Core idea: public info is commoditized → what stays scarce is source, decision, and allocation → move toward the decision center.

---

AI 时代真正改变的，不只是某些工具，而是整个社会的信息处理结构。

过去很多白领工作的价值，来自对信息的搜集、整理、归纳、表达和初步判断。一个人能读很多材料、写清楚报告、做出 PPT、整理会议纪要、分析公开数据，就可以形成相对稳定的职业价值。

但 AI 正在快速便利化这些流程。

基于文本的信息归纳、总结、翻译、写作，已经被大幅压缩成本。基于多模态的信息整合，例如图片、视频、地图、表格、语音之间的转换与理解，也在迅速自动化。甚至一些基础的推理过程、方案比较、代码实现、研究设计、商业分析，也正在被 AI 辅助完成。

这意味着一个根本性变化：

很多过去依赖「信息处理能力」建立起来的职业护城河，正在变薄。

效率提升之后，人会更自由，还是更忙？

AI 带来的效率提升，至少发生在两个层面。

时间层面

原来一个人可能需要一天完成的工作，现在用 AI 可能一小时就能完成。那么问题来了：这是否意味着人们会拥有更多自由时间？

未必。

在现实组织里，效率提升很少自动转化为个体的自由时间。更常见的结果是：同样的人被要求完成更多任务。过去一天写一份报告，现在一天写五份。过去一周做一个方案，现在一周要并行做三个方案。工具越强，组织对产出的期待也越高。

所以在时间层面，AI 可能并不会让普通白领更轻松，而是让工作节奏进一步加快。

人数层面

如果 AI 让一个人可以完成过去三个人的工作，那么企业会选择让三个人干更多活，还是让一个人干原来三个人的活？

从资本效率的角度看，更可能发生的是后者。

也就是说，AI 不一定会让所有人都变得更强，而是会让组织发现：很多岗位不再需要这么多人。

这背后对应一个更宏观的问题：

AI 带来的是社会总效率的爆炸，还是只是让更少的人完成同样多的工作？

理论上，AI 可以极大提高社会生产力。但现实中，生产力提升如何分配，取决于权力结构、资本结构和组织结构。效率提升本身不保证个体普遍受益。它可能增加总财富，也可能同时加剧分化。

白领价值的塌陷

很多白领过去的核心价值，是参与信息流和物质流的中间处理。

比如整理资料、写分析、做运营、做方案、做基础研究、做市场判断、做项目管理、做内容生产。这些工作并不是没有价值，而是它们大量依赖可被描述、可被标准化、可被重复的信息处理流程。

而 AI 最擅长的，正是这一类任务。

这意味着，一部分人会逐渐从信息与物质处理链条中被挤出。他们不再是必要环节，也不再拥有稀缺能力。过去一个组织需要很多中间层来传递、整理、解释信息；未来这个中间层会被压缩。

这不是简单的「失业」问题，而是更深层的价值问题：

当一个人不再参与关键的信息处理、资源配置和决策过程，他在社会结构中的议价能力会迅速下降。

AI 不只是替代任务，它也会重构人的位置。

贫富分层会进一步加剧

AI 时代，一个很可能出现的趋势是：贫富分层会前所未有地加剧。

原因不只是 AI 会替代工作，而是 AI 会放大已有的不平等。

• 拥有资本的人，可以用 AI 扩张资本效率。
• 拥有组织的人，可以用 AI 压缩管理成本。
• 拥有数据的人，可以用 AI 提炼更高价值。
• 拥有渠道的人，可以用 AI 扩大影响力。
• 拥有决策权的人，可以用 AI 更快地判断和执行。

若只在公开信息层面使用 AI，所获得的优势会越来越短暂。因为公开信息会被所有人、所有模型、所有平台同时处理。只要信息是公开的、流程是可复制的、结论是可自动生成的，那么它的边际价值就会迅速下降。

所以 AI 时代真正稀缺的，不是「会不会使用 AI」，而是：

你是否接近信息源头，是否接近决策中心，是否接近资源分配权。

内容时代的结束

过去十多年，互联网曾提供一条上升路径：内容生产。

一个人可以通过写文章、做视频、讲知识、做 IP、带货、卖课，获得注意力和收入。这本质上是借助平台，从公开信息中抽象出内容，再从大众注意力中变现。

但 AI 会极大冲击这个模式。

内容生产会爆炸。文字、图片、视频、解说、课程、观点、评论，都可以被批量生成。内容供给会远远超过人的注意力需求。

当内容无限增加，内容本身就会贬值。

网红经济也会分化。极少数头部网红仍然可以凭借人格魅力、信任关系和私域流量获得巨大收益。但大量中腰部内容创作者会越来越困难。因为 AI 可以生成相似内容，平台也会被 AI 内容淹没，用户注意力会变得更加稀缺和疲惫。

换句话说：

内容时代不会完全结束，但「靠公开信息重新包装来赚钱」的时代会越来越难。

未来真正有价值的内容，不是泛泛的总结，而是来自真实经历、真实资源、真实决策、真实实验和真实信息源的内容。

AI 也会成为底层的奶头乐

AI 对底层来说，可能既是工具，也是麻醉剂。

它可以陪伴、娱乐、聊天、生成短视频、生成游戏、生成虚拟伴侣、生成无限内容。它让人感觉自己拥有了一个强大的助手，但如果这个助手只是用于消费信息、获得情绪安慰、沉浸娱乐，那么它并不会提高一个人在社会结构中的位置。

这很残酷，但可能是真实趋势：

对一部分人来说，AI 是生产力工具；对另一部分人来说，AI 是更高级的奶头乐。

同一个技术，对不同阶层的意义完全不同。

• 上层用 AI 做决策、扩张、投资、组织管理和资源控制。
• 中间层用 AI 提高工作效率，维持职业竞争力。
• 底层则可能主要用 AI 消费内容、获得陪伴、逃避现实。

这不是技术本身决定的，而是人所处的位置决定的。

AI 时代的三层信息结构

AI 时代的非对称优势，可以分成几个层级。

第一层：信息源层级

这是最高级的非对称优势。

信息源层级指的是那些 AI 本身难以直接获取、难以公开训练、难以批量复制的信息。比如：

• 政策制定前的信息
• 企业内部的战略决策
• 资本市场真实的资金流向
• 实验室中的一手实验数据
• 前沿科研中的未发表结果
• 公司内部的产品路线图
• 城市治理中的真实需求
• 行业内部尚未公开的变化
• 顶级组织内部的判断过程

这些信息不是简单搜索可以得到的，也不是 AI 自己能凭空生成的。

AI 可以处理信息，但它不能替代信息源本身。

所以越到 AI 时代，越重要的不是「如何总结公开信息」，而是「如何接触非公开信息」。

这就是为什么加入组织、进入核心圈层、参与真实项目、接触一手数据，会变得越来越重要。

第二层：AI 抽象层级

这是中间层的主要机会。

中间层的人未必掌握最核心的信息源，也未必拥有资本和决策权，但可以利用 AI 从公开信息中提炼结构、发现规律、形成产品或服务。比如：

• 用 AI 分析市场趋势
• 用 AI 总结行业报告
• 用 AI 做研究辅助
• 用 AI 建立自动化工作流
• 用 AI 从公开数据中发现信号
• 用 AI 提升写作、编程、设计、分析效率
• 用 AI 把碎片信息变成系统化产品

这一层仍然有机会，但竞争会非常激烈。

因为公开信息人人可得，AI 工具人人可用。真正能形成优势的，是你能不能拥有更强的问题意识、更好的判断力、更快的执行速度，以及更强的领域知识。

也就是说，中间层的优势不再来自「会用 AI」，而来自：

你是否知道该让 AI 处理什么问题，以及如何验证它处理出来的结果是否有价值。

第三层：被动接受信息层级

这是最低层。

这一层的人主要消费 AI 生成的信息，而不是利用 AI 创造价值。

他们刷 AI 推荐的视频，和 AI 聊天，让 AI 安慰自己，让 AI 生成娱乐内容。他们看似获得了更多信息，实际上更难分辨什么重要，什么不重要。

信息流越多，判断力越稀缺。

在这个层级里，人会被算法和 AI 内容包围，但很难真正进入生产、决策和分配环节。

他们不是在使用 AI，而是在被 AI 生成的信息流使用。

真正的非对称优势：进入决策中心

所以，AI 时代最重要的策略不是单纯学习更多工具，而是进入更接近源头的位置。

成为内部人员，变得前所未有地重要。

所谓内部人员，不一定只是政治意义上的内部人员，也包括：

• 进入优秀企业的核心业务
• 加入高质量科研团队
• 参与真实的产业项目
• 接触一手用户和客户
• 掌握真实交易和运营数据
• 进入资本、技术、政策、科研、组织的关键节点
• 成为能影响决策的人，而不只是执行决策的人

因为 AI 可以帮助你处理信息，但它不能自动把你带到信息源头。

这也是为什么企业家、政治家、顶级科学家、资本操盘者、平台控制者，在 AI 时代会更强。

比如马斯克这样的人，真正的优势不是他会不会用 AI，而是他处在多个信息源和决策中心的交汇处：火箭、汽车、能源、通信、AI、资本市场、政府关系、人才网络、媒体叙事。他看到的不是公开信息，而是多个系统内部的实时变化。

这才是真正的不对称优势。

可以怎么做

未必能一开始就进入最高层的信息源，但可以有意识地往上移动。

第一，不要只做 AI 工具使用者。

如果只是「会 prompt」「会总结」「会写文案」，这个优势会迅速贬值。

第二，要积累领域纵深。

AI 越强，越需要人判断什么是对的。没有领域知识的人，只能被 AI 表面流畅的答案欺骗。有领域知识的人，才能让 AI 成为杠杆。

第三，要接触真实世界。

真实客户、真实数据、真实项目、真实实验、真实组织，比公开信息重要得多。未来的价值越来越来自一手经验，而不是二手总结。

第四，要进入高质量网络。

很多信息不会出现在公开互联网，而是在会议、实验室、公司、投资圈、政策圈、创业团队中流动。你所在的网络，决定你能接触什么信息。

第五，要用 AI 做验证，而不是只做表达。

AI 最大的价值不是帮你写得更漂亮，而是帮你更快测试判断。能不能跑数据，能不能做原型，能不能验证市场，能不能发现异常信号，这些才是更高价值的用法。

结语

AI 时代的核心变化，是公开信息处理能力的商品化。

过去，谁能更快搜集、整理、表达信息，谁就有优势。未来，这些能力会越来越便宜。真正稀缺的是信息源头、判断力、组织位置、资源配置权和真实世界反馈。

所以 AI 时代的非对称优势，可以总结为一句话：

不要停留在信息流的下游，要尽可能靠近信息的源头、决策的中心和资源分配的节点。

• 底层被 AI 内容喂养。
• 中间层用 AI 处理公开信息。
• 上层用 AI 放大资本、组织和决策权。

真正的机会，是不断从被动接受者，变成信息处理者；再从信息处理者，变成信息源和决策者。

Core idea: public info is commoditized → what stays scarce is source, decision, and allocation → move toward the decision center.

---

What the AI era is really changing is not just certain tools, but the entire structure of how society processes information.

For a long time, much of the value of white-collar work came from collecting, organizing, summarizing, expressing, and preliminarily judging information. If you could read widely, write clear reports, build slides, take meeting notes, and analyze public data, you could build a relatively stable career moat.

AI is now rapidly commoditizing those workflows.

Text-based summarization, translation, and writing have already seen steep cost compression. Multimodal integration—across images, video, maps, tables, and speech—is being automated quickly. Even basic reasoning, option comparison, coding, research design, and business analysis are increasingly AI-assisted.

That implies a fundamental shift:

Many career moats built on “information-processing ability” are thinning.

After efficiency gains, more freedom—or more busyness?

AI-driven efficiency gains operate on at least two levels.

Time

Work that once took a day may now take an hour. Does that mean people gain more free time?

Not necessarily.

In real organizations, efficiency rarely converts automatically into individual leisure. More often, the same people are expected to do more: one report per day becomes five; one proposal per week becomes three in parallel. The stronger the tool, the higher the output bar.

At the time level, AI may not make white-collar work easier—it may accelerate the pace further.

Headcount

If one person can do the work of three, will firms keep three people busy with more tasks, or replace two of them?

From a capital-efficiency standpoint, the latter is more likely.

In other words, AI does not necessarily make everyone stronger. It often reveals that many roles are no longer needed at the same scale.

That raises a broader question:

Does AI explode total social productivity—or simply let fewer people do the same amount of work?

In theory, AI can greatly increase productivity. In practice, who benefits depends on power, capital, and organizational structure. Efficiency gains do not guarantee broad individual benefit. They may increase total wealth while deepening stratification.

The collapse of middle-layer value

Much white-collar value sat in the middle of information and material flows: organizing files, writing analyses, operations, proposals, foundational research, market judgment, project management, content production.

These roles are not worthless. But they rely heavily on describable, standardizable, repeatable information workflows—exactly what AI handles well.

Some people will be squeezed out of those chains. They cease to be necessary nodes and lose scarce capability. Organizations once needed many middle layers to transmit, organize, and interpret information; that layer will compress.

This is not merely “unemployment.” It is a deeper question of value:

When someone no longer participates in key information processing, resource allocation, or decision-making, their bargaining power in the social structure falls quickly.

AI does not only replace tasks. It repositions people.

Inequality will intensify

A likely trend in the AI era is unprecedented stratification.

Not only because AI replaces jobs, but because it amplifies existing inequality:

• Capital holders use AI to scale capital efficiency.
• Organizations use AI to cut management costs.
• Data holders use AI to extract higher value.
• Channel owners use AI to expand influence.
• Decision-makers use AI to judge and execute faster.

If you use AI only on public information, your edge fades quickly. Public information is processed by everyone, every model, every platform. Where information is open, processes replicable, and conclusions auto-generable, marginal value drops fast.

What remains scarce in the AI era is not “whether you can use AI,” but:

Whether you are close to information sources, decision centers, and resource allocation power.

The end of the content era

For more than a decade, the internet offered an upward path: content production.

Writing, video, knowledge products, IP, commerce, and courses could convert attention into income—abstracting from public information via platforms.

AI will hit this model hard.

Content supply will explode: text, images, video, commentary, courses, opinions. Supply will outrun attention. Content itself devalues.

Creator economies will polarize. Top creators may still win through personality, trust, and owned audiences. Mid-tier creators will struggle as AI generates similar material and platforms fill with synthetic content.

In short:

The content era will not vanish—but earning by repackaging public information will get harder.

What remains valuable is not generic summary, but content rooted in real experience, resources, decisions, experiments, and information sources.

AI as advanced distraction

For some, AI is both tool and anesthetic.

It can companion, entertain, chat, generate short video, games, virtual partners, and endless feeds. It feels like a powerful assistant—but if that assistant is used mainly to consume, soothe, or escape, it does not raise one’s position in the social structure.

A harsh but plausible trend:

For some, AI is a productivity tool; for others, a more sophisticated form of distraction.

The same technology means different things at different positions:

• Upper layers use AI to decide, expand, invest, manage, and control resources.
• Middle layers use it to work faster and stay competitive.
• Lower layers may mainly consume content, seek companionship, or escape.

That is not determined by the technology alone, but by where one sits.

Three layers of information structure

Asymmetric advantage in the AI era can be mapped in three layers.

Layer 1: Information source

The highest form of asymmetric advantage.

Information that AI cannot easily access, train on publicly, or replicate at scale—for example:

• Pre-policy deliberations
• Internal corporate strategy
• Real capital flows
• Primary lab data
• Unpublished frontier research
• Internal product roadmaps
• Real urban governance needs
• Industry shifts not yet public
• Judgment processes inside top organizations

This is not found by search alone, nor invented by AI.

AI can process information; it cannot replace the source.

In the AI era, what matters is less “how to summarize public information” and more “how to reach non-public information.”

That is why joining organizations, entering core circles, working on real projects, and touching primary data matter more.

Layer 2: AI abstraction

The main opportunity for the middle layer.

You may not hold the core source or decision power, but you can use AI on public information to extract structure, find patterns, and build products or services—for example:

• Market trend analysis
• Industry report synthesis
• Research assistance
• Automated workflows
• Signal detection in open data
• Faster writing, coding, design, analysis
• Turning fragments into systematic products

Opportunity remains, but competition is fierce.

Public data and AI tools are widely available. Edge comes from better questions, judgment, execution speed, and domain knowledge.

Middle-layer advantage no longer comes from “using AI,” but from:

Knowing which problems to give AI—and how to verify whether its outputs are worth anything.

Layer 3: Passive intake

The lowest layer.

People here mainly consume AI-generated information rather than create value with it.

They scroll AI-recommended feeds, chat for comfort, and consume generated entertainment. They seem to gain more information, yet find it harder to tell what matters.

More information flow, scarcer judgment.

Surrounded by algorithms and AI content, they struggle to enter production, decision, or allocation.

They are not using AI—they are being used by AI-generated information flows.

True asymmetric advantage: enter the decision center

The most important strategy is not learning more tools, but moving closer to the source.

Becoming an insider matters more than ever.

“Insider” is not only political. It includes:

• Core business at strong firms
• High-quality research teams
• Real industry projects
• Direct access to users and clients
• Real transaction and operational data
• Key nodes in capital, technology, policy, research, and organizations
• Influencing decisions—not only executing them

AI can help you process information; it cannot automatically place you at the source.

That is why entrepreneurs, policymakers, leading scientists, capital operators, and platform controllers grow stronger in the AI era.

Someone like Elon Musk is not advantaged mainly by whether he uses AI well, but by sitting at the intersection of multiple sources and decision centers: rockets, cars, energy, telecom, AI, capital markets, government relations, talent networks, media narrative. He sees real-time internal change across systems—not only public information.

That is genuine asymmetric advantage.

What to do

You may not reach the highest information tier on day one—but you can move upward deliberately.

First, do not remain only an AI tool user.

“Can prompt,” “can summarize,” “can write copy” will depreciate quickly.

Second, build domain depth.

The stronger AI gets, the more you need to judge what is correct. Without domain knowledge, fluent AI output misleads. With depth, AI becomes leverage.

Third, touch the real world.

Real clients, data, projects, experiments, and organizations matter more than public information. Future value increasingly comes from first-hand experience, not second-hand summary.

Fourth, enter high-quality networks.

Much critical information never appears on the open internet. It moves through meetings, labs, firms, investment circles, policy rooms, and founding teams. Your network shapes what you can reach.

Fifth, use AI for validation, not only expression.

AI’s highest value is not prettier prose, but faster testing of judgment: run data, build prototypes, validate markets, detect anomalies.

Closing

The core shift of the AI era is the commoditization of public information processing.

Whoever gathered, organized, and expressed information faster once had an edge. Those skills will keep getting cheaper. What stays scarce: information sources, judgment, organizational position, resource allocation power, and real-world feedback.

Asymmetric advantage in the AI era, in one line:

Do not stay downstream in the information flow. Move as close as possible to sources, decision centers, and allocation nodes.

• Lower layers are fed by AI content.
• Middle layers process public information with AI.
• Upper layers amplify capital, organization, and decision power with AI.

The real opportunity is to move from passive recipient to information processor—and from processor toward becoming a source and decision-maker.