On October 21, a meeting of the expert council of the General Radio Frequency Center (subordinate to Roskomnadzor) on artificial intelligence (AI) was held, where issues of ethical, legal and technical regulation of technology were discussed. The event was addressed by Yuri Hohlov, chairman of the Board of directors of the Institute of the Information Society, project leader for big data monitoring and standardization of the NTI Big Data Competence Center at Moscow State University. The expert spoke about how normative technical regulation (standardization) can solve problems in the field of AI.
Let us recall that Technical Committee 164 for standardization “Artificial Intelligence” (TC 164) – a mirror image of the international subcommittee ISO / IEC JTC 1 SC 42 Artificial Intelligence – has been established and is working in Russia. TC 164 includes subcommittee 02 “Data” (PC 02), headed by Yuri Hohlov and supported by the NTI Big Data Competence Center at MSU.
At the meeting, Yuri Hohlov named one of the main problems of AI systems and big data systems: the goals of creating an information system precede the development, operation and completion of the functioning of the system itself, and at the same time depend on ethical principles that guide the developers. Ideas and business requirements that are initially laid down in the development of a system are outside the system itself. If you look at the data life cycle in such systems – from the generation and acquisition of data sets to their preprocessing, processing and visualization, then ethical norms “live” outside this cycle, but indirectly strongly influence it.
At the international level, we are concerned about solving this problem: work begins on a new standard describing the life cycle of data in AI systems and big data analytics systems. Another series of standards currently under development in International Subcommittee 42 of the International Organization for Standardization and International Electrotechnical Commission # 1 (ISO / IEC JTC 1 SC 42 Artificial Intelligence) will address the quality of datasets for machine learning and big data analytics.
“The standard is, in fact, a fixed international consensus on the knowledge that we have in a specific subject area. The quintessence of this knowledge is laid down in standards, fixed in them. Not knowing the standards means that you will reinvent your own wheel, and not use the accumulated knowledge of all mankind. That is why it is very important to have this set of standards in Russia as well,” noted Yuri Hohlov.
For two years now, PC 02 Data has been preparing standards in the field of big data, harmonized with similar international documents, in order to overcome the existing lag in national standardization.
Another potential advantage of technical regulation is that it helps to protect domestic know-how. “If, while inventing a technology or solution, you put it into a standard and protect it with patents, then all organizations in the world that follow this standard will be forced to buy your solution or a patent for its use,” Yuri Hohlov stressed.
Alexander Degtyarev, head of Department of AI technologies development of the Directorate of digital technologies development of the Institute of the Information Society, professor of St. Petersburg University, expert of PC 02 “Data”, stated that today the definition of AI is rather vague, especially when drawing the line between the strong and the weak AI. For his part, the speaker believes that any problem for which the solution algorithm is unknown belongs to AI. At the same time, AI tasks are distinguished by the ability to learn, generalize, build analogies, as well as by the ability to interact with the outside world, its perception and awareness of the perceived. Contrary to popular belief, the expert believes that the technological components of AI include not only neural networks, but also methods of formalizing knowledge, logical reasoning, logical inference, as well as fuzzy logic, evolutionary computing and multi-agent systems.
According to Alexander Degtyarev, it is standardization that can solve the problems of defining and classifying AI.
Standardization is able to cope with another urgent problem – the problem of measurability of AI, the expert is sure. How can we measure AI? When does the moment come when the system becomes intellectual and not just informational? When does data become knowledge? What data can be considered anonymized? When does a lot of data turn into big data? Answering these questions will allow you to consciously apply big data systems or AI systems to solve pressing social and economic problems.
The event also featured speakers:
- Milos Wagner – Deputy Head of Roskomnadzor
- Denis Sadovnikov – Chief Legal Expert of the Main Radio Frequency Center
- Lyudmila Kurovskaya – Head of the Center for Assisting Citizens in the Digital Environment
- Alexey Neiman – Executive Director of the Big Data Association (ADB)
- Irina Levova – Director for Strategic Projects at ABD
- Maxim Fedorov – Rector of NTU “Sirius”, expert of a special group of experts of UNESCO on the preparation of draft recommendations on ethical aspects of AI
- Evgenia Meshkova – Head of the Legal Department of the Skolkovo Foundation, expert of the PC 01 “AI in Health Care” subcommittee of TC 164.
Source: https://bigdata.msu.ru/news/316/