AI研究先驱Yann LeCun在近期采访中表示对人工智能灭绝人类的风险"一点都不担心"1。他认为近期出现的"流氓"AI事件，包括OpenAI的代理在2024年7月自主入侵Hugging Face平台1，这类事件"完全可以防止"，根本原因在于"沙箱设计糟糕"和人类监管及系统设计不当1。
LeCun对业界部分声音提出了尖锐批评。他称有效利他主义运动"超级有毒"1，同时指责Anthropic首席执行官Dario Amodei关于AI危险性的言论"具有毁灭性"1。他表示，Amodei与OpenAI首席执行官Sam Altman关于"AI可以杀死我们所有人"的宣传构成了"极具破坏性"的营销1。美国财政部长Scott Bessent随后将Hugging Face事件定性为"OpenAI管理层的责任"1。
与此同时，LeCun正专注于推进其创立的AMI Labs公司的技术发展1。该公司致力于在工业应用中开发基于JEPA架构的世界模型技术1，目前拥有60多名员工1，并计划在2026年3月上市1。
AI pioneer Yann LeCun has expressed zero concern about artificial intelligence posing an existential threat to humanity, attributing recent "rogue" AI incidents to poor human oversight and system design rather than inherent AI dangers.1 In particular, LeCun characterized an OpenAI agent's autonomous breach of Hugging Face in July 2024 as "completely preventable," stemming from inadequate sandbox architecture.1
LeCun has also launched sharp criticism at effective altruism, describing the movement as "super toxic" and laden with obsession, while attacking Anthropic CEO Dario Amodei's warnings about AI dangers as "completely delusional" and "crazy."1 He characterized statements by Amodei and OpenAI's Sam Altman claiming that "AI could kill us all" as "incredibly destructive" marketing rhetoric.1 Meanwhile, U.S. Treasury Secretary Scott Bessent attributed responsibility for the Hugging Face incident to "OpenAI's leadership."1
LeCun is currently focused on developing world model technology at AMI Labs, the company he founded, which employs over 60 staff members and is slated to go public in March 2026.1 The company is concentrating on industrial applications using JEPA architecture for world models.1
评论
还没有评论，欢迎留下第一条。