someday we might drive cars that enjoy doing their work.
What led you to build Creatures?
I was a college dropout but I trained to be a primary school teacher. I was really lazy at it and hated standing up in front of people and speaking. But the reason I got into teaching was not really because I was interested in teaching but because I was interested in children's minds and how people grow and develop.
So you started writing computer programs?
Yes, this was the '70s when I was at college. The microcomputer had just started. I generously spent my girlfriend's life savings on a computer and taught myself to program it. I figured if I couldn't teach children to learn things then maybe I could teach computers to learn things. I thought I'd be reasonably good at that.
What are some of your criticisms of the attempts to create artificial intelligence -- you know, in the last 50 years.
Phew!
Yes, I know, there's a lot there.
It's a big subject. This all started just after the war. Everyone was used to command and control, and logistics, and very centralized views of society. People assumed that the human brain was organized on similar principles, that there was some kind of central controller inside our heads pulling the levers and making us think. The people thinking about these things were often mathematicians who believed that perfectly logical thought was the highest kind of thought. So they tried to make perfectly logical kinds of machines. But "Star Trek" has more or less demonstrated that Mr. Spock isn't always as bright as he looks. It's Captain Kirk who always comes out on top because he's got emotions and common sense and all these other things that Spock doesn't have.
People made a mistake about the machine they should be using for all this. Most people in AI came to the conclusion that the digital computer was the epitome of what the brain was like. Once they decided that, they saw their problem as: How do you make a computer intelligent? There was a kind of breakaway faction in all of this who decided that brains are not a bit like computers, we've got a hundred billion nerve cells in our brains, not one very smart machine. This half thought, "Oh well, the clever machine therefore is the neuron, the brain cell, and all we have to do is grow a few brain cells together in a bucket and that will suddenly become intelligent."
The whole AI world split in two halves. Both of them thought they knew what the magic machine was. I think both were wrong. We don't know what the magic machine is yet. I don't think we even have the faintest idea how the brain works.
