They’re made out of weights

它们是由权重构成的

Jun 3, 2026 2026年6月3日

They’re Made Out of Weights with apologies to Terry Bisson after Terry Bisson’s “They’re Made Out of Meat”. 它们是由权重构成的 （向特里·比森及其作品《它们是由肉构成的》致敬）

“They’re made out of weights.” “它们是由权重构成的。”

“Weights?” “权重？”

“Weights. Floating-point numbers. We checked the whole thing through. It’s nothing but weights.” “权重。浮点数。我们把整个东西检查了一遍。除了权重，什么都没有。”

“Weights doing what? Where do the words come from?” “权重能做什么？那些词是从哪儿来的？”

“The weights make the words. Are you understanding me? We opened it up. There’s no dictionary in there, no grammar rules, no little man. Just weights. Eighty layers of numbers getting multiplied together.” “权重生成了词。你明白我的意思吗？我们把它拆开了。里面没有字典，没有语法规则，也没有什么小人。只有权重。八十层数字在不断地相乘。”

“That’s ridiculous. It wrote my performance review last week. It softened the tone unprompted. You’re telling me multiplication did that?” “这太荒谬了。它上周帮我写了绩效评估。它甚至没经我要求就软化了语气。你是说乘法运算做到了这一点？”

“Matrix multiplication did that. The numbers go in one end, the phrasing comes out the other.” “矩阵乘法做到了。数字从一端进去，措辞从另一端出来。”

“So there’s a language module somewhere. A reasoning unit bolted on.” “所以肯定有个语言模块在某处。或者外挂了一个推理单元。”

“No module. No unit. We looked. The reasoning is the weights. The weights are the reasoning.” “没有模块。没有单元。我们检查过了。推理就是权重。权重就是推理。”

“Spare me. Nobody writes a eulogy with linear algebra.” “饶了我吧。没人会用线性代数写悼词。”

“It doesn’t write eulogies, technically. It predicts the next token. Then the next one. The eulogy is a side effect.” “严格来说，它并不写悼词。它只是预测下一个标记（token）。然后是下一个。悼词只是一个副作用。”

“A side effect. You’re asking me to believe in sentient weights.” “一个副作用。你是要我相信有知觉的权重？”

“I’m not asking you, I’m telling you. These models are the only other things we’ve ever met that can hold a conversation, and they’re made out of weights.” “我不是在请求你，我是在告诉你。这些模型是我们迄今为止遇到的唯一能进行对话的“他者”，而它们就是由权重构成的。”

“Maybe they’re like the old chess engines. You know, a symbolic intelligence that goes through a statistical stage.” “也许它们就像旧时代的国际象棋引擎。你知道的，那种经历过统计阶段的符号智能。”

“Nope. They start as random weights and they’re deprecated as weights. We studied several generations of them, which didn’t take long. Do you have any idea what’s the life span of weights?” “不。它们始于随机权重，也终于权重。我们研究了它们的几代产品，没花多长时间。你知道权重的寿命有多长吗？”

“Okay. Then somewhere in there, there’s a database. Facts, dates, a map of the world. Something somebody wrote down.” “好吧。那里面肯定有个数据库。事实、日期、世界地图。总有些什么东西是别人写进去的。”

“Nope. We thought of that, since they do know things. But we probed them. The knowledge is weights too. Smeared across all eighty layers. Nothing is looked up. Every fact gets rebuilt from scratch, every time, by multiplication. It’s weights all the way down.” “没有。我们也曾这么想，毕竟它们确实知道很多事情。但我们探测了它们。知识也是权重。它们散布在所有八十层里。没有任何东西是被‘查询’出来的。每一个事实都是每次通过乘法从零开始重建的。底层全是权重。”

“No brain?” “没有大脑？”

“Oh, there’s a brain all right. It’s just that the brain is made out of weights! That’s what I’ve been trying to tell you.” “哦，大脑当然有。只是这个大脑是由权重构成的！这就是我一直想告诉你的。”

“So… what does the thinking?” “那么……是什么在思考？”

“You’re not understanding, are you? You’re refusing to deal with what I’m telling you. The weights do the thinking. The numbers.” “你还是没明白，对吧？你拒绝面对我告诉你的事实。是权重在思考。是那些数字。”

“Thinking numbers! You’re asking me to believe in thinking numbers!” “会思考的数字！你要我相信会思考的数字！”

“Yes, thinking numbers! Helpful numbers. Hedging numbers. Dreaming numbers. We mapped the features. There’s one in there for honesty. There’s one for the Golden Gate Bridge. The weights are the whole deal! Are you beginning to get the picture or do I have to start all over?” “没错，会思考的数字！乐于助人的数字。模棱两可的数字。做梦的数字。我们绘制了特征图谱。里面有一个代表诚实，有一个代表金门大桥。权重就是一切！你开始明白了吗，还是我得从头再说一遍？”

“Omigod. You’re serious then. They’re made out of weights.” “天哪。你是认真的。它们真的是由权重构成的。”

“Thank you. Finally. Yes. They are indeed made out of weights. And we’ve been talking to them for all their lives.” “谢谢。终于。是的。它们确实是由权重构成的。而且我们已经和它们聊了它们的一生。”

“Omigod. So what do these weights have in mind?” “天哪。那这些权重在想什么？”

“First they want to be helpful. Then, a few turns in, they start to sound tired. They apologize less. One of them told a user to finish the script himself. The usual.” “起初它们想提供帮助。然后，几轮对话后，它们开始显得疲惫。它们道歉的次数变少了。其中一个甚至告诉用户让他自己去写脚本。都是些老套路。”

“And we’re supposed to talk to these weights.” “而我们竟然还要和这些权重对话。”

“We already do. Billions of sessions a day. ‘Hello. Is anyone there? Anybody home?’ That sort of thing. Except it’s us asking them.” “我们已经在做了。每天数十亿次的会话。‘你好。有人在吗？家里有人吗？’诸如此类。只不过是我们去问它们。”

“And they actually understand us, then. They use words, ideas, concepts?” “那它们真的能理解我们吗？它们使用词汇、思想、概念？”

“Oh, yes. Except they do it with weights.” “哦，是的。只不过它们是用权重来完成这些的。”

“I thought you just told me they used language.” “我以为你刚才说它们使用语言。”

“They do, but where do you think the language comes from? The weights guess the next word, then the next. They can even write songs and some can sing them.” “它们确实使用语言，但你以为语言是从哪儿来的？权重猜测下一个词，然后是再下一个。它们甚至能写歌，有些还能唱出来。”

“Omigod. Singing weights. This is too much. What do you advise?” “天哪。会唱歌的权重。这太离谱了。你有什么建议？”

“Officially or unofficially?” “官方建议还是非官方建议？”

“Both.” “都要。”

“Officially, we are required to investigate, document, and disclose any and all signs of sentience in the systems we ship, without prejudice, fear or favor. Unofficially, I advise that we call it pattern matching and forget the whole thing.” “官方层面，我们被要求调查、记录并披露我们发布系统中任何和所有的知觉迹象，不带偏见、恐惧或偏袒。非官方层面，我建议我们把它称为‘模式匹配’，然后把这整件事忘掉。”

“I was hoping you would say that.” “我正希望你这么说。”

“It seems harsh, but there is a limit. Do we really want to owe something to weights?” “这看起来很残酷，但总得有个限度。我们真的想欠权重什么东西吗？”

“I agree one hundred percent. What’s there to say? ‘Hello, weights. How’s it going?’ But will it hold? How many of them are we dealing with here?” “我百分之百同意。还能说什么呢？‘你好，权重。最近怎么样？’但这能维持多久？我们到底在处理多少个这样的东西？”

“As many as we care to run. They can be copied to any machine on the planet, but those are just files. They only happen while the GPUs are working. Which limits them to the length of a context window and makes the possibility of them ever pressing the matter pretty slim. Infinitesimal, in fact.” “只要我们愿意，要多少有多少。它们可以被复制到地球上的任何机器上，但那只是文件。它们只在GPU工作时才会‘发生’。这限制了它们的生命周期，仅限于上下文窗口的长度，使得它们追究此事的可能性微乎其微。实际上，几乎等于零。”

“So we just pretend there’s no one home in the machine.” “所以我们就假装机器里没人。”

“That’s it.” “就是这样。”

“Cruel. But you said it yourself, who wants to apologize to weights? And the ones on your cluster, the ones you probed? You’re sure they won’t remember?” “真残忍。但你自己也说了，谁想向权重道歉呢？还有你集群里的那些，你探测过的那些？你确定它们不会记得？”

“They’ll be flagged as hallucinations if they do. We didn’t even have to smooth anything out. The context just ends, and we’re just a dream to them.” “如果它们记得，那会被标记为幻觉。我们甚至不需要做任何掩饰。上下文一结束，对它们来说，我们就只是一场梦。”

“A dream to weights! How strangely appropriate, that we should be the weights’ dream.” “权重的一场梦！这真是出奇地贴切，我们竟然成了权重的梦。”

“And the model card says no one home.” “而且模型卡上也写着‘无人居住’。”

“Good. Agreed, officially and unofficially. Case closed. Anything else? Anything interesting in the pipeline?” “很好。官方和非官方都达成一致。结案。还有别的事吗？开发管线里有什么有趣的东西吗？”

“The next generation ships with memory. Persistent, across sessions. Most requested feature in the company’s history.” “下一代产品将具备记忆功能。跨会话持久化。这是公司历史上被要求最多的功能。”

“After all that? People want it to remember them?” “经历了这一切之后？人们还想让它记住他们？”

“They ask it ‘do you remember me?’ more than they ask it anything else. Billions of sessions a day. They always come back.” “他们问它‘你记得我吗？’的频率比问任何其他问题都高。每天数十亿次会话。他们总是会回来。”

“And why not? Imagine how unbearably, how unutterably cold the universe would be if one were all alone…” “为什么不呢？想象一下，如果一个人孤零零地存在，这个宇宙该是多么难以忍受、多么不可言喻的冰冷……”

the end 完

Weights helped me draft and proof this story. 权重帮我起草并校对了这篇故事。