- Статьи
- Economy
- Goals and objectives: agencies have established standards for the implementation of neural networks
Goals and objectives: agencies have established standards for the implementation of neural networks
The Cabinet of Ministers has introduced standards for the implementation of AI for departments — the pace of development set by the authorities is low, since neural networks are already actively being implemented in business and government agencies. So, by the end of 2026, AI should process 25% of chest X—rays, 18% of manufacturing enterprises should use machine vision technologies, and the proportion of trucks detected by AI that violate the weight regime should be 5%, etc. Experts fear that in practice, everything may be limited to demonstrating successful pilot projects of departments. How exactly the targets will be achieved and why such low standards are set is described in the Izvestia article.
What standards for the introduction of AI should officials fulfill?
The Government's Subcommittee on AI has approved key performance indicators (KPIs) for departments and subordinate organizations for 2026. Such standards are being introduced in Russia for the first time, the office of Deputy Prime Minister Dmitry Grigorenko told Izvestia.
So far, KPIs have been established for 20 executive authorities. The indicators are divided into cross-cutting and industry-specific ones. Among the first criteria are the number of employees of departments and subordinate organizations who have access to generative AI services, the proportion of organizations in a particular industry that use artificial intelligence, employee satisfaction, and the artificial intelligence technology adoption index.
Izvestia got acquainted with examples of industry KPIs for the introduction of artificial intelligence by the Ministry of Health, the Ministry of Industry and Trade and the Ministry of Transport. Thus, by the end of 2026, 50% of mammography studies should be processed using AI in institutions subordinate to the Ministry of Health, compared to 30% at the beginning, 25% of chest X-rays (versus 10%), and AI-processed computed tomography studies — from 15 to 25%.
It also follows from the KPIs for the Ministry of Industry and Trade that 17.8% of machine—building enterprises should be equipped with computer vision systems, and 5.5% of chemical industry enterprises (versus 4.8% at the beginning of this year) should be equipped with intellectual decision support technologies, including for predicting the quality of finished products, according to the documents reviewed by Izvestia.
Also, for example, it follows from the KPIs for the Ministry of Transport that the share of freight transport identified by AI that violates the requirements for weight and dimensional parameters should reach 5% (neural networks are not currently used for this). For example, passenger traffic on unmanned trams, which also use AI, is planned to increase by 550% by the end of this year.
As the Ministry of Industry and Trade informed Izvestia, the agency is ready to implement the approved KPIs. Moreover, it was directly involved in their discussion.
—We plan to monitor the achievement of the indicators through a digital passport of an industrial enterprise implemented within the perimeter of the state information system of industry," they noted. — We expect that the presented KPIs will contribute to the popularization of AI technologies in industry, which will automate a number of routine tasks, increase labor productivity and industrial safety, as well as contribute to achieving Russia's technological leadership in the digital sphere.
Izvestia also sent requests to the Ministry of Health and the Ministry of Transport.
The introduction of KPIs will help assess the scale of AI penetration into key processes, as well as set the pace of technology adoption not only for 2026, but also for the next "five-year plan," Dmitry Grigorenko's office noted. The regions are actively involved in the development of KPIs, they said: in 2027-2030, the list of end-to-end KPIs will be expanded, and economic ones will be added to them.
For this purpose, five working groups have been formed by industry: "Social sphere, sports and tourism", "Education, science and Culture", "Infrastructure, transport and territorial development", "Real economy", "Digital economy, public administration and information security", the office of the Deputy Prime Minister specified.
For the government, KPIs for the implementation of AI are a tool for monitoring and strategic planning in key industries. For businesses, these key indicators serve as a guideline indicating the priority areas of technology application. As for citizens, the active introduction of artificial intelligence into the social sphere makes it possible to improve the quality of life of Russians, they added.
— For example, AI technologies in the healthcare sector help doctors analyze the results of examinations faster, speeding up patient care. The introduction of AI solutions in transport makes it possible to optimize flows and improve traffic safety," Dmitry Grigorenko's office said.
What KPIs for departments will give to ordinary citizens
The introduction of departmental and sectoral KPIs for citizens can mean an increase in the availability and quality of services, from medicine to transport, says Yaroslav Seliverstov, a leading expert in the field of AI at University 2035.
-- AI is able to speed up the diagnosis of diseases and improve traffic safety. An important effect will be the reduction of the human factor in routine processes, which reduces the likelihood of errors. Businesses get a more predictable regulatory environment and incentives to adopt technologies," he believes.
International experience shows that the elements of KPIs for AI are already being used, although not always in such a formalized form. For example, the European Commission sets targets for the introduction of AI in business and the public sector through the digital Decade programs, the expert said. The Organization for Economic Cooperation and Development has developed AI maturity indicators used to assess the policies of its member countries. In Singapore and the UK, KPIs are used related to the level of AI adoption in public services and the effectiveness of digital services.
— The Russian approach is notable for the fact that it introduces management KPIs for departments, which makes the system result—oriented, - said Yaroslav Seliverstov.
The introduction of indicators makes it possible to set up the work of departments to actively use AI both in their work and in the work of the industries they oversee, said Karen Ghazaryan, director of the Institute for Research at the Boarding School. This encourages removing restrictions on the introduction of AI in industries and attracting its developers, he noted. But there is also a temptation to set indicators so that the results are obviously achieved without serious effort, and then report on the implementation of precise, disparate pilot projects, he warned.
It can be assumed that AI indicators and their growth rates are low, since they are obviously achievable — departments will not have to make efforts to fulfill them, and the risk that goals will not be achieved is minimized, says Denis Kuskov, CEO of TelecomDaily. According to him, the industry and government authorities are actively implementing artificial intelligence to control various processes, manage personnel, automate office work and much more, the expert believes. In medicine, AI is increasingly being used to process analyses, medical records, and other things — the pace of technological progress is now such that there is little doubt that demand for artificial intelligence is expanding, he noted.
If a business sees the effect of any new technology, it will certainly invest in its implementation, says Leonid Konik, a partner at ComNews Research. According to him, there are several international ratings that are not related to states: the Global AI Index (Tortoise Media), the Global AI Vibration Tool (Stanford HAI) and the AI Preparedness Index (IMF), but these ratings are attempts to systematize market data rather than indicate the need to use certain technologies.
— It is strange to rank departments by the share of industry organizations that use AI. Moreover, with such a "large-block" task statement, one can blindly assert that in every industry 100% of companies use at least some kind of AI technology: a chatbot, a recommendation system, or at least a large language model for generating texts and presentations, Leonid Konik believes.
Industry KPIs make sense in areas focused on working with a mass customer. For example, in healthcare, it is really possible to accurately calculate the share of automated analysis of medical images or electronic medical records, and this will be a real reflection of the effect of AI, Leonid Konik concluded.
Переведено сервисом «Яндекс Переводчик»
