注册并分享邀请链接,可获得视频播放与邀请奖励。

MTS
@MTSlive
Chronicling the singularity New original series @themethodshow
加入 March 2026
1.7K 正在关注    588.9K 粉丝
SITUATION EXPLAINED: OpenAI board member Paul Christiano says we could be 6 months from AI that can overthrow the government. • Christiano heads the Alignment Research Center, helped invent RLHF, led AI safety at the US AI Safety Institute and joined the OpenAI Foundation's board in September, though he stressed he wasn't speaking as a board member • He said there's "a reasonable chance right now" we're in a world where, in 6 months, "there's a disaster that kills very large numbers of people, or where the government is overthrown" • His 1 in 3 figure is for superintelligence arriving within a year if development doesn't slow, while he puts the odds of an AI takeover in that time at about 10%, up from 4% in September • He wants labs to stop improving their algorithms "right now" until they can secure their model weights, so China can't steal a superintelligence mid-intelligence explosion, and says locking them down could take 6 to 12 months • @sebkrier who leads frontier policy development at Google DeepMind, sums up his response as "bullish on rapid capability gains, more skeptical of the leap to takeover, and wary of emergency overreach" @theojaffee: "Do I think we're 6 months away from AI that can overthrow the government? No. Again, there's endogenous societal response and whatnot."
显示更多