JUANA SUMMERS, HOST:
Whatever happens in the conflict with Iran, the face of war itself is rapidly changing. Defense Secretary Pete Hegseth has announced a major new military command to accelerate the use of drones and artificial intelligence in warfare. Meanwhile, Congress is considering restrictions on AI for the military. WUNC's Jay Price reports from Fort Bragg, North Carolina. And we'll note his report contains the sound of gunfire.
(SOUNDBITE OF DRONE BUZZING)
JAY PRICE, BYLINE: This drone is headed directly for a truck driven by, well, itself. An autonomous gun atop its cab shuddered almost nervously as its sensors locked onto the attacker, and in seconds...
(SOUNDBITE OF GUNSHOT)
PRICE: This was part of a demonstration on Fort Bragg of an autonomous system for blasting through minefields. There were two trucks, inexpensive-by-military-standards Ford pickups. One demonstrated the robot gun shooting down two drones with ease. The other drove away on its own, several minutes through rugged terrain. Then it fired a long explosive charge that cut a path through a minefield, an operation that normally risks the lives of soldiers. I was standing only a few yards away when the robot gun shot the drones, but one of the soldiers helping oversee the exercise assured me a human was ready to intervene if something went wrong.
UNIDENTIFIED SOLDIER: So it will identify the target. There's a human on the loop to decide whether to engage the target.
PRICE: Automated weapon systems like these are only part of what militaries around the world are seeking as they move to rapidly adopt AI. Hegseth decreed earlier this year that the U.S. military will become an AI-first war-fighting force. But some members of Congress say there aren't enough guardrails to ensure AI can be used safely.
(SOUNDBITE OF ARCHIVED RECORDING)
MARK KELLY: We need a clear answer on this.
PRICE: Arizona Democrat Mark Kelly during a Senate Armed Services Committee hearing.
(SOUNDBITE OF ARCHIVED RECORDING)
KELLY: Whether or not a human has to make the final decision to strike the target, or can a system execute the engagement autonomously? These are questions we haven't yet dealt with.
PRICE: But now they're trying. The committee added a host of provisions to the annual military policy and spending bill that, among other things, require human oversight of AI-controlled weapons. Paul Scharre, with the Center for a New American Security, authored two books on military AI and led a working group that drafted the Pentagon's first policy on autonomous weapons.
PAUL SCHARRE: The consequences of getting it wrong could be really bad. And not just you bombed the wrong target, which could cause harm to civilians, but it also could create a geopolitical crisis if people are trusting AI and maybe it's not trustworthy yet.
PRICE: Scharre notes that the military for years has had basic forms of automated targeting, typically defensive systems that quickly stop incoming threats.
SCHARRE: And I think what we're likely to see over time is a sort of slow expansion of this speed in warfare such that humans can't really keep up, right? So maybe a human's authorizing a target list of a whole bunch of targets, but the human doesn't really have time to, like, pore through each one individually and analyze it.
PRICE: Indeed, even with a human involved, things can go wrong. The U.S. uses an AI platform called the Maven Smart System to help find and identify targets. In the first hours of the Iran War, it did so at 10 times the speed previously possible, according to analysts at the Center for Strategic and International Studies. But experts say that kind of speed could lead to mistakes by the humans trying to oversee the AI systems. Joe Chapa is a retired lieutenant colonel who helped establish AI policy for the Air Force and Space Force. He says that potential for mistakes makes it crucial for the military to create proper training programs.
JOE CHAPA: Every person who's going to touch an AI system, which very soon is going to be every person, needs at least a baseline education in how statistical AI works.
PRICE: The Senate bill also calls for the Pentagon to keep data on battlefield decisions involving AI. Chapa says that's crucial so the military can understand mistakes and prevent them from happening again, since today's AI models are so complex that even their own developers sometimes struggle to understand their behavior. For NPR News, I'm Jay Price at Fort Bragg, North Carolina.
(SOUNDBITE OF REVERIE SONG, "GIVE IT TIME") Transcript provided by NPR, Copyright NPR.
NPR transcripts are created on a rush deadline by an NPR contractor. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.