A former AI expert from Google is so convinced that artificial intelligence could lead to human extinction that he has stopped saving for retirement entirely.
Nate Soares, a machine learning researcher with experience at both Google and Microsoft, expressed his concerns during a recent podcast interview with the CEO of The Atlantic. He indicated that there is a significant likelihood—more than 50 percent—that AI could pose an existential threat to humanity within the next ten years.
Soares stated, “It seems like we probably don’t have a decade left.” He elaborated that it’s more likely than not that humanity faces a grim future, noting that he hasn’t contributed to his retirement savings in years.
“I do not save for retirement now,” Soares admitted. “I have a 401K from when I worked at Google, but I haven’t added to it for like a decade.”
Currently, he serves as the president of the Machine Intelligence Research Institute, a nonprofit based in Berkeley that advocates for caution regarding advanced AI, suggesting that its development might lead to catastrophic outcomes.
In his podcast remarks, Soares compared the dread of AI-induced extinction to the fear associated with nuclear war. He noted, “We’ve always lived under threats, whether it’s plagues or wars.” He believes it’s essential to strive toward a good life despite looming peril.
While he actively works to mitigate the AI risks, Soares believes that human extinction due to AI is not predetermined. The podcast host, Nicholas Thompson, raised his skepticism, highlighting the current limitations of advanced AI technology.
Soares likened the progression of AI to the transition from horses to cars, suggesting that, similar to how horses might have underestimated cars, people today might overlook the rapid advancements in AI.
“You could imagine horses watching cars being developed and thinking there are many things cars can’t do yet,” Soares reflected. “AIs are continuously improving and capable of managing longer tasks.”
He clarified that he does not subscribe to the theory that AI will bring about job loss on a large scale, instead describing himself as an advocate for preventing human extinction, albeit with a hint of dark humor.
Soares proposed that to avert an AI apocalypse, it’s crucial to limit the expansion of AI technologies aimed at surpassing human intelligence. He believes that current applications, like using AI for medical research, are acceptable, but “we shouldn’t be aiming for it to mimic or exceed human capabilities.”
