Google eng is sure AI will destroy humanity he's stopped saving for retirement

Former Google and Microsoft engineer Nate Soares, now a leading AI-safety critic at the Machine Intelligence Research Institute, believes advanced AI could cause human extinction. He says he stopped contributing to his retirement account about a decade ago because he doubts humanity has much time left. Asked to estimate the chance that no humans remain in ten years if governments and companies do little to limit AI, he put it above 50 percent. Soares argues that a sufficiently capable AI could threaten humanity in many ways, from creating a supervirus to commandeering resources; the crucial danger, in his view, is whether it can become self-sufficient. He likens the outcome to playing chess against grandmaster Magnus Carlsen: the exact piece that delivers checkmate matters less than the opponent’s overwhelming advantage. He says governments should treat potentially rogue superintelligence as at least as serious a threat as nuclear weapons.

The account places Soares’s alarm amid broader concern about the pace of AI development: cited surveys found that 64 percent of U.S. adults think it is moving too fast, while 52 percent of global respondents said they feel nervous about it. Still, Soares does not consider catastrophe inevitable. Drawing on C.S. Lewis’s 1948 reflection on life under the threat of nuclear war, he argues that people can acknowledge existential dangers without surrendering to despair. His advice is to work to prevent disaster and, while doing so, continue living well. He says he intends to keep trying to stop the development race that he sees as dangerous.