Center for Strategic Assessment and forecasts

Autonomous non-profit organization

Home / Science and Society / Analytical work: the experience of Russian and foreign experts / Articles
Hunting for the negative
Material posted: Publication date: 20-03-2019
Artificial intelligence will put you in charge of social media.

If you do not live in social networks, then you probably do not live. Such a disappointing diagnosis you can supply half the world's population. In any case, as according to statistics, today spelled out in social networks. In Russia, the number of networkers is about 30 million people.

About network dependencies, which is compared with a drug, psychologists have written volumes. And the author of the term "virtual reality", by one of its creators, a representative of liberality Silicon valley Jaron Lanier even wrote a book on why you need to escape from the social network. Although it became a bestseller, these calls are unlikely to convince the broad masses of users. Today, the population of social networks is growing rapidly. According to forecasts, by mid-century they span almost the entire population of the planet. Of course, if mankind will not repent.

And to think there is something. In networks today there are clubs not only stamp collectors and theater-goers, but suicides and drug addicts, you can find instructions for beginners terrorists, supporters against the current government and, frankly, negative. For example, you can remember the wave of suicides when there was a boom in one popular multiplayer game, the final goal of which is suicide. With social network users (mostly teenagers) binds "curator" who uses fake accounts, they cannot be identified. First, they explain the rules: "not to tell anyone about this game, always performing the tasks, whatever they were", "for non-job to expel you from the game forever and you expect bad consequences".

And politicians are already talking about the phenomenon of networks: they overthrow the government and even choose presidents. The techniques are surprisingly simple. Network thrown fakes about these or other outrageous actions of the government and there are calls to take to the streets to Express outrage. There are many examples in, say, Egypt, South Sudan, Myanmar.

How to protect society from destructive influence of social networks? This is especially true of the younger generation, which is the vast majority.

In principle, for specialists, particularly psychologists, is not a problem. But the volume of diverse information, including negative, in the networks is huge, with its "manual" processing to handle even the best professionals. Is only technique. And above all artificial intelligence (AI) based on neural networks that mimic the structure and function of the brain. For the solution of this problem came from the team of psychologists of the First St. Petersburg state medical University. Pavlov and specialists in the field of information technologies of Saint-Petersburg Institute for Informatics and automation of RAS.

- It is the identification with AI of different groups of people who are inclined to destructive behavior, - told correspondent "RG" Professor Igor Kotenko, who heads the group of scientists. - Psychologists have offered us seven options for this behavior, in particular, is a negative attitude to life, which can lead to suicide, it's propensity for aggression or protests, threats against a specific person or group of persons, etc. Our goal is to create AI that will be able to work with these signs, figuratively speaking, to "score" them in AI so that he constantly analyze huge information going over the network, to catch different alarming options.

At first glance, the task is not very difficult. After all, every resident of the social network, that is, on sight. On his page it's obvious what he's interested in what he thinks, is talking to whom, what are the actions he called for in others. All at a glance. Watch and take action. But if it was so obvious... the fact that thoughts and behavior in most cases is not as simple and straightforward as it might seem. And in General, the devil is in the details. And where everything is obvious to the expert, the AI may be wrong. So, canadian scientists taught the AI to distinguish a coarse joke or "hate speech" from just insulting. The system was quite imperfect, for example, passed many clearly racist comments.

In short, it all depends on how the AI train. And learning system on examples just like how a baby learns. After all, he did not explain that the cats here are whiskers, these ears, this tail. He often show an animal and say it's a cat. And then he repeatedly repeats this lesson looking at another cat.

Artificial intelligence will lead to a negative situation, and the last word for experts

As they learn and scientists St. Petersburg created a neural network. They will present various options for destructive behavior. Keeping in their "brains" developed by psychologists, the signs, the AI will be to remember them, then to catch in a large flow of social media. Professor Kotenko stressed that the task of the AI to make a guide on destructive, and the last word for experts. Only they can tell, the situation is alarming and it is urgent to take action or a false alarm.

Yuri Medvedev


RELATED MATERIALS: Science and Society