Fluentide

A Chinese podcast · Same story, 4 levels

LinkedIn's AI Slop Button Hits 1 Million Clicks — episode cover art
2:27

LinkedIn 加了一个"这看着像 AI 写的"按钮。三个星期,被点了一百万次。

LinkedIn's AI Slop Button Hits 1 Million Clicks

About this story

LinkedIn's AI Slop Button Hits 1 Million Clicks. HSK 2 Chinese listening practice.

This is an HSK 2 Chinese listening episode that runs about 2 minutes. The full Mandarin script is shown with tap-for-pinyin and a line-by-line English translation, so you can listen and read at once — comprehensible input in the sense of Stephen Krashen's i+1 theory. It teaches 7 key vocabulary words such as 读、字、觉得 and walks through 2 grammar patterns, each explained in English with examples. The same news story is retold at 4 difficulty levels — use the level selector above to find the version that is challenging but still understandable for you.

Read at your level

原文

Read the complete story in Chinese. Reveal pinyin and English only when you need them.

先说 LinkedIn 这个网站。
在 LinkedIn 上,大家写帖子:写自己的工作,写自己想到的事,写这一年做成了什么。
以前这些帖子都是人写的。
这几年,不一样了。
现在 LinkedIn 上有很多帖子,是 AI 写的,不是人写的。
你读的时候,可能看不出来,因为它看上去很正常。
AI 写的帖子,看上去也是一篇正常的帖子。
可是读完你会觉得,好像在哪儿见过。
句子都对,字也没错,可是没有味道。
七月三十号,LinkedIn 做了一件事。
他们在每一个帖子下面,加了一个按钮,很小的一个按钮。
按钮上写的是:这个看着像 AI 写的。
你看到一个帖子,觉得是机器写的,就点一下。
点一下,就是举报。
三个星期以后,一个数字出来了,很大的一个数字。
一百万。
一百万次。
三个星期里,这个按钮被点了一百多万次。
点了有用吗?
LinkedIn 说,有用。
他们把举报和自己的判断放在一起,改了系统。
结果是:被判定为 AI 写的内容,看的人少了四成。
少了四成,是很大的一个变化。
还有一个数字,更让人吃惊。
一家叫 Pangram Labs 的公司,看了很多 LinkedIn 上长一点的帖子。
一百篇里,有四十一篇,是完全由 AI 生成的。
不是人写完 AI 改一改,是从头到尾都是 AI 写的。
所以那个按钮,不只是一个按钮。
一百万个人点它,是在说同一句话:
这样的东西,我不想再看到了。
有意思的是,没有人教过大家,怎么分辨。
没有课,也没有人说过。
大家就是读着读着,觉得不对。
你想一想:
你读一篇文章,能看出来是不是机器写的吗?
English transcript reference

First, about the website LinkedIn.

On LinkedIn, people write posts: about their work, about things they have thought of, about what they got done this year.

In the past these posts were all written by people.

These last few years, that has changed.

Now there are a lot of posts on LinkedIn written by AI, not by people.

When you read one you may not be able to tell, because it looks perfectly normal.

A post written by AI looks like a normal post too.

But when you finish it you feel as though you have seen it somewhere before.

The sentences are fine, the words are not wrong, and yet it has no flavour.

On the thirtieth of July, LinkedIn did something.

They added a button under every post — a very small button.

What it says is: this seems like AI wrote it.

You see a post, you think a machine wrote it, and you press it.

Pressing it is reporting it.

Three weeks later a number came out — a very big number.

One million.

One million times.

In three weeks, that button was pressed more than a million times.

Was pressing it any use?

LinkedIn says it was.

They put the reports together with their own judgement and changed the system.

The result: for content judged to be AI-written, forty percent fewer people saw it.

Forty percent fewer is a very large change.

And there is another number that is more surprising still.

A company called Pangram Labs looked at a lot of the longer posts on LinkedIn.

Out of a hundred, forty-one were entirely generated by AI.

Not a person writing and AI tidying it up — AI from beginning to end.

So that button is not just a button.

A million people pressing it are all saying the same sentence:

I do not want to see this kind of thing any more.

And the interesting part is that nobody taught anyone how to tell.

There was no class, and nobody explained it.

People simply read, and read, and felt something was off.

Think about it:

You read an article — can you tell whether a machine wrote it?

Listen again

Try it without the transcript and notice what sounds clearer.

What vocabulary does this episode teach?

词汇
dúto read

你读的时候,可能看不出来。

zìa written character, a word

句子都对,字也没错 — the sentences are fine, the words are not wrong.

juédeto feel, to think

The whole mechanism is people 觉得 something is off.

xīngqīweek

三个星期里,被点了一百多万次。

ànniǔa button

A very small one, added under every post.

tiězia post (online)

The standard word for a post on a forum or social platform.

jǔbàoto report (something objectionable)

点一下,就是举报 — pressing it is reporting it.

* beyond level超纲词

What grammar patterns appear in this episode?

语法

看不出来

Cannot tell by looking. The 出来 complement means "work it out".

你读的时候,可能看不出来。 — Reading it, you may not be able to tell.

不是……,是……

Not X, but Y — corrects the expected description.

不是人写完 AI 改一改,是从头到尾都是 AI 写的。

Proper nouns

专有名词
LinkedInLinkedInLinkedIn, the professional networking siteAIAIartificial intelligencePangram LabsPangram Labsthe AI-detection company that sampled the posts

Sources

来源

Free account

Keep learning from this story

Create a free account to keep saved words and your preferred level together.

  • Keep words with their story context
  • Remember your preferred level
  • Build your vocabulary over time

Keep learning