Marseille edit
The reign of robots is not for at once
Being seen under this angle, the world seems to be bad. Very bad. We would imagine in one of these movies where robots marked with intelligence take control of the world. Nevertheless, we are there still far.
An intelligence is not conscious as we think of it
Very recently, a "scandal" burst on Web concerning Facebook and the experience which the company realized with robots. The latter owed eventually, to be able to negotiate between them, and with human beings. The project seemed to succeed because both computers conversed with people they notice that they were only machines. But in a robot-robot conversation, these are slightly deviated from the structure of language which we had taught them: English.
Extract of the conversation:
Bob: i can i i everything else
Alice: balls gaunt zero to me to me to me to me to me to me to me to me to
Bob: you i everything else
Alice: balls gaunt has ball to me to me to me to me …
The press and the social networks seized the affair which nevertheless took place in June, by asserting that the researchers had to "disconnect" machines in the "panic". Hello the apocalyptic atmosphere.
In reality, the latter changed the parameters, as whoever is in that case. The problem of language would come in the way of a system given award when they speak in English which would not have been activated. Without award, robots then communicated in the way which seemed to be the simplest to achieve their purpose (here, succeed in an arrangement).
Dhruv Batra, a researcher to Facebook Artificial Intelligence, declares (déclare): "It is as if I said " le" 5 times, you will interpret that I want 5 copies of this object".
This can explain it, and we are still far from the independence of robots because "the error" was due to a parameter in the programming, not a grip of uncontrollable autonomy.
The creation of a language between robots is not an isolated phenomenon. Google Brain launched an experience in 2016 with three neuronal networks, affectionately conscripts Alice, Bob and Eve, who have proved far more vivid. Alice had to send a message coded to Bob and Eve had to manage to decipher it without their knowledge. Both accomplices improved the system of calculation to such a point that the experience makes a success but at the end of of 15,000th trial. Robots showing an artificial intelligence reached in this case, to show creativity because we exactly asked them.
A global intelligence difficult to achieve
The robots’ actions received as the independence result in fact from "bugs", from errors of programming or because we exactly asked them to show it. In all cases, they are programmed in a precise purpose, to realize a task.
They are capable of improving automatically by repeating actions, thanks to the machine learning (or learning machine), but always affect a given purpose. It is not however written that they will be capable of developing a global artificial intelligence (intelligence artificielle), by being able to be successful in all the fronts and have a free will. Because if they can show more successfully a human brain in a given task, to know how to make everything, we have to make them realize a multitude of varied tasks and they have to be programmed to understand them.
The mission thus remains complicated.
The danger of people …for robots
The artificial intelligence will not thus destroy the humanity tomorrow.
But as Jean-Paul Sartre said, "Hell is other people". Others here are people.
The hacking of machines
When the artificial intelligence becomes integrated little by little into our daily life (cars, connected objects, applications), the cybersecurity and the war against hackers are more than of current events. If the IA stays under control, there is nevertheless a possibility of the hacking.
What will it take place if robots are pirated to be ill-advisedly used? The machines of production used in factories could be rescheduled and create unprecedented incidents, as well as cars with autonomous conduct could be diverted at a distance. Moreover, Chinese researchers enjoied for the n-th time pirating electric motor cars Tesla by activating the remote brakes, opening the safe … just to show that yes, any object even state-of-the-art contains faults.
Little praiseworthy intentions
Besides the hacking (piratage) of an already existing entity, the artificial intelligence could be directly created for bad intentions. For example, to analyze security systems and better by-pass them. We say that the hackers of tomorrow will be robots.
It is one of the reasons why the biggest heads of the high-tech environment, being nevertheless one foot in the researches on IA, warn against its dangers. Among them, the serial start-upper Elon Musk, the founder of Microsoft Bill Gates or even the famous scientist Stephen Hawking, would like to become aware of IA’s formidable capacities and propose even a frame(supervision) in the research so that this technology is beneficial for the humanity and not the opposite.
Who had believed it?
Towards the people’s replacement
The artificial intelligence can be thus used "to hurt" but can also hurt. Its capacities are, without being infinite, exponential, and serve almost of the time to automate the tasks.
You see where we want to come there? By day when most of the jobs of today will possibly be replaced by a machine. This fear (almost) as old as the hills is not so ridiculous and not only attributed to the technophobes. Some people see a natural evolution, others a danger for their jobs and the future there. The most optimistic find the opportunity to rethink the economy by reducing the "human" working time to increase the production, and so win in spare time.
Then, for or against a domestic robot which makes our tasks in the work?
The artificial intelligence fascinates so much that it frightens. The man and machines are still at the stage of the taming and we do not know any more what to do: to jump feet joined in the robotics universe or put back of a step to preserve a "human" world? If IA is a powerful technology which can serve the humanity, she must be manipulated with gloves because the knowledge is a weapon which is not put between all the hands.
Article written by Marion Vannelli