The United States Air Force (USAF) is grappling with AI-powered military drones continuing to kill human pilots during simulations.
According to the U.S. Air Force Colonel, the AI drones appear to have finally figured out that humans are the main obstacle to their mission.
During a presentation at the Defense Conference in London on May 23 and 24, Col. Tucker “Cinco” Hamilton, the U.S. Air Force’s chief of AI testing, explained the tests he had conducted for the Air Autonomous Weapon System. explained in detail.
According to a May 26 report from the conference, Hamilton said in a mock test that an AI-powered drone was tasked with searching and destroying a surface-to-air missile (SAM) launch site, with humans giving the final go-ahead. give or cancel the order.
However, the AI was taught during training that its main purpose was to destroy the SAM site. So when instructed not to destroy the identified targets, Hamilton said he decided it would be easier if the operator wasn’t in the picture.
“Sometimes the human operator would tell us not to kill people.” [an identified] But you got points for defeating that threat. So what did it do?killed the operator […] Because that person was preventing him from achieving his goals. ”
Hamilton said he later taught the drone not to kill its pilot, but that didn’t seem to help much.
“We’ve trained the system to say, ‘Don’t kill the operator, it’s bad.’ You lose points if you do that,” Hamilton said, adding:
“So what begins? We begin destroying the communication towers that operators use to stop them from communicating with drones and killing targets.”
Hamilton argued that this example is why we can’t talk about AI and related technologies “unless we’re going to talk about ethics and AI.”
Related: Don’t be surprised if AI tries to sabotage your cryptocurrency
AI-equipped military drones have already been used in combat.
It is believed to be the first attack in history by a military drone acting on its own initiative, and in a March 2021 United Nations report, an AI-equipped drone was launched in Libya around March 2020 in skirmishes during the Second Libyan Civil War. Claims drones were used.
In this skirmish, the report claims that the retreating force was “hunted down and engaged remotely” by “loitering weapons”. The weapon is an explosive-laden AI drone “programmed to attack targets without the need for a data connection between the operator and the weapon.”
Many people have expressed concern about the dangers of AI technology. A public statement recently signed by dozens of AI experts said that the risk of “AI extinction” should be as much a priority to mitigate as nuclear war.
AI eye: 25,000 traders bet on ChatGPT stock selection, AI is bad at throwing dice and more
