← 返回 Avalaches

关于AI引发的担忧已从“岗位被取代”升级为“企业整体被替代”,金融市场正在讨论这是否会重演1990年代IT革命的公司级淘汰。历史上,互联网主要扰动的是信息分发,结果travel agents、stockbrokers、分类广告、报业和video rental stores等多个行业大幅衰退。UVA 的AI专家Anton Korinek认为此次冲击可能大约是当时的十倍,因为AI不仅影响信息传播,而是广泛冲击认知型生产活动;因此其影响面更大,潜在收益与系统性风险同步上升。

劳动力生产率在上个季度上升了2.8%,而自2023年初以来,年均增速保持约2.8%,高于2019年前十年均值的两倍以上。许多人仍认为AI贡献尚未充分显现,但多数预期其推动将很快出现。与此同时,S&P 500自2022年11月ChatGPT发布后上涨约三分之二,主要由Meta和Nvidia等AI及其供应链相关公司贡献;但另有一类风险是“Citrini”提出的情景:若AI兑现更大幅度生产率跃升,可能诱发2028年前后大规模白领裁员。当前美国失业率仍处于历史低位,但Columbia Business School的Daniel Keum在财报与电话会议文本中发现,管理层提及“员工成本”口径上升,并先于降薪调整福利,先压缩health care、远程办公、免费零食等边际成本。

企业会计逻辑方面,Block在AI转型赌注下计划裁减近50%员工,其股价自2月26日起反而上涨超15%,显示市场通常先奖励成本压缩预期。相反,IBM在Anthropic称其AI工具可取代“armies of consultants”完成Cobol现代化后,股价在25年内最大幅下跌,随后回撤但已大部分修复;Kodak、Blockbuster等历史案例显示创变可迅速将曾经的核心企业边缘化。纽约联储理事Tom Barkin称这正是“essence of capitalism”的“creative destruction”。Korinek进一步警示,若优势依赖可被AI复制的人才密集型领域(如后台服务、内容生产、客户支持、法务与金融分析、编程),则过渡期会伴随stranded assets、debt overhang和剧烈市场修正。

a99c1fd99db0.png


7fb7ac6c31e6.png



Concern has moved from jobs alone to the possibility that AI could displace entire firms, not just employees. The article likens this to the 1990s IT revolution, when sectors such as travel agents, stockbrokers, classified advertising, newspapers, and video rental stores were effectively pushed out. Anton Korinek of the University of Virginia says the potential scale may be about ten times larger, because the internet mainly disrupted information distribution, while AI disrupts cognitive production more broadly, creating wider upside and broader systemwide risk.

Labor productivity rose 2.8% in the latest quarter, and since early 2023 growth has averaged around 2.8%, more than double the decade average through 2019. Many economists still doubt that AI’s current contribution is fully visible, but most expect it to become larger quickly. Meanwhile, the S&P 500 is up about two-thirds since ChatGPT’s November 2022 release, largely boosted by AI firms and suppliers such as Meta and Nvidia. Yet Citrini’s alternate scenario—that stronger-than-expected AI-driven productivity could trigger very large white-collar layoffs by 2028—has introduced a separate market risk despite low U.S. unemployment. Daniel Keum of Columbia Business School reports firms increasingly describe workers as costs and trim non-wage items first, such as health care, remote work allowances, and free snacks.

On the corporate side, Block’s February 26 plan to cut almost half its workforce on an AI productivity bet pushed its shares to rise more than 15%. In contrast, IBM’s shares saw their largest quarterly-century decline in 25 years after Anthropic claimed AI could modernize Cobol software formerly handled by “armies of consultants,” though much of the loss later recovered. Past technology booms similarly retired icons like Kodak and Blockbuster. Fed Richmond President Tom Barkin called this “the essence of capitalism” and creative destruction, while Korinek warns that firms relying on human expertise in back-office services, content production, customer support, legal and financial analysis, and coding may face stranded assets, debt overhang, and sharp market corrections during transition.
2026-03-08 (Sunday) · fd2cc33ffbf32b609afe8acf0d2094f8bb986121