- Статьи
- Internet and technology
- Export opinion: speech technologies and computer vision have become the leading branches of AI science in the Russian Federation
Export opinion: speech technologies and computer vision have become the leading branches of AI science in the Russian Federation
The Russian Federation ranks third among the BRICS countries in terms of the number of publications at the world's leading conferences, and in terms of the quality of research, it is ahead of many countries with a much larger volume of work. Computer vision and speech technology have become particularly prominent areas. At the same time, there remains a gap between scientific developments and their practical application, in particular, in the field of AI agents and infrastructure. The Ministry of Finance told Izvestia that they intend to develop demand for domestic AI solutions and accelerate the transfer of scientific developments into practice.
Where has Russian AI proved to be the strongest
Russia ranked 20th among 195 countries in terms of the number of publications at ten leading international A* level conferences on machine learning and artificial intelligence, according to a study by the HSE Science Center. Such conferences are considered the most prestigious scientific events in the field of computer science: the share of rejected applications reaches 75-80%, and the papers presented at them largely determine the directions of development of the global industry.
Russia became the third among the BRICS countries after China and India. At the same time, the average international citation index (FWCI) for domestic research was 3.59, higher than that of India (3.10), despite a significantly smaller number of publications. In total, in 2020-2025, Russian organizations presented 560 studies that received more than 12.3 thousand citations.
— Russia is traditionally strong in theoretical computer science. In BRICS, she ranks first in the field of formal concept analysis, and second in 23 other topics. The core of these competencies is optimization and the mathematics of machine learning," explained the HSE.
The high position of our country testifies to the strength of the national scientific school, despite infrastructural restrictions, sanctions pressure and a shortage of computing power, says Marina Krasnova, managing partner of the Union Group.
"The fact that almost three quarters of the papers have been selected for the main programs of highly selective international conferences confirms the international recognition of the quality of the research presented," she noted.
The family of Russian automatic machine learning tools FEDOT is in demand in the world for building applied AI models. Over the past three years, the framework has been downloaded hundreds of millions of times by users from 42 countries. The United States, Russia and China are the leaders in terms of the number of uses, said Alexander Bukhanovsky, head of the ITMO Institute of AI.
— Over the past year, areas have appeared on the scientific agenda in which Russia does not yet have fundamental foundations, but they open up promising opportunities. These include computing technologies for AI, as well as economic, socio-humanitarian aspects of its application," he said.
Another feature of Russian AI science is the close cooperation between universities and technology companies. More than 40% of domestic publications were prepared with the participation of the corporate sector. Yandex is the leader in the number of publications among companies, HSE and Skoltech are among universities, and AI is the leader among non—profit organizations.
Priorities are gradually shifting from basic research to applied solutions. Marina Krasnova highlighted generative and language models, computer vision for industry and medicine, autonomous transport, robotics, optimization and compression of neural networks, as well as the use of AI in materials science, bioinformatics and geology among promising areas.
"Over the five years of AI's existence, more than 1,400 scientific publications have been prepared, 400 in 2025, more than 130 of them were included in the main programs of the A* and A conferences with the strictest selection, which indicates a focus on quality rather than quantity," said Ivan, Director General of the AI Institute, Dean of the Moscow State University Institute of Physics and Technology The oseledets.
From publications to products
The results of the HSE research are an important indicator of the demand for AI from business and government, according to the Ministry of Finance of the Russian Federation.
— It is necessary to create conditions for the accelerated transfer of the results of these studies into the field of practical application. To this end, the Ministry of Finance, together with interested departments, is working in several areas: improving AI regulation, developing computing infrastructure and staffing, as well as stimulating demand for domestic AI solutions, the ministry's press service told Izvestia.
It is difficult to single out one of the most promising areas: its relevance is primarily determined by the economic impact, said Valentin Malykh, head of the Center for the Development of large language models at MWS AI. The company relies on the development of agent-based capabilities of large language models. Among other promising areas, the expert names audio generation and computer vision.
Some domestic developments are already in demand abroad. Valentin Malykh cites VisionLabs face recognition solutions as an example. At the same time, international scientific collaborations and sanctions restrictions make it difficult to determine the nationality of some of the research, so the actual volume of Russian scientific activity may be higher than official figures.
One of the limitations for the development of AI remains data. IT entrepreneur Mikhail Lisetsky noted that the capabilities of the models depend on the volume, quality and degree of markup of the source information. Russian projects using mainly Russian-language sources have less raw material for tokenization, and obtaining high-quality commercial databases often requires significant costs.
— One of the weakest areas in AI science is natural language processing: in terms of the number of publications at ACL and EMNLP conferences, our country is inferior to the world leaders, and the citation rates here are lower than in fundamental machine learning or computer vision. Only a few business players are actively involved in research activities, while most companies prefer not to invest in AI science," the expert explained.
Sergey Markov, Director of Artificial Intelligence Technology Development at Sberbank, believes that it is important for the country to simultaneously develop its own large generative models, VLA systems for controlling robots and transport, as well as new computing architectures in order to maintain technological independence. According to him, Russian fundamental research in the field of mathematical foundations of deep learning is currently in high demand abroad.
AI is becoming an infrastructure industry
Another critical factor is computing power. Anton Urusov, executive Director of the Russian Direct Investment Fund (RDIF), believes that published statistics alone are no longer enough to assess technological leadership: the volume of research is growing so fast that the very passage of work at a leading conference provides less and less information about its potential.
In his opinion, the assessment should take into account the ability to turn scientific results into competitive models and technologies. Russian research schools are capable of operating at a global level, but the scale of the available computing infrastructure remains one of the limitations for the commercialization of developments.
"Talent is a necessary condition for success, but talent alone is no longer enough,— Anton Urusov added.
AI is becoming a capital-intensive infrastructure industry, which is why Russia needs long-term investments in computing power, personnel, and technological base. RDIF considers AI and related infrastructure as one of its key investment areas.
Russia's strong position can also be used to expand cooperation within the BRICS. Anna Nesterova, Chairman of the Russian part of the BRICS Women's Business Alliance, believes that the results of the study create the basis for the formation of the association's own scientific and technological space. According to her, our country can increase the exchange of competencies, joint research and interaction between scientific schools, laboratories and technology companies.
"This is an opportunity for Russia not only to strengthen its own competencies, but also to become one of the centers of attraction for scientific cooperation within the association," she noted.
Yandex also emphasized the relationship between science and business: the real tasks of the services form new research issues, and scientific developments help to improve products. Partnerships with universities and research centers make it possible to simultaneously develop technologies and a domestic AI school.
Переведено сервисом «Яндекс Переводчик»
