RUSI
Printable version |
Struggling, Not Crumbling: Russian Defence AI in a Time of War
Notwithstanding its difficulties at the front in Ukraine and economic hardships at home, Russia continues to pursue AI and other emerging technologies in a bid to future-proof its armed forces. However, it faces several major hurdles to before it can reach its objective of achieving battlefield superiority in selected areas.
Discussing Russia and high technology these days may seem counterintuitive given the images from Ukraine of the Russian army in disarray, in trenches, using primarily massive artillery rather than ultramodern weapons with a high degree of autonomy. Likewise, the typical Russian tendency for overblown rhetoric has created unrealistic expectations among the broader public of what Russia was ready to deliver on the battlefield. Yet, the war is also providing a testing ground for new weapons systems on both sides. Mounting troubles notwithstanding, Russia continues to prioritise AI and selected emerging and disruptive technology (EDT) programmes not merely despite, but because of the weakening of its conventional forces and the growing capability gap with the West.
Drivers Behind Russian Interest in AI
There are a host of incentives driving the Russian political and military leadership’s continued focus on AI, despite the attention consumed by its war of aggression. AI is seen as a source both of potentially rapid military modernisation and of new vulnerabilities that enemies can exploit.
When Russian authorities launched a large-scale military reform programme in 2008, they expressed the concern that it would be too time-consuming to catch up militarily with the West the traditional way. EDTs, conversely, seemed to offer a potentially rapid, non-linear pathway to narrow – if not close – the capability gap. This logic still applies. Moreover, given the growing asymmetry in conventional power due to Russia’s extensive personnel and material losses in Ukraine, this reasoning appears even more relevant today.
The Russian authorities believe that trends such as the proliferation of autonomous and AI-enabled weapons systems and the convergence between human–machine learning, cyber and AI, coupled with novel operational concepts and force structures, will change the trajectory and character of future warfare and human involvement therein. Consequently, these developments seem to have the potential to undermine fundamental pillars of international security such as deterrence, arms control and strategic balance. Therefore, gaining or losing ground in the contest for cutting-edge military technology appears to have profound consequences for power distribution and Russia’s influence in the international system. Indeed, President Vladimir Putin has consistently spoken about technological development in existential terms: one either succeeds or they will be annihilated. In 2023, he compared AI development to the invention of the nuclear bomb – an event that fundamentally changed the course of history.
Furthermore, Putin and other central players anticipate that defence innovation – and AI in particular – will produce dual-use technology that can drive a nationwide economic growth. This is needed now more than ever given the weight of Western sanctions. The hope is that the domestic defence industry will be able to provide technological solutions that Russia can no longer get from abroad. As put by the head of the AI Department at the Russian Ministry of Defence (MoD), Vasilii Yelistratov, military-civilian transfer takes place in both directions. Likewise, German Gref, the CEO of Russia’s largest state-owned bank Sberbank and one of the key figures in Russian AI development, argues that the introduction of AI could increase Russia’s GDP by 1% by 2025. He has pointed to Sberbank’s increase in labour productivity as a result of implementing AI. Others are even more optimistic. In July 2023, Russia hosted its first Future Technologies Forum, where Putin promised to transition the whole national economy and governance system to a new management model based on AI and Big Data in order to boost labour productivity.
While these promises and benefits are yet to materialise, such statements and assessments clearly demonstrate the importance that Russian authorities attach to AI development, both for the future of the armed forces and the state as a whole.
How is Russia Going About Developing AI?
The Russian approach to defence innovation is based on a traditional top-down, state-driven model, although with important modifications. The objective is to take advantage of the progress being made in the civilian sector by fostering civil-military cooperation. In this way, Russia aims to maximise the state’s access to talent, resources and ideas, thus shortening the time between the generation of an idea and its full implementation. This approach appears to partly emulate US military-civilian sector cooperation, as well as the Chinese model of military-civil fusion, connecting science and technology parks to the campuses of large military universities.
Defence R&D infrastructure is centrally coordinated by the Russian MoD’s Main Directorate of Innovative Development (GUIR). To further strengthen the MoD’s role as an engine for AI implementation, a special AI department was created in 2021. GUIR’s role is to organise, coordinate and support innovation programmes. Russia’s defence R&D infrastructure consists of several hundred research institutes, design bureaus and testing centres that conduct applied research for the needs of the armed forces and the defence industry. In addition, Russia has created a number of AI centres and laboratories at leading academic institutions, such as the Neural Networks and Deep Learning Lab at the Moscow Institute of Physics and Technology; the Higher School of Economics; the Ivannikov Institute for System Programming of the Russian Academy of Sciences; the Skolkovo Institute of Science and Technology; and the National Centre for Cognitive Technologies at the Information Technologies, Mechanics and Optics University in Saint Petersburg.
This ecosystem is supplemented with so called ‘radical innovation centres’, ‘technoparks’ or ‘technocities’ as generators of ideas and dual-use technologies. The Advanced Research Foundation (Fond perspektivnykh issledovanii), created in 2012, focuses on developing new and potentially disruptive dual-use technologies, such as unmanned vehicles (the Marker unmanned ground vehicle and the Udar unmanned tank); autonomous systems and automated decision-making; superconductors (Liman); additive technology of polymetallic products (Matritsa); autonomous deep-submergence vehicles (Vityaz-D); and ultra-thin materials for improving individual camouflage and protection (Tavolga).
Subsequently, in 2018 Russia established the Era Technopolis, which explicitly seeks to develop technology for the Russian armed forces in cooperation with the military-industrial complex and the civilian sector. Currently, there are more than 100 entities involved in cooperation with Era, including top arms manufacturers such as Kalashnikov, Sukhoi and Sozvezdie and dozens of civilian universities and research institutions, including the Kurchatov Institute – which hosts the largest interdisciplinary laboratory that has received a special role in Era, contributing to its management. The structure is supplemented with military scientific units (nauchnye roty), created first in 2013 on the basis of Russian military research and higher educational institutions. As of 2023, eight units are operating as part of Era within their specialised fields of expertise. Staffed by conscripts, Russia hopes that these units may also provide a recruiting ground, allowing it to retain specialists in the armed forces and defence industry.
Among the top priorities is improving command, control, communication and decision-making with AI, seen as critical to gaining and maintaining information superiority
Era has a cluster of 16 prioritised development fields, including AI, pattern recognition, robotics, small spacecrafts, information security, energy sufficiency, nanotechnology, nanomaterials, information and telecommunication systems, information technology and computer science, hydrometeorological and geophysical support, hydroacoustic object detection systems, geographic information platforms for military use, weapons based on new physical principles, radiolocation and targeting for high-precision weapons, and automated control. As Deputy Head of ERA for Scientific and Educational Activities Andrey Morozov explained, AI technology is not so much a product in itself, but rather a foundation intersecting almost all of the military’s EDT programmes.
What are Russia’s Defence AI Priorities?
Russia has a broad spectrum of programmes that are similar to those developed in the US and China, though they are usually smaller in scale. As of September 2022, GUIR has supported over 500 projects, 222 of which were planned for completion and implementation in 2022. Among the extensive list of projects, it is possible to discern several priority areas, including command, control, communication and decision-making; unmanned vehicles; nuclear and high-precision weapons; air defence, early warning, electronic warfare and space-based systems; and cyber and influence operations to shape the psychological domain.
While the list of Russian AI-enabled projects is too extensive for this article, below are some examples that illustrate the areas of Russia's particular interest in AI for defence applications.
Among the top priorities is improving command, control, communication and decision-making with AI, seen as critical to gaining and maintaining information superiority. At a Defence Ministry Board meeting in December 2022, Putin called for the integration of AI technology at ‘all levels of decision-making’ in the armed forces. The National Defence Management Centre, established in 2014 to provide the primary joint all-domain command and control structure, reportedly applies AI to support information collection, analysis and decision-making. Notably, Putin has referred to experiences from the battlefield in Ukraine, which in his assessment show that the most effective weapons systems are those that operate quickly and ‘almost in an automatic mode.’
Unmanned systems, furthermore, stand out as a special priority for AI applications. Before the 2022 invasion, Russia had more than 100 types of unmanned vehicles for a broad spectrum of missions at different stages of research, development and deployment. Many have been tested in Syria and are currently being employed in Ukraine. One of them is the KUB-LA kamikaze drone. Its producer, ZALA Aero Group – a subsidiary of Kalashnikov – claims it is able to select and engage targets with the use of AI technology. Another example is the Lancet-3 loitering munition, also observed in Ukraine. According to the same producer, it is highly autonomous given the use of sensors that enable it to locate and destroy a target without human guidance, even being capable of returning to the operator if a target has not been found. Likewise, Russia is testing several Marker unmanned ground vehicles that aim to use AI-enabled object recognition, process data via neural network algorithms, and employ autonomous driving capabilities. It remains unclear at this stage whether these unmanned systems have been used in this way in Ukraine. More data will be needed to verify their actual capabilities. Nonetheless, these claims indicate the areas of interest and the level of ambition when it comes to AI applications.
According to the Russian authorities, AI and autonomous elements are also being applied in the guidance systems of other key weapons such as the Sarmat intercontinental ballistic missile. Yelistratov has claimed that AI is ‘present in all weapons, especially in high-precision ones’. Russia is also using such technology in experimental weapons systems, such as the Poseidon nuclear-powered and nuclear-capable unmanned underwater vehicle.
The Russian Aerospace Forces, created in 2015 to integrate defensive and offensive capabilities, have been developed to disrupt or degrade the foundations of US and NATO information technology-enabled warfare, including communications, space-based systems, critical networks and infrastructure that developed countries depend on. Despite claims in the first months of the war that Russia was not using electronic warfare in Ukraine, several systems have been confirmed to be operating in Ukraine, some of which allegedly employ AI. Among them is the RB109-A Bylina, which – according to Russian sources – uses AI to collect and prioritise large amounts of data to efficiently jam electronic signals. In June 2023, Russia reportedly used the S-350 Vityaz mobile surface-to-air defence missile system to shoot down a Ukrainian aircraft while operating in autonomous mode – that is, the system detected, tracked and destroyed a Ukrainian air target without human assistance. To verify such claims, more evidence will be necessary. There are plans to use the system to protect Moscow against Ukrainian drones from the end of 2023, which may provide additional information about the system’s actual capabilities.
What are the Obstacles?
The pace of Russian defence AI development varies; some programmes are more advanced than others. Generally, however, Russia lags behind its main competitors, the US and China.
Russia continues to struggle with long-standing structural problems, such as insufficient funding, extensive corruption, inefficient use of resources, poor quality control, low labour productivity across the economy, the reliance of the defence industry on state order, and preferential funding that undermines competition and innovation. Weak rule of law, deficient intellectual property rights, and heavy-handed bureaucratic control are other factors hampering innovation. The education system appears to no longer be a suitable foundation for hi-tech development, as the Soviet system once was. Despite taking a number of steps to improve interest in and the quality of science and IT education, Russia has also been sliding down the Global Innovation Index, going from 43rd place in 2016 to 47th in 2022.
There are also doubts about the efficiency of the Russian defence innovation model itself. The cooperation framework managed by the Main Directorate for Innovation involves more than 1,200 entities including industrial parks, engineering centres, financial development institutions and leading universities and research institutes, and it is growing. One question is whether Russia’s extensive R&D infrastructure can be successfully managed and coordinated centrally by the MoD, not least given widespread corruption.
Statements by Russian authorities indicate that the battlefield advantages provided by AI-enabled weapons systems and infrastructure are among the lessons Russia has been learning in Ukraine
These long-standing impediments are aggravated by the circumstantial constraints generated as a consequence of Russia’s war in Ukraine. Despite Russian authorities’ assurances about the benefits of sanctions in terms of reducing Russia’s dependence on Western technology such as microelectronics, Russia is now growing more reliant on third countries – notably China, which is filling Russia’s most pressing technology gaps. Sanctions are likely to have a long-standing impact on the Russian national economy and tech industry.
Thus far, however, Russia has been able – to some degree – to evade sanctions and exploit loopholes in the exports control regime, using illicit networks to smuggle key components to third countries, from where they can be shipped to Russia. As a result, for instance, Russia has managed not only to continue but to double missile and tank production compared to the numbers before February 2022. Likewise, in August 2023, the Moscow State University launched its new supercomputer, which is to be used for various AI and high-performance computing applications and for training large AI models.
That said, Russia is struggling with other pervasive problems, such as the long-standing decline in professional expertise in the defence sector. The country has witnessed a demographic decline over the past 30 years, further worsened by short life expectancy. The exodus of IT specialists, top scientists and university professors in the wake of the invasion and the high casualties on the battlefield are likely to aggravate the brain drain.
These problems are likely to impact the culture of defence innovation. On the one hand, the brain drain, economic hardships, and limited access to ideas from and cooperation with other countries may further undermine creativity and aggravate existing structural problems. On the other hand, the Russian leadership is determined to rebuild and modernise its ravaged armed forces, and aims to do so rapidly. This, together with the pressure to deliver technological solutions for immediate battlefield deployment, could accelerate the development of AI for defence applications. Statements by the Russian authorities, including Putin, clearly indicate that the battlefield advantages provided by AI-enabled weapons systems and infrastructure are among the lessons Russia has been learning in Ukraine.
Tentative Conclusions
The overall deployment of Russian AI-enabled systems indicates that Russian AI appears to be in the early stages of maturity. The primary focus is on incremental evolution: upgrading legacy systems – nuclear, strategic non-nuclear, and non-military methods and means of warfare – with new technologies. Russia is combining conventional warfare and platforms with innovative technological solutions, including AI in data analysis and decision support, loitering munitions, electronic warfare and communication analysis, and as a component in cyber warfare and information confrontation – to name but a few examples. Simultaneously, Russia is experimenting with selected ‘risky projects’ and novel systems, materials and approaches to warfare that can potentially yield battlefield advantages – if not superiority – in selected areas.
One of the challenges in assessing Russian – and other countries’ – AI-enabled weapon systems is the difficulty of determining when full autonomy has actually been used in a lethal context, as opposed to declarations by authorities or producers that may have agendas other than speaking the truth (such as advertising systems to potential buyers, or portraying the army as ultramodern to impress domestic and international audiences). More data and cross-referencing will be needed to verify such claims. Still, these statements provide important information about the types of capabilities and capacities that the state is interested in and actively pursuing.
The study of Russian defence AI development is further complicated by the ambiguity of the term ‘AI’ in the Russian discourse. According to the official definition provided in the 2019 Russian AI Strategy, AI consists of ‘technological solutions capable of mimicking human cognition and performing intellectual tasks similarly to, or better than, humans’. The Russian definition distinguishes between ‘automation’ (avtomatizatsiya) – that is, automated, remotely controlled and semi-autonomous weapons systems – and ‘intellectualisation’ (intellektualizatsiya), corresponding to integrated machine learning and other sub-elements of AI technology. However, these terms are often used interchangeably – as are automation and autonomy – and without proper precision.
The extensive Russian failures during the 2022 full-scale invasion of Ukraine are likely to prompt major reassessment and reforms in the Russian armed forces. It remains to be seen whether the experience will also result in a major push to accelerate defence innovation. In any case, the Russian regime is unequivocal about where its priorities lie: despite the deteriorating
Original article link: https://rusi.org/explore-our-research/publications/commentary/struggling-not-crumbling-russian-defence-ai-time-war