Drones being evaluated by the US military could soon be equipped with an artificial intelligence that is claimed to be better than humans at identifying targets, although the classified nature of the work makes it difficult to verify this claim.
Stephen Bornstein of Athena AI, the Australian company behind the system, says the AI will assist human drone operators, who can lose concentration after hours spent looking at streaming video.
鈥淭he AI will do a lot of the heavy lifting for them,鈥� says Bornstein. 鈥滻t can look for, identify and geolocate objects on the ground as well as checking if an object is in a no-strike area and the collateral damage risk.鈥�
Advertisement
Although the AI hasn鈥檛 yet been used in combat, Athena is working with Red Cat, a US firm that provides drones to the US military, to implement its technology for the US Army. Red Cat is also sending 200 of its drones to Ukraine, and while they won鈥檛 be equipped with Athena鈥檚 AI system, they could be upgraded to use it at a later date.
An Athena company video shows the AI identifying and tracking tanks and other military vehicles, and spotting people on foot. It can even help determine whether a person looks like an enemy soldier.
鈥淚t can check whether people are wearing a particular type of uniform, if they are carrying weapons and whether they are giving signs of surrendering,鈥� says Bornstein.
This ability means the AI can warn the operator not to attack invalid targets. Bornstein says the AI will improve outcomes and meets Geneva Convention standards for the humanitarian use of new weapons.
鈥淲e have worked extensively with military legal officers and gone through extensive scientific testing, which has shown that our system performed better than a human at identifying targets and non-targets in dynamic targeting scenarios, and resulted in a better legal outcome than human operators alone,鈥� says Bornstein.
Because details of this testing program are classified, New 杏吧原创 hasn鈥檛 been able to verify this claim. Athena was able to share a presentation by the Australian government鈥檚 Defence Science and Technology Group that concluded the AI system gives a tactical advantage and could provide 鈥渓egal and ethical guidance in the heat of battle鈥�.
at the University of California, Berkeley, who campaigns against autonomous weapons, says the claims are difficult to evaluate given the military secrecy involved. 鈥淲ithout seeing their testing methodology and data it鈥檚 hard to say,鈥� he says.
Russell points out that other AIs that were claimed to have better-than-human performance, such as ones that could spot skin cancer from phone pictures, turned out to perform poorly .
He is also concerned that AI recognition systems can be fooled. For example, in 2017 researchers designed a 3D-printed turtle and tricked an AI into identifying it as a rifle. The US military is already working on deploying this type of camouflage.
