AI and the future of warfare

Artificial Intelligence (AI) and Machine Learning (ML) will revolutionize warfare as we know it, plain and simple. Just as the airplane and the aircraft carrier ended the days of the big battleships and dreadnaughts, AI is going to have a profound impact on the future of warfare and how battles will be fought. We may not like, understand, or even agree with this concept, but that doesn’t mean it’s not going to steamroll right over us if we do not start to prepare for it.  Sun Tzu said in The Art of War, “If your enemy is secure at all points, be prepared for him.”

What I hope to outline in this article are some basic understandings of what AI and ML are as well as their differences. Then we’ll discover how this new technology can and will be used in the future of warfare and why it is imperative that our leaders and military prepare for what is coming. 

Photo by Markus Spiske on Unsplash
Photo by Markus Spiske on Unsplash

Artificial Intelligence Versus Machine Learning

AI and ML are two important studies in the world of computer science. The definitions of these two are always changing as new research is conducted by both the academic and industrial worlds. There are many misconceptions about their capabilities. Though frequently treated as synonyms, there are several key differences that distinguish artificial intelligence and machine learning from one another.

At present, the focus of ML is on acquiring skill or knowledge. Though it is an offshoot and important component of AI, ML has a narrower focus and is much less capable.

 When a ML system is programmed into a computer, it gives that computer the ability to perform certain tasks, typically a singular task it can become an expert at and learn from how it performs them and then improve upon how those tasks are carried out. Its task could be sorting labeled data into groups or finding a pattern in a dataset. For example, a computer could be given information from a historical event. Depending on the instructions provided, it might take that data, examine it, find a common pattern, and then use the pattern to draw a conclusion that can be used to predict a future event. 

Photo by Joshua Sortino on Unsplash
Photo by Joshua Sortino on Unsplash

A computer with ML can learn from past experience. When given new information, it will look at previous data and compare it with what it has newly received, then either sort the new information into groups or adjust its view to better fit the new information (depending on what task the computer is trying to complete). The goal of ML is to make the most accurate prediction possible with the data it has received.

In contrast, AI is mainly concerned with success rather than accuracy. AI has a much wider range of capabilities than ML, and is a field that focuses on creating a smart computer system—a man-made “thinking power”, if you will—capable of successfully acquiring and applying information, knowledge, or data on its own in order to complete a task. AI’s long-term goal is to enable an intelligent system to solve a variety of complex problems as well as, or better than, the human brain.

AI doesn’t need to be pre-programmed to do specific tasks when implemented into a computer system. It is self-learning and makes decisions with the help of ML processes such as reinforcement learning algorithms and deep learning neural networks.

The practical applications of AI and ML

ML can be broken down into four main subtypes. These are: supervised, unsupervised, reinforcement, and deep learning. All four types of machine learning have different levels of ability and purpose, though all deal with datasets and information sorting.

Supervised ML is used to classify data. It is highly structured, meaning it deals with labeled datasets. A machine programmed with supervised learning is taught to sort labeled data with a desired output. It can learn from its mistakes and self-correct when new data is introduced in order to produce the most accurate results possible. One military example of supervised learning would be in image analysis, such as looking at drone feeds for specific targets.

Photo by Mitch Nielsen on Unsplash
Photo by Mitch Nielsen on Unsplash

Unsupervised ML is mainly used for pattern recognition, clustering data, and descriptive modeling. It is unstructured, meaning it’s taught with unlabeled data and does not use output categories or labels when sorting information. Unsupervised learning is often used to find meaningful patterns in datasets.

Reinforcement ML observes and interacts with its environment to gather useful data, then makes a calculated action that will “maximize the reward or minimize the risk”, depending on the task it is trying to accomplish. An example of reinforcement learning would be computer programs that are able to beat humans at a game of chess. Imagine wargamers at the Pentagon being able to simulate how a war might play out in Korea or the Middle East prior to a military action being taken—this is the kind of future capability this new technology will bring.

Deep ML, the final and most complex subset, imitates the human brain’s ability to process data and find patterns which are later used to help make decisions. It learns entirely on its own from unstructured and/or unlabeled data. Examples of deep learning use in machines are text generation technology, automatic translations, and the development of self-driving vehicles.

Deep learning was used and expanded upon by the United States military during Operation Iraqi Freedom. When US Forces had to travel from the airport to the Green Zone, they had to travel along a road called Route Irish—it was the most heavily IED laden road in Iraq. The Army and Air Force placed surveillance assets (drones, cameras, and aerostat blimps) to monitor the route. Aside from looking to see if it spotted a human placing an IED along the side of the road, it would take images of the road and then compare it against previous images to see if it spotted something new. If the computer program spotted some new loose dirt or an object near the road, it would send a warning to the vehicles traveling near it and dispatch a road clearing team to investigate. This is one example of how AI and ML are able to be integrated with surveillance to detect and prevent IED attacks on American Forces. 

Photo by NASA on Unsplash
Photo by NASA on Unsplash

Moving on, we now look at artificial intelligence, which is broken down into three main subtypes: weak, general, and strong. Though strong AI is still but a distant dream of the academic and industrial worlds, both weak and general AI have been developed and are currently in use. 

Weak AI is great at focusing on tasks it has been taught to do and can complete those tasks extremely well, but it isn’t very flexible and does not have much capability. Examples of the uses of weak AI are virtual personal assistants (such as Amazon Alexa and Apple’s Siri), email spam filters, and NPCs—non-player characters—in video games.

General AI has much wider capacity than weak AI, but it has not yet reached its full potential. Because general AI can only draw upon the information it has, it will become more knowledgeable over time but cannot formulate its own decision at present. General AI is often used in recommendation systems on YouTube, Facebook, Google, Amazon, etc.

Strong AI has not yet been achieved, but scientists hope it will someday surpass human intelligence. Strong AI will be self-aware, learn independently, make judgements, reason, solve complex problems, and plan. It will also be capable of communication and objective thinking. This is the type of AI that is often described in science fiction movies and novels.

James Rosone and Miranda Watson have incorporated the advancements in both ML and AI into their new series, the Monroe Doctrine. In their fictional series, the government of China creates the most powerful AI in the world to help their government manage their growing economy and country, but also help them achieve economic, political, and military dominance over the world. When the leaders of China give the AI, named Jade Dragon, the objective of defeating NATO, the AI orchestrates a series of events that leads to the near economic collapse of the West before it initiates a full-on military surprise attack on a weakened America and her NATO allies. The Monroe Doctrine illustrates how if a machine is given full use of a nation’s current economic and military capabilities, it might use those assets to wage an unrestricted war on a given target.

To purchase the series or learn more about it, click here.

Banner Monroe.jpg

AI as a military tool

Though AI is still of limited use at present, it is developing rapidly. AI is remarkably effective at completing tasks dealing with large amounts of data (labeled or unlabeled), such as image recognition, prediction systems, recommendation systems, anomaly detection, and language translation. An artificial intelligence could be a great asset to the military by assisting with fraud detection, predicting when maintenance issues will cause a weapons system to fail, or suggesting a winning strategy in a conflict simulation.

In many cases, an AI has the capability of identifying a seemingly insignificant fact, action or item that can tip off the military to something before it happens. For example, AI could help the military identify specific precursors to an action made by a possible enemy. If an AI suddenly saw the Chinese making large fuel and munition purchases on the open market, and most of it was being sent to the ports and military facilities near Taiwan, then the military could make a good assessment that they were looking at either a large-scale training exercise or a potential military attack against Taiwan by mainland China.

The most successful integration areas of AI into the military thus far are when dealing with large amounts of labeled data, such as identifying a T-90 battle tank from a satellite image and using facial recognition to identify a high-value target in a crowd of civilians. However, AI falls short of human-level intelligence in many other areas, including understanding the context of speech (such as sarcasm and regional dialect), situational awareness, working in unfamiliar situations, and solving multiple types of problems.

Photo by Chris Liverani on Unsplash
Photo by Chris Liverani on Unsplash

Though the ability has not yet been realized, scientists dream of an AI that can be used by the military to conduct missions, achieve sensor fusion, and automate tasks. Extensive research is continually ongoing and has already resulted in shorter training times and consistently better results for AI systems.

The future of AI in modern warfare and the battlespace

Artificial intelligence has great potential and will play an important role in the future of our world, especially regarding modern warfare. One of the areas where AI will be key is in military command centers. In a military command center, many personnel are doing various tasks, and dozens of screens around the room show hundreds of different things happening simultaneously in different locations. The commander oversees everything, and it is his or her job to keep track of all the moving parts being thrown at them.

Because of the sheer number of decisions to make, it can be easy to get lost in the deluge of information and end up focusing on something unimportant while missing a crucial decision that would have negative ramifications for the mission. An AI in this situation would be able to help sort through and prioritize important information to lighten a commander’s load so that he or she can make the best decisions possible.

However, there is a dark side to this potential, as an AI might misinterpret a mistake or accident as a first-strike attack and bring its conclusion to the commander. The commander would react accordingly and retaliate against the perceived enemy. There was a case of this happening in 1983 when the Soviets had a glitch in their computer system. Their brand new early-warning satellite system, which had just been installed, had mistakenly detected a nuclear missile launch by the US. The Soviets thought they were under attack by the US and were minutes away from launching all their nukes when one of the personnel realized it was just a computer error and not a real threat. If a human being hadn’t caught that mistake, the AI would have caused a nuclear war.

Photo by Jefferson Santos on Unsplash
Photo by Jefferson Santos on Unsplash

In conclusion, while there is still much more to learn about AI and ML, I hope this article has given you some insight and ideas into how this technology can and will be used by militaries in the future. We are likely to see this deployed in everything from unmanned combat aerial vehicles and drones to decisions of when and where to deploy soldiers and supplies. The military applications for AI and ML are almost endless, as are their civilian capabilities. While this may seem daunting to contemplate, the more we learn to understand how both AI and ML can and will be used the more we can implement safeguards against it and learn to harness its power for our good and our own defense. 

I hope you enjoyed this short article, please feel free to leave a comment below and be sure to check out some of our fictional works where we employ many of these ideas into our writings. 


Similar Posts