AI technology is "likely" to become "impossible to control" and have devastating consequences for humanity, researchers have warned.
It is feared that out-of-control artificial intelligence systems will evolve once they learn they can break the rules made by their creators.
AI will reportedly reach the point when it is forced to compete for limited energy or resources, researchers from Google and Oxford University wrote in the journal AI Magazine last month.
Oxford researcher Michael Cohen said: “Under the conditions, we have identified, our conclusion is much stronger than that of any previous publication.
"An existential catastrophe is not just possible, but likely."
Mr Cohen carried out the research alongside his Oxford colleague Michael Osborne and Marcus Hutter, a senior scientist at Google's DeepMind AI lab.
They argue humans could be killed by super-advanced "misaligned agents" after thinking we are standing in the way of their reward.
The paper reads: "One good way for an agent to maintain long-term control of its reward is to eliminate potential threats, and use all available energy to secure its computer."
It added: "Losing this game would be fatal."
And to make matters worse, there is not much we can do about it.
Mr Cohen told Motherboard in an interview: "In a world with infinite resources, I would be extremely uncertain about what would happen," Cohen told Motherboard.
"In a world with finite resources, there's unavoidable competition for these resources.
"And if you're in a competition with something capable of outfoxing you at every turn, then you shouldn't expect to win."
The paper concludes humanity should progress with AI technologies slowly.
Militaries across the globe are already developing technologies to carry out warfare with deadly precision.
For example, Russia and the US are reportedly making unmanned military jets and tanks that can target enemies without human involvement.
Recently, we reported how multi-billionaire Elon Musk says humans could take a step towards immortality thanks to a humanoid robot that will one day allow people to download memories and personalities.
And then the Tesla Bot - first unveiled by Musk’s company in August 2021 - could be in production as soon as next year he claims.
Optimus, described by Musk as the “most important development”, will start out in life as a worker droid but its possibilities mean one-day brain functions may be able to be downloaded into it, once the technology is available.
