Nations need to work together to train non-hostile Artificial Intelligence, Geoffrey Hinton has said
Artificial intelligence pioneer and Nobel Peace Prize laureate Geoffrey Hinton has urged governments worldwide to collaborate in training AI systems not to harm humanity, warning that the rapidly advancing technology will soon likely surpass human intelligence.
Speaking at the World Artificial Intelligence Conference (WAIC) in Shanghai on Saturday, Hinton said that despite divergent national interests, no country wants AI to dominate humanity.
He noted that international cooperation is unlikely on offensive AI use – such as “cyberattacks, lethal autonomous weapons or fake videos for manipulating public opinion.” However, nations could form a “network of institutions” to guide the development of a highly intelligent AI “that doesn’t want to get rid of people,” Hinton added. He compared this proposed cooperation to Soviet-US collaboration on nuclear non-proliferation during the Cold War.
Hinton, often referred to as the “Godfather of AI,” likened AI development to “raising a tiger cub” that could become dangerous once it matures.
“There’s only two options if you have a tiger cub as a pet. Figure out if you can train it so it never wants to kill you, or get rid of it,” the scientist said.
He explained that AI is likely to increasingly seek more control in order to achieve its assigned tasks as it grows more intelligent, and simply “turning it off” when it outpaces humanity will not be an option.
“We will be like three-year-olds and they will be like adults,” Hinton said.
Speaking to the press later in the day, he noted that it should be relatively easy for “rational” nations to cooperate on the subject, but said it may be “difficult” for the US under “its current administration.”
On Wednesday, the White House announced its “action plan” to achieve “global dominance” in AI through investments, subsidies, and the removal of legal restrictions on the technology’s development.
Beijing has announced its intention to establish an organization to coordinate international cooperation on AI. “We should strengthen coordination to form a global AI governance framework,” Chinese Premier Li Qiang said at the WAIC on Saturday.
