Winners of Nobel Prizes for AI "concerns of 'superhuman AI' out of control"

Winners of the Nobel Prize for their artificial intelligence (AI)-related research have expressed a series of serious concerns about the situation of "out of control," saying that so-called "superhuman AI" will be developed faster than expected.

Jeffrey Hinton, a professor at the University of Toronto in Canada, a co-winner of the Nobel Prize in Physics, said at a joint press conference held at the Swedish Academy, "I used to think that the development of super-intelligence would be much later, but looking at the recent development speed, I think it will be 5 to 20 years."

"We have to seriously worry about how we can maintain control (over AI)," he said.

Professor Hinton, who is called an AI pioneer, said, "I wish I had considered safety early," when asked if there was anything I regret assuming I went back to the past.

Demis Hassabis, CEO of Google DeepMind, a co-winner of the Nobel Prize in Chemistry who was present at the press conference, also said he agreed with Hinton.

"My aspiration is to develop (AI) tools that contribute to scientific discovery," Hassabis said. "I think AI will have great tools to help humans with problems today, such as diseases, energy, and climate."

At the same time, however, we have also been considering the risks involved in developing strong general-purpose technologies, he said, "AI is probably one of the most powerful technologies developed by mankind, so we need to deal with the risks very seriously."

The winners also emphasized the need for AI regulation.

Professor Hinton first pointed out that there is virtually no regulation on the "Lethal Autonomous Weapon Systems (LAWS) that AI technology is applied to.

CEO Hassabis said, "AI regulation is necessary, but it is also very important to regulate it properly," adding, "As technology is evolving at a very rapid pace, the regulatory method we discussed only a few years ago may not be a good regulation to discuss now."