Artificial intelligence in the Russian army
Pavel Luzin on the challenges and idiosyncrasies of the development of advanced military command-and-control systems in Russia
The use of artificial intelligence (AI) is expanding in commercial, scientific and military fields. AI technologies cover the analysis of large volumes of varied visual and speech data, machine learning and the autonomous operation of unmanned aerial vehicles (UAVs) and unmanned ground vehicles (UGVs), among other things. AI is indispensable because of the complexity of modern arms systems, the exponentially expanding amount of information on troop status (in peacetime and on the battlefield) and enemy actions, and, in general, due to the advancement of the network-centric concept of warfare. The latter involves the integration of information on all combat forces and assets, enabling them to operate as a single system that exchanges real-time data.
Frankly speaking, both excessive optimism and pessimism should be avoided when it comes to the prospects of the development of these technologies; AI is neither a panacea nor a magic wand. AI solutions are simply technologies. When used properly, they assist weapons system operators, commanders as well as military and political leadership with the complex and time-consuming calculation of all available quantitative information in the decision-making process. However, they cannot improve military and state governance, nor can they devise political aspects of military campaigns, nor can they substitute the thinking and actions of officers and soldiers on the battlefield.
The Russian Armed Forces have been trying to catch up on modernisation for a decade and therefore, following the lead of the US and Chinese militaries, they have shown great interest in AI, attempting to use it in military communications and control systems. The Russian approach in this area is conservative due to existing problems, the peculiarities of Russia’s political and economic system, Russia’s level of scientific and technological advancement as well as the condition of industry and education. A conservative approach implies that modernisation despite limited resources is coupled with preservation of the existing order in the army and country as a whole. Therefore, let us identify the main areas of AI use in the Russian army and outline the long-term risks.
The origins and main contradictions of the Russian approach
The development and application of automated command-and-control systems (ASUs – avtomatizirovannaya sistema upravleniya) to the implementation of military tasks began in the 1950s in parallel with the development of the Soviet missile programme. In the 1960s and 1980s, ASUs for aviation, navy and ground forces started to appear. However, the limited number of these systems in the Soviet army, as well as a shortage of communications, intelligence and navigation solutions and of high-precision weapons, resulted in the Russian army not relying on ASUs in its military campaigns for two decades after the collapse of the USSR.
And while the development of such systems continued on a limited scale, the Russian Armed Forces began to implement the concept of network-centric warfare introduced in the early 1990s as late as the 2010s. At the same time, the issue of the prospects for the use of artificial intelligence solutions in new weapons and troop control systems became acute. By the end of the decade, this issue had become politicised, and, finding an answer to it had become one of the main imperatives for the military.
By the early 2020s, Russia had established a full range of military command-and-control systems, from individual and tactical ones for different branches of the armed forces and corps to the ASUs in military districts and at the National Defence Centre. Improvements in space, airborne and ground-based electronic and optical reconnaissance solutions, albeit at the catch-up stage and with great technological and economic difficulties, have raised the issue of efficient data analysis. Also, target recognition and classification systems with an automatic shootdown option are being developed for the aviation, navy and ground forces (especially for the purpose of maritime air and missile defence and UAVs).
It would seem that the use of AI has expanded at least in the case of specific tasks, since the automation of entire military command-and-control processes is still being developed on a purely theoretical level. However, communications and command and control still remain the Achilles’ heel of the Russian Armed Forces. The military itself describes this problem as ‘a deep mismatch between the organisational and technical dimensions’ of information support for combat operations. Simply put, the effectiveness of communication and control systems that were deployed in the past decade is limited by the insufficient training of officers and junior commanders as well as the decision-making system. Not only will artificial intelligence be unable to breach this gap but it may even deepen it.
Organisational and technical idiosyncrasies
Any country looking to apply AI in warfare faces the same challenges. First and foremost, they need to build an infrastructure to collect raw data ranging from information on the position of troops and the condition of military vehicles on the battlefield and the work of logistics services to reconnaissance and target classification and real-time information about enemy advancement. Inevitable misinformation due to the technical imperfections of such infrastructure, errors in machine learning itself, etc. must also be taken into account.
However, approaches to solving these problems depend on the system of relations within the armed forces, on the prevailing political and military culture in society as much as they depend on the quality of weapons, the state of the economy and the education system. As a consequence, key factors affecting the prospects of using AI in the Russian army include, for example, the previously described problem of the political leadership’s distrust in the armed forces in general and distrust within the military hierarchy in particular. Technical and legal difficulties are also a huge hurdle. For example, the available tools for military automation are still not very compatible with one another, and the inability to do away with paperwork limits the speed of transmission and processing of classified information. Moreover, there are problems with insufficient progress of R&D for military ASUs using AI solutions as well as their production. On the one hand, the central research institutes of the Ministry of Defence, of which the 27th Central Research Institute is the main one, are dealing with this. Its real scientific potential can be illustrated by one simple fact: at the time of the writing of this article, the institute was looking for a researcher in applied mathematics and ASUs with a monthly salary of RUB 30,000–40,000 ($404–$538). The inevitable shortage of military mathematicians and developers stemming from such an approach appears to be partially compensated by the system of army scientific companies introduced in 2013. These companies enrol conscripted university graduates with majors in line with the profile of a particular military scientific company.
On the other hand, the Russian defence industry is involved in developing AI solutions: these are being developed by Rostec companies (JSC Concern Sozvezdie, CNIIEISU Central Scientific Research Institute for Economics, Informatics and Control Systems), JSC Concern Morinformsystem-Agat, NIISSU JSC Research Institute for Command-and-Control Systems and some other establishments. In addition to the personnel shortages that these companies face, the institutional environment imposes additional constraints on them. The red tape permeating all aspects of their activities makes it impossible for them to be creative and take risks. The shortage of domestically designed and manufactured electronic components coupled with specifically Russian technical standards (due to the lack of international industrial cooperation) limits the potential even further. In other words, imported electronics are difficult to integrate with Russian components, and the supply chains have to be changed frequently because of sanctions. Not only are the obstacles in building the ASUs themselves noteworthy but so are those related to the systems (sensors) that are designed to collect the primary real-time data for their operation.
This ‘conservative’ approach, which envisages not only the development of AI-enabled control systems but also the preservation of the existing organisation of military and economic relations in the country, entails serious long-term risks. First, there is the risk of the excessive and even deceitful formalisation of military management at all levels in the name of automation. Deceitful formalisation means a desire to turn something that cannot be formalised by default into an algorithm, i.e., thinking, operating under the conditions of uncertainty, individual initiative, etc. The desire for such formalisation is based, among other things, on the notion that social processes are fully manageable. This almost religious dogma inherited from the late Soviet period is shared by the Russian political elite as a whole, as well as by the developers of the Russian military’s ASUs.
Secondly, there is a risk of information overload in the central command system and in central ASUs, since, within a unified system, information, along with responsibility for combat decisions, will be constantly delegated from lower-level officers to higher-level officers. For example, the movement of Russian peacekeepers from their base in the Volga region to an airfield for subsequent redeployment to Nagorno-Karabakh in autumn 2020 was managed directly from the National Defence Control Centre in Moscow.
Thirdly, there is the risk that officers will formalistically implement the recommendations of military command-and-control systems built using AI technologies, avoid responsibility for decision-making and ‘delegate’ it to the machine (Russian military theorists are aware of this). The flip side of this coin is that such systems are completely ignored on the battlefield, as they are believed to be created in isolation from reality, with its inherent uncertainty, and from the military’s need for freedom of decision-making and the exercise of human will.
Prospects for application
The main vectors for the use of AI solutions by the Russian army in the foreseeable future are becoming clear. One of them is the improvement of onboard ASUs to expand the effectiveness of aircraft and artillery in the face of a persistent shortage of high-precision munitions. This is intended to ensure, at a reasonable cost and within limited capabilities, the superiority of the Russian military over the enemy in local conflicts. At the same time, a unified information space for active troops will be created, for now, in the framework of individual operations rather than military campaigns as a whole.
Autonomous combat UAVs and stealth UAVs are likely to be exceptional solutions for a long time to come. As for UGVs, they would be suitable for supplying troops on the battlefield and evacuating the wounded along routes laid out by those troops rather than for combat operations due to the extreme complexity of their development.
Another vector is the expansion of the practice of using computer-simulated campaigns. At the very least, creating machine algorithms to calculate the basic parameters of such campaigns seems to be an achievable task today.
Another vector is the use of AI in optimising the supply of troops in peacetime and in preparation for military operations. The fact is that the Ministry of Defence’s reliance on large dual-use logistics hubs instead of hundreds of traditional military warehouses implies improving the quality of logistics automation in the Armed Forces. The negative American experience with the Autonomic Logistics Information System is likely to be taken into account. This system was designed to optimise the supply of spare parts for F-35 fighter jets, but its side effect was the low combat readiness of the aircraft.
When it comes to a large-scale conflict in which Russia might face a superior adversary, the main goal is not to achieve qualitative parity or superiority over it (this is simply beyond its capabilities). The goal is to keep Russian forces manageable and operational until an adversary’s information network is disrupted by electronic warfare.