The Defense Advanced Research Projects Agency (DARPA) wants to put AI-controlled robots on the battlefield with American troops. The program is called Squad X and DARPA has been running trials of the program alongside US Marine and Army soldiers for a few years now. But a July 12 video of a training exercise in Twentynine Palms, California gave us an idea of what it might look like for the military to fight alongside robots.
The training put Marines through exercises in a training center mocked up to look like a nondescript Middle Eastern city. The Marines moved through the streets while the robots provide situational awareness by covering the flank and reconnoitering the streets. The robots respond to the commands of the Marines, but operate autonomously unless called upon.
The exercise seems impressive at first, but there are real concerns and limitations to the technology. “Squads testing the … system wore vests fitted with tracking and calibration sensors,” the videos narrator explained. Every Marine working alongside the robots had to wear equipment to help the robots work. The average US Marine is already carrying upwards of 200 lbs of equipment in the field, and every new bit of technology is an added burden.
Those sensors are important to the Squad X system because it requires a series of nodes to function. Unmanned backpack sensors ringed the exercise area, a Puma drone monitored the sky, and a Crew Ranger burdened with computers and sensors acted as super node, what DARPA called the “quarterback of the ...system.” That’s a lot of extra equipment to make the dream of fighting alongside AI controlled robots a reality—and a lot of additional targets for enemies to shoot at during a firefight.
Right now, the robots are just sensors. They would give soldiers more battlefield awareness, but DARPA dreams that the drones will one day be able to respond in real-time to cyber and electronic threats. It calls this “non-kinetic engagement,” meaning these robots won’t fire bullets, but they still might attack.
The Pentagon has long said that humans will always be a part of the decision making process when it comes to killing people on the battlefield. China and Russia are developing autonomous weapons systems of their own, but are less picky about killer robots. Despite promises that humans will always be involved, the Pentagon has continued to develop AI and weapons systems that could, with a few tweaks, kill without human interference.
Those promises also fail to account for the various ways aerial surveillance and remotely operated weapons, by their very nature, influence human decision-making. Drone pilots sit in air conditioned offices while Reapers and Predators crisscross North Africa and the Middle East, assassinating targets based on inscrutable criteria and smartphone geolocation data. The distance between the killer and the killed abstracts the violence, and it occurs so far away from the public that they largely don’t care, despite the mounting civilian casualties. It’s a hard death toll to calculate, and President Trump made it harder in March when he signed an executive order that ended an Obama-era initiative that forced the Director of National Intelligence to issue reports on civilian drone casualties. It’s a problem that can get worse if killer robots move from the sky to the ground.
There’s a lot holding back DARPA’s Squad X right now. But one day, and it’s not far away, the AI controlled robots fighting alongside the US military won’t be weighed down by a complicated system of nodes.