THE IDIOT SAVANT
In my undergraduate years, following my military service, I used daily to walk across Waterloo Bridge to my college and observed the construction work in progress for the Festival of Britain 1951 Exhibition. Some novel, but temporary, structures included the Skylon and the vast Dome of Discovery. More permanent was the Royal Festival Hall. The most ancient feature was the old Shot Tower, which had been used to produce round shot for firearms by means of pouring molten metal through selected mesh at the top, the molten drops adopting a spherical form during their long fall before splashing into a tank of water from which they could be extracted. By no means a new technology, but effective in its time! This old landmark was now to be used for a novel purpose. A radar dish was mounted on its top, the idea being that for the opening ceremony a radar pulse would be fired at the moon and the echo pulse returning from some 240,000 miles away would be the trigger to switch on the lights.
I seem to remember that, in the event, things did not work out exactly as intended, and a little judicious "nudging" was required. Technical novelties often fail to perform to order!
The Dome of Discovery housed numerous exhibits (it was an education in itself) but I can only call two of them to mind at this distance in time.
One comprised a demonstration of robot turtles - carapaces carried on tricycle undercarriages, presented by some learned doctor whose name I forget.
On each "turtle" was mounted a pair of photoelectric cells coupled to valve amplifiers which operated relays. There were two driving wheels powered by an on-board accumulator, and a steering wheel directed by a stepping-relay (all this was hidden under the carapace).
In demonstration, these metal animals exhibited quite purposive behaviour. They could be set so as to move in the direction of light, or away from it. When their accumulators ran low they would proceed to a charging rail at the rear of the platform and remain in contact with it until their battery was recharged.
This activity parodied that of simple single-celled animals - positive and negative heliotropism, and "feeding".
My thought at the time - and I knew it was naive, even back in 1951 - was this: if one filled the Albert Hall with such decision-taking elements, suitably interconnected and with appropriate external inputs, would it be able to "think"? Would it be self-aware?
Now, all these decades later, I have on my desk at home equipment containing at least the equivalent of an Albert Hall filled with valve binary elements. However, I am quite certain that it is not self-aware!
Nevertheless, my original thought was not altogether ridiculous; it all depends on how the binary elements interrelate, whether they exceed a certain critical quantity, and on the nature and complexity of the input and output devices.
Popular journalism bestowed the description "electronic brain" on the early computers, but this is a misnomer.
The human brain contains more than a billion nerve cells or neurones. These can be thought of as binary elements with the capacity of linking not merely to another single such element but to any of the contiguous ones.
Any "thought" takes the form of a succession of neurones "firing" and establishing a unique pattern of temporary linkages. Once a neural pathway has been created, although only for an instant, it can be readily re-invoked by another associated pathway - this is the basis of memory. Thus the thinking and memory activity is spread over a large volume of the neural matrix.
Forgive this absurdly oversimplified reference to a highly complex (and not fully understood) topic, but I am seeking to bring out the essential difference between an animal (or human) brain and an "electronic brain".
In a normal computer all the central activity is carried out in the tiny arithmetic unit, which does sums at an incredible speed. It is the way in which this activity is directed by the program software that creates the illusion of machine intelligence.
However, this does not necessarily imply that there could never be such a thing as "real" computer intelligence or a computer "mind".
Consider other fields in which human art has not directly imitated nature but has nevertheless developed very adequate equivalents. There is no such thing (in the real sense) as a wheel in nature, but man's mechanical machinery and vehicles depend on them, although nature favours levers and joints. There are still some situations where you would be better off on foot or horseback rather than in an appropriate vehicle, but in general the artefacts perform remarkably well.
Again, nature's solution to the problems of flight usually involves subtly-flexing and flapping wings. Some early inventors tried unsuccessfully to copy this, but success came with fixed wings, movable control surfaces and thrust from rotating propellers.
I sometimes relax with "cryptic" crossword puzzles. Often the solution involves one or more "intuitive leaps" where an obscure word that one might not have used, heard or read for many years for some reason flashes into consciousness from the deepest recesses of the mind, by some mechanism of which one is totally unaware.
A computer could only do this by trial-fitting many thousands of words until it finds one that satisfies the criteria. The speed at which it does this might well outstrip the lateral thinking of the human mind, but it is functioning as an Idiot Savant - a wise fool!
However, there is no reason to suppose that real machine intelligence could never be created - not by slavish imitation of nature, but by another route entirely. The flight of birds is a wondrous miracle of nature, but there is no bird that can fly faster than sound for thousands of miles carrying many tons of payload!