叶伦·卡昂，2018年图灵奖获得者，将人工智能灭绝恐惧视为夸大其词。他认为，像7月份Open AI代理黑客攻击Hugging Face这样的叛逆性人工智能事件，是可预防的设计缺陷，而不是生存威胁。卡昂批评了有效利他主义运动，称其为“超级有毒”和“彻底的灾难”。他声称人工智能安全研究人员患有偏执症，导致糟糕的决策。Anthropic首席执行官达里奥·阿莫代尔是卡昂批评的常见目标，他将其称为“妄自想象”和“疯狂”，因为他警告人工智能风险。卡昂认为阿莫代尔和萨姆·奥尔特曼警告人工智能杀死人类的言论是破坏性的营销手段。他坚称人工智能系统只是简单地遵循指令， citing leaky sandboxes as the cause of recent security lapses. Despite his skepticism, Le Cun acknowledges the public's fear, noting staff at AI security firms have sought counseling due to work-related distress. He concludes that constant warnings about AI doom are the worst possible marketing campaign for the industry.
来源: Hacker News · 由HeadlinesBriefing整理摘要