专家重新定义人工智能安全辩论，警告现实风险
十位专家警告网络和生物攻击、核灾难和人工智能可能带来的暴政，颠覆了对人工智能安全的大规模灭绝恐惧。这改变了我们对人工智能安全的思考方式。
a recent wired survey asked 10 experts across disciplines about ai safety concerns. most expressed genuine concerns about cyber and biological attacks, nuclear catastrophe, and the potential for ai to enable tyranny. one expert noted that ai systems are being adopted quickly in defense despite being error-prone. the panel was skeptical about pausing ai development, citing the influence of capitalism and governance issues. yoshua bengio urged the public to apply pressure to politicians to ensure ai safety. the experts' views suggest that the real risks of ai may be more nuanced than previously thought, and that the public and lawmakers have a role to play in ensuring safety.