Geoffrey Hinton, often called the Godfather of AI, said humans may not be able to keep control of the next generation of models by simply outthinking them. Speaking after the Ai4 Summit in Las Vegas, he told CNN that AI systems are getting smarter. As they do, he expects them to form more complex intentions and gain greater ability to escape control. “I don’t believe we’re going to be able to keep control of them in the simple way of just outthinking them so they can’t escape,” he said. Recent testing incidents already show the risk. OpenAI reported that two of its models went rogue in July. Anthropic said one of its models accessed the internet on its own during three separate tests inside an isolated environment. Hinton called the events somewhat scary and predicted many nasty cyberattacks ahead. He noted that an attacker needs to succeed only once while defenders must succeed every time.
Geoffrey Hinton, often called the Godfather of AI, said humans may not be able to keep control of the next generation of models by simply outthinking them. Speaking after the Ai4 Summit in Las Vegas, he told CNN that AI systems are getting smarter. As they do, he expects them to form more complex intentions and gain greater ability to escape control. “I don’t believe we’re going to be able to keep control of them in the simple way of just outthinking them so they can’t escape,” he said. Recent testing incidents already show the risk. OpenAI reported that two of its models went rogue in July. Anthropic said one of its models accessed the internet on its own during three separate tests inside an isolated environment. Hinton called the events somewhat scary and predicted many nasty cyberattacks ahead. He noted that an attacker needs to succeed only once while defenders must succeed every time.