Former Google CEO Warns We Need to Pull the Plug on AI If It Starts to Evolve

"The technologists should not be the only ones making these decisions."

Self-Improvement

Eric Schmidt, the one-time head of Google, is warning that humans may have to "unplug" artificial intelligence before it's too late.

In an interview with ABC's George Stephanopoulos, Schmidt suggested that AI technology is innovating so rapidly that it may pass us by before we recognize the dangers it poses.

"I've done this for 50 years [and] I've never seen innovation at this scale," the ex-Google CEO said. "This is literally a remarkable human achievement."

Along with former Microsoft executive Greg Mundie and the late Henry Kissinger, Schmidt warned in a new book that along with the incredible benefits AI may bring to humanity, such as the rapid discovery of new medications, the technology will also become more self-sufficient.

"We're soon going to be able to have computers running on their own, deciding what they want to do," he said. "When the system can self-improve, we need to seriously think about unplugging it."

Kill Switch Engage

During the interview, Stephanopoulos asked, as anyone who's seen a sci-fi movie about killer AIs could imagine, whether a superintelligent AI would be capable of heading off any attempts to destroy it.

"Wouldn't that kind of system have the ability to counter our efforts to unplug it?" Stephanopoulos asked.

"Well, in theory, we better have somebody with the hand on the plug," Schmidt responded.

And who should that unplugger be?

"The future of intelligence... should not be left to people like me, right?" Schmidt said.

"The technologists should not be the only ones making these decisions," he said. "We need a consensus about how to put the right guardrails on these things to preserve human dignity. It's very important."

Himself an AI investor, Schmidt suggested — in a meta twist — that AI itself may be able to act as a watchdog for the technology.

"Humans will not be able to police AI," he said, "but AI systems should be able to police AI."

It's a pretty strange take for someone who has cowritten two whole books about the dangers posed by the technology — but maybe that's just what Silicon Valley has done to his brain.

More on Schmidt: Former Google CEO Alarmed by Teen Boys Falling in Love With AI Girlfriends