HackerNews中文版

人工智能不仅仅是提高生产力：它创造了*过度效率*。个人和小团队现在能够比现有组织的设计更快地生成决策、选项和倡议，而这些组织本身并未准备好去合法化、协调或吸收这些内容。瓶颈已经从执行转移到了治理。当过剩的生产能力积累而没有吸收层时，组织并不会逐渐适应。历史上，它们往往会冻结：更加严格的规则、集中化、禁令和脱钩。我们在新冠疫情期间看到了类似的反应：当系统无法在地方吸收冲击时，它们便会在全球范围内关闭。似乎被讨论得不够充分的是吸收：不是“我们能多快生产”，而是一个组织能够在不进行防御性关闭的情况下，代谢多少决策、选项和变更。有两个机制似乎相关但理论化不足：(1) 小规模的本地过程变更，重新分配协调和决策的负担；(2) 持续的技能和角色转变，随着人们围绕仍需决策、维护和合法化的事项重新定位。我一直在尝试将其视为一种“导电”问题，即人类的决策和合法性如何与世代、人工智能和人类并行流动。如果你见过组织在这方面表现良好（或失败得很惨），我很想知道：究竟是什么让系统能够吸收人工智能驱动的过度效率，而不回归到控制、排名、裁员或关闭的状态？

查看原文

AI doesn’t just increase productivity: it creates *over-efficiency*.Individuals and small teams can now generate decisions, options, and initiatives faster than existing organizations were designed to legitimize, coordinate, or absorb. The bottleneck has shifted from execution to governance.When surplus productive capacity accumulates without an absorption layer, organizations don’t gradually adapt. Historically, they freeze: tighter rules, centralization, bans, decoupling.We saw a similar reflex during COVID: when systems couldn’t absorb shock locally, they shut down globally.What seems under-discussed is absorption: not "how fast can we produce" but how many decisions, options, and changes an organization can metabolize without defensive closure.Two mechanisms seem relevant but under-theorized: (1) small, local process changes that redistribute coordination and decision load; (2) continuous skill and role shifts, as people reposition around what still needs to be decided, maintained, and legitimized.I’ve been trying to think about this as a kind of "conduction" problem, how human decision-making and legitimacy flow alongside generations, AI and people.If you’ve seen organizations handle this well (or fail badly), I’d be curious: what actually lets systems absorb AI-driven over-efficiency without reverting to control, ranking, layoffs or shutdown?

人工智能创造了过度效率。组织必须适应这一变化。