This month, the European Parliament called upon all nations to hem in the use of military artificial intelligence (AI) on the battlefield, but also to set definitions along with ethical principles on what decisions AI can make. Among the concerns isn’t the science fiction view that a machine will rise up and turn on its human masters, but that AI could make for simply more powerful weapons that increase the collateral damage, and aren’t bound to accept rules of engagement.
A report published earlier in January by the Center for a New American Security (CNAS), titled “AI and International Stability: Risks and Confidence-building Measures,” also laid out the potential threats that machine learning and AI could bring if not utilized carefully. The authors of the report suggested it is vital to assess how the militarization of AI could affect international stability, and that the adoption of AI into national security applications and warfare could pose a genuine risk.
The Role of AI
The most important consideration when discussing AI and machine learning is exactly what is meant by these terms. It is all too easy again to immediately think of terminators, but also seemingly sentient machines. AI, as it is being used today, is nothing more than computer-generated algorithms that serve to solve rather simple tasks.
Siri on Apple devices and Alexa on Amazon products are very good examples of how AI is being employed today. While it might seem that these voice-activated devices have intelligence, in reality it is code – complex code – but still computer code.
“People think of AI as being self-aware, but that is not really the case today, and may never be so,” explained Jim McGregor, principal analyst at TIRIAS Research.
“In its current applications, AI is just designed to make key decisions and solve relatively simple problems,” McGregor told ClearanceJobs.
Therefore, AI isn’t going to be used to program legions of robotic soldiers – and it is likely we are decades away from even employing reliable robotically controlled vehicles on the battlefield. Russia has already tested some platforms in Syria with middling results. However, AI could have a place with the military in other ways in the not-too-distant future.
“AI could play a transformational role reshaping the nature of warfare at virtually every level at which it is conducted,” said John Hale, Ph.D., chairperson of computer science and professor of bioinformatics and computational biology at the University of Tulsa.
“The first thought is how AI may take soldiers out of the loop through robotics and autonomous vehicles, and indeed things are edging this way,” Hale told ClearanceJobs.
“AI will also support human decision making by providing warfighters with new intelligence from oceans of data gathered by sensors in the theater of combat,” Hale added. “AI also has the potential to lend significant support to tactical and strategic decision making in planning military operations. The technology will inevitably result in new efficiencies and the ability to wage war with greater precision; that much is clear.”
The Risks of AI
Many in the tech sector have also warned that the ethical considerations and questions about the global impact of warfare driven by AI are deep and profound.
“Artificial intelligence is certainly likely to alter the character of warfare over the next several decades, hence the major funding commitments being made by countries like China,” said Christopher Whyte, assistant professor in the homeland security and emergency preparedness online program in the L. Douglas Wilder School of Government and Public Affairs at Virginia Commonwealth University.
AI will certainly have the potential to make militaries far more lethal. The United States and its near-peer adversaries including China and Russia have been developing what could utilize AI to “enhance” current platforms, while using AI and machine learning as a force multiplier in the physical space as well as the cyber environment.
“Drone swarms, advanced sensing and intelligence aggregation from Big Data allows for superior maneuverability in the battlespace,” Whyte told ClearanceJobs. “In areas like cyber conflict, AI promises to be particularly transformative. There, rather than simply enhance attack capabilities; it is likely that AI will allow adversaries to subvert the current state of the art of defensive efforts.”
While AI could certainly reshape and transform the character of warfare, it will have many of the same risks militaries always face when adopting innovative new technologies.
“Technology has to be implemented in ways that avoid some in-built risk of insecurity,” explained Whyte. “Early decisions made by those who built the first computer networks rejected in-built security management, thus creating a path dependent situation that meant vulnerabilities in anything that relies on the Internet to function. With AI, we have to get issues of safety right now rather than give thought to them later, otherwise we might find ourselves with tools that only understand the world via the biases we’ve baked into their design.”
Another risk could be the human factor in this new technology adoption. History is chock full of militaries that fail to marry technological potential with real strategic need suggested Whyte.
“To get AI right means unraveling why military organizations sometimes reject obvious advancements or focus too much on one application when so many others beckon,” he added. “Done right, AI could ensure America’s competitive edge out to mid-century. Done poorly, AI could build our adversary’s abilities to such a degree that it becomes prohibitively expensive to change our approach.”
As for the danger of the smart weapons turning on humans, McGregor dismisses that notion.
“AI is already shaping the future of warfare, and we have already made a leap with drones, so now it is how we can replace the pilot but also smart weaponry,” said McGregor. “There is no risk to lose control as the majority of AI is built around doing those specific tasks.”
The Rewards of AI
Where there is risk there can be reward, and in the case of AI, the greatest reward or benefit could be the ability to keep warfighters out of harm’s way, but also how it can improve efficiency.
“It seems likely that the biggest impact on how militaries operate will be with logistics, the supply chain and long-term planning,” said Whyte.
“Conversations about AI safety and how realistically a smart algorithm can proxy for human expertise are absolutely worth having,” he added. “But the next decade or more is likely to be the tale of mile-wide-but-inch-deep upgrading of any facet of military operation that can be done faster and more adaptively by an AI system.”
That could include a supply apparatus so sophisticated that it uses Big Data to not only predict with some degree of probability the development of some new hot spot, but also to redirect resources to ensure relevant units are optimally deployed nearby with the most appropriate equipment to meet the challenge, suggested Whyte.
“This is the kind of advance that we can build new smart technologies around,” Whyte told ClearanceJobs, “like novel unmanned systems that are automated at every stage from production to deployment to reprovision.”