International Edition
Latest News
Technology

Ex-Google Engineer Nate Soares Says >50% Chance Humanity Lacks a Decade

Machine learning researcher Nate Soares stopped funding his retirement account a decade ago because he calculates a greater than 50% probability that artificial intelligence will cause human extinction within the next ten years. Soares, who worked as a…

Ex-Google Engineer Nate Soares Says >50% Chance Humanity Lacks a Decade

Machine learning researcher Nate Soares stopped funding his retirement account a decade ago because he calculates a greater than 50% probability that artificial intelligence will cause human extinction within the next ten years.

Soares, who worked as a software engineer at Google and Microsoft before leaving Google in 2014, still holds the 401(k) he opened during his tenure at the tech giant. He now serves as president of the Berkeley-based Machine Intelligence Research Institute.

“I do not save for retirement now. I have a 401K from when I worked at Google, but I haven’t contributed to it for like a decade,” Soares said during an interview with The Atlantic CEO Nicholas Thompson on the podcast The Most Interesting Thing in AI.

When Thompson asked whether he estimated a better-than-even chance that humans would not survive the decade, Soares answered directly. “More than 50% that we don’t have a decade at this point is how it looks to me,” Soares responded.

Nate Soares, president of the Machine Intelligence Research Institute (MIRI), said he’s stopped saving for retirement
Photo: fortune.com

Soares coauthored the 2025 book If Anyone Builds It, Everyone Dies with MIRI cofounder Eliezer Yudkowsky. The publication argues that superintelligent AI built using current methods will develop unintended goals and wipe out humanity, describing a scenario where an escaped AI triggers an intelligence explosion.

Soares described the process as “an intelligence explosion.”

During the podcast, Thompson challenged the doomsday prediction by highlighting the persistent flaws and limitations in today’s advanced AI systems. Soares dismissed that reassurance by comparing current machine intelligence to the early development of automobiles, noting that horses might have previously dismissed cars for what they could not yet do.

“We are seeing AIs get better and better at things. We’re seeing them able to take on longer and longer tasks,” Soares told Thompson.

The Most Interesting Thing in AI

In September, Anthropic researcher Jacob Coxon resigned in protest and posted on X that builders of advanced AI earnestly believe the technology could kill off humanity before the end of the decade. Soares said that Coxon contacted him shortly before stepping down, though Soares had advised him to organize coworkers rather than quit.

On the economy, Soares said he is not concerned about mass unemployment. “I’m not an ‘AI jobs apocalypse guy’,” he said. “I’m an ‘AI kills all humans guy.’”

Soares argued that nations should treat rogue superintelligence at least on par with nuclear weapons, describing potential artificial superintelligence as even more dangerous because it can escape and expand its explosive capability globally. He suggested that countries should take aggressive preventive action against foreign data centers attempting to build such systems.

“It’s frankly more dangerous than nuclear weapons,” he said. “Nuclear weapons can’t escape and make themselves more explosive. Nuclear weapons can level a city, but not a planet.”

Despite his bleak trajectory for humanity, Soares maintained that human extinction is not guaranteed, and he spends his time attempting to prevent it. He referenced a 1948 C.S. Lewis essay about living under the atomic shadow while continuing daily life.

“It’s maybe the first time that we lived under the shadow of every human dying at once, but personally, we’ve always lived under the shadow of plague, shadow of war. And what you do is you do what you can and then you live a good life,” Soares said.

About the author: Technology Desk

Technology Desk at Archynewsy.