1.  Calculating machines: “something extraordinary”

This book is about ‘computer intelligence’, in quotes.  It weaves together a set of remarks, pertinent and impertinent, on the subject of artificial intelligence (AI).  The development of machinery that possesses some of the characteristics of intelligence will undoubtedly come to be seen as one of the most significant events of the late twentieth century.  But just how significant will AI prove to be?  Edward Fredkin, then director of the Laboratory of Computer Science at the Massachusetts Institute of Technology, one of the leading AI research centres, was in no doubt:

There are three events of equal importance... Event one is the creation of the universe ... Event two is the appearance of life ... And, third, there’s the appearance of artificial intelligence... There can’t be anything of more consequence to happen on this planet.

Edward Fredkin (1979), quoted in Pamela McCorduck, Machines Who Think, New York: W.H. Freeman.

So, the development of AI may be the last event of consequence – if it were to happen.  If it has not already proven too late, we ought perhaps to form an opinion about AI.

    It seems obligatory to begin with a definition of the term ‘artificial intelligence’ but alas there is no consensus on what precisely it means, as illustrated by the following definitions from three textbooks:

[AI is] the study of how to make computers do things at which, at the moment, people are better.

Elaine Rich and Kevin Knight (1991), Artificial Intelligence, New York: McGraw-Hill.

[AI is] the study of mental faculties through the use of computational models.

Eugene Charniak and Drew McDermott (1985), Introduction to Artificial Intelligence, Reading, Mass.: Addison-Wesley.

AI is the study of agents that exist in an environment and perceive and act.

Stuart Russell and Peter Norvig (1995), Artificial Intelligence: A Modern Approach, Englewood Cliffs, N.J.: Prentice-Hall.

These definitions immediately identify some key points of controversy: Does AI necessarily involve comparisons between humans and computers?  Is AI primarily concerned with ‘doing’ rather than ‘thinking’?  And with all this mention of intelligence and agents, is AI about espionage?   Rather than embark on an abstract discussion of these issues at this stage, we will, with the help of many commentators, present a review of the development of AI up to the present and then speculate upon its (and our) future.

    There seems to be something inherently contradictory in the juxtaposition of the words ‘machine’ and ‘intelligence’:

machine n vt,  1. apparatus for applying mechanical power .. 2. person who acts mechanically and without intelligence.

Concise Oxford Dictionary.

However, not all man-made devices are for applying mechanical power.  Are the telescope, the sundial and the abacus considered to be machines?:

The abacus is a hand-operated calculating machine ...

Donald Clarke, ed.  (1982), The Encyclopedia of How it Works, London: Cavendish.

The abacus is thought to have been developed in the Orient over five thousands years ago but its use in many widely separated cultures (China, Egypt, Greece, Mexico, Peru and so on) might indicate that it was independently invented in several places.  Numbers are represented by beads strung on rods or wires set in a rectangular frame and calculations are carried out by sliding the beads along the wires.  The abacus is basically for adding and subtracting and there is no one-stage process for multiplication.  This was, of course, a difficulty for early arithmeticians, who had numeral systems that did not allow them to write numbers in columns and so to perform multiplication on paper the way we did until recently.

    The abacus does, however, lack one important characteristic that we associate with machines – it does not possess the ability to perform apparently autonomous actions.  The calculations are performed by the user moving the beads: the beads never move ‘of their own accord’.  Consider the difference between a ‘gun’ and a ‘machine-gun’.  A gunfires a single bullet when we pull the trigger but a machine-gun fires long bursts, that is, one action by the user sets in train a series of actions carried out by the machine.  We set a prototypical machine into action by pulling a lever or pressing a button and thereafter it performs the actions it has been designed to perform.  So a ‘calculating machine’ designed to perform addition would be one which, after we set it off, would autonomously carry out a sequence of actions through which it could determine the result of adding the two numbers we had previously specified.

    The French mathematician and philosopher, Blaise Pascal, modestly introduced such a device in 1642:

I submit to the public a small machine of my own invention by means of which alone you may, without effort, perform all the operations of arithmetic, and may be relieved of the work which has often times fatigued your spirit.

Blaise Pascal (1642).

The invention of the automatic calculator was for a long time attributed to Pascal but it is now known that a presumably even more modest German named Wilhelm Schickard, professor of astronomy at Tübingen, had designed and built such a machine in 1623.  These devices could only perform addition and subtraction but nonetheless they created quite an impression, particularly with those such as astronomers, engineers and accountants whose livelihoods depended on making numerous tedious and detailed calculations.  Others, however, were disconcerted:

The mind had somehow been taken over by the machine ... it could do any kind of calculation without error, which was something extraordinary to be able to do without a pen, but even more so without even knowing arithmetic.

Pascal’s sister (regrettably unnamed), quoted in L. Perrier (1963), Gilberte Pascal, Bibliographie de Pascal.

This is probably the first ever comment on the subject of AI.

    Over the next two centuries the technology of calculating machines was considerably refined, most notably through the introduction of multiplication and division by the German philosopher Gottfried Wilhelm Leibniz (1646-1716).  It is difficult today, in the age of cheap calculators, to appreciate what a bane something like multiplication must have been but a diary entry of Samuel Pepys (1633-1703) gives an idea:

... By and by comes Mr Cooper, of whom I intend to learn mathematiques, and do begin with him today, he being a very able man.  After an hour’s being with him at arithmetique (my first attempt being to learn the multiplication table); then we parted until tomorrow.

Samuel Pepys (1662), Diary, July 4.

At that time, Pepys, who was as well educated as anyone of the time, was in charge of the Contracts Division of the Admiralty – and yet he needed expert tuition on multiplication.

    The advent of machines capable of carrying out arithmetical operations had two outcomes.  On the one hand, awe-ful ignorance provoked extremes of exaggeration, such as this description of a machine displayed at the Paris Exhibition of 1857:

This machine, among the most ingenious, solves equations of the fourth degree and of even higher orders ... scientists who vaunt their calculating powers, as a divination of the laws of nature, will be advantageously replaced by a simple machine, which, under the nearly blind drive of an ordinary man, of a kind of movement, will penetrate space more surely and profoundly than they.  Any man knowing how to formulate a problem and having the machine of the Messieurs Scheutz at his disposal for solving it will replace the need for the Archimedes, the Newtons or the Laplaces.  And observe how in the sciences and arts, all is held together and intertwined: this nearly intelligent machine not only effects in seconds calculations which would demand an hour; it prints the results that it obtains, adding the merit of neat calligraphy to the merit of calculation without possible error.

Baron L. Brisse (1857), Album de l’exposition universelle.

On the other hand, “nearly intelligent” machines did not impress the realists:

That arithmetic is the basest of all mental activities is proved by the fact that it is the only one that can be accomplished by a machine.

Arthur Schopenhauer.

These two reactions we will see repeated throughout the history of AI.  By the way, Schopenhauer (1788-1860) was a gloomy German metaphysician who was happy to introduce to the West various Eastern philosophies that supported his insistence on the universality of suffering.  He wrote abusive texts On Women and also On University Philosophy, the latter after he had deliberately timed his lectures to coincide with those of Georg Hegel – and had then found that students preferred Hegel.  His comments on arithmetic are mild in comparison.

  On to Chapter 2
 Back to Drakkar Press home page
 Back to Whoever Said .. home page