Go to ...

The Newspapers

Gathering and spreading news from various Russian Newspapers

The Newspapers on Google+The Newspapers on LinkedInRSS Feed

Wednesday, March 21, 2018

How dangerous artificial intelligence, and whether to fear him

Artificial intelligence (AI) is a topic that has long not come from the pages of popular science magazines and recurs in movies and books. The more specialists are developing this area of science, the more myths it covered.

The development and future of artificial intelligence concerned with those who are at the helm of the state. Not so long ago, Russian President Vladimir Putin visited the office of Yandex in the day of the 20th anniversary of the company where he explained, when AI will surpass human intelligence.

Anyone who is even slightly penetrates into the essence of the potential of artificial intelligence, you realize that to ignore this subject is impossible. This is not only an important topic for discussion, but probably one of the most important in the context of the future.

What is artificial intelligence and what is actually afraid of people told, “Reedus” expert on methods of machine learning Sergey Markov.

As said John McCarthy, the inventor of the term “artificial intelligence” in 1956, “as soon as he earned, nobody calls it AI”. AI is already in the entire reality: calculators, Siri, self-driving cars, etc. and it still did not believe. Why is it that people deny the existence of AI?

Mainly because of terminological confusion, as different people interpret the term “artificial intelligence” a completely different meaning.

In the science of artificial intelligence is a system designed to automate the solution of intellectual tasks. In turn, under the “intellectual challenge” understand the problem which people solve using your natural intelligence.

Easy to see that such a definition of artificial intelligence is extremely broad — it covers even the basic calculator because the arithmetic task, in fact, too intelligent, a man decides them with his intellect.

Therefore, within the concept of “artificial intelligence” was an important boundary that distinguishes applied or, as they say, “weak” artificial intelligence, designed to solve any intellectual problem or small, from a hypothetical strong AI, also known as universal artificial intelligence (eng. — Artificial general intelligence).

This system, once established, will be able to solve unlimited a wide range of intellectual tasks like human intelligence. From this point of view, the calculator is able to count much faster than a human, or a program that benefit from human chess is an application AI, the same hypothetical super-intelligence of the future strong AI.

When you read about the different discovery and development in the field of AI, you know that basically all is happening in the US or Asia. How are things in Russia? We have any developments?

— Area of computer science in our days is international, many of our specialists work on the creation and development of various machine learning models, consisting of both Russian and international teams. We have traditionally strong mathematical and algorithmic school, established research centres of world level as in the leading universities and some private companies.

But let’s face it — budgets in our country on science and education, do not go to any comparison with the research budgets of most developed countries. Budget revenues of the Russian Federation in 2016 amounted to about 200 billion U.S. dollars, while as soon as the defense spending amount three times more than the entire Russian budget.

The entire budget of the Russian science is comparable with the budget of only one University from the ivy League. The cash-strapped 90s the country has left many top professionals, continuity was broken a number of scientific schools. It was also almost lost his own electronics manufacturing.

At that time, as the world’s it leaders are in a race to create specialized processors for training neural networks, the development of algorithms and software. However, in this area the groups had achieved very impressive successes.

For example, a team led by Artem Oganov has created a system USPEX is capable of predicting the crystal structure of chemical compounds, which led to a real revolution in modern chemistry.

The team of Vladimir Makhnicheva and Victor Zakharov Moscow state University computational mathematics and Cybernetics, with systems created by them, as well as supercomputers Lomonosov and IBM Blue Gene/P, to calculate a 7-figure chess endings.

Neural network “Yandex” recognize and synthesize speech, generate music in the style of “Civil defense” and the composer Scriabin. A strong team of specialists in AI and machine learning are created and in the “Sberbank”.

In short, the notable success we have in the country.

The faster evolving technology of artificial intelligence, the more people captures the fear and how quickly they will be left without work. Is it really that bad?

Yes and no. Humanity has repeatedly faced with the emergence of technologies that revolutionary changed the entire manufacturing sector.

So it was with a steam engine during the industrial revolution, virtually destroying many jobs (mostly related to primitive physical labor), so it was with the electronic computing machines that have replaced human tasks based on the production of mathematical calculations.

In the XV—XVIII centuries of ecah when in England “sheep ate people”, the social consequences were really catastrophic. England lost by different estimations, from 7 to 30% of its population. The power elite of the time were seriously concerned about what to do with extra people. Jonathan swift responded to the quest of a humorous pamphlet, in which he proposed to use poor children as food.

Nowadays, however, we see that instead of extinct professions came new, and the Earth’s population far more than in the eighteenth century. In the twentieth century the consequences of automation was not as catastrophic from a social point of view. However, to underestimate the risk of still not worth it.

“In 30 years, robots will be able to do almost everything that people can do, — such forecast was made by Moshe Vardi (Moshe Vardi), Professor of computational engineering and Director of the Institute of information technology Ken Kennedy (Ken Kennedy Institute for Information Technology) at rice University (William Marsh Rice University). This leads to the fact that more than 50% of the world population will become unemployed”.

Robots take work

The other day the Chairman of the Duma Committee on information policy, information technologies and communications Leonid Levin said that Russia is an important problem of displacement of labour by artificial intelligence.

Sooner or later people will replace the automated system, and the market will spill 2% of the working population of the country. That’s why about how they employ, those who lose their jobs due to the development of digital technologies, we need to think now said Levin.

According to the Chairman, in the near future we will face rising unemployment. But whether or not robots will “take away” our jobs and should I be concerned about it said “Reedus,” a specialist in machine learning Sergey Markov.

-Sergei, even now, is already “dead profession” that require no human labor, although it would seem, 10 years ago, no one thought that, for example, the conductor will soon become unnecessary. What other profession will supplant technology?

We are approaching the time when machines surpass humans in almost any case. I believe that society needs to confront this problem before it will stand upright. If machines can do almost everything that able people that will do? said Moshe Vardi, Professor in computational engineering and Director of the Institute of information technology the Ken Kennedy at rice University.

For a long time on the road to automation was standing technological limitations of the machine could not recognize images and speech, could not speak, could not well enough to understand the meaning of statements in natural language, did not have sufficient data in order to learn many of the usual human things.

Thanks to recent advances in the field of artificial intelligence many of these restrictions were actually removed. In addition, many professions themselves have undergone a transformation, making them more convenient for automation.

For example, a modern office clerk is in correspondence not on paper, but in electronic form, the accountant performs a transaction not on paper, but in the accounting program, the machine operator controls the machine is often not with arms, but with the help of the control program. So now the task of automation in many professions ceased to be scientific and became a purely engineering.

The truth is that the production area is associated with AI, but rather creates jobs — we need experts in the field of machine learning and data preparation staff for the layout of the training arrays, the introduction etc. But at some point electrohouse will definitely start eating people, and the consequences need to be taken care of now.

It is important to understand that stopping technological progress is impossible, and the attempt to do so will result in much more catastrophic consequences.

— Can we ever fully trust the robots (AI), or still in any case must be the human factor?

This issue has several aspects. On the one hand, people in the past were beware of almost any vehicle. The first Elevator, the first car of the first train or plane — all that was once unusual, it seemed to many threat. Yes, largely dangerous it was — man-made disasters have claimed many lives.

Nevertheless, in our days these things become commonplace and no longer arouse strong fear. In this sense, our descendants will treat AI systems more easily. People sometimes tend to mystify things they do not understand. The savage thinks that the engine lives the evil spirit and modern man in the street thinks that our AI systems are conscious, although it is not so.

On the other hand, I don’t think that a universal system of AI will ever be a part of our production sector. In my opinion the future is more for synthetic systems — that is, a Union of man and machine into a single organism. In this sense, the artificial intelligence of the future will be enhanced human intelligence.

By the way, human intelligence is not quite correct to call natural. A child from birth has no intelligence, all it teaches society, parents, environment. In this sense, we are all essentially “artificial intelligence”, and our fears related to AI, are largely the fears themselves.

Recently, many scientists, e.g. Stephen Hawking, bill gates or the same Elon Musk, began to panic that the AI would doom humanity to destruction, and the future they see some kind of dystopia. Whether to take such predictions seriously?

Honestly, I wouldn’t be seriously afraid of these statements. Stephen Hawking, of course, is not a specialist in the field of AI, as, in General, and Elon Musk.

On the other scale statements of people, such as Andrew ng is an American computer science researcher, associate Professor at Stanford University, researcher, robotics and machine learning, leading specialist of the laboratory of artificial intelligence of the Chinese Corporation Baidu.

Eun, speaking about the problem of AI safety, compares it with the problem of overpopulation of Mars — of course, we sometime we colonize Mars, and then maybe at some point there will arise the problem of overpopulation. But should we do it today?

Mark Zuckerberg also quite skeptical of the statements made Mask. “Artificial intelligence will in the future make our lives better, and to predict the end of the world is very irresponsible,” he said.

Personally, I think saying the Mask should be considered in a pragmatic manner — Musk wants to stake out this topic and, ideally, to receive from the state the means for its development.

Is it all so rosy and nothing to worry about?

The real danger associated with the development of AI, are, in my opinion, in a completely different plane than it used to be. The main risks are connected not with the fact that we create Skynet, which will enslave mankind. Risks from the introduction of technologies of AI and machine learning is much more prosaic.

Trusting the important issues of the different mathematical models, we can suffer from errors made during their development. Artificial intelligence, reproducing the actions of people-experts, will inherit their mistakes and prejudices. Shortcomings in the management systems of production or transport can lead to disaster.

Interference of criminals in the work of the vital systems in terms of total automation may lead to dangerous consequences. The more complex the system, the more they can be potential vulnerabilities, including those associated with the specifics of those or other artificial intelligence algorithms.

Of course, to manage these risks should create a legal framework, reasonable safety regulations, special techniques to identify vulnerabilities. Some AI systems will be used to control other. Perhaps code vital systems will be required to be published for an independent audit. In short, the experts in this field have a lot of work.

Related posts:
The Americans accused the cats in approaching environmental disaster
China launched a satellite relay data "Tanglang-1"
Scientists told about the beneficial effects of sex on sleep
Who said that increases the risk of seven cancers


More Stories From Science