Industry experts agree that the future of warfare is automated, with a single human operator teaming with autonomous systems to carry out missions. But what steps need to be taken to get there?
AI and defense experts shared their insights into that question at a Hudson Institute event on March 31, saying that even though we have a long way to go until AI can be operationalized, efforts related to policy, data, and acquisition can move the United States closer to that capability.
Scott Gilloon, sector vice president of Air Force Strategic Development at General Atomics Aeronautical Systems, said that one of those key steps is policy.
“I think when we start talking about a legacy system and where we were with automation, it is very important to look at it in terms of that was done because of policy,” said Gilloon. “It’s not that industry said, hey, here is a … state of the art [system], it’s [government saying] hey, here’s what I want you to do, industry. So, requirements matter, policy matters.”
While policy can slow AI adoption in warfare, Dan Javorsek, president of EpiSci, explained that focusing on battlefield outcomes – such as faster decision-making and operational efficiency – rather than perfecting individual AI systems, can help bypass bureaucratic delays.
Outside of policy matters, technical and operational steps also can be taken to deploy AI in command-and-control (C2) environments – which require a larger breadth of decision-making and capabilities than current AI systems are typically tasked with, Gilloon said.
One of those steps includes merging operator expertise with data science capabilities.
“We’re trying to execute within the commander’s intent, to affect some strategic goal – that’s a pretty big sort of compilation of activities that’s going on,” said Gilloon. “It also requires a lot of … contextual elements … that’s the piece that … we’ve been approaching to try and understand.”
Addressing the Federal government’s hesitation to fully leverage commercial technology can also improve the military’s acquisition and deployment of AI.
“The government has been saying that they want to leverage commercial for over a decade, and yet VC [venture capital] backed companies haven’t really penetrated into the defense market much,” said Javorsek. “We haven’t been thinking about leveraging commercial the right way, which is to minimally modify a commercial stack and then rapidly iterate that in the environment.”
“So, when it comes to the C2 piece it’s not a technical problem, it is a trust and a policy and a confidence in what is capable at the tactical edge,” he continued.
Another critical step in deploying AI in warfare is overhauling how the military collects and leverages data. Unlike commercial tech companies, which excel at data-driven innovation, the military remains stuck in outdated and siloed collection methods, said Kevin Chlan, senior director for Air Dominance & Strike at Anduril Industries.
“There’s no way at a data level to collect [information on certain interactions in the field], and I don’t know how much further along we are today than we were eight years ago,” said Chlan.
“I don’t know what that strategy should look like … but I do think that in order for us to truly capitalize on some of these emerging technologies that we see coming out in the commercial sector around us, there are ways that they have figured out how to harvest that data and exploit it in really novel and interesting ways, and we have not yet done it,” he explained.