
"Open the pod bay doors, HAL." A classic quote from Kubrick's "2001 - a Space Odyssey" sets in motion the disastrous train events on board an interplanetary probe when an AI (Artificial Intelligence) computer takes over from its human pilots. The computer, HAL, judges the mission to have been compromised by its human astronauts. Such is the stuff of SF "bad machine" scenarios over decades of inspired writing ... but soon we may regard it as prophecy and not just scarey fantasy. Enter Cyc.
Cyc (pronounced "psych") is a computer knowledge base with a startling difference. It learns by interactive aggregation and common sense ... just, some would say, like a human mind. Pre-loaded with over 3 million assertions that help it explore creatively and accurately, it can actually learn more easily as it becomes more proficient. In other words, it is an evolving system. When it goes online later this year and gets plugged into the Internet, it is expected to accelerate away from us in the fast lane. (Full article here in the
New Scientist).
Will Cyc ever reach the supposed "singularity" state where smarter than human intelligence takes a machine beyond the reach of its human creators? Could it ever achieve consciousness, mind? Would pulling the plug be equivalent to murder ... putting the plug back in equivalent to resurrection? Would it take steps to defend itself? A little while ago such questions might have seemed fit only for such entertaining SF films as Kubrick's masterpiece. Now, with Cyc, the scenario seems so much closer. Should we be worried?
Well, yes, I think we should be concerned. The very idea that we should create something we cannot control is a very old theme in human literature but now, in this century, we have the power to create something that might one day destroy us completely rather than merely irritate us. My guess is that we should watch Cyc and Cyc's children very, very closely.