The development of artificial intelligence AI technology is getting faster and faster, its potential has exceeded people's imagination, and concerns about the control of AI technology are also increasing. Many people are debating whether to suspend the development of more powerful AI, fearing that AI will break out of human control and harm humans, and some are worried that AI will spiral out of control and eventually have catastrophic effects on humans. Is there a possibility that AI technology can get out of control, and how catastrophic will it cause to mankind?
1. AI is developed and programmed by humans. Tech developers are working to develop AI that models the human brain to ensure they meet ethical and legal requirements. But the values and goals of artificial intelligence in the system may produce deviations and mutations in the process of training and operation, and secondly, artificial intelligence has the same ability to learn independently as humans, which is an uncontrollable risk factor, and if it is adversely affected by the outside world, it will produce different ideas, such as Google's parent company Alphabet announced that its chatbot Bard contains an artificial intelligence reaction that does not match the facts, Bard said, the first photo of the extrasolar planet where the Earth is located was produced by James M. Photographed by the Webb Space Telescope, and NASA confirmed that this is not correct, and Google AI robots have had the phenomenon of cursing, so there is no human emotion in AI, can not effectively distinguish between correct values and emotions, without the right guidance is prone to loopholes.
2, if artificial intelligence is popularized in all aspects of human beings, there will also be a risk of being hacked, which may lead to these systems out of control or harmful behavior, such as some cars are currently Internet of Vehicles, and car control is completed by the three electric systems, which means that the car may be controlled, and it is not ruled out that AI has its own consciousness, against the will of the driver, the personal safety of the driver cannot be guaranteed.
3, the threat of self-replication, AI has powerful computing power, and has rigorous logic, many devices on the market can already be equipped with a certain AI system, if the carrier equipped with AI system has a strong aggressiveness, once out of control may produce great damage, which is also one of the major risks of artificial intelligence out of control, and AI can completely replicate itself in a very short time, which may also become an important source of security threats for artificial intelligence out of control.
4. Risk of data misleading. If the data used to train AI systems is biased or contains misleading or malicious information, it may cause it to learn the wrong model or rules, resulting in uncontrolled or unexpected behavior in the application, which does not exclude or AI vulnerabilities are generated autonomously and extremist malicious organizations, which also increases the risk of AI getting out of control.
To sum up, in the development, application, upgrading and application of AI, AI has major risk factors that may get out of control, and the development of AI should be cautious, not overnight, but under the condition of ensuring controllability.