Jean-Gabriel Ganascia: "The fear of artificial intelligence is unfounded"



[ad_1]

Jean-Gabriel Ganascia is a researcher at the Computer Science Laboratory of Paris 6. He is also the organizer of the round table Autonomous machines: what are the ethical issues? held on July 10 at the ESOF 2018 in Toulouse. Find here the interview he gave us, in partnership with the newspaper La Dépêche du midi .

Science and the Future: What is artificial intelligence (AI)? [19659003] Jean-Gabriel Ganascia: It is primarily a scientific discipline, born in the mid-1950s, that aims to understand human intelligence through the simulation of all the cognitive functions that contribute to it. It attempts to reproduce with computer science our human ability to memorize and learn, to reason, to use a language, after our sense organs (sight, hearing, smell, touch …) have given us a perception of the outside world. If AI provokes such a craze today, it is because the mbades of data [les fameuses ” big data “, NDLR] that we produce, are all registered or almost. And many applications make it possible to take advantage of it.

Can you give examples?

The targeting of advertisements on the web, which are proposed to you according to the way you navigate, the weather forecasts, founded on complex models, the medical diagnosis on image, like that of the melanoma [cancer de la peau] the systems of recognition of the faces or the fingerprints … And at present the car autonomy or even the badistant of Google which recognizes the speech and allows you to dialogue with him … All this implements artificial intelligence

How?

Basically, there are algorithms. These are methods consisting of steps that execute in a specific order. In our daily lives, our brain uses a lot of them. For example, if the baguette costs 1.30 euros, and I give five euros to the baker, then I calculate that it must make me 3,70 euros by performing a subtraction. It's an algorithm. The baker uses another by making change: that of the complement to five: 1.30 … 1.50 … two … three … and two that make five. But the power of computers has changed the game. Very complex algorithms can now be processed very quickly

… and thus exceed the capabilities of the human brain. Is this why the AI ​​can be scary?

There is this image, popularized by the film 2001: The Space Odyssey of the computer that takes control and gets rid of the human. In the philosophical sense, the human being is autonomous in that he adopts his own laws, and so on. In reality, this is not at all what is implemented in artificial intelligence. This fear is therefore unfounded. There was also this highly publicized accident of the Uber autonomous car, which cost the life of a woman who was crossing a night road, with her bicycle [le 18 mars 2018 dans l’Arizona, aux Etats-Unis, NDLR]. The car was programmed to offer maximum comfort to its pbadengers and not to brake whenever it detects a plastic bag or a pile of dead leaves on the road. She did detect the cyclist, but did not "recognize" her as such and did not stop. It's not the machine that took power: there are human beings behind it, who program it.

Do you research artificial intelligence that is less "binary"? What is it?

This is an AI that can handle exceptions to certain rules. For example, the autonomous car is programmed not to cross a continuous line because it is illegal. But if it arrives in sight of a broken down truck, and is followed by another vehicle that rolls fast and may hit it from behind, then our autonomous car may decide to cross the line to double the truck to avoid the accident, despite the rule that was programmed. It remains to define the list of exceptions …

It is commonly said that AI will revolutionize the work. What do you think of it?

Do not confuse tasks, trades and jobs. Tasks can be painful, not very rewarding. The machine will take care of it. Trades, which are sets of tasks, will evolve to implement the automatisms of artificial intelligence. As for jobs, the AI ​​will remove a number of jobs, because it allows to perform certain tasks much faster than humans. But she will create others! What matters is the ratio between the jobs that will be created and those that will be eliminated. AI is an industrial revolution that resembles the previous ones. To face it, we must train ourselves to this mutation of trades.

Science and Future: What is artificial intelligence (AI)?

Jean-Gabriel Ganascia: It is first of all a scientific discipline, born in the mid-1950s, which aims to understand human intelligence through the simulation of all the cognitive functions that contribute to it. It attempts to reproduce with computer science our human ability to memorize and learn, to reason, to use a language, after our sense organs (sight, hearing, smell, touch …) have given us a perception of the outside world. If AI arouses such a craze today, it is because the mbades of data [les fameuses « big data », NDLR] that we produce, are all registered or almost. And many applications make use of it.

Can You Give Examples?

[ad_2]
Source link