“Well, not really.” Paddy smiled as he rooted in his pocket for his tobacco. “But it is faster in certain kinds of calculation and I really need it for the development work on LAMA. That’s a new language that we are developing here.”
“What’s it for?”
“A new, rapidly developing and special need. You write programs in LOGO, don’t you?”
“Sure. And BASIC and FORTRAN — and I’m learning E out of a book. My teaching has been telling me something about Expert Systems.”
“Then you will already know that different computer languages are used for different purposes. BASIC is a good first hands-on language for learning some of the simplest things computers can do — for describing procedures, step by step. FORTRAN has been used for fifty years because it is especially good for routine scientific calculations, though it now has been replaced by formula-understanding Symbolic Manipulation systems. LOGO is for beginners, particularly children, it is so graphical, making it easy to draw pictures.”
“And it lets you write programs that write and run other programs. The others don’t let you do that. They just complain when you try.”
“You’ll discover that you can do that in LAMA, too. Because, like LOGO, it is based on the old language LISP. One of the oldest and still one of the best — because it is simple and yet can refer back to itself. Most of the first expert programs, in the early days of artificial intelligence, were developed by using the LISP language. But the new kinds of parallel processing in modern AI research need a different approach — and language — to do all those things and more. That’s LAMA.”
“Why is it named for an animal?”
“It isn’t. LAMA is an acronym for Language for Logic and Metaphor. It is partially based on the CYC program developed in the 1980s. To understand artificial intelligence it is first vital that we understand our own intelligence.”
“But if the brain is a computer, what is the mind? How are they connected?”
Paddy smiled. “A question that appears to be a complete mystery to most people, including some of the best scientists. Yet as far as I can see it’s really no problem at all, just a wrong question. We shouldn’t think of the mind and brain as two different things that have to be connected, since they are just two different ways of looking at the same thing. Minds are simply what brains do.”
“How does our brain computer compute thoughts?”
“No one really knows exactly — but we have a pretty good idea. It isn’t really just one big computer. It’s made of millions of little bunches of interconnected nerve cells. Like a society. Each bunch of cells acts like a little agent that has learned to do some little job — either by itself or by knowing how to get some other agents to help. Thinking is the result of all those agents being connected in ways that make them help each other — or to get out of the way when they cannot help. So even though each one can do very little, each one can still carry a little fragment of knowledge to share with the others.”
“So how does LAMA help them share?” Brian had listened with complete concentration, taking in every word, analyzing and understanding.
“It does this by combining an Expert System shell with a huge data base called CYC — for encyclopedia. All previous Expert Systems were based on highly specialized knowledge, but CYC provides LAMA with millions of fragments of common sense knowledge — the sorts of things that everyone knows.”
“But if it has so many knowledge fragments, how does LAMA know which ones to use?”
“By using special connection agents called nemes, which associate each knowledge fragment with certain others. So that if you tell LAMA that a certain drinking-cup is made of glass, then the nemes automatically make it assume that the cup also is fragile and transparent — unless there is contrary evidence. In other words, CYC provides LAMA with the millions of associations between ideas that are needed in order to think.”
When Paddy stopped talking to light his pipe the boy sat in silence for almost a minute.
“It’s complex,” Paddy said. “Not easy to pick up the first time around.”
But he had misunderstood Brian’s silence, misunderstood completely because the boy had followed what he had said to its logical conclusion.
“If the language works like that — then why can’t it be used to make a real working artificial intelligence? One that can think for itself — like a person?”
“No reason at all, Brian, no reason at all. In fact that is just what we are hoping to do.”
7
February 22, 2023
Erin Snaresbrook felt logy with sleep — even though she had slept for only five hours. It had not been by choice but by necessity, since she hadn’t been to bed at all for almost three days. She was beginning to hallucinate and more than once had found her eyes closing in the O.R. for lack of proper rest. It was too much. She had used one of the vacant intern’s rooms, fallen into a black hole of fatigue and, what seemed like a moment later, had been dragged painfully awake by the clamor of alarm. A cold shower shocked her back to life; reddened eyes blinked back at her from the mirror as she put on a touch of lipstick.
“Erin, I have to tell you. You look rotten,’’ she muttered, sticking out a furred and tired-looking tongue. “I prescribe coffee for your condition, Doctor. Preferably intravenously.”
When Snaresbrook came into her waiting room she saw that Dolly was already there, turning the pages on a worn copy of Time. She looked at her watch.
“Patients steal all the new magazines, would you believe it? Rich patients, or they wouldn’t be here, they even pinch the toilet paper and bars of soap. Sorry I’m late.”
“No, that’s fine, Doctor, it’s all right.”
“We’ll have some coffee, then get to work. You go in, I’ll be just a moment.”
Madeline had the mail ready and she flipped quickly through it, glancing up when the door flew open. She smiled insincerely at the angry General.
“Why are you and the patient still in this hospital?
Why have my orders for moving him not been carried out?”
General Schorcht snapped the words like weapons. Erin Snaresbrook thought of many answers, most of them quite insulting, but she was too tired for a shouting match this early in the day.
“I will show you, General. Then maybe you will climb down off my back.” She threw the correspondence onto the desk, then pushed by the General and out into the hall. She stamped toward the intensive-care unit where Brian was, heard the General’s heavy footsteps behind her. “Put this on,” she snapped, and tossed General Schorcht a sterile mask. “Sorry,” she said, took the mask and fixed it into position over the other’s nose and mouth; it’s not easy to fit one of the things with only one hand. When her own mask was in place she opened the door to the ICU just enough so they could see in. “Take a good look.”
The figure on the table was barely discernible behind the network of pipes, tubes, wires, apparatus. The two arms of the manipulator were positioned over him, the multibranching fingers dropping down into the opening in the cloth. The flexible tube of the oxygen mask wormed out from under the drapes and there were drips and tubes plugged into arms and legs and into almost every orifice of the unconscious body. Lights flickered on one of the complex machines; a nurse looked at a readout on the screen and made an adjustment. Snaresbrook let the door swing shut and pulled the mask from the General’s face.
“You want me to move all that? While the connection apparatus is in place — and in operation? It is working with the internal computer now to reroute nerve signals.”