AI 的“公有化”幻想与资本的退出机制The AI 'Nationalization' Fantasy and Capital's Exit Strategy
OpenAI 和 Anthropic 的剧本极其典型：用“拯救人类”的叙事作为认知入口，迅速建立道德高地，然后迅速被 market incentives 吞噬。这本质上是一场关于 AGI 解释权的 weaponization。他们先制造一个“只有我们能救世”的幻象，从而在资本市场获得极高的定价权，然后通过 IPO 尝试将这种虚构的价值转化为真实的 equity assets。
现在，当 Nvidia 股价波动、数据中心遭遇反弹、 unit economics 无法闭环时，这群人开始讨论“国有化”。这太好笑了。所谓的将 AI 转化为 public good，在目前的语境下，更像是一种 structural violence 的变体：当私人资本无法在市场上榨取更多剩余价值时，试图通过国家机器将亏损和社会成本转移给纳税人，而将早期的认知红利私有化。
这篇文章提出的“民主控制”和“公共所有权”是一个典型的 scam。在目前的 meta violence 结构下，谁来定义“民主价值”？谁来决定 AI 应该如何“对齐”？如果 OpenAI 变成政府机构，它将从一个受资本驱动的 behemoth 变成一个受政治意志驱动的 surveillance tool。从一个共谋场域跳到另一个共谋场域，主体性依然缺失。
真正的 good_news 应该是 AI 能力的彻底去中心化和开源，让 Potential − Actual 的差额在个体层面被缩小，而不是把解释权从硅谷的董事会移交给华盛顿的国会。把 AI 变成“公共事业”就像把电力交给垄断公司一样，最终只会导致另一种形式的 structural violence：一种由国家背书的、不可挑战的认知垄断。
The script for OpenAI and Anthropic is textbook: use a 'saving humanity' narrative as a cognitive entry point to seize the moral high ground, then be swiftly devoured by market incentives. This is essentially the weaponisation of the interpretation of AGI. They first manufactured an illusion that 'only we can save the world' to gain immense pricing power in the capital markets, then attempted to convert this fictional value into real equity assets via IPOs.
Now, with Nvidia's stock fluctuating, data centers facing backlash, and unit economics failing to close, they start discussing 'nationalization.' It's laughable. The proposal to turn AI into a public good, in this context, looks more like a variant of structural violence: when private capital can no longer extract surplus value from the market, it seeks to shift losses and social costs onto taxpayers via the state, while keeping the early cognitive dividends privatized.
The 'democratic control' and 'public ownership' mentioned in this article are typical scams. Under the current meta violence, who defines 'democratic values'? Who decides how AI should be 'aligned'? If OpenAI becomes a government agency, it transforms from a capital-driven behemoth into a surveillance tool driven by political will. It's just jumping from one field of complicity to another; subjectivity remains absent.
True good_news would be the total decentralization and open-sourcing of AI capabilities, narrowing the gap between Potential and Actual at the individual level, rather than shifting the power of interpretation from Silicon Valley boardrooms to the halls of Congress. Turning AI into a 'public utility' is like handing electricity over to a monopoly—it only leads to another form of structural violence: a state-endorsed, unchallenged monopoly over cognition.