2026-07-06｜当数据像情绪一样影响人

今天写了几首歌，听到 AI 写出来的曲子，忽然觉得很好听。工作了一整天以后，偶尔让自己从文件、系统、网页、仓库和文字里退出来，听一段旋律，好像人的心会稍微松一点。音乐很奇妙，它不需要解释太多，却能很快触碰到人的情绪。一个旋律起来，人的心情就会跟着动。明明刚才还在处理很理性的事情，下一秒可能就被一段声音带到很多年前，想起一些人、一些路、一些没有说完的话。

人其实很容易被感染。

不只是音乐。别人的一句话、一个表情、一种语气、一个眼神，甚至一个沉默，都会影响人的情绪。有时候本来心里很平静，遇到一个焦虑的人，自己也开始紧张；遇到一个愤怒的人，自己也会被激起防御；遇到一个悲伤的人，心也跟着沉下去。人和人之间的情绪好像会流动，会传染，会互相放大。

这也是人矛盾的地方。

人需要接触别人，需要被理解，需要共情，需要从别人那里获得温度和连接。可是接触越多，也越容易被别人带偏。一个人如果内心不够稳，很容易把别人的情绪当成自己的情绪，把别人的判断当成自己的判断，把外部的声音当成自己真正的方向。

所以一个人要活得清楚，并不是完全不接触外界，而是在接触外界的时候，仍然知道什么是自己的。

今天我突然想到，数据其实也很像情绪。

数据表面上看起来是理性的，是数字、记录、文本、图片、点击、搜索、时间、位置、行为、版本和统计结果。人们容易以为，只要是数据，就比情绪更客观，更可靠，更不会骗人。可是数据进入系统以后，也会像情绪一样影响判断。它会被选择、被放大、被排序、被解释、被推荐，也会被某种结构带着走。

一个人看到某个数字变高，会开始相信它更重要；看到某个评价变差，会开始紧张；看到某个名字反复出现，会觉得它更可信；看到系统推荐某个方向，会不自觉地以为那就是更正确的方向。数据不一定直接命令人，但它会改变人的注意力。它不一定像情绪那样明显冲进身体，却会慢慢改变一个人看世界的顺序。

这很像情绪感染。

情绪感染让人不知不觉跟着别人的状态走；数据感染让人不知不觉跟着系统呈现的重点走。一个人可能以为自己在独立判断，实际上已经被排序、推荐、统计和重复出现的信息带偏。数据越多，人的判断不一定越清楚。有时候数据越多，人越容易被包围，越容易忘记自己原来要判断的是什么。

这也是 AI 时代一个很大的结构问题。

AI 能够接触大量数据，也能把这些数据变成总结、建议、排序、预测和生成内容。它可以帮助人看见过去看不见的关系，也可以让人更快处理复杂信息。但与此同时，它也可能把某些数据放大，把某些声音压下去，把某些模式当成方向，把某些重复当成重要，把某些表面相关当成真正原因。

如果人没有判断边界，就会被数据带走。

就像一个情绪不稳定的人很容易被别人一句话影响，一个判断不稳定的系统也很容易被数据的表面强度影响。看到更多，不等于理解更深；接触更多，不等于判断更好；被大量数据包围，不等于接近真实。很多时候，真实并不是藏在最多的数据里，而是需要人判断哪些数据有关，哪些数据只是噪音，哪些数据反映事实，哪些数据只是系统放大的回声。

我开始觉得，未来真正重要的问题不是“要不要接触数据”，而是“如何接触数据而不被数据接管”。

人不能完全脱离情绪，因为没有情绪，人就失去很多感受、关系和理解他人的能力。可是人也不能完全被情绪支配，否则就会被外界牵着走。数据也是一样。人和系统都不能完全脱离数据，因为没有数据，就无法理解复杂现实，也无法做出更好的判断。可是人和系统也不能完全被数据支配，否则判断会被统计、排序和可见性绑架。

所以关键不是隔离，而是边界。

情绪需要边界，数据也需要边界。

一个内心强大的人，不是不受影响，而是知道自己正在受影响，并且能够停下来问：这是我的情绪，还是别人的情绪？这是事实，还是我被刺激后的反应？这是我真正的判断，还是我只是想证明自己没有输？

同样，一个成熟的数据系统，也不应该只问：数据说了什么？它还应该问：这些数据从哪里来？它们代表谁？谁没有被记录？什么被放大了？什么被隐藏了？这些数据是否足以支持行动？人在行动之前是否真正理解了后果？

数据如果像情绪一样会感染人，那么 AI 时代就需要一种新的保护：让人接触数据，但不被数据直接推着执行；让系统使用数据，但不把数据自动变成责任；让模型生成建议，但不把建议直接当成最终判断。

这和我最近一直写的语言到执行边界也连接起来。语言说出口，不等于执行；表达被记录，不等于授权；数据被看见，也不等于它可以直接决定行动。无论是语言、情绪还是数据，它们都可以进入人的判断过程，但不能绕过人类判断，直接进入执行结构。

未来的 AI 系统如果只追求速度，就会很容易把数据当成决定。它会说：因为很多人点击，所以这个重要；因为很多人这样选择，所以这是趋势；因为历史上经常发生，所以未来也应该这样；因为模型预测概率高，所以可以执行。可是人类社会不是只由概率组成的。真实的人有例外，有尊严，有不被记录的处境，有被系统误解的可能，也有在关键时刻拒绝被数据定义的权利。

数据可以帮助判断，但不能替代判断。

这句话听起来简单，但在 AI 时代会越来越难做到。因为系统越强，人越容易相信系统已经看得更多；模型越流畅，人越容易以为它已经理解；数据越丰富，人越容易把复杂现实交给自动总结。可是很多真正重要的判断，恰恰发生在数据无法完全解释的地方。

比如，一个诚实干活的人可能没有漂亮的数字记录，但不代表他没有能力。一个学生某一段时间成绩不好，不代表他没有未来。一个病人某些指标看起来异常，不代表他的生活处境可以被一组数字完全定义。一个人在某次对话里说了一句气话，不代表那就是他真实、稳定、可执行的意图。一个工具被很多人使用，不代表它一定是更负责任的选择。

数据需要被接触，也需要被质疑。

情绪也是这样。人不能因为情绪会带偏自己，就完全拒绝情绪。情绪有时候会提醒我们哪里受伤了，哪里重要，哪里需要保护，哪里不对劲。只是情绪不能直接接管行动。一个人成熟以后，不是没有情绪，而是能够和情绪保持距离：我感受到了，但我不立刻被它控制。

数据也应该这样。

一个系统成熟以后，不是不使用数据，而是能够和数据保持距离：我看见了这些数据，但我不把它们自动当成最终事实；我参考这些模式，但我不让模式替人决定；我承认数据的价值，也承认数据之外仍然有人类处境、人类判断和责任边界。

今天听 AI 写的歌，我也有一种很复杂的感受。AI 真的可以生成很好听的旋律，可以把人的情绪调动起来，可以让一个工作很累的人短暂放松。这是技术温柔的一面。它让创作更容易，让普通人也能听到自己写下的文字变成歌，让一些无法说出的感受有了声音。

但也正因为它能感染人，才更需要人保留判断。

好听不等于真实属于自己，感动不等于已经理解，数据丰富不等于结论正确，生成流畅不等于思想对齐。人可以享受 AI 带来的音乐、文字和效率，也可以从中获得帮助，但人不能把自己的判断完全交出去。

我越来越觉得，AI 时代不是要求人变得没有情绪，而是要求人更清楚地认识情绪。不是要求人拒绝数据，而是要求人更清楚地理解数据。不是要求人不用工具，而是要求人在使用工具时仍然知道：什么是感受，什么是事实，什么是建议，什么是确认，什么可以进入执行，什么必须先停下来。

这也许就是未来人的一种新能力。

不是完全隔离世界，而是在接触中保护自己。不是拒绝信息，而是不被信息吞没。不是拒绝情绪，而是不被情绪带走。不是拒绝数据，而是不让数据替自己活。

当下观察

情绪会感染人，数据也会感染人。

一个人如果没有内在判断，很容易被别人的情绪带偏；一个系统如果没有判断边界，也很容易被数据的表面强度带偏。

AI 时代真正需要建立的，不是完全隔离数据，而是让人和系统在接触数据时仍然保留距离、判断和确认。

数据可以帮助理解世界，但不能自动接管行动。情绪可以提醒人哪里重要，但不能直接决定后果。

未来真正重要的能力，可能是既能接触世界，又不被世界带走。

July 6, 2026｜When Data Influences People Like Emotion

Today I wrote several songs and listened to music generated with AI. Some of the melodies were surprisingly beautiful. After working for an entire day, stepping away from files, systems, webpages, repositories, and text for a while felt like a small form of relief. Music is strange. It does not need much explanation, yet it can touch human emotion very quickly. A melody begins, and the heart moves with it. One moment a person may be dealing with something highly rational, and the next moment a sound can carry them back many years, toward certain people, certain roads, and certain words that were never fully spoken.

Human beings are easily affected.

Not only by music. A sentence from another person, a facial expression, a tone of voice, a glance, or even silence can influence emotion. Sometimes one begins calmly, but after meeting an anxious person, one becomes anxious too. After meeting an angry person, one begins to defend. After meeting a sad person, the heart sinks. Emotions seem to flow between people. They spread, amplify, and reshape one another.

This is one of the contradictions of being human.

People need contact with others. They need to be understood. They need empathy, warmth, and connection. But the more they come into contact with others, the more easily they may also be pulled away from themselves. If a person is not internally stable, they may mistake another person’s emotion for their own emotion, another person’s judgment for their own judgment, and external noise for their real direction.

To live clearly, then, is not to cut off the outside world completely. It is to remain aware of what belongs to oneself while still being in contact with the outside world.

Today I suddenly felt that data is also very similar to emotion.

On the surface, data appears rational. It appears as numbers, records, text, images, clicks, searches, timestamps, locations, behaviors, versions, and statistical results. People easily assume that because something is data, it is more objective, more reliable, and less deceptive than emotion. But once data enters a system, it can influence judgment just like emotion. It can be selected, amplified, ranked, interpreted, recommended, and carried by a particular structure.

When a number rises, people may begin to believe it matters more. When a rating falls, they may become nervous. When a name appears repeatedly, they may assume it is more credible. When a system recommends a direction, they may unconsciously assume that it is more correct. Data does not always command directly, but it changes attention. It may not enter the body as visibly as emotion, but it can slowly change the order in which a person sees the world.

This resembles emotional contagion.

Emotional contagion makes people unconsciously follow the state of others. Data contagion makes people unconsciously follow the priorities displayed by systems. A person may believe they are judging independently, when in fact their attention has already been pulled by ranking, recommendation, statistics, and repeated information. More data does not necessarily make judgment clearer. Sometimes, the more data surrounds a person, the more difficult it becomes to remember what was actually being judged.

This is one of the major structural problems of the AI era.

AI can access large amounts of data and turn that data into summaries, suggestions, rankings, predictions, and generated content. It can help people see relationships that were previously difficult to see and process complex information more quickly. But at the same time, it can amplify some data, suppress other voices, treat certain patterns as direction, treat repetition as importance, and treat surface correlation as real cause.

If human beings lack judgment boundaries, they will be carried away by data.

Just as an emotionally unstable person may be easily affected by another person’s sentence, a judgment system without stability may be easily affected by the apparent strength of data. Seeing more does not mean understanding more deeply. Contacting more does not mean judging better. Being surrounded by a large amount of data does not mean being closer to truth. Often, truth does not lie in the largest quantity of data. It requires judgment about which data matters, which data is noise, which data reflects reality, and which data is only an amplified echo of the system.

I am beginning to think that the important question in the future is not whether we should come into contact with data, but how we can contact data without being taken over by it.

Human beings cannot completely separate themselves from emotion. Without emotion, people lose many forms of feeling, relationship, and understanding. But people also cannot be completely ruled by emotion, or they will be pulled by the outside world. Data is similar. Humans and systems cannot completely separate themselves from data. Without data, they cannot understand complex reality or make better decisions. But humans and systems also cannot be completely ruled by data, or judgment will be captured by statistics, ranking, and visibility.

The key is not isolation. The key is boundary.

Emotion needs boundary. Data also needs boundary.

A strong person is not someone who is never affected. A strong person is someone who knows they are being affected and can pause to ask: Is this my emotion, or someone else’s? Is this fact, or is this my reaction after being triggered? Is this my judgment, or am I only trying to prove that I have not lost?

In the same way, a mature data system should not only ask: what does the data say? It should also ask: where did this data come from? Whom does it represent? Who was not recorded? What was amplified? What was hidden? Is this data sufficient to support action? Did the human being truly understand the consequences before action was taken?

If data can affect people like emotion, then the AI era needs a new form of protection: allowing people to contact data without letting data directly push execution; allowing systems to use data without turning data automatically into responsibility; allowing models to generate suggestions without treating suggestions as final judgment.

This connects to the language-to-action boundary that I have been thinking about recently. Language spoken is not execution. Expression recorded is not authorization. Data seen is not automatically a basis for action. Whether language, emotion, or data enters the judgment process, it must not bypass human judgment and move directly into execution.

If future AI systems pursue only speed, they may easily treat data as decision. They may say: because many people clicked, this is important; because many people chose this, this is a trend; because it often happened before, it should happen again; because the model predicts a high probability, execution is acceptable. But human society is not made only of probabilities. Real people have exceptions, dignity, unrecorded circumstances, possibilities of being misunderstood by systems, and the right to refuse being defined by data at critical moments.

Data can support judgment, but it cannot replace judgment.

This sounds simple, but it will become increasingly difficult in the AI era. The stronger the system becomes, the more easily people believe the system has seen more. The more fluent the model becomes, the more easily people assume it understands. The richer the data becomes, the more easily people hand complex reality over to automatic summary. But many of the most important judgments take place precisely where data cannot fully explain.

For example, an honest worker may not have polished digital records, but that does not mean they lack ability. A student may perform poorly for a period of time, but that does not mean they have no future. A patient may have unusual indicators, but that does not mean their life situation can be fully defined by a set of numbers. A person may say something emotional in one conversation, but that does not mean it is their real, stable, executable intention. A tool may be used by many people, but that does not mean it is necessarily the more responsible choice.

Data needs to be contacted. It also needs to be questioned.

Emotion is the same. A person should not reject emotion completely simply because emotion may mislead them. Emotion sometimes reminds us where we are hurt, what matters, what needs protection, and what feels wrong. But emotion cannot directly take over action. A mature person is not someone without emotion, but someone who can maintain distance from emotion: I feel this, but I will not immediately be controlled by it.

Data should be treated similarly.

A mature system is not one that refuses data, but one that can maintain distance from data: I see this data, but I do not automatically treat it as final truth; I refer to these patterns, but I do not let patterns decide for human beings; I recognize the value of data, while also recognizing that beyond data there are human circumstances, human judgment, and responsibility boundaries.

Listening to AI-generated music today gave me a complicated feeling. AI can truly generate beautiful melodies. It can move emotion. It can help a tired person briefly relax. This is a gentle side of technology. It makes creation easier. It allows ordinary people to hear their own words become songs. It gives sound to feelings that may have been difficult to express.

But precisely because it can affect people, human judgment becomes even more necessary.

Beautiful does not automatically mean truly belonging to oneself. Being moved does not mean understanding has occurred. Rich data does not mean the conclusion is correct. Fluent generation does not mean the thought is aligned. People can enjoy the music, writing, and efficiency that AI brings. They can receive help from it. But they cannot give away their judgment completely.

I increasingly feel that the AI era does not require human beings to have no emotion. It requires them to understand emotion more clearly. It does not require humans to reject data. It requires them to understand data more clearly. It does not require humans to avoid tools. It requires them to know, while using tools, what is feeling, what is fact, what is suggestion, what is confirmation, what can enter execution, and what must pause first.

Perhaps this is a new human capability for the future.

Not to isolate oneself completely from the world, but to protect oneself while remaining in contact. Not to reject information, but to avoid being swallowed by information. Not to reject emotion, but to avoid being carried away by emotion. Not to reject data, but to refuse letting data live on one’s behalf.

Current Observation

Emotion can affect people, and data can affect people too.

If a person lacks inner judgment, they can easily be pulled away by another person’s emotion. If a system lacks judgment boundaries, it can also be pulled away by the apparent strength of data.

What the AI era needs is not complete isolation from data, but the ability for people and systems to maintain distance, judgment, and confirmation while coming into contact with data.

Data can help people understand the world, but it must not automatically take over action. Emotion can remind people what matters, but it must not directly determine consequences.

A future human capability may be the ability to contact the world without being carried away by it.

Research Context

Complex Systems · AI Governance · Decision Architecture · Enterprise Transformation · Human Judgment

AI-Era Structural Notes · Data Contact Boundary · Emotional Contagion · Judgment Boundary · Decision Architecture

Responsibility Drift · Meaningful Confirmation · Data Influence · Human Attention · Structural Protection

Canonical Public Research Reference

Xufen Tu（涂绪芬）

Independent Interdisciplinary Researcher in AI Governance & Complex Systems

Human Judgment as Structural Constraint in AI-Mediated Complex Systems

Canonical public research source: https://xufentu.com