马克·扎克伯格（Mark Zuckerberg）正在人工智能竞赛中孤注一掷，他押注无限资本、顶尖人才与原始计算能力将确保其获胜。尽管扎克伯格所追逐的制胜法宝——人工智能“超级智能”——仍属模糊的理论概念，但Meta首席执行官周一公布的一系列惊人举措，瞬间改变了其他所有参赛者的现实处境，尤其是那些缺乏Meta资源的前沿人工智能初创公司。
扎克伯格周一宣布对公司的人工智能业务进行重大重组，将旗下所有人工智能业务和项目整合到新成立的部门“Meta超级智能实验室”（Meta Superintelligence Labs，简称MSL），并任命数据标注初创公司Scale AI前首席执行官汪滔（Alexandr Wang）为Meta首位首席人工智能官。
扎克伯格在《财富》杂志获取的一份内部备忘录中写道：“随着人工智能的发展步伐加快，开发超级智能已近在眼前。我相信这将开启人类新纪元，我将全力以赴，确保Meta引领这一进程。”他补充道，自己即将分享的细节，正是围绕实现公司“为每个人打造个人超级智能”这一愿景所付出的努力。
扎克伯格在备忘录中写道，新成立的超级智能实验室“涵盖了所有基础研究团队、产品团队以及基础人工智能研究团队，还包括一个专注于开发下一代模型的全新实验室。”
他还确认，前GitHub首席执行官兼投资人奈特·弗里德曼（Nat Friedman）也已加入Meta，将与汪滔共同领导MSL，负责Meta在人工智能产品和应用研究方面的工作。扎克伯格写道：“奈特过去一年来一直担任Meta顾问小组成员，因此已深入了解我们的路线图以及所需推进事项。”《财富》杂志上周报道称，弗里德曼还与汪滔和Scale存在关联——他是Scale的长期活跃投资人，并于2023年11月在犹他州共同主持了秘密的Scale AI安全峰会。
Meta在过去几周掀起了一场大规模招聘热潮，作为与Scale达成的143亿美元交易的一部分，汪滔也加入了该公司，该公司还以传闻中高达1亿美元的薪酬待遇从OpenAI招募顶尖研究人员。扎克伯格在周一的内部备忘录中列出了十一位从OpenAI、Anthropic和谷歌加盟Meta的顶尖研究人员。
对于在2022年末凭借ChatGPT掀起生成式人工智能热潮的OpenAI而言，Meta的激进招聘行动构成重大威胁。OpenAI首席研究官陈信翰（Mark Chen）将这种情况形容为仿佛有人“闯入我们家中”，称人才流失无异于“盗窃”。OpenAI表示已着手重新调整薪酬结构并制定“创意”留任方案以保持竞争力。
尽管OpenAI与微软（Microsoft）长期保持合作关系，后者已向OpenAI投资超130亿美元，但据报道，两家公司的关系在过去一年趋于紧张。然而，随着Meta不断进军OpenAI所深耕的领域，由山姆·奥特曼（Sam Altman）领导的这家初创公司可能需要寻找财力雄厚的盟友。
据媒体报道，OpenAI最近开始使用谷歌生产的人工智能芯片，这表明两家公司之间的联系日益紧密，尽管谷歌的Gemini大型语言模型与OpenAI直接竞争。
Meta的人工智能豪赌
凭借每季度超400亿美元的互联网广告业务，Meta有能力为一场不计代价的人工智能攻坚战提供资金支持，即便短期内无法盈利。
然而，Meta的最新举措仍潜藏诸多风险。
在任命汪滔为首席人工智能官时，Meta选择了一位并无计算机科学背景的人士来领导其全部人工智能工作——这一决定可能不会得到Meta众多人工智能科学家和博士的认可，许多人已经选择离职。
扎克伯格在备忘录中提到，他与汪滔已合作多年，并表示“我认为他是这一代最令人印象深刻的创始人”。
尽管“超级智能”尚无公认的正式定义，但通常指在几乎所有领域都远超人类能力的智能，包括科学创造力、通用智慧和社交技能，即在认知层面全面超越人类。超级智能通常被视为超越通用人工智能（AGI）的概念，尽管后者的定义也较为模糊，但通常指在广泛工作任务中具备人类水平智能的人工智能系统。也就是说，它能像人类一样进行推理、规划、解决问题、理解语言并以可推广的方式学习。
扎克伯格宣称，Meta“在为全球提供超级智能方面具备独特优势”，并指出其正在全力打造的数据中心可提供超越小型实验室的计算能力——目前该公司在数据中心方面的投入已达数百亿美元，且仍在持续加码。据《金融时报》上周报道，Meta正寻求从私人资本公司融资290亿美元，以全力推进其人工智能数据中心建设。
当然，微软和谷歌也已投入数百亿美元用于构建其人工智能基础设施。OpenAI表示，未来数年将与包括软银（Softbank）在内的合作伙伴共同投资5000亿美元，以打造“星际之门”人工智能数据中心网络。
倘若将超级智能竞赛视作一场意志力与资本的激烈角逐，那么扎克伯格似乎在押注自己能比竞争对手坚持更久。（*）
译者：中慧言-王芳
马克·扎克伯格（Mark Zuckerberg）正在人工智能竞赛中孤注一掷，他押注无限资本、顶尖人才与原始计算能力将确保其获胜。尽管扎克伯格所追逐的制胜法宝——人工智能“超级智能”——仍属模糊的理论概念，但Meta首席执行官周一公布的一系列惊人举措，瞬间改变了其他所有参赛者的现实处境，尤其是那些缺乏Meta资源的前沿人工智能初创公司。
扎克伯格周一宣布对公司的人工智能业务进行重大重组，将旗下所有人工智能业务和项目整合到新成立的部门“Meta超级智能实验室”（Meta Superintelligence Labs，简称MSL），并任命数据标注初创公司Scale AI前首席执行官汪滔（Alexandr Wang）为Meta首位首席人工智能官。
扎克伯格在《财富》杂志获取的一份内部备忘录中写道：“随着人工智能的发展步伐加快，开发超级智能已近在眼前。我相信这将开启人类新纪元，我将全力以赴，确保Meta引领这一进程。”他补充道，自己即将分享的细节，正是围绕实现公司“为每个人打造个人超级智能”这一愿景所付出的努力。
扎克伯格在备忘录中写道，新成立的超级智能实验室“涵盖了所有基础研究团队、产品团队以及基础人工智能研究团队，还包括一个专注于开发下一代模型的全新实验室。”
他还确认，前GitHub首席执行官兼投资人奈特·弗里德曼（Nat Friedman）也已加入Meta，将与汪滔共同领导MSL，负责Meta在人工智能产品和应用研究方面的工作。扎克伯格写道：“奈特过去一年来一直担任Meta顾问小组成员，因此已深入了解我们的路线图以及所需推进事项。”《财富》杂志上周报道称，弗里德曼还与汪滔和Scale存在关联——他是Scale的长期活跃投资人，并于2023年11月在犹他州共同主持了秘密的Scale AI安全峰会。
Meta在过去几周掀起了一场大规模招聘热潮，作为与Scale达成的143亿美元交易的一部分，汪滔也加入了该公司，该公司还以传闻中高达1亿美元的薪酬待遇从OpenAI招募顶尖研究人员。扎克伯格在周一的内部备忘录中列出了十一位从OpenAI、Anthropic和谷歌加盟Meta的顶尖研究人员。
对于在2022年末凭借ChatGPT掀起生成式人工智能热潮的OpenAI而言，Meta的激进招聘行动构成重大威胁。OpenAI首席研究官陈信翰（Mark Chen）将这种情况形容为仿佛有人“闯入我们家中”，称人才流失无异于“盗窃”。OpenAI表示已着手重新调整薪酬结构并制定“创意”留任方案以保持竞争力。
尽管OpenAI与微软（Microsoft）长期保持合作关系，后者已向OpenAI投资超130亿美元，但据报道，两家公司的关系在过去一年趋于紧张。然而，随着Meta不断进军OpenAI所深耕的领域，由山姆·奥特曼（Sam Altman）领导的这家初创公司可能需要寻找财力雄厚的盟友。
据媒体报道，OpenAI最近开始使用谷歌生产的人工智能芯片，这表明两家公司之间的联系日益紧密，尽管谷歌的Gemini大型语言模型与OpenAI直接竞争。
Meta的人工智能豪赌
凭借每季度超400亿美元的互联网广告业务，Meta有能力为一场不计代价的人工智能攻坚战提供资金支持，即便短期内无法盈利。
然而，Meta的最新举措仍潜藏诸多风险。
在任命汪滔为首席人工智能官时，Meta选择了一位并无计算机科学背景的人士来领导其全部人工智能工作——这一决定可能不会得到Meta众多人工智能科学家和博士的认可，许多人已经选择离职。
扎克伯格在备忘录中提到，他与汪滔已合作多年，并表示“我认为他是这一代最令人印象深刻的创始人”。
尽管“超级智能”尚无公认的正式定义，但通常指在几乎所有领域都远超人类能力的智能，包括科学创造力、通用智慧和社交技能，即在认知层面全面超越人类。超级智能通常被视为超越通用人工智能（AGI）的概念，尽管后者的定义也较为模糊，但通常指在广泛工作任务中具备人类水平智能的人工智能系统。也就是说，它能像人类一样进行推理、规划、解决问题、理解语言并以可推广的方式学习。
扎克伯格宣称，Meta“在为全球提供超级智能方面具备独特优势”，并指出其正在全力打造的数据中心可提供超越小型实验室的计算能力——目前该公司在数据中心方面的投入已达数百亿美元，且仍在持续加码。据《金融时报》上周报道，Meta正寻求从私人资本公司融资290亿美元，以全力推进其人工智能数据中心建设。
当然，微软和谷歌也已投入数百亿美元用于构建其人工智能基础设施。OpenAI表示，未来数年将与包括软银（Softbank）在内的合作伙伴共同投资5000亿美元，以打造“星际之门”人工智能数据中心网络。
倘若将超级智能竞赛视作一场意志力与资本的激烈角逐，那么扎克伯格似乎在押注自己能比竞争对手坚持更久。（*）
译者：中慧言-王芳
Mark Zuckerberg is stacking the deck in the AI race, betting that unlimited capital, top talent, and raw computing power will ensure victory. If the winning hand Zuckerberg is chasing—AI “superintelligence”— is still very much a vague and theoretical concept, the Meta CEO’s remarkable series of moves unveiled Monday instantly changed the reality for everyone else in the game, particularly the pioneering AI startups without Meta’s resources.
Zuckerberg announced a major revamp of its AI operations on Monday, putting the company’s collection of AI businesses and projects under the umbrella of a newly created organization called Meta Superintelligence Labs, or MSL, and appointing Alexandr Wang, the former CEO of data-labeling startup Scale AI, as Meta’s first ever Chief AI Officer.
“As the pace of AI progress accelerates, developing superintelligence is coming into sight,” Zuckerberg wrote in an internal memo obtained by Fortune. “I believe this will be the beginning of a new era for humanity, and I am fully committed to doing what it takes for Meta to lead the way.” He added that the details he would share were about building towards a company vision of ‘personal superintelligence for everyone.”
The new superintelligence lab, Zuckerberg wrote, “includes all of our foundations, product, and FAIR teams, as well as a new lab focused on developing the next generation of our models.”
He also confirmed that former GitHub CEO and investor Nat Friedman has also joined Meta to partner with Wang to lead MSL, heading Meta’s work on AI products and applied research. “Nat has served on our Meta Advisory Group for the last year, so he already has a good sense of our roadmap and what we need to do,” Zuckerberg wrote. Fortune reported last week that Friedman is also connected with Wang and Scale – he is a longtime and active Scale investor and co-hosted the secretive Scale AI Security Summit in Utah in November 2023.
Meta has embarked on an extraordinary hiring spree in recent weeks, bringing Wang on board as part of a $14.3 billion deal with Scale, and recruiting top researchers from OpenAI with rumored $100 million compensation offers. In Zuckerberg’s internal memo on Monday, he named eleven top researchers who had joined Meta from OpenAI, Anthropic and Google.
For OpenAI, which kicked off the generative AI craze with the release of ChatGPT in late 2022, Meta’s aggressive hiring spree represents a critical threat. OpenAI’s chief research officer Mark Chen described the situation as feeling like someone “breaking into our home,” calling the talent loss “theft.” OpenAI said it had begun recalibrating compensation and crafting “creative” retention packages to stay competitive.
While OpenAI has a longstanding partnership with Microsoft, which has invested more than $13 billion in OpenAI, the relationship between the two companies has reportedly grown strained over the past year. As Meta increasingly seeks to move into OpenAI’s territory however, the Sam Altman led startup may need to find allies with deep pockets.
OpenAI recently began using AI chips made by Google, according to media reports, signaling a growing bond between the two companies, even though Google’s Gemini LLMs compete directly with OpenAI.
Meta’s AI gambit
With an internet advertising business that generates more than $40 billion every quarter, Meta can afford to bankroll a no-holds-barred AI batter even it doesn’t immediately deliver a profit.
Still, Meta’s latest moves carry plenty of risks.
In making Wang chief AI officer, Meta has chosen someone who is not a computer scientist to lead all of its AI efforts—a choice that may not go over well with Meta’s deep bench of AI scientists and PhDs, many of whom have already decamped.
Zuckerberg noted in his memo that he and Wang had worked together for several years and said “I consider him to be the most impressive founder of his generation.”
There is also no agreed-upon formal definition of ‘superintelligence,’ though it is typically refers to an intelligence that vastly surpasses human capabilities in virtually all domains, including scientific creativity, general wisdom, and social skills—exceeding human cognition across the board. Superintelligence is generally perceived as going beyond artificial general intelligence, or AGI, which, though also vague, typically refers to an AI system with human-level intelligence across a wide range of work-related tasks. That is, it can reason, plan, solve problems, understand language, and learn in a generalizable way, much like a human.
Zuckerberg claimed that Meta is “uniquely positioned to deliver superintelligence to the world,” pointing to its efforts to build out data centers supporting more computing power than smaller labs – it is currently spending tens of billions on data centers and is raising more. The Financial Times reported last week that Meta is seeking $29 billion from private capital firms for its all-in push to build AI data centers.
Of course, Microsoft and Google are also spending tens of billions of dollars in cap ex to build out their AI infrastructure. And OpenAI has said it intends to invest $500 billion with partners including Softbank in the coming years to build out its Stargate network of AI datacenters.
If the race to Superintelligence is a test of wills and capital, Zuckerberg seems to be betting that he can outlast the competition.