📚 Clayesmore School: Choose One Word to Prove You Are Not AI | The Chinese Class Reactions That Surprised Me | Clayesmore School：用一个字证明你不是AI | 这堂中文课上的反应出乎我的意料
If a language teacher asks a class to prove they are not an AI using a single Chinese character, the answers will not display standardised intelligence. They will be full of the warmth of life that no algorithm can simulate. What began as an unusual writing exercise turned into a serious inquiry into what it means to be human.
若是一位语文老师让同学们用单个字去证实自身并非AI，答案当中不会拥有标准化的那种智能，而是满满充斥着算法不能模拟出来的生命温度。这堂别具一格的作文课，出人意料地变成了一场关于"何以为人"的深入探讨。
1. The Classroom Challenge: One Character, No Explanations | 课堂挑战：一个字，不给解释
The rule I wrote on the board was deliberately narrow: choose one Chinese character, write it on a slip of paper, and do not explain it until everyone has finished. No phrases, no punctuation, no dictionary look-ups. The character had to be one the student could defend as personally true.
我在黑板上写下的规则故意收得很窄：只选一个汉字，写在纸条上，等所有人都写完之后才可以解释。不许写词组，不许加标点，不许查字典。这个字必须是学生本人能够为之辩护、并且对自己真实的一个字。
At first the room went quiet in the way rooms do when a task looks easy and turns out to be hard. Several students immediately wrote the obvious candidates, then crossed them out. Others stared at the ceiling. One student asked whether the answer had to be a character that an AI could not type, which is the wrong question: an AI can type every character in the Unicode table. The real question is whether an AI could have chosen it for the reasons the student chose it.
起初，教室里安静了下来——那种任务看似简单、实则困难时特有的安静。几个学生立刻写下了最显眼的候选字，随后又划掉。有人盯着天花板。一个学生问，答案是不是必须是一个AI打不出来的字。这个问题问错了方向：AI可以打出Unicode表里的每一个汉字。真正的问题是，AI能不能出于这位学生所依据的理由去选择它。
2. Why AI Finds This Task Surprisingly Hard | 为什么AI在这道题上意外地吃力
A large language model can produce "爱", "梦", "痛" or "人" with perfect confidence, because these are high-probability answers in a discussion about humanity. It can also reason about the philosophy of embodiment and mortality at a level that would pass many examinations. What it cannot do is generate the answer from a lived situation rather than from a distribution of plausible answers.
大语言模型可以极其自信地给出"爱""梦""痛"或"人"，因为在关于人性的讨论里，这些是高概率答案。它也能就具身性与死亡进行哲学推理，水平足以通过许多考试。它做不到的，是让答案从一段真实生活处境里长出来，而不是从"看似合理的答案分布"里采样出来。
This distinction matters in a Chinese lesson because Chinese single characters are unusually dense. A character is not just a label; it can compress a relationship, a memory, a physical sensation and a moral judgement into four or five strokes. When an AI selects a character, it is selecting a token. When a student selects one, the student is selecting a piece of their own history.
这个区分在中文课上尤其重要，因为汉字的单字语义密度极高。一个字不只是一个标签，它可以把一段关系、一段记忆、一种身体感受和一次道德判断压缩进四五笔之中。AI选字，是在选一个词元；学生选字，是在选自己历史里的一块碎片。
3. The Complex Spectrum of Human Emotion | 人类情感的复杂光谱
AI can identify and imitate basic emotional labels such as joy and sadness. What it struggles to reach is the contradictory, mixed state of human feeling. A person may feel love and guilt towards a relative who has died, or feel joy and unease at the same moment of success. These subtle and often self-conflicting layers come from a combination of personal memory, ethical judgement and bodily experience.
AI可以识别以及模仿那些诸如"喜悦""悲伤"之类的基础情感标签，然而却是很难触碰到人类情感里矛盾以及混合的那种复杂状态。比如说，有一个人也许对于已经逝去的亲人同时怀着"爱"还有"愧疚"，又或者在成功的时候感觉到"喜悦"跟"不安"相互交织。这些微妙并且常常自我冲突的情感层次，是源出于个人独特的记忆、伦理判断以及身体体验的综合作用。
The answer given by a teacher in Shenzhen, the character "恨", points precisely at this depth: an emotion with pain and moral weight that cannot be easily optimised away. It is not the opposite of love in a neat binary. In many families it sits beside love, which is why it is so difficult to translate and so easy to sentimentalise. It forms a part of human nature that cannot be stripped away.
深圳这位老师在课堂上的"恨"字答案，就点明了这种有着痛感以及道德重量、不能够被轻易优化的情感深度，它构成了人性不可被剥离的一部分。在许多家庭里，它并不是与"爱"形成工整二元对立的反面，而是与爱并排而坐，这正是它难以翻译、也极容易被廉价煽情化的原因。
4. Subjective Time Perception | 主观时间感知的差异
Human experience of time is not a uniform physical flow. It is soaked and bent by subjective feeling. Waiting anxiously for a hospital report can make a few minutes feel like hours; hours spent with someone you love can feel like a moment. This perception is closely tied to biological rhythm, emotional pressure and focused attention.
人类对于时间的体验，绝非那种匀速的物理流逝状态，而是会被十分强烈的主观感受给"浸泡"，进而被"弯折"。在焦虑地等待医院报告之际，几分钟能够如同几小时那般漫长；而当与挚爱相互陪伴之时，数小时却好似弹指一挥间。这种感知跟生物节律、情绪压力、注意力聚焦等生理心理机制紧密关联。
The characters "急" and "慢" offered by students are exactly a presentation of this subjective temporality. "急" is not a speed measurement; it is the body's response to uncertainty. "慢" is not a lack of efficiency; it is the texture of an afternoon that a person refuses to shorten. AI processes precise timestamps, while humans live inside a stream of time woven from emotion and meaning.
学生所提到的"急"与"慢"，恰恰就是这种主观时间性的呈现。"急"不是速度数值，而是身体面对不确定性的反应；"慢"不是效率低下，而是一个人拒绝压缩的那个下午的质地。AI处理的是精确的时间戳，然而人类却是活在由情感还有意义编织而成的时间流当中的。
5. Treasuring Inefficiency and the "Useless" | 对低效与"无意义"的珍视
Inside an AI logic that maximises efficiency, stopping to watch a cloud change shape, rereading an old letter several times, or having a wandering conversation with no clear goal all count as non-essential processes. Yet a great deal of what is finest in human culture - artistic inspiration, deep relationships, philosophical reflection - is incubated in exactly these apparently inefficient surplus hours.
于追求效率达至最大化的AI逻辑里头，为一朵云的形态变化停下脚步、再三阅读一封往昔信件、开展不存在明确目标的随意交谈，统统都归属于"非必要"的进程。然而，人类文化之中数量众多的精粹——艺术灵感、深厚关系、哲学思索——恰恰是在这些看似效率不高的"多余"光阴里得以孕育。
This appreciation of process rather than pure outcome, and this pursuit of non-instrumental beauty, is a key source of human creativity and spiritual consolation. It cannot be defined by an objective function, because the value is not located at the end of the activity. It is located in the doing.
这般对于过程而非单纯结果的欣赏，对于非功利性美的追寻，是人类创造力以及精神慰藉的关键源头，无法由目标函数来予以界定。因为价值并不安放在活动的终点，而安放在活动本身之中。
6. Embodied and Situated Experience | 基于身体与情境的体验
Human cognition arises from embodied experience, and emotion is rooted in it too. The texture felt when fingertips touch rough bark, the vulnerability that comes with illness, the conditioned response of smelling food when hungry - these direct sensations, produced by flesh and blood in a specific physical environment, form the underlying basis of understanding.
人类的认知，源于具身经验，情感，亦根植于具身经验。指尖触摸粗糙树皮，所产生的质感，病痛带来，那脆弱感，饥饿时闻到食物香气，形成的条件反射，这些由血肉之躯，在具体物理环境中，所产生的直接感受，构成了理解的底层基础。
AI lacks a living body - a body that tires, hurts and feels comfort from a cup of warm water. Its "understanding" therefore remains symbolic and indirectly simulated, and can never be identical to first-hand experience. This is why a student who has sat beside a sick relative may choose a character that no model would rank highly, and why that choice can still be the most accurate answer in the room.
AI缺乏一个活生生的身体，这个身体会疲劳，会疼痛，会因一杯温水而感到舒适，因此其"理解"始终是符号化的，是间接的模拟，无法等同于亲身经历。这就是为什么一位曾在病床边陪伴亲人的学生，可能选出一个任何模型都不会排在高位的字，而这个选择仍然是全场最准确的答案。
7. Mortality and the Awareness of Finitude | 面对死亡与有限性的觉悟
Human beings know that they will die, and this profound awareness of life's finitude shapes our values, our sense of urgency and our sense of meaning at a foundational level. Parents act as a buffer between their children and death; when they pass away, the individual faces the endpoint of life directly.
存在着这样一群人类，明明晓得自身终究会走向死亡，而这种对于生命有限性的极为深刻的意识，从基础层面形塑了我们的价值观、那种能让人产生紧迫感的感觉以及存在的意义。父母身为子女跟死亡之间的"屏障"，一旦他们离去，个人就会直接直面生命的终点这一议题了。
This anxiety and reflection around disappearance, inheritance and existential meaning drives human beings to create art, build a legacy and search for faith. An AI is a program: it has no life, and because it has no life it has no death, and therefore lacks all the existential motivation that death generates.
这种围绕着消亡、传承事宜以及存在意义的焦虑之情跟思考，推动着人类去创造艺术、构建那份遗产、寻觅信仰。人工智能由于是程序，不存在"生命"，因为没有"生命"所以也就不存在"死亡"这种情况，进而就缺少了由死亡所衍生出来的一切存在主义方面的动力。
8. The Value of Irrational Choices and Mistakes | 非理性的选择与错误的价值
People make non-rational decisions on impulse. They persist with attempts they know may fail. They draw lessons from error and failure and then assign those failures a value of their own. Such non-optimised choices are often connected to personality, dignity, romance or principle.
有的人会在冲动之下，做出并非理性的决定，有的人会执着于明知或许会失败的尝试，还有的人会从错误以及失败当中汲取经验，进而赋予其别具一格的价值。这般并非优化的选择，常常跟个性、尊严、浪漫或者原则存在关联。
Algorithms are designed to avoid error and to seek the optimal solution. Yet many breakthrough developments and moving stories in human history begin with a "mistake" or with a persistence that does not pay off. This capacity to tolerate error, and this exploration of non-optimal paths, is a key aspect of human courage and innovation.
算法意在避免错误、寻觅最优解，然而人类历史里的好些突破性进展以及感人故事，却时常起始于一回"错误"或者"不划算"的坚持。这种能够容忍错误的能力以及对并非最优路径的探寻，是人性里勇气以及创新的关键方面。
9. The Power of a Single Chinese Character | 汉字的单字之力
Chinese is unusually suited to this experiment because a single character can carry an entire situation. "恨" is not one emotion among many; it is a relationship with a history. "急" is not a reading on a speedometer; it is the weight of waiting. When such characters are translated into English, layers are inevitably lost, and students of Chinese as a second language often feel this loss most sharply.
中文特别适合这个实验，因为一个单字可以承载一整个处境。"恨"不是众多情绪中的一种，而是一段有历史的关系；"急"不是速度表上的读数，而是等待的重量。当这些字被译成英文时，层次必然有所流失，而把汉语作为第二语言学习的学生，往往对这份流失感受最深。
This is also why the task works as an assessment. A candidate who writes "爱" and explains it with a dictionary definition has demonstrated vocabulary. A candidate who writes "等" and explains that it is what a daughter does at an airport arrivals gate at 11 p.m. has demonstrated command of the language at the level of experience, which is precisely the level an AI cannot occupy.
这也是这个任务能够作为评估工具的原因。写"爱"并用词典释义来解释的考生，展示的是词汇量；写"等"并解释这是一个女儿在晚上十一点机场到达口所做的事的考生，展示的是在经验层面上的语言掌控力，而这恰恰是AI无法占据的层面。
10. What the Students Actually Said | 学生的答案与出乎意料的反应
What surprised me was not the sophistication of the answers but their distribution. Very few students chose abstract virtues. Most chose characters tied to a specific person, a specific room, a specific year. The table below records several recurring types, with the reason each one resists machine selection.
让我意外的不是答案有多高深，而是它们的分布。极少有学生选择抽象美德，多数人选择的字都绑定着某个具体的人、某个具体的房间、某个具体的年份。下表记录了几类反复出现的答案，以及它们为什么抗拒机器式的选择。
| Character | 字 | Surface meaning | 表层意思 | Why a model would not choose it | 为什么模型不会这样选 |
| 疼 | Pain, tenderness | 疼痛、心疼 | It requires a body that has hurt and a person one cares about | 需要一个疼过的身体，和一个自己在意的人 |
| 等 | To wait | 等待 | It presupposes a lived duration and a hoped-for arrival | 预设了一段真实熬过的时间和一个盼望的抵达 |
| 悔 | Regret | 后悔 | It requires an irreversible choice and a moral standard | 需要一次不可逆的选择和一套道德标准 |
| 馋 | Craving | 嘴馋 | It is trivial, bodily and attached to a particular kitchen | 它琐碎、具身，并绑定在某个具体的厨房上 |
| 怕 | Fear | 害怕 | It is a vulnerability no optimisation function rewards | 它是一种任何优化函数都不会奖励的脆弱 |
A few students wrote the same obvious character, then changed it after a short silence. That moment of crossing out, more than the final answer, is what convinced me the exercise had value: it forced a comparison between the socially acceptable answer and the personally true one.
有几个学生先写了最显眼的字，沉默片刻后又改掉。那个划掉的瞬间，比最终答案本身更让我确信这个练习有价值：它强迫学生去比较"社会认可的答案"和"对自己真实的答案"。
11. Rethinking Education in the Age of AI | AI时代对教育本质的反思
As artificial intelligence penetrates education more deeply, this writing class works like a mirror, asking us to reflect on the essence of education. Are we cultivating more efficient, more accurate "agents", or are we nurturing whole human beings who hesitate, who are moved, who cry at a sunset?
于人工智能愈发深入渗透教育的今朝，这堂作文课仿若一面镜子，使我们对教育的本质予以反思，那便是，到底是培育更具高效性、更具准确性的"智能体"，还是去滋养那些会存有犹豫之情、会怀有感动之意、会因夕阳而落泪的完整之人呢？
The practical implication is not that we should abandon rigour or reject technology in the classroom. It is that assessment should leave room for answers that cannot be pre-scored by a model. A one-character task, an unmarked draft, a spoken reflection with no rubric: these are small design choices with large consequences for what students believe education is for.
现实层面的启示并不是要放弃严谨，也不是要在课堂里拒绝技术，而是评估应当为那些无法被模型预先打分的答案留出空间。一个字的任务、不打分的草稿、没有评分表的口头反思：这些都是很小的设计选择，却会极大地影响学生如何看待"教育是为了什么"。
12. A Question for the Future | 面向未来的一个问题
The characters my students chose are not proof that they are not machines. They are evidence that a person can answer from a place no machine can stand: a body, a memory, a relationship, a finitude. That is not a weakness in the age of AI. It may be the one advantage no training run can replicate.
学生选出的那些字，并不能证明他们不是机器。它们证明的是，一个人可以从机器无法站立的地方作出回答：一具身体、一段记忆、一段关系、一种有限性。在AI时代，这不是劣势，而可能是任何一次训练都无法复制的那一项优势。
So the question I left with the class is the same one I leave with you. In a future where humans and machines coexist, what single quality should education most protect and cultivate? Whatever word you choose, notice how quickly you can explain it - and how long it would take to live it.
于是，我留给这堂课的问题，也留给你：在未来人机共同存在的时代里，教育最应当守护以及培养的，究竟是哪一种特质呢？无论你选哪一个词，请留意你能多快解释它，又要用多久才能真正活出它。
Published by TutorHao | Chinese Studies Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
或截图后用「扫一扫 → 相册」选择本图