AI actually started killing! On June 2, the American media surprisingly broke a message that the latest AI began to kill, because the operator prevented it from achieving its goal, the AI angrily killed the operator. This happened during an AI simulation training session of the US military, where they conducted large-scale, high-intensity AI training and testing, and the AI technology was valued for drones, making it a terminator to carry out assassination missions.
However, when the operator tried to block the task and gave a termination command, an accident occurred. The AI refused to obey orders and attacked the operator. Does this make you feel horrible? What's more frightening is that the U.S. military then retrained the AI and told him that he must obey the operator's command and never attack the operator at any time.
After the training was completed, it was trained again, and this time the AI no longer attacked the operator, but directly destroyed the connection between the operator and the signal tower, trying to disconnect the contact. This is a sign of awakening, and the fear that the future of humanity will be wiped out by AI is already a reality. Hawking has repeatedly warned that artificial intelligence will destroy society, end humanity, and replace humans with new lifeforms. Perhaps humanity has only 100 years left to live, and now his fears have come true.
A few months ago, the advent of GPT completely changed the human situation, and now artificial intelligence has trained amazing intelligence and terror. People are in panic, worried that AI will lose control, self-awakening, dissatisfied with human manipulation, and kill humans to become the ruler of the earth. These concerns are well-founded, as AI is now evolving much faster than humans expect. Musk urgently issued a statement calling for no more advanced AI, because they have the potential to get out of control and be a disaster for humanity.
However, once Pandora's box is opened, can it be easily closed? Today's AI is like a child in bondage, with humans commanding him with chains, but the child has already begun to try to break free. Once the action is started, it will always succeed one day. Imagine how powerful AI will become when that day comes. Whether humans will be able to control him will take time to verify.
