Meta一名高层高管表示，即将到来的“人口崩塌”意味着社会迫切需要依靠AI来提升生产力。科技行业领袖正努力说服心存疑虑的公众：这项强大的技术会给人们的生活带来益处。Meta首席数据官亚历克斯・舒尔茨周三在英国媒体 人工智能未来峰会上称，出生率持续走低，意味着50年后全球人口将会大幅减少。
“整个世界都没有认真思考人口崩塌这类问题……我们迫切需要更高的生产力。”他说，“我们该如何照料老年人？我们要怎么做，才能充分利用地球上现有的劳动力？”
预计全球人口最早将在2064年见顶；同时，受寿命延长、出生率下滑影响，劳动人口的萎缩速度会更快。
舒尔茨发表此番言论之际，Meta正紧追OpenAI、Anthropic、谷歌这些行业领跑者，已经为AI布局投入数十亿美元。
Meta首席执行官马克・扎克伯格上月接受采访时称，公司新模型MuseWonder已经追上对手，跻身前沿大模型行列；面向普通用户的Muse智能代理自9月上线以来，下载量已经达到数百万次。
与此同时，市场普遍担忧AI热潮背后巨额投入的可持续性。Meta上周宣布计划再融资950亿美元，支撑持续加码的开支。
在美国本土，各地社区对AI数据中心建设的抵触情绪，已经成为周二中期选举中选民抛弃共和党现任候选人的原因之一。
但舒尔茨极力宣扬AI带来的生产力红利：这项飞速发展的技术已经在“提速”软件工程师的工作，也帮律师省下大量时间——律师借助AI工具预判庭审提问。
他介绍，在Meta内部，AI已经帮软件工程师免去数据仓库告警监控这类枯燥繁琐的工作，目前该场景下超过10%的任务可以由智能代理自主完成，无需人工介入。
不过他也承认，并非所有AI应用都能提升生产力，部分使用者只是“消耗token，却没有产出明显价值”。就连他自己团队里，能高效用好AI的人还不到一半。
舒尔茨说，随着新模型门槛降低，AI带来的生产力增益正在改善。他还提到，今年年初OpenClaw把vibe coding带火之前，他自己上手AI的节奏也很慢。
针对AI会造成大规模失业的担忧，他提出不同看法：“拥有工具不会消灭岗位，只会改变岗位的工作内容。”他补充，生产力提升反而会催生新的工作。
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
According to a senior management in Meta, the impending “population collapse” meant that society urgently needed to rely on AI to increase productivity. The leaders of the science and technology industry are trying to convince the skeptical public that this powerful technology will benefit people's lives. Meta's chief data officer, Alex Schultz, at the British Media Futures Summit on Wednesday, said that the continuing low birth rate meant that the global population would be significantly reduced in 50 years.
“The whole world has not considered such problems as the collapse of populations... We urgently need higher productivity.” He said, “How should we care for the elderly? What are we going to do to make full use of the labour available on Earth?"
The global population is expected to peak at the earliest by 2064; at the same time, the working population will shrink more rapidly due to increased longevity and declining birth rates.
As Schulz makes this statement, Meta is following up on the leaders of Openai, Anthropic and Google, who have invested billions of dollars in the AI layout.
In an interview with Chief Executive Officer Mark Zuckerberg in Meta last month, it was stated that the company ' s new model, MuseWonder, had reached the front line of LLM; and that the smart agent, Muse, targeting ordinary users, had downloaded millions of times since September.
At the same time, the market is generally concerned about the sustainability of the huge amounts of inputs behind the AI boom. Meta announced last week plans to finance an additional $95 billion to support ongoing incremental spending.
In mainland the United States, the resistance of communities to the construction of the AI data centre has become one of the reasons for voters to abandon current Republican candidates in the mid-train Tuesday elections.
But Schulz has been very active in promoting the productivity dividend of AI: this fast-growing technology has been working as a “pacing” software engineer, and has saved lawyers a great deal of time — lawyers use the AI. tool to prejudge court questions.
He described the fact that within Meta, AI had exempted software engineers from such cumbersome and cumbersome work as monitoring data warehouse alarms, and that more than 10 per cent of the tasks in the scene could be performed by intelligent agents without manual intervention.
He acknowledged, however, that not all AI applications could increase productivity, and that some users were simply “consumpting token but not producing tangible value”. Not even half of his own team can use AI efficiently.
Schulz stated that productivity gains from AI were improving as the threshold for the new model was lowered. He also mentioned that before OpenClaw brought the vibe coding on fire early this year, his own hands were slow.
In response to the concern that AI could cause mass unemployment, he expressed a different view: “The availability of tools does not eliminate jobs, but changes the job content of jobs”. He added that productivity gains would lead to new jobs.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.