Nobel Physics Prize laureate Hinton Geoffrey, honorary professor at the University of Toronto in Canada /Courtesy of University of Toronto

Jeffrey Hinton, an emeritus professor at the University of Toronto known as the "AI lend," criticized Silicon Valley corporations for prioritizing profit over AI safety.

In an interview with CNN in the United States on the 28th, Hinton was asked whether AI corporations mostly act for profit and said, "Early OpenAI paid attention to the risks of AI, but gradually drifted away from safety and focused on revenue." He added, "It varies by corporation, but Meta was fixated on revenue from the start and did not care much about safety."

Among AI corporations, he assessed Anthropic as "the company that, if anything, values safety the most, but they, too, pursue revenue."

Hinton said, "AI has advanced faster than expected," and noted, "AI is driving innovation in a wide range of fields, including health care, education, climate change response, and new drug development, but there is a lack of effort to reduce the accompanying risks."

In particular, he argued that governments should introduce regulations to reduce such AI risks. He said, "At a minimum, we should require thorough safety measures from big corporations releasing AI chatbots so they do not encourage children to take extreme measures," and criticized, "President (Donald) Trump is trying to block any regulation, which I think is really crazy."

Hinton predicted that the advent of AI will bring changes on a scale similar to the Industrial Revolution. If the Industrial Revolution made physical strength meaningless by making it impossible to get a job just by being strong, AI will render human intelligence meaningless, he said.

He went on, "AI has become better at tasks like reasoning and at deceiving people," warning, "If you try to eliminate AI, it will devise a plan to deceive you to stop it."

Hinton also predicted that AI will continue to advance next year and threaten jobs. He said, "AI will gain the capability to replace numerous jobs," adding, "In the past it could only handle one minute of coding, but now it can execute one-hour projects, and in a few years it will be able to handle projects that take several months."

※ This article has been translated by AI. Share your feedback here.