AI, history the Turing test passed

General scientific debates. Presentations of new technologies (not directly related to renewable energies or biofuels or other themes developed in other sub-sectors) forums).
Christophe
Moderator
Moderator
posts: 79118
Registration: 10/02/03, 14:06
Location: Greenhouse planet
x 10973

AI, history the Turing test passed




by Christophe » 09/06/14, 18:35

A first, a computer program has just passed the Turing test: http://www.slate.fr/story/88233/turing-ordinateur-test
0 x
User avatar
sen-no-sen
Econologue expert
Econologue expert
posts: 6856
Registration: 11/06/09, 13:08
Location: High Beaujolais.
x 749




by sen-no-sen » 09/06/14, 19:14

Worrisome ... but so fashionable!
the first AI worthy of the name should see the light of day within twenty years according to specialists ...
0 x
"Engineering is sometimes about knowing when to stop" Charles De Gaulle.
Christophe
Moderator
Moderator
posts: 79118
Registration: 10/02/03, 14:06
Location: Greenhouse planet
x 10973




by Christophe » 09/06/14, 20:43

I don't know what you don't hear "worthy of the name", can you elaborate?

ps: definition of the Turing test: http://fr.wikipedia.org/wiki/Test_de_Turing
0 x
User avatar
sen-no-sen
Econologue expert
Econologue expert
posts: 6856
Registration: 11/06/09, 13:08
Location: High Beaujolais.
x 749




by sen-no-sen » 09/06/14, 21:07

Christophe wrote:I don't know what you don't hear "worthy of the name", can you elaborate?

ps: definition of the Turing test: http://fr.wikipedia.org/wiki/Test_de_Turing


That is to say systems endowed with capacities allowing the resolution of complex problems.
0 x
"Engineering is sometimes about knowing when to stop" Charles De Gaulle.
Christophe
Moderator
Moderator
posts: 79118
Registration: 10/02/03, 14:06
Location: Greenhouse planet
x 10973




by Christophe » 09/06/14, 21:32

Well, complex problem solving is already a bit the case right now, isn't it?

For me, an AI worthy of the name is taking initiatives not provided for in the code! What the Turing test does not foresee (but issued a long time ago!)
0 x
User avatar
sen-no-sen
Econologue expert
Econologue expert
posts: 6856
Registration: 11/06/09, 13:08
Location: High Beaujolais.
x 749




by sen-no-sen » 09/06/14, 21:48

Christophe wrote:Well, complex problem solving is already a bit the case right now, isn't it?

For me, an AI worthy of the name is taking initiatives not provided for in the code! What the Turing test does not foresee (but issued a long time ago!)


Today's supercomputers are not artificial intelligences per se, they are only exosomatic devices allowing human brains to increase their capacities.
We can indeed talk about AI when the systems can be ingenious ... unexpected!
It is not impossible that this will happen soon ...
I made a post on the work of S. Wolfram on cellular automata and the emergence of complexity from simple laws here:
https://www.econologie.com/forums/automates-cellulaires-elementaires-t13075.html
0 x
"Engineering is sometimes about knowing when to stop" Charles De Gaulle.
Christophe
Moderator
Moderator
posts: 79118
Registration: 10/02/03, 14:06
Location: Greenhouse planet
x 10973




by Christophe » 10/06/14, 12:35

A complement in PourLaScience:

No, the Turing test has not passed!

On June 8, newspapers around the world announced that the "Turing Test has been passed" The announcement is spectacular ... but excessive and dishonest!


http://www.pourlascience.fr/ewb_pages/a ... -32984.php

(...)
In the system which on June 7, 2014 allegedly passed the test, the imitated human subject is 13 years old and the English language is not his mother tongue (which makes it possible to excuse his faults).

Nowhere does Turing speak of a 13-year-old child!

Nowhere does Turing contemplate that the imitated subject can only master rough English!

In conclusion, the announcement is misleading and to pass the Turing test as conceived by Turing, the system will have to be improved on four points:

(1) The system will have to trick its interlocutors not in 30% of the cases, but in 50% of the cases.

(2) The duration of dialogue between a judge and the system should not be limited to five minutes, but may be extended for a long time (say at least an hour).

(3) The computer system should imitate not a 13 year old child, but an adult (if possible not an idiot, ignorant of everything, having no idea about anything!)

(4) The system must be expressed in correct English (or another language).

We're not there!

I specify that I am of those who think that it will be possible one day, but who also think that before that we will have developed machines which on many problems requiring intelligence will have become stronger than us ... and that it has largely started.
0 x
User avatar
sen-no-sen
Econologue expert
Econologue expert
posts: 6856
Registration: 11/06/09, 13:08
Location: High Beaujolais.
x 749




by sen-no-sen » 10/06/14, 20:40

It's a matter of a few months anyway!
0 x
"Engineering is sometimes about knowing when to stop" Charles De Gaulle.
moinsdewatt
Econologue expert
Econologue expert
posts: 5111
Registration: 28/09/09, 17:35
Location: Isére
x 554




by moinsdewatt » 12/06/14, 22:06

Artificial intelligence: is it really important to pass the Turing test?

By Raphaële Karayan, published on 12/06/2014

Jean-Gabriel Ganascia, expert in artificial intelligence, sets the record straight and explains why the experience where a computer program pretended to be a teenager, about whom much has been said in recent days, has no significant value.
..............

The Turing test, which aims to assess the ability of a computer to imitate human conversation, is considered a foundational benchmark in artificial intelligence. But critics have appeared about the validity of this experiment, its seriousness and its results. To understand what the Turing test really consists of, put it in its historical context and analyze the scope of this weekend's experience, we interviewed Jean-Gabriel Ganascia, artificial intelligence expert and philosopher, professor at the Pierre and Marie Curie University.
..................


http://lexpansion.lexpress.fr/high-tech ... d7a33878_0
0 x
Christophe
Moderator
Moderator
posts: 79118
Registration: 10/02/03, 14:06
Location: Greenhouse planet
x 10973

Re: IA, history the Turing test passed




by Christophe » 10/12/17, 10:06

About the little-known, tragic life of Alan Turing: https://lejournal.cnrs.fr/articles/alan ... stin-brise

This week, the film "Imitation Game" is released on French screens, which traces the life of mathematician Alan Turing. The opportunity to reread our portrait of this genius with a tragic destiny, inventor of the computer and hero of the Second World War. Sentenced in 1952 for homosexuality, he was pardoned posthumously at the end of 2013 by Queen Elizabeth II.
On June 7, 1954, Alan Mathison Turing, mathematician of genius newly elected member of the Royal Society of London, inventor of the computer, precursor of artificial intelligence, is found dead, lying on his bed, at the age of 41 years. Near him, an apple poisoned with cyanide is half-opened. According to official sources, this is suicide. For two years, the man had been humiliated by the British justice system, which had spared him the prison only on condition that he underwent chemical castration by hormone treatment. His fault: being homosexual in a still ultra-conservative England where the law condemns such mores. During the Second World War, this outstanding scientist had helped to break the secret code of the German navy and thus avoid the invasion of England.

A childhood far from his family

Alan Turing was born in London in 1912. His father, then a colonial administrator, had to return to his post in India, but the climate in Madras was considered unfavorable for the health of children. Still an infant, Alan is therefore placed with his older brother near Hastings, in the south of Great Britain, in the family of a colonel where it is believed that the education of boys must make them men, tough guys , real ones. Turing sons will only see their parents occasionally during their childhood.

At 10, young Alan devours the Wonders of nature that every child should know. "The book tells how the embryo develops from a fertilized cell following the laws of physics and chemistry," comments Jean Lassègue, philosopher at the Center for Research in Applied Epistemology1, in Paris, and author of 'a biography of Turing. It is a revelation: the body is presented as a gigantic machine. Alan realizes that it is possible to determine the laws that govern its construction using the natural sciences. From then on, he never ceased to be passionate about decoding the secrets of life and the nature of the mind

Lightning strike at the boarding school

At the age of 14, the adolescent joined the boarding school of the very strict Sherborne Grammar School, where one privileges "the values ​​proper to the education of gentleman, Greek and literature over scientific subjects, specifies Jean Lassègue. The young man's results are poor. He is considered rough and clumsy and, the height of scandal in an environment imbued with classical culture and Victorian morals, he does algebra during religious instruction classes. ” Not very sociable, Turing finds comfort only in science lessons. Impressed by this student who, in his spare time, develops very complex calculations, his math teacher pleads his case with the school principal and saves him from repetition.

The following year, in 1927, Alan made a decisive encounter at school. His name is Christopher Morcom, is a year older than him and has a passion for science and math. It's love at first sight. With this relationship, undoubtedly platonic, the young Turing becomes aware of his homosexuality, then severely repressed in England at the start of the 19th century. But Christopher died of bovine tuberculosis at XNUMX years old.

"Alan Turing's vocation is linked to this drama: he undoubtedly thought that he had to embody the scientific destiny that was promised to his deceased friend", analyzes Jean Lassègue. According to the philosopher, this astonishing idea will haunt him all his life. And we can glimpse there, implicitly, the source of his future work, in which he invents a concept that has become absolutely obvious today: it is possible to dissociate a machine and the programs that make it work. Just as the young man had to imagine that his body, his machine, could carry, like a program, the spirit of Christopher…

The time of the first articles

Between 1931 and 1935, Turing pursued studies of pure mathematics at King's College, Cambridge. He published his first articles, obtained a research grant and solved the thorny problem of decision, formulated by David Hilbert in the late 1920s. Thanks to his article on this problem, which he finalized in 1936, Turing specifies in a completely original way the notion of calculation. "He manages to establish a limit between what is calculable and what is not", explains Jean Lassègue. What is calculable can be predicted: the result will always be the same. The non-calculable is what resists this determinism, it is what can evolve unpredictably, as sometimes happens in physics.

"Above all, Turing shows that what can be calculated can be broken down into a finite number of steps, each of which can be performed by a machine," continues Jean Lassègue. This famous Turing machine, which only exists on paper, is none other than the computer (literally "calculating machine") or computer. Alan is only 24 years old. He then joined Princeton University in the United States, where he prepared a doctorate in mathematical logic under the supervision of Alonzo Church. John von Neumann, the authority on mathematics and physics, asked him to become his assistant, but the researcher preferred to return to England. We are in 1938. The threats of a conflict with Germany are becoming clearer.

At the service of information

"Upon his return to England, says Jean Lassègue, Alan Turing is recruited by the British cipher service, the Government Code and Cypher School, which has just moved to Bletchley Park, near Oxford". The goal of this unit: to decrypt the radio messages that the Nazis exchange with their formidable fleet of submarines. Thanks to the ingenuity of Turing and his colleagues, most of the German intercepted messages were decrypted in 1942. According to historian Philippe Breton, from the Culture and Society in Europe laboratory2 in Strasbourg, tens of thousands of human lives have been spared thanks to this feat. For Shaun Wylie, a friend of Alan met at Princeton and found at Bletchley Park, "it was better that security did not know anything about his homosexuality: he would no doubt have been fired and we would have lost the war ..." 3.

After the war, Turing can finally devote himself to the materialization of the ideal machine designed in his article of 1936. He entered the National Laboratory of Physics, near London, where he wrote in three months the project to build a prototype of 'Automatic Computing Engine (ACE). It is not a simple calculator, as in competing American projects, but a machine capable of processing any type of data. With the plans quickly completed, Turing let programmers and engineers build the ACE, which did not come into service until 1950.

Meanwhile, in October 1948 he joined Max Newman's team at the University of Manchester. And it is in fact there that, in June of the same year, the first program in the world runs on a computer also designed largely according to Turing's ideas. Paradoxically, in this period of concretization of his work, the scientist begins to lose interest in this emerging computer science. "Because his deep ambition went far beyond," reveals Jean Lassègue. He wanted to build a brain equivalent capable of thinking. "

The thinking machine

Turing is now focusing on the conceptual possibility of lending intelligence to machines. In October 1950, he published his article entitled "The computer and intelligence" in the philosophical review Mind. Often considered as laying the basis for what will become artificial intelligence, the text opens with the game of imitation, in which the researcher imagines the way for a machine to pretend to be a human being.

"Besides, after having built the computer, a deterministic machine which exploits in all its extension the domain of what is calculable, Turing comes up against the biological processes which maintain with determinism a very different relationship", notes Jean Lassègue. It is impossible, for example, to predict the vagaries of the evolution of forms (leaves of a plant, animal jaws, etc.) over random micro-changes spread over thousands of millennia. By questioning the regularity of the distribution of the leaves of a plant and its mathematical rules, Alan Turing revives his childhood passion. The secret which he then endeavors to unravel becomes that of matter: how does it know what form to take? In 1952, he published an article on this question in the journal Philosophical Transactions of the Royal Society, which would become a founder in theoretical biology.

A tragic end

But the last two years of his life precipitated his tragic end. An adventure with a young man takes him to court, where he pleads guilty to "repeated indecent practices in the company of another man". Considering that he has too much work to go to prison, he accepts the alternative that is offered to him: chemical castration. They can do whatever they want with his body, he thinks, his mind, his software, will remain intact ... "But, under the effect of hormones, he is transformed into what he imagines to be a almost woman. What if thought and body had deeper connections than he thought? What if he was wrong? Asks Jean Lassègue, imagining the last thoughts of the mathematician. On June 7, 1954, Alan Turing was found stiff and cold in his bed. The cyanide apple has done its work.


0 x

Back to "Science and Technology"

Who is online ?

Users browsing this forum : Forhorse, Remundo and 164 guests