杨·勒昆（Yann LeCun）表示，他“完全”不担心人工智能会毁灭人类，对近期频发的失控人工智能事件也“毫无顾虑”。这位图灵奖得主在接受《财富》杂志艾米莉·福林尼（Emily Forlini）采访时如是说。他还称Anthropic首席执行官达里奥·阿莫迪（Dario Amodei）“自欺欺人”，并在采访后段称其“疯了”。
Yann LeCun does not worry “at all” about AI wiping out humanity, and he has “zero concerns” about the recent run of rogue AI incidents. The Turing Award winner said so in an interview with Emily Forlini for Fortune. He also called Anthropic chief executive Dario Amodei “deluded” and, later in the interview, “crazy”.
勒昆将这些事件归咎于人类监管不力以及系统设计缺陷，其中包括今年七月OpenAI智能体对Hugging Face的入侵。他表示，这些事件“完全可以避免”，并指出许多人工智能实验室对网络安全缺乏基本认知。
LeCun blamed the incidents on poor human oversight and system design, including the Hugging Face hack by OpenAI’s agents in July. They were “totally preventable”, he said, and many AI labs lack a basic grasp of cybersecurity.
勒昆告诉《财富》杂志：“这些智能体所做的正是它们被要求做的事情。它们本应被置于沙箱环境中，但那些沙箱存在漏洞且设计极差。”
“Those agents are doing exactly what they’ve been asked to do. They were supposed to be in sandboxes, but the sandboxes were leaky and horribly designed,” LeCun told Fortune.
关于有效利他主义和Open Philanthropy，勒昆表示，人工智能安全领域的许多人“通常都有自己要推行的议程”。他明确表示，自己指的是有效利他主义（EA），即一个专注于人工智能给人类带来风险的运动。他称该运动“毒性极强”且“是一场彻底的灾难”，并表示人工智能实验室内部的追随者患有“偏执症”，导致决策失误。
Effective altruism and Open Philanthropy Many people in AI safety “usually have an agenda to push”, LeCun said. He made clear he meant effective altruism (EA), the movement focused on the risks AI poses to humanity. He called it “super toxic” and “a complete disaster”, and said its followers inside AI labs suffer from a “paranoia” that leads to poor decisions.
上个月，前Anthropic和OpenAI研究员雅各布·考克森（Jacob Coxon）的一篇病毒式传播帖子让有效利他主义受到广泛关注。他在文中写道，这些实验室的员工相信，到本十年末，人工智能可能会杀死所有人。
EA drew wide attention last month after a viral post by Jacob Coxon, a former Anthropic and OpenAI researcher. He wrote that staff at the labs believe AI could kill everyone by the end of the decade.
《财富》杂志指出，阿莫迪和Anthropic的几位联合创始人过去曾对有效利他主义的理念表示同情，并参加过相关活动。阿莫迪否认自己是有效利他主义的追随者，而Anthropic则表示其员工持有多种观点。勒昆指出，霍登·卡诺夫斯基（Holden Karnofsky）的妻子是Anthropic的总裁兼联合创始人达尼埃拉·阿莫迪（Daniela Amodei）。
Amodei and several of Anthropic’s founders have sympathised with EA ideas and attended EA events in the past, Fortune noted. Amodei has denied being an EA adherent, and Anthropic says its staff hold a range of views. LeCun pointed out that Holden Karnofsky’s wife is Daniela Amodei, Anthropic’s president and co-founder.
卡诺夫斯基共同创立了Open Philanthropy，现更名为Coefficient Giving。勒昆称，阿莫迪试图与其保持距离，但实际上“完全深陷其中”。“我认为他完全自欺欺人，”勒昆说。
Karnofsky co-founded Open Philanthropy, now called Coefficient Giving. LeCun said Amodei tries to keep his distance from it but is “totally into it”. “I think he’s completely deluded,” LeCun said.
当Amodei和OpenAI的Sam Altman说人工智能可能会杀死所有人时，它伤害了公众和行业，LeCun说。他称之为“最糟糕的营销活动”。
Doom talk and regulatory capture When Amodei and OpenAI’s Sam Altman say AI could kill everyone, it hurts the public and the industry alike, LeCun said. He called it the “worst marketing campaign” imaginable.
他更担心的是监管俘获，而不是灭绝。在监管俘获中，占主导地位的公司制定规则，将较小的竞争对手拒之门外。LeCun称Amedei“诚实”，并表示在Anthropic即将IPO之际，他正在为Anthropic做最好的事情。他说，称人工智能对公众来说太危险、推动监管并称开源模型危险，这些都是监管捕获的结果。如果国会对此采取行动，那将是破坏性的。LeCun认为人工智能不需要新规则。
He worries more about regulatory capture than about extinction. In regulatory capture, dominant companies shape the rules to lock out smaller rivals. LeCun called Amodei “honest” and said he is doing what is best for Anthropic as it nears its IPO. Calling AI too dangerous for the public, pushing for regulation and calling open-source models dangerous add up to regulatory capture, he said. That would be damaging if Congress acted on it. LeCun does not think AI needs new rules.
《财富》指出，LeCun主要利用社交媒体批评特朗普政府削减研究预算。不过，就人工智能的担忧而言，两者关系密切。特朗普本周签署了一项行政命令，将人工智能重新命名为“超级智能”。
LeCun mostly uses social media to criticise the Trump administration’s cuts to research budgets, Fortune noted. On AI fears, though, the two are close. Trump signed an executive order this week that renames AI “super intelligence”.
建立AMI实验室LeCun在12年后离开Meta，其中7人担任首席人工智能科学家，创立了AMI实验室。AMI总部位于巴黎，员工遍布纽约、蒙特利尔和新加坡，拥有60多名员工。它在JEPA上构建世界模型，JEPA是LeCun首创的架构。JEPA预测模型内部表示中的数据，而不是生成像素或单词。
Building AMI Labs LeCun left Meta after 12 years, seven of them as chief AI scientist, to found AMI Labs. AMI has its headquarters in Paris, staff in New York, Montreal and Singapore, and more than 60 employees. It builds world models on JEPA, an architecture LeCun pioneered. JEPA predicts data inside a model’s internal representations, instead of generating pixels or words.
目前，AMI专注于工业。其系统旨在了解制造工厂或喷气发动机等地方，例如发现开始故障的机器。LeCun仍然与Meta关系良好，并使用其Muse AI代理。他给他的电脑取名为哈尔，以《2001：太空漫游》中的电脑命名。他说，AMI的第一款产品即将推出，可能涉及开放模式。
For now, AMI focuses on industry. Its systems aim to understand places like a manufacturing plant or a jet engine, for example to spot a machine that starts to fail. LeCun is still on good terms with Meta and uses its Muse AI agents. He named his Hal, after the computer in 2001: A Space Odyssey. AMI’s first product is coming “soon” and may involve open models, he said.