Experts from the St. Petersburg Federal Research Center of the Russian Academy of Sciences (SPC RAS), a member of the Consortium for Research on Artificial Intelligence (AI) Security, took part in a foresight game and a strategic session to form the image of the future Russian registry of AI technologies trust.
Right now, the artificial intelligence technologies rapidly get into various spheres of life and economy of all globally leading countries, including: industry, healthcare, finance, transport, education and many other areas. Artificial intelligence allows for automating processes, improving work efficiency, analyzing large amounts of data and making better reasoned decisions.
However, the widespread use of AI technologies bears many threats that should be accounted for at the level of relevant government agencies. Say, many AI algorithms (especially deep neural networks) are essentially “black boxes” - this means that they can make decisions that cannot be accurately predicted. AI technologies also depend on the correctness of the training data, ethical justification for the use of artificial intelligence, etc. Overall, all these problems come down to the quite a fragmented formal regulation at the AI technologies use.
To develop proposals for arranging the Russian Registry of AI Technologies Trust at the «MIREA — Russian Technological University» (RTU MIREA, Moscow), experts engaged in the Consortium for Research on AI Security held a strategic session and foresight game. The leading experts of SPC RAS, as well as of the following companies Gazinformservice, Yandex, Positive Technologies, Gazprombank, Code of Security, Foundation for Assisting the Development of Secure Information Technologies, National Technological Center for Digital Cryptography, Raft Digital Solutions and Swordfish Security were among the participants.
One of the session's moderators was Andrey Chechulin, Head of the International Center for Digital Forensics at SPC RAS.
"The implementation of AI technologies can no longer be stopped, since the current efficiency of entire sectors of the economy depends on them. However, artificial intelligence can work not only for the good, but also to the advantage of malefactors. Therefore, we must learn not only to control as well as to distinguish dangerous AI technologies from the trusted ones. As part of the event, the experts did their best to compile a first approximation for the concept of what requirements artificial intelligence must meet so that it can be trusted. The ultimate goal of our work is to draw up a detailed register of the AI technologies trust," explains Andrey Chechulin.
After discussing the key aspects of the future register, the experts started working in the format of a foresight game. This approach allowed for modeling various scenarios regarding the register's development and proposing recommendations for its effective functioning. Splitted up into teams, the experts analyzed the interaction of all stakeholders: developers, suppliers, users, regulators, and information security specialists. The use of specially selected roles ensured a comprehensive analysis of risks and opportunities.
The result of the work was the formation of primary principles and criteria that AI technologies must meet to be included in the register. Participants outlined key requirements for data, models and software, and developed the preliminary mechanisms for certification and evaluation of trust. The results received will form the basis for a Registry of AI Technologies Trust, expected to become in the future a tool for increasing the transparency, reliability and security of artificial intelligence in Russia.
Consortium for Research on AI Security was established on May 21, 2024, under the initiative of the National Technology Center for Digital Cryptography, the Academy of Cryptography of the Russian Federation and the V.P. Ivannikov Institute for System Programming of the Russian Academy of Sciences with the support and active participation of the Ministry of Digital Development of the Russian Federation. The objective of the Consortium is to consolidate the efforts of leading expert centers in the long-term development of AI technologies and the implementation of secure development principles to solve the problems set in the national projects of the Russian Federation.