"Trinary" Computing May Be Key to Artificial Intelligence

It's been said that even the most powerful computers today have the intelligence of a not-too-bright insect, and that we are nowhere near true artificial intelligence. That may be because, in order to build computers that think and reason, we are going about it wrong.

Research by MIT neuroscientist Gousong Liu suggests that neurons operate not on binary principles as do computers, but through a trinary system of ones, zeroes and negative ones. This allows them to process infinitely more complex levels of information, leading to the ability to learn, make judgments and selectively ignore information. Liu has also found that neurons are composed of multiple modules... giving us clues as to how to proceed with quantum computing.

Drew Endy, another researcher at MIT, suggests that DNA could be harnessed as a programming language of sorts, allowing us to crate nano-scale objects and pursue "synthetic biology."

NASA and the US military, meanwhile, are pursuing AI technology in the present. The military has developed a smarter unmanned plane that can respond to real-time situations and make some decisions independently. NASA is applying AI principles to two new rover designs, called K9 and Scorpion.

The Scorpion Rover

Their on-board computers are 10 times faster than those aboard the spectacularly successful Spirit and Opportunity rovers currently operating on Mars; they also allow K9 and Scorpion to multitask and make independent decisions. And unlike past rovers, K9 and Scorpion walk on legs rather than ride on wheels, which should prevent them from getting stuck. K9's developers say that 75% of its parts are off-the-shelf materials available at any computer hardware store.

NASA may use the technologies in the K9 and Scorpion rovers in its next unmanned mission to Mars, scheduled for launch in 2009.

Sources: NewsFactor, SFgate, KurzweilAI.net