If intelligent machines ever write a history of their evolution, Hans Moravec will figure prominently in it. At Stanford University in the 1970s, he developed a mobile robot that could navigate its environment, albeit slowly. In 1980 he cofounded Carnegie Mellon University's Robotics Institute, where he is now working on machines that can make realistic 3-D maps of their surroundings. That small but complex step, he believes, could within 50 years lead to robots that are smarter than we are. He discussed his electric dreams with Discover associate editor Fenella Saunders. When will we have robots that can help with everyday chores? Within three years we'll be able to build a "navigation head" about the size of a basketball. We could attach such a head to a self-propelled cleaning machine or a security robot and train it to follow new routes through a factory. Making a smart factory vehicle would raise enough income and credibility for the really big product--a robot vacuum cleaner with about the intelligence of a guppy. What about further ahead?
Next we'd build universal robots that can do almost any basic chore. Assuming you have a vigorous industry, utility robots will lead to a general-purpose robot that can adapt to its surroundings. Then comes a third generation that could analyze its own behavior. Do you think robots will ever attain true consciousness?
A third-generation robot will need an internal psychological model so it can rehearse its actions, instead of having to make mistakes to learn. It will also ask how you feel to update its psychological model of you. There's no reason it couldn't respond to questions like that itself. It might say, "I'm not happy because everything I tried didn't work out right, but I've gone over it in my head and I think I can do better." I think it's conscious for the same reason humans are conscious: The root of our consciousness is, we think about how we feel. So does this robot. Why would an intelligent robot continue...
Please join StudyMode to read the full document