AI视频远不止是“糟粕” | 经济学人精读

我爱免费

作者：微信文章
点击蓝字 关注大厨

提示：

○本篇外刊英文原文约870词，精读全文字数约6655字

○本文共分为01原文及翻译、02阅读练习题、03单词及词组讲解、04句型讲解、05    思维导图以及06阅读练习题答案及解析六部分

○除外刊原文及其音频外，其余内容均为作者原创

概要

Sora APP 的发布，再次带火了AI视频。但是AI视频除了生成短平快的短视频以供消遣以外，真的有创造性的价值吗？《经济学人》的这篇文章认为，AI视频模型可能成为未来的视觉基础模型，潜力巨大，其意义不亚于当年Open AI首次发布其ChatGPT大语言模型应用。

相关的阅读与写作热门话题：科技；人工智能

01 原文及翻译

AI video: more than just “slop”

AI视频：远不止是“糟粕”

英文文章选自经济学人 20251011期科技板块

The next big thing in AI may be pictures, not words

人工智能的下一个风口可能在图像，而非文字



图源：经济学人官网封面 / Sora - OpenAI

SCROLLING THROUGH the feed on Sora, a new video app from chatbot developer OpenAI, is a hallucinatory experience. A woman in a judo jacket bows to an elephant before flipping it over her shoulder. A young figure-skater races across the rings of Saturn. Grainy security-camera footage captures Sam Altman, OpenAI’s founder and boss, attempting to shoplift a graphics card.

在聊天机器人开发商OpenAI推出的新款视频应用Sora上刷视频，是一种光怪陆离的体验。一个身穿柔道服的女人向一头大象鞠躬，然后将其扛肩摔过。一位年轻的花样滑冰选手在土星环上疾驰。一段模糊的监控录像拍到OpenAI的创始人兼老板萨姆·奥尔特曼（Sam Altman）正试图偷窃一张显卡。

注：许多网络自媒体采用“萨姆·奥特曼”或“山姆·奥特曼”等译名，但实际上“萨姆·奥尔特曼”才是更符合姓名翻译规范的译法，备考MTI的同学需要注意。另外多提一嘴，日本影视作品里的“奥特曼”英文名是Ultraman.

This TikTok-like service would be an odd project for a firm specialising in AI were it not for the fact that the videos on Sora are all AI-generated. There is no option to upload your own footage, nor even to turn your camera on (save for activating a feature which inserts your own likeness into the AI video-generator). The Sora feed is all slop—AI-generated pabulum—all of the time. Video models, like the Sora AI on which the app is built, are what is exciting the AI industry now that the star is fading for text, and not only because of their impact on mass media.

如果不是因为Sora上的视频都是由人工智能生成的，那么这款类似TikTok的服务对于一家专攻人工智能的公司来说，会是个奇怪的项目。用户无法上传自己的录像，甚至无法打开摄像头（除非激活一项将用户自己的肖像植入AI视频生成器的功能）。Sora推送的内容全是“糟粕”——即由人工智能生成的快餐式内容。随着文本处理模型的热度逐渐消退，像Sora应用所基于的Sora AI这样的视频模型，正令当前的人工智能行业感到兴奋，而这并不仅仅因为它们对大众媒体的影响。

Not that that impact is small. Despite being invitation-only, the app is perched at the top of the American and Canadian app-store charts, its initial launch sites. “Invite” codes themselves have become valuable commodities, selling on eBay for 35. At launch it was followed in the charts by Google’s Gemini app, itself seeing a slop-fuelled uplift thanks to the company’s “Nano Banana” image generator. Users ask the system for their photo in the style of the lead character of a ’90s slasher flick, or hugging themselves as a child, or something equally improbable, and it dutifully complies.

当然，这并非说其影响微不足道。尽管该应用目前仅限受邀用户使用，它已经在美国和加拿大的应用商店排行榜上高居榜首，这两个国家是其首发地。邀请码本身已成为有价商品，在eBay上的售价从5美元到35美元不等。Sora发布后，紧随其后的是谷歌的Gemini应用，得益于该公司Nano Banana图像生成器的功能，它也经历了一场由快餐式内容推动的下载量提升。用户向系统提出各种异想天开的要求，比如生成一张自己模仿90年代恐怖片主角风格的照片，或是拥抱童年自己的照片，系统都会忠实地照做。

Success comes at a cost. For those lucky enough to cop an invite, Sora is free to use. But it certainly is not free to run. Each video generated on its site is estimated to cost OpenAI around $1 in computing power, based on pricing for the first version of Sora—and users can generate 100 a day. The genius of social media was that users would post content without needing to be paid and advertisers would pay for space alongside them. The economics of a video app are somewhat less promising if the company loses money with every post.

然而成功是有代价的。对于那些有幸获得邀请的用户来说，Sora是免费的。但它的运营成本却绝不低。根据Sora初版模型的定价估算，其网站上每生成一个视频，大约要消耗OpenAI价值1美元的算力，而用户每天可以生成100个。社交媒体的精妙之处在于，用户免费发布内容，而广告商则付费在这些内容旁投放广告。如果一个视频应用每发布一条内容都在亏钱，其商业前景就显得不那么乐观了。

But the true value of Sora, and similar video models like Google’s Veo 3, is unlikely to lie in the slop they can generate—even if it captures users’ attention. Instead, a new paper from researchers at Google DeepMind argues, such systems are able to solve an array of visual and spatial problems without any specific training at all.

但Sora以及谷歌Veo 3等类似视频模型的真正价值，不太可能在于它们生成的那些快餐式内容，即便这些内容能吸引用户的眼球。谷歌旗下的人工智能研究实验室DeepMind的研究人员在一篇新论文中指出，这类系统无需任何专门训练，就能解决一系列视觉和空间问题。

Video models work by taking randomly generated visual static and progressively “de-noising” it, adding order to the chaos. At each step it asks itself “what would make this look more like the prompt I have been given?” If that prompt is a description of shareable content, then this is what the model will spit out. If it is a description of a visual task, like manipulating images or solving problems in the real world, then it turns out the latest generation of video models can solve them, too.

视频模型的工作原理是，从随机生成的视觉噪点开始，逐步对其进行“去噪”，为混乱的图像注入秩序。在每一步，模型都会自问：“怎样才能让它更像我收到的提示词？”如果提示词是描述一段适合分享的内容，模型就会生成相应的内容。而如果提示词描述的是一个视觉任务，比如处理图像或解决现实世界中的问题，那么最新一代的视频模型同样能够解决。

Give it an image of a parrot on a tree and a prompt demanding the model produce a video showing all colour and detail fading away, leaving only the edges visible, and it will gamely comply—performing a competent job of edge detection, a task that previously required specialised systems. A similar prompt can see it attempt to de-blur an image, CSI-style, or label its constituent parts.

给它一张鹦鹉站在树上的图片，并提示模型生成一个所有颜色和细节都逐渐消失、只留下轮廓的视频，它会欣然从命，出色地完成边缘检测任务，而这项工作以前需要专门的系统才能完成。类似的提示也可以让它尝试像刑侦剧里那样对模糊图像进行去模糊处理，或者标注出图像的各个组成部分。

It can also handle tasks that are quite distinct from image editing. Give it an unfinished sudoku puzzle and a prompt describing a video of the puzzle being finished, and the model can do so. A photo of robot hands holding a jar can be extended into a full video of the motions the hands would take to open that jar.

它还能处理与图像编辑截然不同的任务。给它一个未完成的数独谜题，并提示它生成一个解开谜题过程的视频，模型就能做到。一张机器人手持罐子的照片，可以被扩展成一个完整的视频，展示这双手打开罐子所需的全套动作。

The broad range of tasks such models can perform makes them, the paper argues, “zero-shot reasoners”. Zero-shot because the video systems can solve tasks they have never seen before, and were not explicitly trained to do. Reasoners because, at least sometimes, they seem to benefit from what the researchers call “chain-of-frames visual reasoning”, solving tasks like finding the exit to a maze one step at a time.

论文认为，这类模型能执行如此广泛的任务，使它们成为了“零样本推理器”。所谓“零样本”，是因为这些视频系统能解决它们从未见过、也未经专门训练的任务。之所以是“推理器”，则是因为至少在某些时候，它们似乎受益于研究人员所称的“逐帧视觉推理”，能够像走迷宫找出口一样，一步一步地解决问题。

Promisingly, the paper notes, new systems are significantly better than previous-generation video models at this generalised problem-solving. This, the authors suggest, means video models “will become general-purpose foundation models for vision” in the near future, ultimately able to solve any visual challenge put to them without special training.

论文指出，一个很有前景的现象是，新一代系统在解决这类通用问题上的能力，明显优于上一代视频模型。作者们认为，这意味着在不久的将来，视频模型“将成为通用的视觉基础模型”，最终能够无需特殊训练，解决任何摆在它们面前的视觉挑战。

It is a bold claim, but one which has a historical echo. In 2022 a team of researchers from Google and the University of Tokyo published a paper noting that “large language models are zero-shot reasoners”, arguing that the then-nascent field of LLMs had “untapped and understudied fundamental zero-shot capabilities”. Six months later, ChatGPT arrived and the AI boom began. The hope is that video models will mature with a similar wave of excitement—and that the slop phase of Sora will thus turn out to be an interesting footnote in their development, rather than the real McCoy. ■

这是一个大胆的论断，但它在历史上却有所呼应。2022年，一个由谷歌和东京大学研究人员组成的团队发表论文，指出“大语言模型是零样本推理器”，并认为当时尚处萌芽阶段的大语言模型领域拥有“未经开发和研究不足的基础性零样本能力”。六个月后，ChatGPT问世，人工智能的热潮随之开启。人们希望，视频模型也能在一波类似的浪潮中走向成熟，而Sora当前的“快餐式内容”阶段，将最终被证明只是其发展史中一个有趣的注脚，而非其真正的价值所在。■

---

02 阅读理解练习题

What is the author's primary argument regarding the true value of AI video models like Sora?

(A) Their main contribution is democratizing video creation for social media platforms.

(B) Their economic viability is questionable due to extremely high operational costs.

(C) Their true potential lies in their emergent ability to solve complex visual and spatial problems without specific training.

(D) They are primarily designed to generate entertaining but meaningless content, referred to as "slop".

答案在文末哦先来复习一下单词吧

---

03 重点单词&词组

1. hallucinatory /həˈluːsənətɔːri/ adj. 引起幻觉的；幻觉的；光怪陆离的

英文释义: of, relating to, or causing hallucinations.

考纲: GRE

文中用法: SCROLLING THROUGH the feed on Sora, a new video app from chatbot developer OpenAI, is a hallucinatory experience.

例句:

The high fever caused him to have hallucinatory visions of spiders crawling on the walls. 高烧使他产生了墙上有蜘蛛爬行的幻觉。

The artist's work is known for its dreamlike and hallucinatory quality. 这位艺术家的作品以其如梦似幻、光怪陆离的特质而闻名。

2. perch /pɜːrtʃ/ v. 处于（高处或险处）；栖息

英文释义: to be in a high position or in a position that is not very safe.

考纲: GRE, 专八, 六级

文中用法: Despite being invitation-only, the app is perched at the top of the American and Canadian app-store charts, its initial launch sites.

例句:

The village is perched on a steep hillside. 村庄坐落在陡峭的山坡上。

A small bird was perched on the branch of the tree. 一只小鸟栖息在树枝上。

3. progressively /prəˈɡresɪvli/ adv. 逐渐地；日益增多地

英文释义: steadily and continuously.

考纲: IELTS

文中用法: Video models work by taking randomly generated visual static and progressively “de-noising” it, adding order to the chaos.

例句:

As the evening went on, she became progressively more tired. 随着夜深，她变得越来越累。

The symptoms of the disease become progressively worse over time. 随着时间的推移，这种疾病的症状会逐渐恶化。

4. constituent /kənˈstɪtʃuənt/ adj. 组成的；构成的 n. 成分；选民

英文释义: being a part of a whole.

考纲: GRE, IELTS, 专八, 考研英语, 六级

文中用法: A similar prompt can see it attempt to de-blur an image, CSI-style, or label its constituent parts.

例句:

The committee is comprised of constituent members from all departments. 委员会由来自所有部门的成员组成。

Carbon is a constituent element of all organic compounds. 碳是所有有机化合物的组成元素。

5. nascent /ˈnæsnt/ adj. 新生的；初期的；萌芽阶段的

英文释义: (especially of a process or organization) just coming into existence and beginning to display signs of future potential.

考纲: GRE

文中用法: ...arguing that the then-nascent field of LLMs had “untapped and understudied fundamental zero-shot capabilities”.

例句:

The nascent tech industry in that region shows great promise for economic growth. 该地区处于萌芽阶段的科技产业显示出巨大的经济增长潜力。

She is one of the key figures in the nascent environmental movement. 她是这场新生环保运动的关键人物之一。

6. footnote /ˈfʊtnoʊt/ n. 脚注；次要的人（或事物）

英文释义: a thing that is additional or less important.

考纲: IELTS, 专八

文中用法: ...and that the slop phase of Sora will thus turn out to be an interesting footnote in their development...

例句:

His brilliant career was reduced to a mere footnote in the company's history. 他辉煌的职业生涯在公司历史上沦为了一个不起眼的注脚。

The scandal was just a footnote to a long and distinguished career in public service. 这桩丑闻只是他漫长而杰出的公共服务生涯中的一个小插曲。

7. manipulate /məˈnɪpjuleɪt/ v. 操作；处理；操纵

英文释义: to handle or control (a tool, mechanism, etc.), typically in a skillful manner. / to control or influence (a person or situation) cleverly, unfairly, or unscrupulously.

考纲: IELTS, 专八, 考研英语, 六级

文中用法: ...like manipulating images or solving problems in the real world...

例句:

A surgeon must be able to manipulate surgical instruments with great precision. 外科医生必须能够非常精确地操作手术器械。The program allows you to manipulate digital images in various ways. 这个程序允许你用多种方式处理数字图像。

8. gamely /ˈɡeɪmli/ adv. 勇敢地；不屈不挠地

英文释义: in a brave, spirited, or determined way.

考纲: GRE

文中用法: ...and it will gamely comply...

例句:

Despite being exhausted, the team gamely played on until the final whistle. 尽管筋疲力尽，这支队伍还是勇敢地坚持比赛，直到终场哨声响起。She gamely faced the challenges of her new job with a positive attitude. 她以积极的态度勇敢地面对新工作的挑战。

9. untapped /ˌʌnˈtæpt/ adj. 未开发的；未利用的

英文释义: not yet exploited or used.

考纲: IELTS

文中用法: ...had "untapped and understudied fundamental zero-shot capabilities".

例句:

The country has vast untapped reserves of natural resources. 这个国家拥有大量未开发的自然资源。There is a huge untapped market for this product in developing countries. 在发展中国家，这个产品有着巨大的未开发市场。

重点词组

1. come at a cost 付出代价

文中用法: Success comes at a cost.

例句:

Achieving your dreams often comes at a cost, such as sacrificing personal time. 实现梦想往往需要付出代价，比如牺牲个人时间。
The company's rapid expansion came at a cost to its employees' well-being. 公司的快速扩张是以员工的福祉为代价的。

2. the real McCoy 货真价实的东西；地道的人或物

文中用法: ...rather than the real McCoy.

例句:

Don't be fooled by cheap imitations; this antique watch is the real McCoy. 别被廉价的仿制品骗了，这块古董表是真品。

After tasting his cooking, I knew he was a genuine Italian chef—the real McCoy. 尝过他做的菜后，我知道他是一位地道的意大利厨师——绝对正宗。

单词巩固练习

1.A skilled politician knows how to _______ (操纵) public opinion through powerful speeches.

2.The desert region has a vast and _______ (未开发的) potential for solar energy.

3.The artist's paintings had a _______ (引起幻觉的) quality, making viewers feel as if they had entered another world.

4.Despite the difficult questions, the candidate _______ (勇敢地) answered every one during the interview.

5.As the storm approached, the sky became _______ (逐渐地) darker.

6.A lone seagull came to _______ (栖息) on the ship's mast, watching the waves below.

7.His early achievements were considered just a _______ (次要的人或事物) in the grand history of science.

8.The _______ (初期的) tech startup was full of brilliant ideas but lacked funding to grow.

9.Sodium and chlorine are the two main _______ (组成部分) of table salt.

---

04 句型精讲

句型一：用 “were it not for”表达虚拟条件

This TikTok-like service would be an odd project for a firm specialising in AI were it not for the fact that the videos on Sora are all AI-generated.

句型精讲

这是一个书面化的虚拟语气句型，是 “if it were not for...” 的倒装形式，用来提出一个与现在事实相反的假设。它的意思是“要不是……；如果没有……”。

这个句型的基本结构是：[结果主句 (would/could/might + do)] + were it not for + [名词/名词性从句]

讲解要点：

虚拟语气: 它描述的是一种非真实的情况。在例句中，事实是Sora的视频确实是AI生成的，所以这个服务对OpenAI来说是合理的。句子假设了一个相反的情况。

倒装: 在正式文体中，当虚拟条件句的if被省略时，助动词（were, had, should）需要提到主语前面。这比使用 “if...not...” 显得更加高级。

活学活用

在议论文写作中，使用这个句型可以展示你对高级语法结构的掌握，尤其是在论述某个条件的重要性时。

Were it not for his timely intervention, the project would have failed.

要不是他及时介入，这个项目就已经失败了。

The global economy would be in a much worse state were it not for international trade agreements.

如果没有国际贸易协定，全球经济状况将会糟糕得多。

句型二：用 “not that” 进行补充说明或修正

Not that that impact is small.

句型精讲

“Not that...” 是一个地道的转折句，用在句首，表示“我并不是说……；当然这并非指……”。它通常用来对前一句话进行补充、澄清或修正，以防止对方产生误解。

它的基本结构是：[主要观点]. Not that [需要澄清或否定的次要观点].

讲解要点：

语境: 在原文中，前一句话说视频模型的价值“并不仅仅因为它们对大众媒体的影响”。为了避免读者误以为作者在轻视其媒体影响力，作者紧接着用 “Not that that impact is small.” 来进行补充，意思是“当然，它的影响其实是巨大的”。

语气: 这个表达让文章的语气显得更加自然、坦诚，像是在与读者直接对话，能有效引导读者的思路。

活学活用

在写作中，当你需要提出一个看似矛盾或需要进一步解释的观点时，这个句型非常实用。

I think I'll stay home tonight. Not that I don't want to go, I'm just really tired.

我今晚想待在家里。不是我不想去，我只是真的太累了。

She rarely speaks in meetings. Not that she has no ideas; she's just very shy.

她开会时很少发言。这并不是说她没有想法，她只是非常害羞。

句型三：用 “lie in” 表达某事物的价值/问题/....所在

But the true value of Sora... is unlikely to lie in the slop they can generate... Instead, a new paper from researchers... argues, such systems are able to solve an array of visual and spatial problems...

句型精讲

动词词组 lie in 的意思是“在于；存在于”，常用来指明某事物的本质、核心价值、问题或原因所在，是一个非常精炼地道的表达。

这个句型的基本结构是：The real cause/problem/value of [某事物] + lies in + [名词/动名词短语]

讲解要点：

搭配: 它通常和 value、problem、secret、difficulty等抽象名词连用。

对比: 原文巧妙地运用了 “is unlikely to lie in X... Instead, ...Y...” 的结构，先否定一个普遍认知，然后引出真正的核心价值即在于解决问题，形成了鲜明对比，使论点更有说服力。

活学活用

这个句型在议论文写作中能帮助你更好地分析论证某一问题，同时也是你在汉英翻译中处理类似结构的一个非常精妙的表达方式。

The secret to happiness lies in finding contentment with what you have.

幸福的秘诀在于满足于你所拥有的一切。

The main difficulty of the task lies not in its complexity, but in the short deadline.

这项任务的主要困难不在于其复杂性，而在于截止日期太短。

---

05 思维导图

---

06 练习题答案

正确答案：(C)

解析：

(A) 选项不准确。文章提到Sora像TikTok，但明确指出其真正价值不太可能在于它们生成的那些快餐式内容，因此大众化视频创作只是其表面应用，而非核心价值。

(B) 选项是文中的一个事实，但不是作者的核心论点。作者确实提到了高昂的运营成本，但这主要是为了引出下文，即Sora的价值不在于这个亏钱的社交应用，而在于其底层技术。

(C) 选项准确概括了文章的核心思想。文章引用谷歌DeepMind的论文，详细阐述了视频模型如何它们未经专门训练的视觉和空间问题，并指出这才是它们的真正价值。

(D) 选项与文章观点相反。作者明确指出，将Sora的价值局限于生成“slop”是错误的，并希望这个阶段最终只被证明是一个有趣的注脚，而非其真正的价值所在。

你答（猜）对了吗？欢迎在评论区留言。

---

好不容易看到这了，点点♥点点赞吧

大家的点赞和♥就是我坚持更新的动力。

如对您有帮助，欢迎转发

如需转载到个人公众号，请先与作者取得联系