Woe from intelligence – Newspaper Kommersant No. 65 (7510) of 04/14/2023
[ad_1]
United Russia is preparing a bill to regulate artificial intelligence (AI). The document should define responsibility and exclude cases of use of AI by fraudsters. Kommersant’s sources suggest that within the framework of regulation, a ban on the use of AI in critical industries may be introduced. The authorities also plan to assess the risks to the labor market associated with the spread of technology. She is most concerned about the creative professions: voiceover artists, artists and musicians fear loss of work and copyright infringement. But companies are not ready to abandon neural networks that help optimize costs.
On April 13, Anton Gorelkin, Deputy Chairman of the State Duma Committee on Information Policy, said on United Russia’s Telegram channel that the faction is working on a bill that will regulate the activities of AI: “Together with the ministries, we plan to assess changes in the labor market associated with the spread of AI solutions . Now technologies that were supposed to save a person from unnecessary physical labor are trying to take over the solution of creative tasks instead.
The report says that there is no talk of banning neural networks: the bill should “update existing acts, determine responsibility and exclude the use of AI by fraudsters.”
The deputy’s office refused to clarify the details.
Meanwhile, the government is working rather in the opposite direction. The Ministry of Economy told Kommersant that, together with the Competence Center for the implementation of the federal project “Artificial Intelligence” (Sberbank) and the National Center for the Development of Artificial Intelligence under the Government of the Russian Federation (NRU HSE), on behalf of Deputy Prime Minister Dmitry Chernyshenko, “they are conducting a study to identify barriers to regulatory -technical regulation preventing the introduction of AI technologies. Based on the results, an “AI regulatory plan” will be prepared.
Other countries are also trying to regulate AI. For example, in the UK, the government proposed the creation of a closed pilot project, during which models of the practical application of AI legislation for business will be developed (see “Kommersant” dated March 29). China wants to test the security of generative AI systems before they are released. The US plans to develop regulations for AI. Italy and Germany introduced temporary restrictions in March on the use of the ChatGPT chatbot to process user data. First of all, we are talking about the regulation of advisory services that target and offer information to the user based on algorithms, says Alexandra Orekhovich, Director for Legal Initiatives of the Internet Initiatives Development Fund: “In addition, with the increasing popularity of neural networks, the question arises of the limits of their application. AI is able to influence the formation of user opinion and even, according to some experts, impose it.
A Kommersant source in the market believes that a ban on the use of AI in critical industries may be introduced: “For violations, fines can be set. In addition, a single state operator of the largest AI system can be selected to provide services to companies.”
There are several aspects that may be subject to regulation in terms of the distribution of neural networks operating on the ChatGPT model, says Alexander Zhuravlev, managing partner of the EBR law firm, in particular, “issues of automating the adoption of legally significant decisions, generating AI of illegal content and its distribution, for example, for preparation of dangerous deepfakes”. In addition, he notes, the issue of copyright infringement and data collection for AI training looks sensitive. Thus, one of the world’s largest record labels, Universal Music Group, demanded that streaming services block access to tracks hosted there for developers of AI-based chatbots, the Financial Times reported on April 13.
In the Russian Federation, AI tools are beginning to be actively introduced, for example, in the book market. So, “Strings” and “LitRes” are already using AI to dub audio books (see “Kommersant” dated January 23). This has already caused dissatisfaction among professional announcers: in January, their union proposed to the State Duma to establish regulation of voice synthesis using AI (see “Kommersant” dated January 27). In the House of Stories, they told Kommersant that they were going to republish the books of science fiction writer Sprague de Camp, the covers for which were made by AI. The industry believes that AI will help save on the services of illustrators. Boris Kuznetsov, director of the Rosmen publishing house, estimates the maximum cost of a cover at 50,000 rubles: “For the publishing market, this is a very serious amount.” However, he noted that there may be problems in the field of copyright: “They relate to the source code that uses neural networks, but in the future can be identified by the copyright holder.”
Sofya Sharypova, legal adviser in the intellectual property practice at the EBR law firm, notes that neural networks do use the works of artists for training, but do not copy them: “However, claims against AI developers are growing. For example, Getty Images has already filed two lawsuits against Stability AI (the creator of Stable Diffusion), and three artists have also filed a lawsuit against Stability AI, DeviantArt Inc and Midjourney.
AI developers themselves, however, are skeptical about the prospects for regulating the industry.
“It is very strange for me to hear when a number of government officials are trying to regulate AI, which is, in fact, matrix multiplication,” emphasizes Dmitry Markov, Executive Director of MTS AI. The situation, in his opinion, looks “as if, from the point of view of the law, they want to regulate how mathematically defined multiplication will work for us.”
[ad_2]
Source link
