Issues of application of artificial intelligence technologies and systems in the military sphere
The current state of research and development in the field of artificial intelligence (AI) is characterized by extensive penetration of AI technologies into various spheres of human activity.
This is due, on the one hand, to the achievement of such a level of development of this scientific direction, which ensures the effective use of AI systems to solve a variety of practical tasks that are currently accessible only to humans. And on the other hand – the emergence of such tasks, the solution of which is simply impossible in other ways.
SECURITY OF BANKING SYSTEMS AS AN EXAMPLE OF CONFRONTATION
One of the most characteristic examples illustrating the latter circumstance is the confrontation in cyberspace, which manifests itself in a concentrated form, for example, in the field of ensuring the cybersecurity of banking systems and services that form the basis of the economy of any developed state.
So, in particular, the information infrastructure of "Sber", the largest financial and economic conglomerate in Russia, unites more than 300 thousand workstations of employees, 80 thousand ATMs, 20 thousand means of network education and information protection. This infrastructure provides uninterrupted service to about 140 million individual and 3 million corporate clients in conditions of continuous pressure from the global computer underground and special services of states hostile to the Russian Federation.
During periods of aggravation of the international situation imposed on Russia by the collective West, this pressure acquires the character of massive, time- and space-distributed information technology impacts (cyberattacks) aimed at system destruction.
At the same time, the Sbera Security Management Center provides detection and neutralization of more than 100 high–tech cyberattacks and 14 thousand attempts of cyber fraud every day - based on complex intellectual analysis in hard real time of more than 0.5 billion messages about signs of unauthorized access to information and technological resources of Sbera detected by local information security means (SPI) with the purpose of stealing funds or blocking (destroying) financial services.
The successful confrontation of the cybersecurity service "Sber" with the most highly intelligent hacker groups in the world, constantly developing and applying increasingly sophisticated technologies of covert preparation and organization of cyber attacks, is provided by a multi-level AI system that implements the above-mentioned complex analysis of the flow of low-informative messages from local SPI.
In turn, the cognitive basis of this system is a continuously updated distributed knowledge base (BZ) containing logical and semantic relationships used for this analysis. The actualization of this BZ is based on the use of AI technologies to identify these relationships by analyzing "big data", which in this application are multi-petabyte arrays of fragments of network traffic passing through various points of the information infrastructure of the "Sber" accumulated over a long time.
The given example makes it possible to assess the nature of practical tasks that, due to their intellectual complexity, the volume and intensity of the receipt of processed data and strict restrictions on the timeliness of their processing, in principle, cannot be solved without the use of AI technologies.
DIRECTIONS OF POSSIBLE APPLICATION OF ARTIFICIAL INTELLIGENCE TECHNOLOGIES AND SYSTEMS
The saturation of troops (forces) with robotic (unmanned, unmanned) means and the emergence of deeply integrated network-centric robotic and hybrid groups operating in real time in material space and under conditions of considerable uncertainty about the state and intentions of the enemy, created a situation where the use of AI technologies as the basis for the effective combat functioning of such groups is not only appropriate, but and necessary.
On the other hand, the development of these technologies and the understanding of their capabilities to improve the quality of solving traditional tasks of managing troops (forces) and various types of support (combat, logistical, medical), as well as operational and combat training tasks, made it quite natural to saturate automated systems for various purposes with software and hardware complexes that implement certain approaches from the field of AI.
Let's look at some of the most interesting examples of the use of AI technologies in the armed forces of leading foreign countries.
In modern conditions, based on the analysis of plans, programs and specific measures for the development of AI technologies in China, the USA, India, France and in a number of other countries, it is possible to identify the following areas (spheres) of possible application of AI technologies and systems.
DECISION SUPPORT SYSTEMS
One of these areas is the creation of decision support systems (DSS) in the interests of combat management (at the tactical, operational and strategic levels) and operational training of command and control bodies of troops (forces). The history of the development and more or less successful application of DSS in the military sphere has been going on for more than a dozen years, but only with the advent of methods and means of computer implementation of operations on arrays of incomplete information characteristic of armed confrontation, such systems are able to give a new quality to decision-making processes.
This new quality is associated with the creation, among other things, of promising training systems with AI for conducting tactical and operational level war games. As well as strategic (political-military) level games with full-scale consideration, including the possibilities of informational and psychological confrontation and struggle in cyberspace.
The US Department of Defense, along with the traditional areas of application of decision support systems, is considering the creation of a DSS for humanitarian operations and the actions of contingents of the US armed forces when they are involved in the fight against natural disasters.
IMPROVING THE EFFICIENCY OF RECONNAISSANCE AND TARGETING
In this area, AI technologies have reached a level where it can be possible to provide a comprehensive intellectual analysis in real time of intensive flows of information coming from geographically dispersed stationary and mobile sensors of various types of basing (land, sea, air, space) operating in various physical fields and providing the extraction of heterogeneous information about the manifestations of controlled objects and sources of radio emissions serving them.
This task has been known for as long as the task of creating a DSS. However, only with the advent of methods of complex intelligent analysis of large arrays of incomplete, unreliable and contradictory information, the accuracy and speed of determining the location of objects for targeting weapons can reach the required values.
INCREASING THE COMBAT CAPABILITIES OF DRONES
Another task is to increase the combat capabilities of unmanned aerial vehicles (UAVs) and swarming robotic groups by increasing situational awareness of UAV operators.
In the United States, one of the most advanced in this area is considered to be a large-scale project "Maven", which started several years ago. Within the framework of this project, using artificial intelligence technologies, the analysis of large volumes of video materials coming from reconnaissance UAVs is carried out, its generalization and presentation to the operator-analyst of the processed information for the subsequent formation of a "database of targets" intended for destruction. The final information should come in the form of target designations for shock UAVs.
INCREASING THE CAPABILITIES OF STRIKE AIRCRAFT
The next task is to increase the capabilities of manned strike aircraft, including by creating a fundamentally new integrated human–machine air combat control system using AI.
In this direction, the MMT program of the Ministry of the Armed Forces of France is of undoubted interest. It includes such components as "virtual assistant" – "smart cabin", "human-machine interaction", "mission management", "intelligent sensors", "robotic support and maintenance", etc.
A large number of sensors should be involved in this system, providing full-scale data on the technical condition of the components and assemblies of the aircraft, on the air situation, on the condition of the crew members. It is assumed that such a system will have the potential for development and "deep learning".
The crew should have a "virtual assistant" who will be able to make up for "low-level operations", unloading pilots from a number of secondary tasks. This should allow the crew to focus on tactical solutions for the implementation of the combat mission, having maximum awareness of the situation.
An important role is assigned to "intelligent sensors", which must adapt to changes in management needs associated with dynamic changes in the environment and operational needs.
AND OTHER TASKS
We will also highlight such a task as ensuring more efficient operations in cyberspace. Many foreign experts note that AI technologies could realize an increasingly significant part of the work of highly qualified specialists in "cyber warfare", since AI tools are potentially able to find vulnerabilities in computer networks much faster and provide accelerated development of special information technologies that use these vulnerabilities to solve intelligence and impact tasks.
Another task is to create a "digital assistant soldier" ("virtual instructor") and personal automated translators. Including oral speech – to communicate with the local population in the appropriate languages.
The next task is to create simulator simulators using virtual reality technologies that are as close as possible to real conditions. Such simulators are needed to train Air Force pilots, SAM operators, tankers, etc. And also for officers – operators for the management of heterogeneous forces and means.
A significant achievement of Israel is considered to be the creation of tactical and operational level simulators based on artificial intelligence technologies (Command and Staff Training – CST) for conducting "joint operations", including in very difficult urban conditions. Up to operations without the direct use of weapons of destruction ("operations outside the war").
The next task is predictive maintenance. The use of data from sensors installed on tanks, cars, armored personnel carriers and other types of equipment, about their condition. Identification of non-functioning and non-standard systems. Assistance in maintaining the optimal level of functionality of these systems. Proactive notification of the need to repair specific units of equipment or their individual units and parts.
Finally, we will highlight such a task as improving the efficiency of logistics systems. Including optimization of the replenishment cycle of warehouses, taking into account the cost of maintaining a certain level of inventory.
HUMAN CONTROL AND INTERPRETATION
Many experts reasonably pay attention to the fact that there is still a need for interpretation by a person (a highly qualified operator) the results that are obtained as a result of the functioning of AI systems.
This once again underlines the importance of developing such a direction as "human-machine interaction", the need for active training of relevant personnel, since AI systems for use in the armed forces must correspond to the level of preparedness of the relevant operators. And vice versa – AI systems should not become a "black box" for the operator.
The issues of human (operator) control of AI systems, ensuring the reliability and safety of the use of weapons using AI can be considered among the most difficult when considering the problems of widespread introduction of AI technology and systems.
This is especially true of the means of destruction. The overwhelming majority of experts and officials agree that any decisions on the use of shock (lethal) means should be in the hands of a person exclusively.
* * *
The authors are sincerely grateful to Academician of the Russian Academy of Sciences I.A. Sokolov for very useful comments that contributed to improving the quality of this article.
Andrey Kokoshin
Andrey Afanasyevich Kokoshin is an academician of the Russian Academy of Sciences; Igor Anatolyevich Sheremet is an academician of the Russian Academy of Sciences.