We remember Pic de la Mirandole as the man who knew everything about the knowledge of his time. But Einstein warns us: “imagination is more important than knowledge”. Knowledge alone is sterile and does not allow progress. The instrument of knowledge is intelligence or reason, that of imagination is instinct or the heart, according to Pascal's own words, and by this yardstick of imagination, the great thinker makes figure of a visionary as evidenced by all of his contributions in science as well as in philosophy.
“Intelligence has nothing to prove, it has to clear away. She is good only for menial tasks. » Simone Weil, Gravity and Grace
Pascal physicist and engineer
In 1631, the young Blaise is 19 years old and the family settles in Rouen where his father is named by the cardinal of Richelieu deputy commissioner for the collection of the tax and the lifting of the sizes, very unpopular taxes of the Old Regime. . In order to facilitate the tedious paternal work, Blaise invents the first calculating machine which makes it possible to mechanically carry out additions and subtractions of two numbers as well as multiplications and divisions by repetition.
A Pascaline, signed by Pascal in 1652, at the Museum of Arts and Crafts of the National Conservatory of Arts and Crafts in Paris.
David.Monniaux/Wikimedia, CC BY
It works thanks to an original system of six gear wheels and includes a pawl or jumper which automatically transfers the deductions from the operations. The machine will be called the Pascaline, we see a copy at the Museum of Arts and Crafts in Paris. A dozen machines were built, but, too expensive to sell, the invention will prove a commercial failure.
In 1654 Pascal initiated a new discipline which he called the geometry of chance. It is a method that sheds light on the problem of luck in games of chance and it constitutes the first step in the statistical calculations of which he can be considered the pioneer.
Before him, Galileo had already rubbed shoulders with it, responding to the Grand Duke of Tuscany who asked why, when you roll 3 dice, the sum 10 comes out more often than 9? With equiprobable faces, we find a probability of 0.116 of rolling 9 and 0.125 of rolling 10; the Grand Duke had an eye for everything.
In his Treaty of the arithmetic triangle, Pascal explains the technique of calculations called today probabilities, he introduces in particular the reasoning by recurrence ubiquitous in mathematics.
A computer on the one hand, a method of statistical calculation on the other hand, here are the two necessary and mandatory ingredients to develop an artificial intelligence (AI).
What is AI?
AI is the ability of an artificial machine, that is to say built by man, a computer therefore, to demonstrate properties that equal or even exceed the human brain, even that of Einstein.
AI is all the rage. Television, in its evening news, recently boasted of its benefit in optimally managing the round of truck transport between various delivery points. The intelligence needed here is equivalent to that of a colony of ants. Optimization programs in a multiple choice problem have been known for a long time.
To master the complexity of the world, physicists, who have the upper hand in statistical calculation techniques, have developed the so-called Monte-Carlo simulation, a name which obviously recalls the game of roulette and the green carpet. It is the method of choice for evaluating the possibility of carrying out a problem governed by a large number of correlated parameters, each of which is forced to follow a complicated law of probabilities.
In practice, we already know that a computer beats the world champion in chess, because it can calculate all possible moves at high speed and select the winning choice without firing a shot.
AI is found in facial recognition or in driving an autonomous car. In a more socially committed way, a Chinese court uses him as a prosecutor. A machine is capable of identifying various common offences: theft, fraud, dangerous driving… Trained on 17,000 files, it can lay charges with an accuracy of 97% on the basis of a verbal description of the case. This will reduce the burden on the judges by the same amount, but who will bear the responsibility in the event of an error?
Note that AI analyzes our behavior, for example through our use of the Internet, to then promote our tastes or our habits and thus channel our future choices, which ultimately constrains our freedom.
So can AI emulate humans in all their activities? Our nervous system operates like a super computer that receives, stores and retrieves information. But he can also learn and design. Already, some machines use feedback mechanisms that analyze a trend to adopt new behaviors that are more apt to the success of their task. This is the case of stochastic cooling where information on the dynamics of a beam of particles accumulated in a circular ring is used to improve the subsequent qualities of the beam. Thus CERN has developed a device to produce intense threadlike beams of antiprotons. In a way, the system is evolutionary and creative, it learns, but the mechanism stems entirely from the intelligence of the person who devised the clever scheme and Simon Van der Meer received a Nobel Prize in 1984 for this invention.
All of these examples rely on increasingly sophisticated algorithmic manipulations. The problem is whether AI can go beyond simple calculations to exercise the imagination which is, as Einstein reminded us, the essential engine for advancing ideas.
Imagination is an art hidden in the depths of the human soul, and we will no doubt wrest from Nature only with difficulty the secret of its true movements. Immanuel Kant
Can AI be creative?
The ambition of the promoters of AI is neither more nor less than to clone the human brain. However, this one is capable not only of manipulating knowledge, but he also knows how to conceive new thoughts, which allow him to explore the fields of art, mathematics, morality, the concept of God… which makes him differs a priori from a computer which only performs the operations that are submitted to it. If there is a creation of thought at the end of the chain of algorithms, the operating method of a machine copies that of the nervous system and a supercomputer can develop an AI. Otherwise, the AI will be limited to the servile intelligence of which Simone Weil speaks.
Note that the creativity that originates in intuition frees itself from the strict determinism of the laws of classical physics to give a chance to chance, at least apparent. Chance seems to be the instrument of creativity, but there is chance in human life, but a priori there is none in a computer, except that of electrical breakdowns that one can hardly imagine to be creative.
The search engines available today already contain infinitely more knowledge than the brain of Pic de la Mirandole. It remains to be seen whether one day a highly evolved robot will be able to compete with Pascal in terms of inventiveness. Let him have the last word. In his Pensées, he writes, taking on the accents of a prophet: “The arithmetic machine produces effects which come closer to thought than anything animals do. But she doesn't do anything that can be said to have willpower, like animals”. But it is the will, criterion of the freedom of the living, which leads to both artistic and scientific creation. Thus for Pascal, the stated objective of the AI seems a decoy.