by Giancarlo Maragucci


Today Google is giving more than 540 million results when searching for artificial intelligence (AI), and all over the world many countries and organizations invest billions of US dollars in it. Nevertheless, it is still the initial phase of what could be a new chapter of human kind, where real and electronic lives coexist in symbiosis.


Big data, machine learning, deep learning, so many language variations trying to explain that a machine, using a sophisticated algorithm, can manage a lot of information at light speed – and learn from experience – in order to imitate human brain processes and actions. Many applications are already in place in healthcare, banking, education: all implementing some sort of algorithm to help improving the efficiency of the processes.


The military community is also working on implementing AI but obviously the pace of introducing AI is slower than other fields, and this is due to the nature of the activities, related to war and warfare, to violence and death.


Before elaborating on AI for the military, it is important to define war and warfare, even if both are apparently and instinctively clear.  The term war is still not universally defined even if, in the popular sense, it is a conflict between political groups involving hostilities of considerable duration and magnitude. In social science, certain qualifications are added. Sociologists usually apply the term to such conflicts only if they are initiated and conducted in accordance with socially recognized forms. They treat war as an institution recognized in custom or in law. Military writers, on the other end, usually confine the term to hostilities in which the contending groups are sufficiently equal in power to render the outcome uncertain for a time. We need to agree that war is fight by nature.


About 180 years ago, von Clausewitz observed: “war should never be thought of as something autonomous but always as an instrument of policy”, and in 1939 the British Edward Carr divided international political power into three categories: military power, economic power, and the power over opinion. During the Cold War, the United States and its armed forces expanded those categories and developed a four-element schema known as DIME (Diplomacy, Information, Military, Economy). DIME expresses the power of a State (or organization or a Coalition) and it is commonly applied in modern operation planning processes, where all powers collaborate through separated and coordinated lines of effort to reach unanimously the end state. Besides the different views or definitions though, the nature of war remained unchanged through the centuries.


Warfare is in a continuous evolving status, where technology and brain process persist to race looking for new systems, sensors, weapons, in order to empower the fighters and the decision makers. New systems mean new training, improved TTPs (Tactics, Techniques and Procedures), in order to achieve new capabilities. The way we fight war, warfare, makes the diffrence every time we have a new tank, or ship or aircraft, and when we change doctrine or organization. Then warfare should define only HOW we fight, but nowadays the term is also associated with one of the 5 domains (air, land, sea, space, cyberspace) intending also WHERE we fight. If space warfare is how we fight in space and cyber warfare is how we fight in the cyberspace, how do we define, for example, Electromagnetic Warfare (EW) and Information Warfare (IW)? In these cases, warfare is associated with the instrument or group of instruments, for example the Electromagnetic Waves or the media systems, so we are defining essentially WHAT (we use for fighting).


Unfortunately, war and warfare are often confused and used inappropriately, for instance we hear a lot about generations of war, but it should be described as generations of warfare since we describe the evolution of fighting.


Where is then AI standing? AI cannot be considered a new domain (a new WHERE) and all AI applications are empowering present weapon systems, enabling a faster information analysis for more efficient and effective military operations. AI is then enhancing HOW. Till we get to autonomous operations.


Scenario: the Commander approves the list of effects on a decision point during the planning process, then the computer decides the actions (kinetic and non-kinetic), what manned and unmanned assets to task and the time to strike. The Commander will be notified about the assessment when mission is accomplished, and the effect achieved. Apparently, what has been described looks very similar to the existing process, maybe with a little of more automation and less human effort, but what about if on the enemy side there is also an AI driven system? In this case, both systems will start a real race in the information environment, trying also to disrupt the enemy’s AI processes with proper counter-AI operations.


AI versus AI: how do we call it? And when does it apply? AI will be implemented in many tactical systems, improving accuracy and capacities, but the key functions are mainly two:

         1. Operations Planning and,

         2. Command & Control (C2).

Here the AI will make the difference between the opponents, since the advantage will be measured based on the confidence level of information. Planning and C2 are both dependent on accuracy of information, in terms of intelligence and situation of the operational environment, and wrong plans and/or disrupted C2 can easily lead to mission failure.


It is then necessary to define AI Warfare, which could be “the ability to process autonomous decisions at operational and component level in planning and C2 of operations, disrupting enemy planning and actions”. AI warfare could impose own will to the enemy without shooting a single bullet or flying a sortie.


Therefore, if war is an act of violence intended to compel the opponent to fulfil our will then AI will probably lead to a possible new Cold War era, a war with no hostilities, where deterrence will not be dictated by the nuclear arsenal but by the fastest and most trained and networked super (maybe quantum) computers.