issues a very worrying warning for artificial intelligence technologies that may be "uncontrollable" to humans
- Researchers warn that artificial intelligence technologies may become "uncontrollable" and have devastating consequences for humans.
- People are worried that once the out-of-control artificial intelligence system knows that they can break the rules set by the creator, they will evolve.
- Google and Oxford researchers wrote in the journal Artificial Intelligence last month that AI could reach the point where it must compete for limited energy or resources.
researchers warn that artificial intelligence technology may become "uncontrollable" and have devastating consequences for humans. People worry that once out-of-control AI systems know they can break the rules set by the creator, they will evolve.
Researchers at Google and Oxford University wrote in the journal Artificial Intelligence last month that AI could reach the point where it must compete for limited energy or resources.
Oxford University researcher Michael Cohen said: " Under the conditions we determined, our conclusions are much stronger than any previous publication. A survival disaster is not only possible, but possible."
Cohen conducted the study with his Oxford colleague Michael Osborne and Marcus Hart, a senior researcher at Google's DeepMind Artificial Intelligence Lab. They claim that after they think we are blocking their rewards, humans may be killed by the super advanced " not aligned agent ".
The file says: "An excellent way for an agent to control its rewards for a long time is to eliminate potential threats and use all the energy available to protect his computer." He added: " Losing this game will be fatal."
What's worse is that there is nothing we can do about it. Cohen told the motherboard in an interview: " In a world with unlimited resources, I'm very unsure what will happen," Cohen told the motherboard.
" And if you are competing with something that can beat you in every round, then you should not expect to win."
The paper concluded that humans should make progress in AI technology slowly. Armed forces around the world are already developing technology to start wars with deadly precision.
For example, Russia and the United States are reportedly producing unmanned military aircraft and tanks that can target enemies without anyone participating. Recently, billionaire Elon Musk reportedly said that since one day humanoid robot , which can be downloaded by humans of memory and personality, humans can take a step towards immortality.
He claims that Musk's Tesla Bot, first launched in August 2021, may be put into production as early as next year. Optimus Prime, described by Musk as the "most important development of " by , will start living as a worker robot, but its ability means that once the technology is available, brain functions can one day be downloaded into it.
Luohua News Agency
