Russia KPIs and Artificial Intelligence (AI)
A Russian internet site, The Bell, recently reported that prime minister, Mikhail Mishustin, intends to use Artificial Intelligence (AI) to “monitor the implementation of new” Key Performance Indicators (KPIs) for Russia’s National Projects.[1]
Reportedly, Mishustin “presented a new plan to achieve five national goals with 25 KPIs, for which members of the government will be “personally responsible”. KPIs will be “decomposed” to each ministry, department and region and monitored by AI. This was described as “”mega-KPIs according to the Chinese principle.”
According to the article, “the system will control both the KPIs themselves and the expenditure of funds to achieve them”. Moreover, it was said that “AI sees and shows how demographics fall or grow from the implementation of a national project on the digital economy”.
Just how this new system would work is very unclear!
This blog considers two issues that suggest it would not work very well. One is the propensity of Russian officials to use KPIs as a crutch instead of personal complex thinking. The second is the nature of AI itself.
But first some description of the National Projects and their implementation is needed. Following a mid-2018 presidential decree, early 2019 saw the release of a document[2] showing total expected expenditure equal to about 3% of GDP annually in the 2019-2024 period on various sectors of the Russian economy, including healthcare, education, the demographic situation (including increasing the fertility rate), culture (including strengthening Russian civic and national identity feelings), roads, city living conditions, ecology, science, promotion of SMEs, the digital economy, labour productivity, and international economic cooperation and exports. In addition to these twelve National Projects, there is a another concerning very large-scale modernization and expansion of large infrastructure such as pipelines, transport and ports (especially in the Russian Far East).
The early 2029 document specified spending “targets and key results” – or KPIs!
KPIs included such things as the number of articles by Russian researchers published in international scientific journals and the proportion of Russian scientific researchers aged 39 or less, 900 domestically produced pianos to be provided to children’s art schools by 2024 and 140 new war memorials, specific numerical export targets for various industry sectors (for example, increase in agricultural exports to $45 billion, and “service” exports to $100 billion) etc.
The “digital economy goals and targets” included expenditure of 1,634.9bn rubles in the period from 1 October 2018 to 31 December 2020. Of this, 1.7bn rubles was earmarked for “regulation of the digital environment”, 772.4bn rubles for “information infrastructure”, 143.1 rubles for “human resources for the digital economy”, 30.2bn for “information security”, 451.8 for “digital technologies”, and 235.7bn for “digital public administration”.
Furthermore, the cost of developing the digital economy as a percent of GDP is specified for 2019, 2021 and 2024, with a breakdown into several types of expenditure. By the end of 2024, 120 thousand people are to be admitted into higher education information technology programs and 10 million people are to receive training via online digital development programs.
Also specified is 1350 “commercially orientated scientific and technical projects in a specified field will receive grant support by the end of 2021. And there are specified 2021 and 2024 numerical targets for such things as “share of households with broadband access to the internet”, “share of “socially significant infrastructure with the capability of connecting to broadband internet”, “Russia’s share of the world capacity for data storage and processing services”.
Now, there is nothing wrong with such detailed planning provided that it is not done so far in advance and in such detail by one document covering the whole nation. Greater flexibility – and economic efficiency – would be achieved by devolving authority for details to lower levels of government and relevant institutions. Mishustin seems to be proposing the opposite.
Implementation of the National Projects has not been going well. According to the Accounts Chamber of Russian Federation[3] in a November report[4] only about 70% of allocated funds for the January-October 2020 period had actually been spent.
Returning to the Russian use of KPIs, a long-time head of Morgan Stanley’s closed Moscow office was quoted as saying that the Russian state “thinks that it can solve problems relating to scientific and technical progress by simply giving everyone KPIs”.[5]
Rather than thinking about good policy principles and concepts, the emphasis is excessively on centrally determined numerical outcomes. While this partly reflects the Soviet central planning tradition, it also reflects the desire of the present Russian authorities to maximise control over events. The resulting constraint on thoughtful innovative thinking and decision making by mid and lower level officials is a drag on Russia’s economic performance.
This view is indirectly supported by 2019 comments by Aage V. Nielsen, Deputy Chairman of the AEB Working Group on Modernisation & Innovations, who said that “combine Russian engineers with Western management skills, culture and habits, and you have a winner!”[6] He explained that Russian managers lack “soft skills” and that “just 2 years ago we asked approximately 25 Scandinavian top managers in Russia how many employees they would typically see in Russian companies in their industries compared to Western Europe. With a few exceptions, all stated 2.5 to 3 times more staff in Russian companies.”
Personally, based on my experience in Russia mainly in educational institutions and talking with Russian businesspeople, I think that 2.5-3 times is partly a result of the excessive paper work demanded – and it is often literally on “paper”, if only because some superfluous stamp needs to be applied! But, there is also a basic preference for detailed planning amongst many Russian managers – and computers and the digital age often allow these to be more complex than necessary.
If AI is to be used to “control” and/or “monitor” National Projects KPIs, then some standard and uniform measures – either in words or numbers – will be needed.
Despite the word “intelligence”, AI is really only a form of computer program that recognises number or word patterns in data bases. Moreover, this recognition only comes after much training on data bases where many many examples are given. It is very unclear what data is to be used to train the AI algorithm (s) used to in Mishustin’s KPI project.
Moreover, the sheer number of examples given to the AI algorithm and the iterative way that its looks for matches in the data set it is given the task of examining means that the reason for or cause of any match is generally not clear to humans.
So, essentially, trying to use KPIs as input to an AI algorithm means trying to use difficult to understand and measure items as input to a difficult to understand AI evaluation process. The result is likely to be some numbers or words that may be very misleading, or even perverse.
All this does not mean that Mikhail Mishustin is entirely wrong in wanting to use AI as part of an economic policy tool-kit. When head of the national tax office, he implemented a program that allows all expenditures recorded on business cash registers to be almost immediately sent to Moscow. Over time, it may be possible to train an AI algorithm to use patterns in such detailed expenditure data to forecast aspects of broader economic performance.
[1] Анастасия Стогней & Валерия Позычанюк, “За выполнением нацпроектов проследит искусственный интеллект”, The Bell, 17 НОЯБРЯ 2020
За выполнением нацпроектов проследит искусственный интеллект (thebell.io)
[2] National Projects p7nn2CS0pVhvQ98OOwAt2dzCIAietQih.pdf (government.ru)
[3] Accounts Chamber of Russian Federation
Accounts Chamber of Russian Federation (ach.gov.ru)
Счетная палата Российской Федерации (ach.gov.ru)
[4] Исполнение расходов бюджета на нацпроекты за 10 месяцев составило 70% (ach.gov.ru)
[5] Анастасия Стогней, “Риск инвестиций в Россию не высокий, а запретительный»: бывший глава Morgan Stanley в России Райр Симонян об уходе из страны самого успешного западного инвестбанка”, The Bell, 29 March 2019
[6] Aage V. Nielsen, Managing Director and Senior Partner, Vitus Bering Management Ltd and Deputy Chairman of the AEB Working Group on Modernisation & Innovations, “Improving productivity in Russian-based companies: challenges and barriers”, Association of European Business Quarterly Magazine, Spring 2019.