Will future battlefields be filled with killer robots? It’s hard to say, but artificial intelligence (AI) is already being used in the United States military. Killer robots aren’t on the scene yet, but they could be one day.
But it’s not all about autonomous weapons. AI has many other uses in the military. In this article, we take a look at some of the projects in development, and examine the strengths and weaknesses of this technology.
What is Artificial Intelligence?
According to the Encyclopedia Britannica, artificial intelligence is “the ability of a digital computer or computer-controlled robot to perform tasks commonly associated with intelligent beings.” However, definitions of AI can vary.
AI is usually focused on mental processes like reasoning, learning, problem-solving, perceiving, and using language.
What is the Role of AI in Modern Warfare?
Although most people likely picture scenes from the Terminator movies when they think of AI in warfare, we aren’t there quite yet.
AI is widely used in civilian applications from phones to banking. It’s similarly prevalent in the military. AI is being used and researched in many areas of defense. Here are the main categories.
AI may aid in military intelligence, including surveillance and reconnaissance.
According to a 2020 Congressional Research Service report, “The Central Intelligence Agency alone has around 140 projects in development that leverage AI in some capacity.”
One controversial use of AI for these purposes is the “digital patterns of life” analysis, which uses AI to bring together a person’s entire digital footprint. The above report indicated “servicemembers, suspected intelligence officers, government officials, or private citizens” could be the target of these investigations.
Logistics is a crucial part of military life. People need to be fed and clothed, things shipped from one place to another, and vehicles maintained. AI may help make these processes more efficient.
The U.S. Air Force is starting to use AI for predictive aircraft maintenance which creates tailored service schedules for each vehicle. The Army is also researching the use of AI to analyze shipping flows and costs.
Cyberspace is the newest battlefield. With cyber attacks becoming more sophisticated every year, cyber defense and offense is an important part of modern-day military strategy.
The U.S. Department of Defense is researching the use of artificial intelligence to defend its networks against cyber attacks. This includes identifying malicious code.
AI may also be used to defend against adversarial information operations. With AI becoming more sophisticated, it’s easier to change faces and even voices in photos and videos online. AI could potentially help the military identify these deep fakes.
AI may also aid in bringing data from disparate areas of the military together for increased coordination.
One example is the Defense Advanced Research Projects Agency (DARPA) Mosaic program. Using AI, the program seeks to consolidate information from Army, Navy, and Air Force systems in order to coordinate attacks from all these branches of the military.
Self-driving cars aren’t just for civilians. The Department of Defense (DOD) is also researching the use of AI for both autonomous and semi-autonomous vehicles. These include air, ground, and naval vehicles.
Both the Air Force and the Army are experimenting with autonomous “wingmen” –
vehicles that assist pilots or soldiers. In the Air Force, an older fully automated plane would act as support for a manned fighter jet. Similarly, the Army is working on a remote combat vehicle (RCV) that flanks an “Optionally Manned Fighting Vehicle.”
And yes, there are also AI weapons systems in use and in development with varying degrees of autonomy.
In terms of lethal autonomous weapons systems (LAWS), the Department of Defense directive that covers LAWS states, “Autonomous and semi-autonomous weapon systems will be designed to allow commanders and operators to exercise appropriate levels of human judgment over the use of force.” Some critics point out the directive doesn’t outline what an appropriate level of human judgement is.
There are also some signals the DOD may be tiptoeing closer to more automation. The Congressional Research Service report notes, “Although the United States does not currently have LAWS in its inventory, some senior military and defense leaders have stated that the United States may be compelled to develop LAWS in the future if U.S. competitors choose to do so.”
Israel has developed a fully autonomous weapon called HARPY that destroys radar emitting targets. Drones developed by some countries including the U.S., China, and Russia navigate independently and are very close to being fully automated.
AI-Assisted vs. Autonomous Weapons Systems
In terms of AI weaponry, there are varying degrees of human control. In his conversation with RSnake, Charlie Burgoyne, founder and CEO of AI firm Valkyrie, made the distinction between the AI in Star Trek and Star Wars. He explained that in Star Wars, robots are fully sentient and act independently. In Star Trek, they simply assist their human operators and don’t think for themselves. Autonomous weapons are closer to Star Wars, AI-assisted are more like Star Trek.
Today, we’re closer to Star Trek than Star Wars, but which path militaries around the world will take going forward is still being determined.
The vast majority of AI weapons systems have a human in the loop. An example is fire-and-forget missiles. A human fires them and the missiles use digital imaging to zero in on their targets. The drones many countries have developed are similar, but unlike missiles they can independently loiter, and even gather data, for longer periods of time.
A fully autonomous weapons system would act with little or no human involvement. A human might start or power it up, then it would complete its mission unassisted. It’s possible some drones are already working this way, though this is a matter of debate.
The Types of AI Used in Warfare
Many different processes are involved in creating an AI that can carry out the tasks people have set for it.
In many cases, data needs to be collected. On the battlefield, this is often done using sensors in things like drones, radar systems, and even smart textiles. Then data is evaluated. A number of processes may be involved in this including rule engines, machine learning, data management, and descriptive analytics.
Machine learning (ML) involves “teaching” an AI to recognize patterns for a specific outcome. For example, in order for an AI system to recognize a vehicle, programmers would gather as many pictures of vehicles as possible, then using a machine learning model, let the AI train itself using these photos.
More complicated AI systems like autonomous vehicles require deep learning. This is an artificial neural network that can perform multiple tasks.
Depending on the application, a number of other processes are used including symbolic logic, neural language processing, speech, and vision.
The Advantages of AI in Warfare
AI has the potential to make military operations many times faster and more efficient.
First, there is the ability to bring together data from multiple sources and consolidate it. AI has the potential to unite weapons from land, sea, and air, and help them work together seamlessly. This is difficult for humans to do, especially in the heat of battle. Picture any Vietnam movie you’ve ever seen – with soldiers shouting into malfunctioning walkie-talkies and hoping the helicopter arrives soon.
Computers can react faster than humans. An AI developed for the Air Force trounced human pilots in a simulated dogfight. The humans couldn’t even get a shot in.
Another key advantage to AI is efficiency. AI can sort through data much more quickly than humans can, saving months of man-power.
AI-sorted data can yield valuable insights. An AI system helped the U.S. government accurately predict Russia’s invasion of Ukraine, for example.
Computers are also more consistent than humans in some ways. They never need coffee breaks or have a bad day due to a breakup. And they can retain knowledge and skills better than humans.
The Challenges of AI in Warfare
But there are serious downsides to consider.
An AI system requires a lot of data. When that data is sensitive and highly classified, it can be difficult to share and process it. The enormous amounts of data required can also slow systems down.
Then there are problems with the ability to adapt to new environments. AI systems often work well in test conditions, but have difficulty with the unpredictability of the real world.
They can also be inaccurate. Facial recognition systems are known to have increased errors depending on the race and gender of the subject.
The issue of accountability is also significant. If an AI mistakenly kills someone, who is to blame? It’s no longer a person who made a bad judgement call, it’s simply a machine that malfunctioned. There’s no way to make an AI face consequences for its actions.
There are also ethical implications. As author Kai-Fu Lee pointed out in The Atlantic, “Autonomous weapons lower the cost to the killer. Giving one’s life for a cause – as suicide bombers do – is still a high hurdle for anyone. But with autonomous assassins, no lives would have to be given up for killing.”
There is no cost to a machine for taking a human life. And a machine can’t evaluate ethical choices.
Human beings are facing some difficult questions as our technology becomes more advanced. How we answer them going forward will have far-reaching implications for our future.