Almost all military specialists associate the prospects for the development of VVST primarily with informatization, robotization, automation of command and control and weapons. In all cases, this inevitably predetermines the creation of military computer systems that handle the processing of gigantic amounts of information and the development of optimal solutions in accordance with the dynamics of military operations. But even the highest automation of the commander’s work will not replace the provision of its artificial intelligence (AI) systems.
We will immediately determine the difference between the automation of the processes of command and control and the use of artificial intelligence systems. In the first case, we are talking about computers equipped with a set of algorithms for collecting, classifying, structuring information, which is then used as an initial data system for solving combat missions using formalized methods. Quite another is artificial intelligence, which can independently develop ready-made solutions, figuratively speaking, think for a commander.
Man or car?
At present, the use of VVST samples equipped with automated control systems is algorithmically provided to a greater extent than troop control. This is due to the narrower range of options for the combat use of weapons and equipment, when control is impossible, for example, in emergency situations. In most cases, the incompleteness of information on the combat situation does not allow for the correct execution of troop command and control tasks, which significantly reduces the adequacy of the decisions made or does not allow them to be carried out at all. If it is not algorithmically provided, then the automated system is useless.
In the course of combat operations, similar situations do not recur, therefore it is almost impossible to create algorithms suitable for all cases of command and control of troops. As a result, the automation of these processes is so far only a tool for preparing initial information for decision-making by the commander.
The commander can make decisions, knowing not only the operational environment, the strength and means of the enemy, but also the peculiarities of his psychology and mentality of his subordinates. Thus, the main differences of intellectualization in relation to automation can be called the realization of the ability to make decisions in conditions of considerable uncertainty, based on heterogeneous information, frequently changing situations. Self-learning and adaptability are also very important as the ability of the system to independently improve the software embedded in it, including self-programming in situations that are not algorithmically provided for.
There is currently no established definition of artificial intelligence, but it can be said that AI is the ability of a computer to make decisions in infinitely varied situations, similar to humans. The famous scientist Turing formulated a test, which, in his opinion, will allow to determine the presence of artificial intelligence in the car. Briefly, its essence is that a person blindly communicating with a machine and another person should not determine who is who.
Currently, the most perfect ACS is unable to pass such a test, since in the absolute majority such systems are objectively rigidly oriented, and the number of tasks they solve is finite. The more algorithms for solving heterogeneous questions introduced into the computer's operating system, the more it will resemble an AI system. But there is no need to turn a computer into a person, providing the opportunity for him to equally equally solve the problems of managing technical objects and theatrical productions.
AI will always be subject-oriented, but such qualities as adaptability, self-learning and intuition will remain the main differences between systems with AI and ACS. Simply put, if we have a complete idea of what actions the automated system will take, then in the case of AI, there will be no such understanding. Self-learning, the computer programs its own work. Self-programming is the main distinguishing feature of AI.
The Agency for Advanced Research Projects of the US Department of Defense (DARPA) intends in four years to create for military needs an artificial intelligence of a new generation, as close as possible to human. DARPA experts formulated the basic requirements in the terms of reference for the L2M project (Life Learning Machines - “Infinitely Learning Machines”): an advanced AI must be able to make decisions independently, respond quickly to changes in the environment, remember the results of its previous actions and be guided by them in their future work.
Technical giants such as Google, Apple, Salesforce and IBM, realizing the promise of artificial intelligence systems, are seeking to acquire companies engaged in AI (from 2011, about 140 of them have already been acquired). Characteristically, at present, emphasis is being placed on AI for solving ground transportation problems, such as the creation of cars without drivers. Already in the near future, this promises a significant return on capital invested in public transport due to low operating costs, a small number of accidents and environmental cleanliness.
The experience gained will allow these firms to move on to the next step - the development of three-dimensional space, that is, the creation of an AI for controlling aircraft. The US Navy Admiral Ray Mabus, speaking at a conference in 2015, said that the F-35 aircraft should and almost certainly will be the last manned fighter-attack aircraft that the naval department will purchase or use. If we consider that deliveries of F-35 to the Air Force are planned up to 2037, and they should be written off to 2070, then we can assume: by the middle of the century, the United States plans to build fully unmanned combat aircraft equipped with AI systems. In 2016, the artificial intelligence to control the ALPHA fighters won a landslide victory over the former pilot-ace of the American army in a virtual air battle. AI ALPHA - joint development of the University of Cincinnati, industrial enterprises and the US Air Force. Moreover, in one of these battles against the ALPHA, two pilots fought on two fighters at once. Artificial intelligence won, simultaneously controlling four aircraft.
Brain attack
Another area of application of AI is medicine, where it is possible to move from existing computer decision support systems in the process of making a diagnosis and choosing a treatment regimen by a doctor to the creation of autonomous doctors.robots, including surgeons for complex operations. The advantages are obvious: minimization of medical errors in diagnosing diseases and prescribing drugs, selection and flawless implementation of the optimal algorithm of surgical operations, absence of fatigue during long operations, and higher speed of their implementation.
As for the fighting, it is the ability to ensure effective resuscitation of the wounded, quick localization of negative consequences with the unpredictable nature of injuries. Achievements in the field of AI will create a system of rehabilitation of the wounded due to the management of the affected internal organs of the person, neuroprocessing of prostheses with loss of limbs.On the basis of all this, a number of basic problems can be singled out, the solution of which is capable of ensuring the creation of AI systems applicable to military activities.
1. Representation of knowledge - the development of methods of structuring, classifying and formalizing knowledge from various problem areas (political, military, military-technical, psychological, organizational, etc.) to develop solutions in the pre-war period.
2. Modeling reasoning (decision-making processes) - the study and formalization of various schemes of human reasoning based on heterogeneous information for combat operations, the creation of effective programs for the implementation of these schemes in computers.
3. Creation of dialogue procedures for communication in natural language, providing contact between the intellectual system and a human specialist in the process of solving problems, including the transmission and reception of non-formalized commands in extreme situations involving a risk to life.
4. Planning of combat activities - development of methods for constructing control algorithms based on knowledge of the problem area, which is stored in an intelligent system and is continuously received from various and heterogeneous sources of information: reconnaissance, geodesic, topographic, meteorological, hydrographic, etc.
5. Training and updating of intellectual systems in the course of their activities, the creation of means of accumulation and synthesis of skills.
Each of these problems is extremely complex. Suffice it to say that the Pentagon created a new division, Project Maven (“Project Expert”), only to solve one subtask in 2016, which is developing AI for analyzing intelligence information from unmanned aerial vehicles operating in Syria and Iraq. The staff of the analytical centers that do this cannot cope with the processing and analysis of the huge amounts of data. Up to 80 percent of their working time takes just a look at the frames. It is assumed that with the help of AI, military objects that are dangerous for their troops will be identified, a sequence of actions on the ground to prevent terrorist acts, plans of militants will be identified.
In August 2016, Amazon, Nvidia, DigitalGlobe and a special unit of the CIA CosmiQ Works began developing artificial intelligence that can recognize objects in satellite images. It is supposed to use AI also in such an area of interstate confrontation as information warfare. In 2014, the Associated Press announced that from now on most Newsrelated to company earnings will be created using robots. In 2016, the Associated Press reporter robots expanded somewhat. They began to be entrusted with the preparation of small news notes related to the US Major League Baseball.
It uses journalist robots and the magazine Forbes, for which the company Narrative Science has created the appropriate specialized platform. In November 2015, a similar development direction was opened by the Russian company Yandex. While artificial intelligence "Yandex" produces only short notes about the weather and the situation on the roads, but in the future its representatives promise to expand the list of topics for publications.
Narrative Science co-founder K. Hammond believes that by 2025, 90 percent of all texts in the world will be prepared by artificial intelligence. The algorithms developed for these purposes can also be effectively used to collect intelligence information regarding countries, organizations and individuals, analyze it and prepare various kinds of materials, including in the interests of information warfare. In particular, to discredit the actions of the country, its government, party leaders and movements in the international arena. Moreover, this kind of action has already been taken in the preparation of almost all the "color revolutions", but at the same time human intelligence was used. AI will do this much faster and more massively. In a letter to the UN, a well-known American businessman, Elon Musk, described this danger as threatening humanity, which could provoke a war when AI creates fake news and press releases, falsifies email accounts and manipulates information. Other scientists have expressed similar concerns.
We emphasize this aspect of capabilities, which should be implemented in AI, as self-study. It is laid by American specialists in the basis of the development of the so-called concept of counter-autonomy. Its essence is that the attacked AI system must be quickly trained: to draw comprehensive conclusions from the fact and method of attack, to evaluate the characteristics of the technical means used in this process, to determine how to effectively counter. That is, each attack of the enemy will make the AI system even more effective if it does not destroy it the first time or does not radically change the methods of attack.
Attempts at the implementation of this concept are stated by the statement of DARPA director Arati Prabhakar, who in 2016 reported on a project to combat programmable radars in Russia (Heaven M is mentioned) and China: which we are going to resolve with the help of cognitive electronic weapons. We use artificial intelligence to study the actions of enemy radar in real time, and then create a new method of jamming the signal. The whole process of perception, study and adaptation is repeated without interruption. "
Thus, the intellectualization of military activity has practically become a fact. Systems for various purposes, equipped with artificial intelligence, are actively being created. However, there are a number of philosophical questions along this path. We cannot always truly explain the thought processes of our own and other people, but we intuitively trust or do not trust actions. Will it also be possible when interacting with machines that think and make decisions on their own and it is not entirely clear how? How pilots, crews will feel tanks and other military equipment working with aircraft and robotic tanks, whose actions are unpredictable? How will a robot behave, whose “brains” will be shaken by electronic warfare, blows of explosive waves, bullets and fragments, how will such a “shell shock” affect their behavior? Finally, is an intelligent robot capable of losing control?
There are many similar questions, but there is no clear answer to them. It seems that humanity acts here, guided by the rule of Napoleon: the main thing is to get involved in a fight, and then we'll see.