AI will be accepted into United Russia and taught to write laws

The world revolution came from where they did not expect. Artificial intelligence has made an unprecedented breakthrough, even according to the boldest expectations of analysts. Experts talk about a revolutionary change in the model of the world due to the widespread introduction of AI. Neural networks are already everywhere: they issue loans, teach schoolchildren to write essays, and come up with catchy headlines for the media. Even politicians found themselves in the “networks” of the machine mind. United Russia is going to use AI for the 2023 regional elections. But experts look at the brave new world, entangled in neural networks, with anxiety.

Without the use of artificial intelligence technologies, it will be impossible to win now. Such a statement at the start of the election campaign in the regions was made by the deputy. Secretary of the General Council of “United Russia” Sergei Perminov. Party members plan to use AI in work with the electorate. According to Perminov, neural networks will help attract new supporters and present the party’s plans. United Russia will work with available public developments in the field of AI. At the moment, we can talk about foreign neural networks ChatGPT 4 and Midjourney, as well as domestic Kandinsky 2.0. The latter will just help with pictures, and the first two can do analytics and make forecasts.

In addition, the party wants to use AI “for creative prompts”, including a technological assault, which such systems cope with in a matter of seconds, ”RBC quotes Sergey Perminov as saying. Recall that earlier the Liberal Democratic Party announced plans to create the Zhirinovsky neural network. As planned, the entire array of works and speeches by Vladimir Volfovich will be loaded into AI. If the project is implemented, the late politician will find a second life in digital and, perhaps, will delight us with his unmistakable forecasts (or, more likely, scare us) more than once.

The prospects and, most importantly, the risks of such a large-scale use of AI were discussed on Monday at the site of the Federation Council. Senator Irina Rukavishnikova recalled, for example, that even Vladimir Putin spoke about the role of artificial intelligence for security and ensuring the sovereignty of the country. In the planning horizon for the next 10 years, Russia’s goal is to massively introduce AI into all spheres of life and the economy, the senator emphasized.

So, for example, AI may well give free legal advice instead of lawyers. But so far only for simple standard cases. Difficult ones will still have to be drawn out by the lawyer himself: the neural network does not yet have the necessary experience. Plus, AI has no prejudices, no emotions, no empathy. Whether that’s good or bad is up to the customers to decide. But experts and senators have already come to a common denominator: the main thing is to prevent AI from making decisions. In law, politics, medicine – anywhere.

Moreover, machine intelligence seems to be able to replace even the lawmakers themselves. Not the fact that completely. But the concept of “machine-designed legal acts” in the discussion in the upper house sounded active.

Pavel Elsukov, an expert at the National Center for the Development of Artificial Intelligence under the Government of the Russian Federation, complained that now many laws and documents of a lower rank from the pen of “live” officials and parliamentarians are simply unreadable. If a machine creates them, maybe the result will be better. But there is a nuance: in order for AI to write a high-quality law, it must be taught this. And how to teach if the “food base” for the neural network, to put it mildly, is not entirely useful?

“There is a risk that we will get a “trash in – trash out” situation (garbage at the entrance – garbage at the exit), Pavel Elsukov explained. – It is necessary to carefully select the material that is given to artificial intelligence for training.

But poorly written laws are far from the main danger posed by AI. According to Roman Torshin, adviser to the Commissioner for Human Rights in the Russian Federation, the GPT-4 neural network, the most famous today, has already managed to seriously violate human rights several times due to the spread of outright fakes. For example, she hung accusations on innocent people and gave it out in response to user requests, as a result of which the victims filed lawsuits against the network administration.

— There is a risk of manipulation of public consciousness and political censorship. I asked the neural network about Trump and Biden, and the answers were radically different due to political censorship in the United States. The risk is also the uncontrolled explosive growth of the audience of neural networks. The number of users is increasing every month. As a result, GPT-4 is approaching the Internet itself in terms of audience coverage, and the neural network itself has turned into a powerful mass media,” Torshin said.

At the same time, AI has already learned to cope with non-standard creative tasks. As Roman Torshin said, for the sake of a joke, the ombudsman’s office asked the neural network to analyze children’s fairy tales, compare them with basic human rights and find exactly those works in which certain rights are best displayed.

Here’s what happened: “Kolobok” – the right to freedom of movement. “Little Red Riding Hood” – the right to protection from violence. “Golden Cockerel” – the right to freedom of thought.

In addition, the massive introduction of AI will inevitably reformat the labor market, the Ombudsman’s adviser stated. Moreover, the neural networks themselves do not hide who exactly they can “move”. When asked by analysts, AI replied that advertisers, marketers and tutors would be the first to disappear. But there will be a demand for trainers for AI itself. Within the walls of the Federation Council, an idea was voiced to create a pool of mentors for machine intelligence from experienced journalists and copywriters. The goal is to teach AI not to lie and not to hang frank noodles from “deepfakes” that cannot be distinguished from the truth on the audience’s ears.