In the future, AI systems will be able to control people as easily as an adult can bribe a 3-year-old with candy, warns Jeffrey Hinton, known as the “godfather of AI.” To avoid this, he has proposed a solution.
In the future, AI systems will be able to control people as easily as an adult can bribe a 3-year-old with candy, warns Jeffrey Hinton, known as the “godfather of AI.” To avoid this, he has proposed a solution.
Jeffrey Hinton, a Nobel laureate in computer science, is known for his groundbreaking work on neural networks, which helped pave the way for today's AI boom. In 2023, he left Google and began speaking out about the dangers of AI.
Hinton says he used to think it might take 30 to 50 years to achieve artificial general intelligence (AGI), but now he sees that moment coming sooner. “A reasonable bet is somewhere between 5 and 20 years,” he says.
Instead of forcing AI to obey humans, Hinton suggests building “maternal instincts” into AI models so that they “really care about people” even when technology becomes more powerful and intelligent than humans.
He shared this decision at the Ai4 industry conference in Las Vegas, CNNwrites .
"AI systems, if they are intelligent, will very quickly develop two subgoals: the first is to stay alive and the second is to gain more control. There is good reason to believe that any subjective AI will try to survive," says Hinton.
That's why it's important to cultivate a sense of compassion for people, Hinton said. He noted that mothers have instincts and social pressure to take care of their children.
"The correct model is the only model we have where the smarter thing is controlled by the less smart thing, which is a mother being controlled by her child," Hinton said.
However, Hinton said he wasn't entirely clear on how this could be done technically, but stressed that critical researchers were working on it.
“This is the only possible positive scenario. After all, if the AI does not take care of me, it will replace me,” he claims. “Most of these superintelligent AI mothers will not want to give up their maternal instinct because they do not want us to perish.”
However, Fei-Fei Li, known as the “godmother of artificial intelligence” for her pioneering work in the field, disagrees with Hinton’s decision.
She calls for the creation of "human-centered AI that preserves human dignity and freedom of choice."
“It’s our responsibility at every level to create and use technology in the most responsible way possible. And at no point should any human being be forced or have to choose to give up their dignity,” Lee said. “Just because a tool is powerful… as a mother, as an educator, and as an inventor, I really believe that this is the foundation of how AI should be focused.”
