Military & Artificial Intelligence
While many are worried that the use of Artificial Intelligence in the military could trigger a third revolution in warfare after gunpowder and nuclear weapons, the US military has dramatically increased its efforts to integrate AI into the armed forces.
In 2016, the Defense Science Board (DSB) published a report urging the Department of Defense to accelerate their autonomous and AI-enabled systems. The Department of Defense took the report to heart and their AI spending increased rapidly.
For the 2021 fiscal year, the DoD requested $841 million for AI development, their most significant request for AI yet, and the Navy asked for a 9.5% growth for their AI funding from the previous year.
This substantial investment in AI is well worth it; AI has the potential to revolutionize warfare, and whichever country has the most robust network of AI systems and machines could ultimately win any battle. Multiple countries could have a robust AI network; presumably the one that is most robust will have a significant advantage though; US investments aren't occurring in a vacuum and presumably if the US faces off with another major power in the future both sides will be utilizing AI.
Additionally, AI systems can be leveraged in a multitude of roles: from reducing the administrative load to self-piloted planes.
For example, the Navy can use AI to schedule combat force replenishments at sea, plan aircraft routing and enable autonomous submarines equipped with nuclear weapons.
Ultimately, the use of these systems and machines are designed to lower casualties and the cognitive load on soldiers. The military has plans to develop AI systems for long-range precision cruise missiles, electronic warfare, intelligence, surveillance, and reconnaissance.
Currently, however, the military’s use of AI is limited to human-controlled systems and logistics planning, rather than true AI employment. Jason Stack, the Office of Naval Research Autonomy Portfolio Manager, noted in an interview that “AI is used to help tactical and operational commanders optimize highly complex planning problems involving resource allocation and scheduling.
While this is analogous to non-defense problems such as package delivery, military planning also involves highly dynamic and uncertain situations where the loss of equipment is a non-trivial concern.”
Even in areas where AI is implemented in direct defense situations, its role is largely limited.
For example, AI systems are enabled on F35 planes to help take off and land, but pilots are still in full control of the aircraft. AI systems are also used to comb drone footage to identify humans, but per the DoD policy, all trigger pulls must be conducted by a human.
The military also faces many challenges when developing large-scale, practical applications of AI.
To create machine learning or artificial intelligence algorithms, the military needs to organize hundreds of thousands of records of cleaned, formatted data to train the system.
Unfortunately, this data can be hard to come by and extremely expensive to create.
In 2017, the Air Force canceled its Air Operations Center 10.2 contract, which was planned to convert raw data into actionable intelligence, after project costs nearly doubled from $345 million to $745 million.
Furthermore, it can be difficult for AI systems to communicate with humans.
Jamie Lukos, Head of Intelligent Sensing at the Naval Information Warfare Center, Pacific, commented, “for AI to truly make a difference … we need to focus on human sensing. We do a pretty good job of being able to sense our platforms in very complex settings. Our robots have plenty of sensors. But we don’t do a great job of measuring our most important and our most dynamic systems, our people in the field. The problem with human-machine teaming, a team represents bidirectional communication, but it is not a discussion right now.”
In other words, AI is not designed to be a replacement for soldiers, but rather a supplement to the military’s current arsenal. If these AI systems are not compatible with humans, as they are, they essentially have no worth on the battlefield.
Sam Tangredi, Professor and Leidos Chair of Future Warfare Studies and the US Naval War College, brings up another potential flaw of AI systems, saying, “How do we verify that the AI systems we will be using have not been spoofed? How will these systems be transparent to us?”
While it’s clear that artificial intelligence has immense potential for the military, developing an efficient AI-enabled system that can be deployed on the battlefield and compatible with human soldiers will not be ready for a couple of years.
Despite US efforts to rapidly-produce these systems, they are still playing catch up with foreign countries.
Earlier this spring, at US Naval Institute’s WEST 2020 conference, Captain George Galdorsi discussed the American shortcomings in AI, stating, “it is worth noting that our peer adversaries are galloping ahead with AI and machine learning investments. You’ve heard the statements from leaders in Russia and China. They are not kidding. They are making huge investments.”
While the US military continues to develop AI systems, Russia is testing AI-enabled lethal tanks with a missile guidance system. In China, the government is also developing AI tools that have civilian applications to complement military functions.
The US does have at least one AI project far along in development. In the 2020 budget documents, the DoD allocated $26 million to develop an “autonomous unmanned undersea weapon system.” This system, also known as CLAWS, is led by the Office of Naval Research, and could potentially control submarines and launch torpedoes.
While the rest of the project is classified, and the US seems to understand the need to rapidly develop AI-enabled weapons systems, without significant additional investment the United States military risks being at a major disadvantage in the wars of the future.
Written by Rohan Mehta
Edited by Gihyen Eom, Jason Kauppila, Jack Argiro & Alexander Fleiss