It's easy enough to say that you can feed into the neural blocks of our cybernetics brain what is euphemistically called the sum of the world's knowledge. It's easy to say it, but impossible to realize as a fact.
Because much of what you feed into that electronic memory is going to be wrong. And if not wrong, it will be distorted or misleading. The world's knowledge? No one knows what it is. No one can verify its accuracy because much of it is opinion. Feed a computer with the sum of knowledge on the characteristics of the terrestrial environment, request the computer to come up with an answer to a scientific enigma, and if the brain could do it, it would, as Selig Albracht once said in sympathy, "throw up in the programming room." Because we don't know that much about the terrestrial environment to answer many of the questions we have. And the computer can't perform miracles.
We fed into 79 everything one could imagine. Every encyclopedia, textbook, reference, almanac, document, report, and so on, on which we could lay our hands. And we crossed fingers and held our breath and stroked rabbits' feet, and knew that no matter what we did we would have a hell of a time unraveling the inconsistencies and the outright contradictions. Because man's grasp of his physical world is too often a sometimes thing, with huge gaps in his cherished book of facts.
Nonetheless we had at it. Under the sociological aspects of programming we poured theories and the theorems, the contents of magazines and newspapers (carefully selected and then assiduously edited)—the gamut. We fed it music—both on tape in the original and broken down into mathematical symbols so that the computer could compare the mathematical notations against auditory output.
Interesting: I wish I could have listened in on 79's "comprehension capabilities" when that program was under way!
In effect we were learning as much as we were teaching. There's a vital distinction—the difference between pouring in data and teaching. One is sheer and often meaningless ingestion, a memory bank that answers when you give it an electrical jolt. Everything that went into the memory neural blocks of 79
went in with a proviso that this was never the latest word; that the population, the Gross National Product, anything, was a fluid fact and always subject to revision and updating. This had to be done with extraordinary care, or else the revising, the input of additional data to the subjects already fed to the cybernetics brain, would become hopelessly unwieldy. We recorded within 79 the records and teachings of the world's great religions, specifying carefully as we did so that in this respect 79 was to function as a repository rather than to employ the material for data-seeking and solving problems. Spare us the reeking contradictions of the theological tomes! But for reference purposes—well, in they went.
Hundreds of specialists kept up the input. From all over the United States the data poured in with a steady flood of material on every subject conceivable. As the special direct-link lines went up and were activated, the task became simpler in terms of physical handling. The tendrils reached out across the length and breadth of the country not only for the primary data but also for the results of new experiments and tests, the conclusions of programs and projects, the creation of new theories as well as standing actualities. Into 79 went technical journals, statistical compilations ... a thundering cascade of data reduced to tiny electronic echoes always available for recall and cross-checking. Day and night, month after month, the work had gone on, and all of us waited for that day when the computer could make that wondrous transformation when data would become knowledge.
But any learning process is terribly intricate, a matter of a search for knowledge crossing its own path again and again and again. The computer is the recipient of data, but it must know that it operates under severe restrictions—its data are never infinite, never definite, never really conclusive.
It must know when to stop solving a problem.
There are other asides. ... A cybernetics system may comprehend excitement as it is registered through glandular changes and physiological alterations (we did this by linking the computer input to human subjects under hypnosis and then running them through a bewildering gamut of emotions while 79
recorded the physiological changes) . . . but—and it's the grandest but of them all—how do you record the tremulous fluttering of a young girl's heart when the stimulus is puppy love?
Time and again old truths came home to roost in the midst of our gleaming new world. None was more applicable to everything we did than the sign that Selig Albracht had hung in the most conspicuous part of his office. And he'd placed a bright light on the sign so it couldn't be ignored. The sign read: INSOFAR AS MATHEMATICS APPLIES TO REALITY IT IS NOT CERTAIN, AND SO
FAR AS MATHEMATICS IS CERTAIN IT DOES NOT APPLY TO REALITY.
Selig wouldn't let anyone get away without realizing the sign or its message.
"You numbskull!" he would roar at the hapless visitor who failed to give the quotation more than a passing glance. "Have you any idea of who said those words?"
Invariably the answer would be No, they didn't know.
And then Selig would lean back in his chair and lower his voice to a growl, and he would tell them who said those words.
Albert Einstein.
It disturbed us that more and more of our staff were beginning to refer to 79 as the "talking computer." It disturbed us because it was true and it really didn't mean very much. Direct verbal conversation—two-way conversation—had been one of the experiments programmed for 79, and we had reached that point in the activation of the organism as a whole where we were able to run our tests.
The idea of the two-way exchange so fascinated me that I insisted I be given the project, much to the relief (I discovered later) of Dr. Vollmer. He had been assigned the task, and he didn't want it, because he thought the whole thing was idiotic. He also thought, and told me with brutal candor, that I was a mathematical mutant. "Anyone who will depart from the pure language of mathematics to indulge in acoustic babble must have, somewhere in his education, a vast and gaping hole."
"You seem to do rather well yourself at such acoustic mayhem," I shot back.
We grinned at each other, and I went off for my session of what Dr. Vollmer considered to be
"acoustic babble."
For a long time it was exactly that. We worked with one of our neural-block packages that, when we completed the input, would be linked to the "main brain" of 79. I can't deny this was more of an experimental program that anything else, but since it fitted neatly into the two-way communications involving bio-cybernetics I managed to get everything I needed. Including Kim to maintain the official control from Vollmer's Bionics Division.
As it turned out, we were wildly successful. In matters where hard data were not concerned, 79
could sustain a verbal conversation. There were gaps, to be sure, but as Kim pointed out, wasn't that a problem faced every day of the week by people?
Into the neural package, which could be considered a separable part of the cybernetics brain, we fed the sum and substance of the language. We started with the alphabet, the method in which the twenty-six basic letters could be arranged (mathematically, that was as easy as pie for 79), and the method in which the words and word groups were utilized to convey data. When the computer neural block had ingested the working foundation, we began the matter of converting data into acoustics.
The process wasn't so much complicated as it was vastly time-consuming, and for this purpose we used several universities around the country. More than a thousand people were involved in the effort, which all went down on tape. The printed letter A, for example, had a numerical value to it. Then, on the same tape, went the vocalization of A. And so on through the alphabet, and from this initial step into groups of letters and their vocalizations, and into basic words. On and on we went through the myriad, complicated contradictions of the language. Unbelievably sensitive in distinguishing the output of the vocalizations, the computer began to distinguish acoustic patterns with letters and words. The engineers built i
nto the service-output systems of 79 an acoustic ability, an electronic-mechanical throat, so to speak. With precision engineering equipment, exquisitely controlled energy output, with plastics allowing flexibility—all of it in a modular package that could be removed, and a standby or even an improved modular throat system plugged back in—79 learned to talk.
It was, I suppose, an artificial speech because it was based on mechanical-mathematical interpretations of comparison—comparing what it obtained on tape both in numerical value and in acoustic value. Linguistics experts were a vital part in the effort; enunciation and pronunciation were absolutely critical.
Finally, because of the long-term effort and the great number of people involved, 79 had in its memory bank, in the neutral-block package, a numerical value and acoustic equivalent for every spoken word of the English language. If switching and feedback worked, if the computer could scan every book ever published (applicable to this program), could maintain its speed-of-light scanning of its memory blocks, it should be able to talk. Often, the results were hilarious; but with experience, order emerged from acoustic chaos, and intelligible speech rose from mechanical vocalization.
If we asked 79, "How are you feeling today?" the computer would not reject the question out of hand as a meaningless query. It interpreted the question on the basis of its own values. Since the term
"feeling" was hardly applicable to something not of flesh and bone and the assorted aches and pains that make up our biological existence, it couldn't answer "Fine," in the sense in which we would interpret this word. (Of course, when it did use the word it had to distinguish, through swift scanning, the meaning of fine as applying to a fine margin or a fine edge; or fine as the levy of a financial punishment; and so forth.)
But it could interpret "feeling" as applying to its electronic and mechanical systems, and come back with an answer that it was functioning with 100 percent efficiency. And it would answer in such a manner!
"How are you feeling today?"
"Today" was this moment, that moment when the question was asked.
"You" was obviously the cybernetics organism itself.
As to the "How" of it, this became what percentage of maximum possible efficiency of all systems.
And the answer to the question would be: "Seven nine point three eight six," and it might throw in a
"thank you" because it had a means of scanning electronically meaningless but biologically desired phrases to be selected from its neutral memory for just such occasions! The answer that at this moment the organism had exactly 79.386 percent of all its systems functioning told us vocally what we could determine with a glance at the systems-monitoring panels.
But when you heard that deeply resonant voice (a matter of nothing more than engineering) with its perfect language acoustics, well, at times it made the hair on the back of your neck go straight up.
The baseball season produced some wild sessions, and Selig Albracht made a small killing on side bets. 79 gave him the probabilities, and he delighted in a back-and-forth session with the computer verbally; the computer, of course, could accept everything available on baseball, work out batting averages, percentages, percentages of injuries, weather of the moment and how weather had affected the team in the past, wind direction and velocity—oh, Albracht had it all down, all right. When he got his answers he added to them. He threw in his own factors about the main hitter of the team being embroiled in a divorce, and he pondered whether it would make the guy sick to his stomach or madder than hell, and he put it all together, and while it was light-years away from being a sure thing, Albracht's predictions for baseball were nothing less than phenomenal.
The main thing was, however, that despite all the obvious problems and others we failed to anticipate, after several months you could hold conversations with the cybernetics brain.
I had some great chess games, in which we called out the moves vocally.
I never won a game, I might add.
As the multiple programming continued and we approached the end of the long preparatory and testing phases, we began to release the tight pressure on the mental reins we had held for so long. We were starting to give 79 its head, turn loose its own capabilities for decision-making. Because now we were ready to see just how well 79 would respond to the unknowns in the problems we were facing ourselves. More and more scientific and technical and sociological groups came secretly to us with their problems. At first they compared what they had already learned through their own, limited, cybernetics systems against the answers that 79 provided for them.
It didn't take long. Project 79 was exceeding our hopes for it, and it was doing so far ahead of schedule.
From the moment our cybernetics brain changed from its role of student to that of equal, it began to rush ahead to a level of intellectual capability that had never before existed in the history of man.
The computer now began to make its own demands. We knew it would do so in response to discovering gaps in its programming. As it mulled over problems and found information to be absent, it began to request specific items.
That's when we knew we had won.
Our cybernetics organism, in every sense of the word, was thinking. Really thinking. When faced with the "insufficient data" obstacle, it didn't sulk or repeat idiotically that it needed further information; it requested what it needed to have. And if that wasn't available, it approached the subject not only mathematically but also in the heuristics evaluation: It played a hunch that was based on the greatest possibility of being right.
And what does man do?
Exactly that.
Which meant we were ready to begin the bio-cybernetics tests.
15
No one spoke. Not a sound from any one of the forty people in the survey room. All eyes looked through the separating glass into the Contact Cubicle.
We had our miracle.
A human brain and a cybernetics organism were in direct contact with each other. They were communicating.
They talked. Not with words or with sounds, but with their mathematical equivalents. With the alpha-wave pattern of the man and the electronic sensitivity of the cybernetics organism. They communicated. The man queried the machine, demanded information.
The machine received the request, evaluated it, scanned its neural banks, returned the information requested.
The man was Maurice Levy, BC Test Subject 83. Kim and I had worked with him for months. He was our most promising key to institute the first manufactured miracle of its kind in history.
Levy was a Navy signalman assigned to Project 79. More than a year before we had begun the search for men and women who could alter, deliberately, the alpha-wave patterns of the brain. Ours was a hit-or-miss quest that had paid off. Each human being was tested with scalp electrodes to determine his specific alpha-wave signature. We studied the variations in alpha rhythm, and we separated those few individuals who could, by intent, alter their alpha-wave pattern.
I didn't know how it was done—Dr. Vollmer and Kim shared some theories into which I refused to be drawn. My only interest was its application.
Being able to control the alpha-wave pattern wasn't enough. It had to be controlled in a specific manner, a sort of on-off-on capability. We found that perhaps one person in every two hundred had some controlled alteration of his alpha-wave pattern. And out of that new group of one in two hundred, only one in every twenty or thirty had the potential, with training, for getting the on-off-on ability down pat.
That began our training program. For with the rare test subjects we had to teach them to control the alpha-wave pattern in such a way that we could adapt their talent to direct two-way communications with 79. Our cybernetics brain had absolute control of its alpha-wave patterns; it was, after all, an interpretation of energy output.
How to carry on the initial tests? Not even the optimists among us hoped for more than basic, crude communications until we had time enough in which to gain experience—the experience that
would give us some sound working foundations on which to continue.
The on-off-on sequenced interruption held the answer. If the test subject could do this—and with increased training they became quite skillful at it—the next step was to teach the test subject Morse code.
The computer learned it in no more time that it took to feed Morse code into its neural cells. We programmed 79 to use Morse code at a speed tremendously reduced from its capability. Electronically, 79 could spit out volumes in Morse code, but we could never utilize the data without processing them.
What we wanted initially was evidence that our program hopes were valid. So the computer responded in Morse code at a preselected speed.
It looked simple enough on the face of it, but then we ran into other problems. Learning Morse and thinking rapidly in Morse aren't the same. A person can learn Morse code, can communicate with the system, but it's halting and awkward and too damned mechanical. Where could we find people who thought as naturally in Morse code as if they had slipped into another language? Which is what Morse code was—a digital language.
That's when we went to the Navy. They had Morse code operators from way back when. They had men who could think in Morse as well as any of us thought with the English language. They didn't have that many who could handle Morse and who could alter their alpha-wave patterns.
Then we found Maurice Levy. He was fifty-two years old, and he had his first homemade telegraph key when he was eight. He had always been wild about Morse. He could think almost as fast in Morse as he did in everyday word language.
And luck smiled upon us. He was an alpha adept—one of the few who had, with training, an almost uncanny ability to control his alpha-wave pattern. He didn't know it, of course. In fact, he'd never heard of alpha-wave patterns within the brain until we told him about the results of his tests. The Navy assigned him (and another thirty as well) to us indefinitely. Maurice Levy—"Call me Manny," he insisted—well, Manny couldn't have been happier about his selection.
The God Machine Page 10