Neuropolitical technologies – Newspaper Kommersant No. 67 (7512) of 04/18/2023

The Liberal Democratic Party announced the launch of a project to create the Zhirinovsky neural network, which will become “the world’s first political algorithm.” The neural network, created on the basis of speeches and publications of a politician, will have to give forecasts and advice on managing the country. Other parliamentary parties, as Kommersant found out, are also trying to use the capabilities of artificial intelligence (AI). Experts assess the prospects of the LDPR project ambiguously, but consider it possible for parties to use neural networks, in particular during the election campaign.

The leader of the LDPR, Leonid Slutsky, spoke about the new project on April 6, on the anniversary of the death of the founder of the party. “This is an ambitious futuristic project, which is designed to preserve in the AI ​​shell the possibilities of that unique forecasting that Vladimir Volfovich possessed. Of course, we do not entertain the illusion that at the current level of AI development we will be able to recreate a person, especially a person of genius. But we will be approaching with months, with years, the solution of this problem,” said Mr. Slutsky.

The neural network is planned to be created on the basis of more than 18 thousand hours of interviews and speeches by Vladimir Zhirinovsky for more than 30 years, as well as 50 volumes of his writings. The development will be carried out by specialists from the University of World Civilizations (UMC) established by the politician. With the help of a neural network, the LDPR is going to write a party program in 2024. State Duma Deputy Vladimir Koshelev (LDPR) also suggested that by 2030 the algorithm will be connected to all available big data and “become a full participant in the country’s governance – a consultant and assistant to the head of state.”

Deputy Chairman of the State Duma Committee on Information Policy Andrey Svintsov (LDPR) explained to Kommersant that Leonid Slutsky would personally supervise the project, and volunteers, UMC students, would do the basic work, that is, digitization. “The team of the algorithm itself are young programmers from large IT companies. We finance the creation of the alpha version through private donations from supporters of the Liberal Democratic Party. About the rest I can’t say yet, except that the scale will surprise you. Will be cool!” the deputy promised.

Experts interviewed by Kommersant consider the creation of the Zhirinovsky neural network to be a very real task, but they doubt that it will be able to produce unique ideas and impromptu, which the late LDPR leader was famous for. “In general, it’s not a problem to create something like this, you need two things: a text model and a voice,” says Vladimir Sveshnikov, co-founder of the Robot Vera company. , which formed the basis of the first version of the popular ChatGPT bot; the model consists of a large number of neural networks that can process, understand, analyze, interpret and generate texts in natural language.— “b”). You can make it generate something in the style of Zhirinovsky. Of course, there will not be any super-unique ideas. But the fact that it will be very similar and it will be possible to mislead people with this thing is one hundred percent. No problem and make a voice. But it is difficult to say whether the neural network will be able to produce super-impromptu, as Zhirinovsky himself did: this is an area not particularly studied.

According to Mr. Sveshnikov, several thousand dollars are enough to make a voice synthesizing simple phrases. Solutions for the text model are also available in the public domain: for example, you can take the development of Sberbank as a basis. The expert believes that parties can use neural networks in the preparation of any text materials, even campaign verses. Also, AI can help in the processing of citizens’ appeals – both at the State Services and in banks, where requests are already partially processed by the machine.

IT expert, associate professor at the Higher School of Economics Oleg Artamonov is skeptical about the LDPR project. “It is important to understand that AI is not sentient in the human sense. None of the existing systems, including the most powerful ones, such as ChatGPT and GPT-4, has consciousness, even the most minimal signs of it,” the expert explains. “A talking AI is just an algorithm that is very good at predicting which word in text should be as follows. He was taught this by downloading hundreds of gigabytes of text – thousands of times more than a person would read in a lifetime. He answers even complex logical questions by simply sorting through words, and quite often gives rise to so-called hallucinations, that is, outwardly looking convincing, but inside they are completely false statements. AI does not have the ability to plan or predict, it is only able to endlessly reassemble what was loaded into it during training. Therefore, what the Liberal Democratic Party will get will be nothing more than a parrot with a huge amount of memory, remembering thousands of phrases and able to take these phrases into pieces and glue them in a different order. This, of course, can be a fun fair entertainment, but it has nothing to do with politics and will not have anything to do with it.

A member of the State Duma Committee on Information Policy, coordinator of the federal party project “Digital Russia” Anton Nemkin (“United Russia”), on the contrary, treats the plans of colleagues from the Liberal Democratic Party with interest and believes that their developments will serve science. “We are now witnessing a real technology race in the field of AI, and the winner will be the one who invests the most resources. If the UMC can develop a new efficient model and test it, then this will go beyond the specific case of preserving the legacy of a major political figure and can be very useful for creating new trends in the development of the Russian IT industry,” the deputy believes. He also sees many opportunities for using AI in party work, from applied ones (automating knowledge bases, training election observers) to more conceptual ones, such as developing algorithms for reports, programs, and perhaps even campaign materials.

Other parties are also trying to “make friends” with the AI. Thus, the Communist Party of the Russian Federation holds a series of seminars where the possibilities of using AI in election campaigns are discussed: on April 7, such an event was held in Yakutsk, on April 14 – in Ulyanovsk, Ivanovo is next in line. Following the results of the seminar in Ulyanovsk, Secretary of the Central Committee of the Communist Party of the Russian Federation, State Duma deputy Sergei Obukhov wrote on the Telegram channel that so far “lawyers are calling for caution”: “The party, of course, will actively use the creative developments of neural networks in the period before the elections. The use of AI for design, layout, chat bots for the accepted concepts of election campaigns should be actively pursued. But after the announcement of the elections, there are many risks when using neural network products.”

Vice Speaker of the State Duma from the New People, Vladislav Davankov, called the LDPR experiment interesting, but noted that “no neural network can convey Zhirinovsky’s sincere love for people, his originality and patriotism.” “Neural networks are able to successfully compile the products of creativity of different people, but you should not expect new ideas, empathy, and even more so a look beyond the horizon. I support the arrival of new technologies in politics, but neural networks will not be able to replace or even come close to public policies. I am sure that there will be serious discussions on the prospects for the development of language models, because they are capable of both changing reality and creating new threats,” said Mr. Davankov.

State Duma deputy, political strategist Dmitry Gusev (A Just Russia – For Truth) told Kommersant that he uses neural networks in his work every day, and his party will also use them in the election campaign. “When we do, we will tell about it,” Mr. Gusev promised.

Ksenia Veretennikova