The explosive introduction of neural networks will give work to thousands of humanitarians: they will teach the machine to speak

Nanny for a neural network – thousands of humanitarians can get such an unexpected specialization. And more recently, there was talk that the “lyrical” professions are dying. Editors, philologists, journalists, translators, copywriters and other pen workers will no longer be in demand with the advent of neural networks. But it turned out that AI still needs to be taught and taught so that a clear, reliable narrative can be obtained from the output content. Therefore, today the leading Russian IT companies, social networks and providers are preparing vacancies for writers. There is only one task – to teach electronic brains to compose meaningful texts.

The developers themselves call such computer tutors AI trainers (from Artificial Intelligence, artificial intelligence – Auth). Like, AI-training is the profession of the future. Today, their main task is to write reference texts that bots will be equal to in their creations.

“Almost everyone, probably, has already tried to work with a neural network out of curiosity,” says editor Svetlana Dmitrieva. – And they noticed that as a help they are still nothing, although they can confuse the facts. But they can’t work on their own. The bot reminded me of an obstinate, stubborn junior student of orthodox views. He knows something, still knows little, self-confident, likes to argue. And he refuses a number of topics out of hypocrisy.

“Large neural networks are trained on information from the Internet,” explains Alexei Gusakov, head of the machine intelligence and research department of the information company. And the Internet, as you know, is a “big garbage dump.” Not even every living person knows where to look for reliable verified facts – fake news and viral mailings are evidence of this. And what can you want from a newborn AI? ..

Therefore, many good quality content specialists are required to train neural networks. There are hundreds of vacancies in one company alone. Analysts of the labor market predict a rush demand for authors who write well in Russian and who are able to perform a clearly defined task. The expected salary for such specialists is from 150 thousand rubles.

“Everyone said that there would be an ambush in Russian,” says programmer Ivan Berezin. – If only because we have long words, “a lot of bukoff”. If you have already worked with the bot, you probably noticed that if you set a complex request, it often hangs. This is because of tokens, this is such a sequence of characters, an element of word formation in programming. There are 2 tokens in the same English word, 10 in Russian. But, of course, this is only part of the problem. Here are the usual difficulties of translation, because the bot is in English, it first translates from Russian into English, and then vice versa. Shades of meaning, allusion, humor are often lost with such a system. Not to mention phraseological turns, sometimes they are “misinterpreted” by the system in a very funny way.

By the way, about humor. In a professional public of IT specialists, companies are warned not to hire humanities students to train AI. And then they, “out of hooligan and gouging motives”, will teach this … “Teach the system everything and go nowhere, you are no longer needed,” adds another programmer. “This is how cashiers will train self-service checkouts.” Other IT employees note that it’s not so bad in their companies with the humanities: “there are different art directors, creatives, PR people also have a non-technical education.” There is nothing, they say, to recruit from the outside, we must use our own personnel.

Valentin Makarov, president of the Software Developers Association, points out another facet of the problem, the psychological one. He believes that in order to “bring AI closer to a person”, the competencies of sociologists, psychologists, philologists, even artists will also be needed … The same “emotional intelligence” is almost unknown to artificial intelligence. Bots and neural networks cannot read the emotion of a request and almost do not distinguish between connotations – what high-quality response texts can be without this?

Therefore, “the demand for humanitarians can be great and reach thousands of people.” “Once standards, sustainable business models and regulations for its use are developed, the hype will subside,” Makarov said. “But the need of the IT market for such specialists will remain.”

In general, it is clear that the case with the use of neural networks with AI in Russian is still completely new, unclear and dark. Experts are slow to judge whether a bot will ever get a hint or a joke. But he almost learned to understand Russian phraseological units.

For example, “MK” asked the bot about the meaning of the saying “to fish in troubled waters.” Answer: “Like a skilled fisherman who can fish in troubled waters, so is a person who can achieve his goals in difficult and incomprehensible situations. The expression means the ability to succeed in difficult conditions, when everything is unclear and confused. And even better AI coped with the rare proverb “Paris is noisy, but Kurmysh lives.” Without hesitation, the bot gave out a “decoding”: “This proverb means that great and famous cities can make noise and attract attention, but in fact life can be full and interesting in small and unknown places.”

Surprisingly, the AI ​​correctly understood the proverb “You can’t wash a white dog white” and even such a forgotten idiom as “saryn on a kitchka”. But still broke down on the expression “stream and plunder.” “This is a mistranslation of the English expression “The slump and the raid”, which describes the situation of the economic crisis,” writes the bot. And here is the lie. This is a type of punishment according to Russkaya Pravda from the 11th-12th centuries for serious crimes: murder, arson, robbery, horse theft, etc. It was expressed in the deprivation of all property, exile and sale into slavery, sometimes with the whole family. Explanatory dictionary “Literary speech”, 1933.