It seems, too, that the human brain is vastly superior to a computer, if only because it can accommodate about 10,000,000,000 circuit elements within the volume of a cabbage, while so far it has not been possible to put more than 10,000 to 100,000 transistors into a computer as big as a room. Remarkable progress has been made in the effort to squeeze more and more elements into smaller and smaller spaces, but it will be a long time before we even begin to approach the economy with which the space available in our skulls is utilized.
Scientists have drawn parallels between the storage capacity of the human brain and the core storages of a computer. But all such comparisons fail, for there is a fundamental difference between the two. The electronic memory stores according to addresses, and the same statement is always to be found under a certain address (provided that it is not deliberately altered). The human brain, however, knows nothing about addresses of this kind. It stores its data here, there and everywhere in accordance with the general principle of the association of ideas, in a network of processes of combination and comparison which is anything but
systematically planned. That is why the human memory is far less reliable than the electronic one, though on the other hand it can establish many more connections than a computer can between the contents of memories. So the human "arithmetic unit" can make intellectual connections on a numerical scale which cannot easily be reached by a computer.
As regards speed, though, the computers are better than human beings in every way. In the copper wires of a computer, current is propagated at a speed of about 186,000 miles a second, while in human nerve fibres it covers a distance of only about 100 yards during the same time. This comparatively slow conduction rate is the reason why a human, even if he is capable of very concentrated thinking, "switches" much more slowly than an electronic brain.
The relative sluggishness of man's thinking processes makes him inferior
to the computers in other ways too. Nobody can consciously absorb more than
about 30 bits of information per second. For that reason alone, the amount
of knowledge we have at our disposal at the end of our fives is limited; we
can never collect more than 50 thousand million bits. In a single second an
electronic computer can take up about a million bits over the magnetic tape.
And for a computer, 50 thousand million bits are an almost ridiculously small
quantity: they can run through a television receiver in thirty minutes. This
aspect of the matter is rather depressing. What is homo sapiens?
Thirty minutes of television time!
You will remember that we wrote at the beginning of Chapter 2 that the world could be considered as an arrangement largely composed of information. That, it now seems, was no frivolous statement. Information theorists take such a view of man, at any rate.
After all, it took quite a time before we arrived at the idea that the concept of "information" can yield a great deal more than the information technologists expect from it. Living things have been exchanging information ever since the first gregarious animals made warning noises to each other. People have been making use of converted and coded news information for thousands of years; the smoke and fire signals of primitive peoples are examples. We have been able to transmit information electrically for more than a hundred years. But only a few years ago did we accustom ourselves to the novel idea that information is being communicated when the brain gives the order: "Forward march!" to the foot. When this notion struck us, it naturally led to the question of what the difference actually is between the various kinds of information involved, and whether it would not be possible to build devices or models which would imitate the biological processing of information. Such models are the electronic fox mentioned in our first chapter, the moth and the tortoises, the chess-playing automaton and the computer that plays Nine Men's Morris.
A hundred years ago men believed that mechanistic principles had at last given them understanding of the world and of the people who live in it. Human beings, too, were envisaged as mere mechanical models. A famous pathologist triumphantly announced after he had dissected a corpse that he had found no sign of a soul. But men soon became aware that this world of pure mechanisms was one that was confined within very narrow limits. So materialists called the notion of energy to their aid and declared that man is a state of energy. Today, we have taken a further step and are saying: "Man is a being made up of information, filled with information, impelled by information."
Has that brought us to our goal?
If we have reached the stage of regarding man pre-dominantly as the sum of pieces of information, as an ideal cybernetic case, and if we further consider the electronic information machines as a superior kind of cybernetic model - a further question pops up: Is electronic information-processing a model of human information-processing? Will an electronic automaton some day really be able to think as a human being thinks?
Theoretically, so the scientists say, this is possible. But so far no machine exists which can process information with anything remotely resembling the universality and perfection attainable by a human being.
Nor is there likely to be such a machine in the foreseeable future. Perhaps none will ever be built, because it is conceivable that the human intellect is inadequate to discover the nature of the ingenious circuits which are the distinctive characteristics of the human brain.
But as you have read, there are electronic computers capable of carrying out weighty economic and commercial calculations and plans even today - and who would dare to say that the professor of economics or businessman wasn't using his brains if he as a mere human made similar decisions?
Another example: I.B.M. prepared a program which states independently whether any theorem of elementary geometry is true or untrue. In other words, a machine programmed in this manner demonstrates geometrical truths all on its own. But Pascal, the mathematician, was regarded - and rightly - as highly talented and particularly intelligent because he could prove geometrical theorems even when he was a child . . .
It is understandable that almost everyone feels an unpleasant sensation if someone says to him: "See that machine over there? It can think!' If we are reluctantly convinced that the machine can reason more quickly and more dependably than man, our instinctive reaction is one of sheer indignation. In fact, most of the arguments advanced against thinking machines do not come from the realm of the exact sciences but out of emotional distaste - because we feel that that which must not be cannot be.
Because there are now mechanisms which behave intelligently - the artificial fox, the synthetic tortoises, the chess-playing machines - and because nonetheless tribute must be paid to our tender feelings, the idea of "instrumental conditioning" has been invented. The question now is no longer: "Can a machine think?" but: "What similarities exist between human intelligence and instrumental conditioning?"
But despite this pious watering-down, "instrumental conditioning," we must recognize the final aim of all cybernetic research, realize what the artificial moths, foxes and tortoises are driving at: a machine that thinks. A machine that apes man.