“Sounds simple enough.”
Brian laughed. “It is one of the most complex things that we have ever taught computers to do. The system has to take each Japanese element of speech and compare it to stored networks of information about how each English word, phrase or expression is used. Thousands of man-hours of programming have been done to duplicate what our brains do in an instant of time. When I say ‘dog’ you know instantly what I mean, right?”
“Of course.”
“Do you know how you did it?”
“No. I just did it.”
“That I just did it is the first problem faced in the study of artificial intelligence. Now let’s look at what the computer does when it hears ‘dog.’ Think of regional and foreign accents. The sound may be closer to dawg, or daw-ug, or any other countless variations. The computer breaks down the word into composite phonemes or sounds, then looks at other words you have recently said. It compares with sounds, relationships, and meanings it holds in memory, then uses a circuitry to see if its first guesses make sense; if not it starts over again. It remembers its successes and refers back to them when it confronts new problems. Luckily it works very, very fast. It may have to do thousands of millions of computations before it types out ‘dog.’ ”
“I’m with you so far. But I don’t see what is expert about this voxfax system. It doesn’t seem to be any different from a word processing system.”
“But it is — and you have put your finger on the basic difference. When I type the letters D-O-G into an ordinary word processor, it simply records mem in memory. It may move them around, from line to line, stretch them out to fit a justified line or type them out when so instructed — but it is really just inflexibly following unchanging instructions. However, your languaphone and your voxfax program are teaching themselves. When either of them makes a mistake it discards the mistake, then tries something else — and remembers what it has done. This is a first step in the right direction. It is a self-correcting learning program.”
“Then this is your new artificial intelligence?”
“No, this is only a small step that was made some years ago. The answer to developing true artificial intelligence is something completely different.”
“What is it?”
Brian smiled at the boldness of the question. “It is not that easy to explain — but I can show you what I have done. My lab is right down here.”
He led the way through the connected laboratories. It all appeared very unimpressive to Beckworth, just a series of computers and terminals. Not for the first time he was more than glad to be at the business end to this enterprise. Much of the apparatus was turned on and running, though unattended. As they passed a bench mounted with a large TV screen he stopped dead.
“Good God! Is that a three-dimensional TV picture?”
“It is,” McCrory said, turning his back on the screen and frowning unhappily. “But I wouldn’t look at it for too long if I were you.”
“Why not? This will revolutionize the TV business, give us a world lead…” He rubbed his thumb along his forehead, realizing that one of his very rare headaches was coming on.
“If it worked perfectly, yes, it should certainly do just that. As you can see it apparently works like a dream. Except that no one can watch it for more than a minute or two without getting a headache. But we think we have a good way to fix this in the next model.”
J.J. turned away and sighed. “What did they use to say? Back to the drawing board. Anyway, perfect this one and we own the world.” J.J. shook his head and turned back to Brian. “I hope you have something to show us that works better than that.”
“I do, sir. I’m going to show you the new robot that will overcome most of the limitations of the older AI machines.”
“Is this the one that can learn new ways to learn?”
“That’s it. It’s right over there. Robin-1. Robot Intelligence number 1.”
J.J. looked in the indicated direction and tried to control his disappointment.
“Where?”
All he could see was an electronic workbench with various items of some kind on it, along with a large monitor screen. It looked just like any other part of the lab. Brian pointed to an electronic instrumentation rack about the size of a filing cabinet.
“Most of the control circuitry and memory for Robin-1 is in there. It communicates by infrared with its mechanical interface, that telerobot over there.”
The telerobot did not look like any robot J.J. had ever seen. It was on the floor, a sort of upside-down treelike thing that stood no higher than his waist. It was topped by two upward-reaching arms that ended in metallic globes. The two lower branches branched — and branched again and again until the smaller branches were as thin as spaghetti. J.J. was not impressed. “A couple of metal stalks stuck on two brooms. I don’t get it.”
“Hardly brooms. You are looking at the latest advance in microtechnology. This overcomes most of the mechanical limitations of the past generations of robots. Every branch is a feedback manipulator that enables the management program to receive input and—”
“What can it do?” J.J. said brusquely. “I’m very pressed for time.”
Brian’s knuckles whitened as he made hard fists. He tried to keep his anger from his voice. “For one thing, it can talk.”
“Let’s hear it.” J.J. glanced obviously at his watch.
“Robin, who am I?” Brian said.
A metallic Ms opened in both of the erect metal spheres. Tiny motors hummed as they turned to face Brian. They clicked shut.
“You are Brian,” a buzzing voice said from the speakers also mounted on the spheres.
J.J.’s nostrils flared. “Who am I?” he asked. There was no response. Brian spoke quickly.
“It only responds when it hears its name, Robin. It also would probably not understand your voice, since it has only had verbal input from me. I’ll ask. Robin. Who is this? Figure next to mine.”
The diaphragms opened, the eyes moved again. Then there was a faint brushing sound as the countless metallic bristles moved in unison and the thing moved toward Beckworth. He stepped backward and the robot followed him.
“No need to move or be afraid,” Brian said. “The current optic receptors only have a short focus. There, it has stopped.”
“Object unknown. Ninety-seven percent possibility human. Name?”
“Correct. Name, last, Beckworth. Initial J.”
“J. J. Beckworth, aged sixty-two. Blood type O. Social Security number 130-18-4523. Born in Chicago, Illinois. Married. Two children. Parents were…”
“Robin, terminate,” Brian ordered, and the buzzing voice stopped, the diaphragms clicked shut. “I’m sorry about all that, sir. But it had access to personnel records when I was setting up some identification experiments here.”
“These games are of no importance. And I am not impressed. What else does the damned thing do? Can it move?”
“In many ways better than you or I,” Brian replied. “Robin, catch!”
Brian picked up a box of paper clips — and threw them all toward the telerobot. The thing whirred in a blur of motion as it smoothly unfolded and rearranged most of its tendrils into hundreds of little handlike claws. As they spread out they simultaneously caught every one of the paper clips. It put them all down in a neat pile.
At last J.J. was pleased. “That’s good. I think there could be commercial applications. But what about its intelligence? Does it think better than we think, solve problems that we can’t?”
“Yes and no. It is new and still has not learned very much. Getting it to recognize objects — and figure out how to handle them — has been a problem for almost fifty years, and finally we have made a machine learn how to do it. Getting it to think at all was the primary problem. Now it is improving very rapidly. In fact, it appears that its learning capacity is increasing exponentially. Let me show you.”