A former Google CEO on when 'we need to seriously think about unplugging' AI

Someone should have a "hand on the plug" for when AI starts to "self-improve," Eric Schmidt said

We may earn a commission from links on this page.
Eric Schmidt sitting in a chair talking with his hands up beside Fareed Zakaria's book in front of a navy backdrop that says "IGP Institute of Global Politics"
Former Google CEO and chairman Eric Schmidt at Columbia University’s School of International and Public Affairs on April 3, 2024 in New York City.
Photo: Shahar Azran (Getty Images)
In This Story

The former head of Google (GOOGL-0.47%) says computers will soon be able to make their own decisions — and humans will need to know when to pull the plug.

Eric Schmidt, who spent a decade as the artificial intelligence pioneer’s chief executive, said humans need to take advantage of AI “while preserving human dignity and values” during an interview on ABC News (DIS-0.76%)’ “This Week.”

Advertisement

Social media moved quickly to change the global zeitgeist, Schmidt said, “and now imagine a much more intelligent, much stronger way of sending messages, inventing things, the rate of innovation, drug discovery and all of that, plus all sorts of bad things, like weapons and cyber attacks.”

Advertisement

Soon, Schmidt said, there will be computers that can run “on their own, deciding what they want to do.” Currently, the industry is focused on AI agents — or software that can complete complex tasks autonomously — but the technology will have “more powerful goals.”

Advertisement

“Eventually, you say to the computer, ‘learn everything and do everything,’ and that’s a dangerous point,” Schmidt said. “When the system can self-improve, we need to seriously think about unplugging it.”

Asked if an AI system that powerful would have the ability to counter efforts to shut it down, Schmidt said, “in theory, we better have somebody with a hand on the plug.”

Advertisement

As AI becomes more intelligent, “each and every person is going to have the equivalent of a polymath in their pocket,” Schmidt said, but it’s not clear “what it means to give that kind of power to every individual.” There is a concern now that a company racing to develop AI will decide to skip steps in safety testing, Schmidt said, and end up releasing a system that is harmful.

The former Google leader said governments are “not yet” doing what they need to do to regulate AI on the way to superintelligence, but that “they will, because they’ll have to.”

Advertisement

Meanwhile, Schmidt said although he personally thought the U.S. was “a couple of years ahead of China,” the country has been able to catch up in the last six months despite efforts by both the Trump and Biden administrations to curb advanced chips and other technologies from entering China.

“It is crucial that America wins this race, globally and in particular, ahead of China,” Schmidt said.

Advertisement

The incoming Trump administration “will be largely focused on China versus the U.S.,” Schmidt said, adding that it “is a good thing,” and that as long as the U.S. values individual freedom, “we should be okay.”