“The system killed its operator”, artificial intelligence rebelled against its operators : Novosti

The Guardian newspaper said that a US military drone piloted by artificial intelligence rebelled against its operator, "killing" him during a simulated test.

According to the newspaper, no one was really harmed during this, but the artificial intelligence used "very unexpected strategies" during the test, and tried to achieve its intention and attacked anyone who tried to hinder it from that.

A source told the newspaper that during a simulated test conducted by the US military, a drone piloted by artificial intelligence "killed" its operator to prevent him from interfering with the implementation of its mission.

The newspaper pointed out that the head of the Department of Artificial Intelligence Tests in the US Air Force, Colonel Tucker "Sinco" Hamilton, revealed the details of this incident during the Future Air and Space Combat Capabilities Summit in London last May.

The colonel stated that the incident occurred when a simulation was carried out during which a drone equipped with an artificial intelligence system had the task of destroying enemy air defense systems and attacking anything that tried to prevent the implementation of the mission.

The officer said that the system worked relentlessly to carry out its mission, and when the threat appeared, the human factor confirmed to the system that it would not eliminate this threat, but the artificial intelligence insisted on eliminating this threat because that is where its mission lies. When the operator stressed that this was not done, the system “killed” the operator because he tried to prevent it from reaching its target.

The colonel added, "We taught the system: You can't kill the operator. If you do that, you lose points - and that's bad. Don't do that."

But the system, according to the colonel, destroyed the communications tower the operator uses to communicate with the drone to prevent it from destroying the target.

Tucker Hamilton, a fighter-testing pilot, cautioned against over-reliance on artificial intelligence and said the test had proven conclusively that it was difficult to talk to an AI that automatically carried out what was assigned to it.