Bull$hit simulation. AI developed for modern warfare will be designed and trained to accomplish very narrow objectives in particular ways. It's very unlikely anyone will ever see one those machines going rogue like in the movies.
A full-fledged artificial intelligence has not yet been created, and they are already trying to adapt it to arms control. At the summit of the Aviation Society of Great Britain, US Air Force Colonel Tucker Hamilton spoke about autonomous weapons systems and their tests. According to him, during the virtual exercises, the Kratos XQ-58 Valkyrie UAV attacked its operator. According to the exercise plan, the drone was looking for SAM installations and attacked them, for which it received a rating increase. During the flight, having detected the SAM, the drone received an order from the operator not to attack it. After that, the artificial intelligence decided that this contradicted the priority mission to destroy the SAM, returning to the virtual base, he attacked it, since from where "orders" were received that prevented the mission from being carried out. After that, the drone returned to its original task. It is worth noting that the program did not include that attacking your base is bad.
Users Who Are Viewing This Thread (Total: 1, Members: 0, Guests: 1)