The neural network will do things quickly, efficiently and impartially. Including on a national scale. The programmers promised to raise the high art of politics to unheard-of heights.
Recently, the Japanese authorities announced that they intend to introduce artificial intelligence (AI) into the political decision-making system. To begin with, innovation will touch on defense, national security and trade, and then, you see, it will reach the rest. Ours are also not lagging behind the advanced Japanese - a week later they announced a collection of seven departments (the Ministry of Energy, the Ministry of Industry and Trade, the Ministry of Culture, the Ministry of Education and Science, Rosreestr, Rosselkhoznadzor and the Federal Tax Service) and a review of AI projects for each of them. And in Ukraine, they even suggested using neural networks to fight lobbyists in the Verkhovna Rada. In order, therefore, to quickly identify, and then (what the hell is not kidding), maybe even to stop.
The rest of the civilized world, too, apparently does not snap its beak. The Pentagon is allocating nearly a billion dollars for the use of AI for military purposes. In China, he is encouraged to predict cyberattacks, rallies and search for "persons who undermine stability." In New Zealand, they boast that they are creating the world's first virtual politician, although in the same Japan they have already done this long ago. In 2018, the virtual robot Michihito Matsuda officially competed for the post of mayor of the city of Tama and even took an honorable third place with 4,000 votes. AI in the shell of a robogirl (her appearance was bought from the international stock of 3D models for a democratic $ 179) promised the residents of Tama justice and positive changes, a future-oriented policy, as well as taking into account their wishes and constructive dialogue - in general, everything that promises in such cases, the average warm-blooded fighter for the power seat.
Except that the candidate's appearance did not look very standard, and the desire to mathematically analyze the petitions submitted to the municipality. By the way, Michihito still maintains his Twitter (@tama_ai_mayor) and website on the Internet and, it seems, is amassing his political experience with might and main to participate in the new battle for the post.
Elections in general turned out to be a fertile field for the use of AI. For example, in 2012, he helped Barack Obama to lead the White House. The AI tirelessly (66,000 times a day!) Ran voter information through simulated elections, figuring out whether a particular person would come, vote for Obama, and, if not, change his mind after the conversation. Its theme, of course, was also picked up by the AI. As a result, the volunteers knew exactly who it was worth spending their time on, and to whom it made no sense to meddle.
Trump is said to have helped with the election, too. Privately held Cambridge Analytica, which took part in more than 40-plus large-scale votes (including Brexit and Kenya's presidential election), attributed its success to its "revolutionary approach to data-driven communication." Having gained access to 87 million Facebook profiles (how - for a long time to tell), the AI very quickly realized which strings of the soul should be played in each specific case in order to win over a potential voter in the right direction. For example, people with high anxiety were given to understand that the candidate would defend the right of everyone to bear arms, and people with a heightened sense of justice were slipped information about his opponent, who had stupidly spent charitable millions on absolutely useless things.
But if everything is more or less clear with the elections (after all, everyone has long been accustomed to the streams of molasses and incriminating evidence pouring on voters at this time, and the majority have already managed to develop immunity), then what about other aspects of big politics, if you give them at the mercy of AI?
“As far as I know, we already have AI systems to support medical and managerial decision making, and all this works well,” says political scientist Igor Lavrov. - Will it be the same with politics? I do not know. For example, every president of a large country has probably at least once in his life faced the dilemma “bomb not bomb”. Now such decisions are made on the basis of advice from a dozen specialized “top people” and experts. Many also turn to their priest about this, and go to their wife for advice. But the problem is that these decisions are often made emotionally anyway.
You know, at the time of Kennedy in the United States there was such a Secretary of Defense, Robert McNamara, who tried to apply the "methods of smart statistics" to the reality of that time. I took a bunch of historical data and looked from the standpoint of modernity, which decisions were correct and which were not. And so he, being essentially the brain of the Vietnam War, eventually resigned because of the horrors of the machine that he himself helped create.
Later, he admitted that politics is such a complex thing that the human brain is not able to adequately assess all existing circumstances and future consequences. And in this regard, AI, probably, could help politicians. But the problem is that AI is created by people, it depends on them what the machine considers good and what is bad, which means that the objective picture will never be completely objective. The political scientist did not see the benefits of politicized AI and for the planet as a whole:
- Every country that will adopt this technology will program AI to protect the interests of that country. That is, the global situation will not change in any way - there will be the same confrontation between the powers.
And, by the way, I personally would be much more comfortable if AI was developed, say, not by Japan, but by some more peaceful country. From time immemorial, it was a rather aggressive state with colossal ambitions. Of course, a lot has changed in the post-war years, but Japanese culture is built on military glory, and there is no getting away from it. So an involuntary question arises - what kind of political decisions will be made there with the help of AI? Especially if you remember their claims to our islands.
Social psychologist Alexei Roshchin saw a cunning plan in the politicization of AI:
- It seems to me that we are dealing with a way of curtailing democracy. They are trying to convince us that political decisions will now be made not by people who may be corrupt or not very competent (which means that these decisions can be challenged), but by some kind of omniscient and unbiased AI. That is, it will become much more difficult to challenge the decision - the machine has counted, it knows better. And, I think, we will very soon be faced with the fact that wishes of very specific people will be given out for computer conclusions, who will thus avoid responsibility and discussion of these very conclusions. They will tell us: AI has calculated that an urgent need to double the prices for housing and communal services. And there is nothing to argue, because the piece of iron sees a thousand moves ahead. In addition, where another person is stopped by intuition, a thorough knowledge of all non-obvious realities or taking into account subtle psychological moments, AI can easily go ahead and break the woods.
But you don't have to make a dull sawmill out of a high-tech product. If you use it exclusively as a source of knowledge in a particular area. Typically, coaching a productive AI occurs in two stages. First, they create a so-called pre-trained model - not a very big AI mind, able to answer not very confused questions quickly and accurately.
Something like the talkative columns of Yandex. Station, Google Home or Amazon Echo, which have gotten crazy on open data.
Then our homebrew smart guy is sent to some ministry or department, where they are already pumping up special data on the profile of the institution, after which the sharply wiser AI is shoved into the devices of officials, each of whom receives a trouble-free assistant analyst as a result.
- Why is AI called intelligence? - asks Laurent Hakobyan, head of the MIPT Intelligent Cryptographic Systems Laboratory. - Because, just like your human intelligence, it can be deployed wherever you want. Whether in the column, even in the smartphone, even on the walls or somewhere else. AI is not possible without the Internet of Things, the idea of which is that ordinary objects become interactive. The minister or prime minister has no time to turn on the computer, scour the sites or wait for the referents to study the problem.
Therefore, it will be easier, faster and more efficient for them to ask an AI question and instantly receive an exhaustive answer to it, if desired, with the output of the necessary numbers, tables and graphs on the walls of the office. Moreover, this answer will be based not on someone's subjective conclusions, but on accurate data. Let me explain. A big politician makes decisions based on personal experience and the reports that are provided to him.
Let's say Trump didn't believe in global warming, but then changed his mind. Apparently, he got acquainted with the report from people whom he trusts. AI, before giving an answer whether there is global warming, will analyze all the reports that exist in the world.
At the same time, he will ignore interpretations and retellings, and will take into account only the data obtained during the experiments. Moreover, if the experiment was carried out incorrectly, say, the measurements were carried out incorrectly, these data will not be taken into account. It turns out that AI will reject all the hype, heresy and information noise, adopting only objective reality.
But here we again run into the issue of training. Who will teach AI to understand what is good and what is bad, which sources can be trusted and which cannot? The same people (they are called assessors) who may have their own subjective views on the subject.
“In order to mark up the data, you need a lot of assessors,” Hakobyan explains. - The more of them, the more they are distributed around the world, the more correct the markup. After all, it is easier to buy up all the media on Earth than to bribe 2000 people who are not connected with each other in different parts of the planet. Besides, when creating an AI, it makes no sense to make it initially deceitful. Therefore, most likely he will tell the president the truth. Another question is how he will act in the end.
Now, according to Laurent Hakobyan, they are learning to listen to the truth at least in the municipalities of Yekaterinburg, Tula and Novosibirsk. If the experiment is successful, it will be extended further:
- It seems to me that it is worth distinguishing between an electoral policy, when a person is fighting for power, and a regional one. In the first case, the use of AI can hardly be called a good deed. In the second, the digital assistant can become a good help not only to the manager, but also to all administration staff, because it will allow them to work more efficiently. For example, do you know that finding unfinished buildings in Siberia is a difficult logistical task? They drive airplanes for reconnaissance, survey the taiga. AI, having analyzed the construction documentation, can answer the question instantly.
Or why are most politicians over 50? Because it takes tens of years for a person to accumulate experience, to build the necessary neural connections in the head.
AI neural networks can accumulate more experience in two years than a person in his entire life. In general, here, as with any thing: in someone's hands, AI can become an instrument of evil and suppression, and in someone else's - an excellent assistant working for the benefit of people. It remains to figure out how to put it in the right hands.
NOT PEOPLE IN GOVERNMENT
In 1964, Nobody claimed the presidency in the United States. His campaign posters announced that he was the most honest of all politicians: “Who will take care of you? Nobody!"
In 1997, residents of Tokitna, Alaska, elected Stubbs, a tailless cat, as mayor. In 2012, he was re-elected, and a year later, an attempt was made on the cat - the mayor was hospitalized with many serious wounds. Also in the official's record is an escape to the outskirts of the city in a garbage truck.
In 2014, a dog named Duke won the election for mayor of the urban settlement of Cormorant (Minnesota) and continued to be elected to this post until 2018, when he retired for health reasons. Residents appreciated Duke for his contribution to the safety of the city - he greeted strangers and passing cars with loud barks, forcing the latter to slow down.
BTW
In April, the Vatican, Microsoft and IBM unveiled a Code of Ethics for AI. In their opinion, among other things, he should respect the private life of a person, work reliably and impartially, and the principle of his decision-making should be transparent. At the same time, the pontiff emphasized the danger of using AI for commercial and political purposes.