夜雨聆风最近注意到一个现象：最近某些“skill”类产品非常火。有人把公司里已经离职同事留下的资料整合起来，做成所谓的 colleague skill；也有人把知名人物的公开内容、表达风格甚至“作战思维”打包，做成可被调用的数字分身。
这让我再次思考一个越来越现实的问题：当AI开始复制人的经验、语言风格与思维方式时，它究竟是在赋能人类，还是在吞噬人的主体性？
一、从效率工具到人格提取：边界正在被改写
早期人们谈论AI，更多是把它理解为一种工具：
帮助处理重复劳动 提升效率 释放时间 让人从低价值工作中解放出来
这是一种非常理想主义的愿景：
机器负责劳动，人负责创造；技术负责执行，人负责生活。
但现实的发展路径似乎并不总是如此。
今天看到一些案例，AI不再只是帮你完成任务，而是开始“代替某个人完成任务”。它调用的不只是知识，而是某个人的经验、语气、判断模式，甚至职业方法论。
也就是说，AI开始从“自动化流程”走向“人格化生产力”。
这一步，意义重大，也问题重重。
二、最尖锐的问题：AI让人自由，还是让人被替代？
今天看到一个访谈讨论，里面有一句观点让我印象深刻：
最初使用AI，是为了让人从工作中脱离出来；
后来却可能变成一部分人利用AI，让另一部分人陷入新的被支配状态。
这确实是值得深思的命题。
技术本身并不天然带来平等。
谁拥有技术、谁掌握资源、谁制定规则，决定了技术最终服务于谁。
于是，AI可能出现两种完全不同的未来：
路径一：解放型未来
人们减少机械劳动 有更多时间陪伴家人、学习、创造 个体获得更高自主性 社会整体效率提升
路径二：控制型未来
少数人拥有AI系统与数据资源 多数人被低成本替代 劳动价值被压缩 人被算法化管理与评价
同一种技术，可能通向两种文明。
三、“活人炼化”的隐喻：当人的经验被无限复制
同时，想到一个很有意思也很尖锐的表达：某些skill的逻辑，像是一种“活人炼化”。
当然，这是一种比喻。意思是：
一个真实的人还活着，还在思考、成长、工作；
但他的知识、表达风格、职业经验，已经被提取、封装、商品化，并被无限调用。
这背后牵涉几个关键问题：
1. 授权问题
这个人是否同意自己的内容被训练、被复制、被商业使用？
2. 收益问题
若其人格特征产生商业价值，收益如何分配？
3. 失真问题
AI模仿出的“他”，还是他本人吗？
4. 替代问题
当数字分身足够便宜，真人是否会被边缘化？
四、活人与古人的区别：文明传承与人格剥削之间
我也想到另一面。
如果对象是古人、历史思想家、经典作者，借助AI重建其思想框架，用于学习、讨论、比较研究，这往往具有很高价值。
例如：
与古代哲人进行思想对话 模拟历史人物立场辩论 用经典文本训练学习助手 让知识传播更生动
这更接近一种文明传承与教育创新。
但即便如此，也仍需尊重文本来源、版权边界与学术诚信。
所以关键不只是“能不能做”，而是：
对象是谁 是否授权 用途为何 是否透明 是否公平
五、我今天的结论
AI真正的伦理问题，也许不是它是否足够聪明，
而是它如何对待“人”。
如果技术的发展让更多人拥有尊严、时间与创造力，它就是文明进步。
如果技术的发展只是让少数人更强、让多数人更廉价，它就可能成为新型枷锁。
未来真正需要被守护的，不只是效率，
而是人的主体性。
Recently, I have noticed an interesting phenomenon: certain “skill”-based AI products have become extremely popular. Some people take the documents and materials left behind by former employees and turn them into so-called colleague skills. Others package the public content, speaking style, and even strategic thinking of well-known figures into digital replicas that can be invoked on demand.
This has led me to reconsider an increasingly urgent question:
When AI begins to replicate human experience, language style, and patterns of thought, is it empowering humanity—or consuming human subjectivity?
I. From Efficiency Tool to Personality Extraction: The Boundary Is Being Rewritten
In the early days, AI was mainly understood as a tool:
to handle repetitive tasks to improve efficiency to free up time to liberate people from low-value labor
It was an idealistic vision:
Machines would handle labor; humans would create.
Technology would execute; people would truly live.
But reality has not always followed that path.
Today, AI is no longer merely helping people complete tasks. In some cases, it is beginning to replace a specific person in completing tasks. It draws not only on knowledge, but also on someone’s experience, tone, judgment patterns, and even professional methodology.
In other words, AI is moving from automating workflows to becoming a form of personalized productive power.
This shift is significant—but deeply problematic.
II. The Sharpest Question: Will AI Liberate People or Replace Them?
I recently heard a striking idea in an interview:
AI was first intended to free people from work.
But it may eventually become a tool through which some people use AI to dominate others.
This is a profound question.
Technology does not naturally produce equality.
Who owns the technology, who controls the resources, and who writes the rules determines whom the technology ultimately serves.
As a result, AI may lead to two very different futures:
Path One: A Future of Liberation
People spend less time on mechanical work More time is available for family, learning, and creativity Individuals gain greater autonomy Society becomes more efficient overall
Path Two: A Future of Control
A small minority owns AI systems and data resources The majority are replaced at low cost The value of labor is compressed People are managed and evaluated through algorithms
The same technology may lead to two entirely different civilizations.
III. The Metaphor of “Refining the Living”: When Human Experience Is Infinitely Replicated
I also thought of a sharp metaphor: some AI skills resemble a kind of “refining the living.”
Of course, this is symbolic language. It means:
A real person is still alive, still thinking, still growing, still working—
yet their knowledge, speaking style, and professional experience have already been extracted, packaged, commercialized, and endlessly reproduced.
This raises several key issues:
1. Consent
Did this person agree to have their content trained on, replicated, or commercially exploited?
2. Compensation
If their personality traits generate economic value, how should that value be shared?
3. Distortion
Is the AI-generated version of “them” still truly them—or only a hollow imitation?
4. Replacement
When a digital clone becomes cheaper and more scalable, will the real person be marginalized?
IV. The Difference Between the Living and the Dead: Cultural Inheritance vs. Personality Exploitation
There is, however, another side to this issue.
If the subject is an ancient thinker, a historical figure, or a classic author, then using AI to reconstruct their intellectual framework for learning, dialogue, or comparative study can be highly valuable.
For example:
engaging in dialogue with ancient philosophers simulating debates between historical figures training learning assistants on classical texts making knowledge transmission more vivid and accessible
This is closer to cultural inheritance and educational innovation.
Even so, it still requires respect for:
source materials copyright boundaries academic integrity
So the key question is not simply whether we can do it, but:
Who is the subject? Was consent given? What is the purpose? Is the process transparent? Is it fair?
V. My Conclusion
The true ethical question of AI may not be whether it is intelligent enough.
It is how it treats human beings.
If technological progress gives more people dignity, time, and creative freedom, then it is a sign of civilization advancing.
If technological progress only makes the powerful stronger while making ordinary people cheaper and more replaceable, then it may become a new kind of chain.
What must be protected in the future is not efficiency alone—
but human subjectivity itself.
