By Marc Stevens
Characters like Commander Data of “Star Trek” and David from the new Steven Spielberg movie “A.I.: Artificial Intelligence” have sparked the imagination of viewers everywhere. Robots, androids, automatons, and the possibilities of artificial intelligence fascinate scientists and drive blockbuster sci-fi movie plots.
Audiences may wonder how close we are to creating the “artificial people” we see in movies and on television.
“I think we”re a long way from it,” said Dan Ventura, assistant professor of computer science. Robots can perform specialized tasks such as building cars or computer chips, but general-purpose “people” are mostly science fiction.
“The things you”re seeing in the movies and in Star Trek – we”re not anywhere near that at all,” Ventura said.
Research into artificial intelligence is constantly advancing, but often claims too much too soon. The field has not advanced as quickly as people expected.
Richard Helps, associate professor of electronics and information technology, said the idea behind movies like “A.I.” is as much fantasy as Frankenstein.
“We are breathing and feeling and thinking, our blood is flowing – it will probably be a long while before artificial systems can do that,” Helps said.
Current technology may be far from Hollywood”s vision of artificial intelligence, but BYU researchers are making progress.
Tony Martinez, professor of computer science, is the head of BYU”s Neural Network and Machine Learning Laboratory. He studies how to teach computers to mimic brain functions and to create computers that learn to solve problems through a process of trial and error without having to be programmed.
Computer processors can be connected in a manner similar to human brain cells, creating a “neural network.”
But it is not as easy as it sounds.
“A big challenge is that we”re still learning how the brain works,” Martinez said. “There are so many intricacies to the brain that right now we only mimic some of the more obvious ways it functions.”
Here at BYU, researchers train computers to produce a correct answer on their own when presented with a problem they have never seen before. They also work to make neural networks faster and more resistant to “noise” that is presented to the network in the form of missing or erroneous data.
In addition, the lab has managed to teach computers how to better understand continuous speech and recognize written characters.
Neural networks are not even close to the artificial intelligence we see in the movies, but they are capable of learning how to make limited subjective decisions, Martinez said.
Organizations across the country are taking advantage of the neural network”s capabilities.
The National Human Genome Research Institute recently used gene technology and a neural network to diagnose four similar types of childhood cancer.
America Online has switched to a neural network to run its parental controls, which used to depend on human reviewers. The neural network looks for inappropriate words and measures the context they are used in.
And the U.S. General Services Administration uses a neural network to operate face recognition technology. The network provides split-second recognition, and can adjust for slight variations in a person”s face, such as a smile or aging features.
Even these advances are far from the thoughtful, feeling boy depicted in Spielberg”s “A.I.”
Helps said more than 99 percent of all software is not capable of learning and imitating judgmental human behavior – it simply follows programming instructions.
“Take an automated teller machine. It really seems to work well with you, but it”s not artificial intelligence at all,” Helps said.
“Somebody thought about what you would need to do in that very narrow instance and wrote a program to take you through a process, giving you the choices you would want to have,” he said.
The possibility of a robot like David that is capable of showing love is even more remote.
“Scientists could, theoretically, somewhere down the line, make a machine that could mimic the way a loving person acts,” Martinez said. “Right now we”re more concerned with developing the foundation that”s needed to move A.I. forward.”
As intriguing as the movies are, artificial people like Commander Data and David may have to remain science fiction.
“I don”t think we”ll ever get to the point where you can say we have actual intelligence in machines,” Ventura said.
