There's an interesting division among journalists covering autonomous technology in the defense industry. One faction is adamant that attempts to create "virtual crewman" to replace human pilots in tanks and ground vehicles is distinct from so-called "killer robots."
The other faction thinks this is all the first step toward a Terminator-like dystopia.
The latest chapter in the ongoing debate concerns an automated tank turret. Under direction from top brass, the Army's Night Vision and Electronic Sensors Directorate -- an Army development laboratory -- is now working on an automated turret for live-fire testing next summer. The news was first reported by Breaking Defense, which follows the defense industry.
The turret will use the Army's controversial Artificially Intelligent Targeting System (ATLAS), which can detect and aim a 50mm gun while autonomously determining whether a target is hostile. The system is not set up to pull the trigger, a step that still must be taken by a human actor.
ATLAS, announced earlier this year, is an ambitious program to combine computer vision, AI, and machine learning to create ground vehicles capable of making targeting decisions and taking appropriate preparatory action at superhuman speeds. In a sign that some things haven't changed since the Old West, the idea is that the quickest draw will typically win a fight.
Officially, autonomous and semi-autonomous weapons systems are subject to strict limitations under Department of Defense Directive 3000.09, which requires a human in the loop in use-of-force situations and prescribes limitations on autonomous weapons to prevent them from firing in case communications are lost.
But critics argue that these restrictions, which, it bears pointing out, are subject to being changed through another DoD directive, don't go far enough to safeguard against the proliferation of autonomous weapons. The Campaign to Stop Killer Robots is one group that wants more stringent assurances that humans will continue to exercise total control over the use of force, ideally in the form of a global autonomous weapons ban.
In 2017, following a letter from more than 100 notable technology personalities, including Elon Musk, the UN's Review Conference of the Convention on Conventional Weapons began formal discussions on the threat of autonomous weapons.
Nevertheless, development of such systems has been speeding along. Last April, robots cleared obstacles and breached barriers for manned tanks in a joint U.S.-British training exercise and demonstration in Germany, the first time robots have been used for that purpose. Another Army program called Operation Wingman has included range tests of self-driving vehicles equipped with fully autonomous weapons systems.
Mirroring the nuclear buildup of the 1950s and later, the quandary of autonomous weapons systems is further blurred with the understanding that countries like China and Russia are developing their own unmanned weapons platforms. That could mean we're already well on the way to a future in which robots are commonplace on the battlefield. One potential consequence of that outcome, as I've written, could be that operators in the field are less likely to hesitate before engaging an unmanned system, upsetting long-standing operational restraints that serve to avert escalation to armed conflict.
According to an interview with Breaking Defense, the Army's acquisition chief, Bruce Jette, has left the door open to a less hands-on form of control, wherein humans are supervising a larger number of autonomous guns, such as the ATLAS-powered turrets.