Name: 為什麼大家開始像 ChatGPT 一樣說話？ | Adam Aleksic | TED (Why Are People Starting to Sound Like ChatGPT? | Adam Aleksic | TED)
Uploaded: 2025-12-31T13:49:31.000Z
Duration: 5 min 12 s

How sure are you that you can tell what's real online?

你有多確定自己能分辨網路上的真實內容？

You might think it's easy to spot an obviously AI generated image, and

你可能會覺得要辨識出明顯是 AI 生成圖像很容易，而且

you're probably aware that algorithms are biased in some way.

你大概也知道演算法在某些方面是有偏見的。

But all the evidence is suggesting that we're pretty bad at understanding that on a subconscious level.

但所有證據都顯示，我們在潛意識層面其實很不擅長理解這點。

Take, for example, the growing perception gap in America.

舉個例子，美國日益擴大的知覺差距。

We keep over and overestimating how extreme other people's political beliefs are.

我們不斷地高估別人政治信仰的極端程度。

And this is only getting worse with social media, because algorithms show us the most extreme picture of reality.

而這點隨著社交媒體只會越來越糟，因為演算法會給我們看最極端的現實圖像。

As an etymologist and content creator, I always see controversial messages go more viral because they generate more engagement than a neutral perspective.

作為一位詞源學家和內容創作者，我總是看到爭議性訊息更容易病毒式傳播，因為它們產生的互動比中立觀點還多。

But that means we all end up seeing this more extreme version of reality, and

但這也意味著我們最終都看到這個更極端的現實版本，而且

we're clearly starting to confuse that with actual reality.

我們顯然開始把它和真實的現實混淆了。

The same thing is currently happening with AI chatbots, because you probably assume that ChatGPT is speaking English to you a except it's not speaking English in the same way that the algorithm's not showing you reality.

同樣的事情現在也發生在 AI 哈拉機器人身上，因為你可能認為 ChatGPT 跟你說英文，但它說英文的方式跟演算法呈現現實的方式不一樣。

There are always distortions depending on what goes into the model and how it's trained.

模型裡面的東西和它的訓練方式，總是會造成一些扭曲。

Like, we know that ChatGPT says Delve at way higher rates than usual, possibly because OpenAI outsourced its training process to workers in Nigeria who do actually say delve more frequently.

像是，我們知道 ChatGPT 說 "Delve" 的頻率比平常高很多，可能是因為 OpenAI 把訓練過程外包給奈及利亞的員工，他們確實比較常說 "delve"。

Over time, though, that little linguistic overrepresentation got reinforced into the model even more than in the workers own dialects.

但隨著時間過去，這種小小的語言過度代表性，在模型裡被強化得比員工自己的方言還要嚴重。

Now that's affecting everybody's language.

現在這已經影響到每個人的語言了。

Multiple studies have found that since ChatGPT came out, people everywhere have been saying the word delve more in spontaneous spoken conversation.

多項研究發現，自從 ChatGPT 出現以來，各地的人們在隨意的口語對話中，說 "delve" 這個詞的次數都變多了。

Essentially, we're subconsciously confusing the AI version of language with actual language.

基本上，我們在潛意識層面把 AI 版本的語言和真實的語言搞混了。

But that means that the real thing is, ironically getting closer to the machine version of the thing.

但諷刺的是，這意味著真實的事物反而越來越接近機器的版本。

We're in a positive feedback loop with the AI representing reality, us thinking that's the real reality, and

我們正處於一個正向的循環，AI 代表著現實，我們認為那就是真實的現實，然後

then regurgitating it so that the AI can be fed more of our data.

再把它們吐回去，這樣 AI 就能接收到更多我們的數據。

You can also see this happening with the algorithm through words like hyper pop, which wasn't really part of our cultural lexicon until Spotify noticed an emerging cluster of similar users in their algorithm.

你也可以看到演算法透過像 hyper pop 這樣的詞語發生同樣的現象，這個詞在 Spotify 發現演算法中出現了一個新興的相似用戶群集之前，其實並不是我們文化詞彙的一部分。

As soon as they identified it and introduced a hyper pop playlist, however, the aesthetic was given a direction.

然而，一旦他們識別出來並推出了 hyper pop 播放清單，這種美學就有了方向。

Now people began to debate what did and did not qualify as hyper pop.

現在人們開始爭論什麼才符合 hyper pop 的資格。

The label and the playlist made the phenomenon more real by giving them something to identify with or against.

這個標籤和播放清單透過提供一個可以認同或反對的對象，讓這個現象變得更真實。

And as more people identified with hyper pop, more musicians also started making hyper pop music.

隨著越來越多人認同 hyper pop，越來越多音樂人也開始製作 hyper pop 音樂。

All the while, the cluster of similar listeners and the algorithm grew larger and larger, and

同時，相似聽眾的群集和演算法也越來越大，

Spotify kept pushing it more and more because these platforms want to amplify cultural trends to keep you on the app.

而 Spotify 也持續推廣，因為這些平臺想要放大文化趨勢以留住你。

But that means we also lose the distinction between a real trend and an artificially inflated trend.

但這也意味著我們失去了真實趨勢和人為膨脹趨勢之間的區別。

And yet, this is how all fads now enter the mainstream.

然而，這就是所有流行趨勢進入主流的方式。

We start with a latent cultural desire, like maybe some people are interested in Matcha or Labubu or Dubai Chocolate.

我們從一個潛在的文化渴望開始，像是有些人對抹茶、Labubu 或杜拜巧克力感興趣。

The algorithm identifies this desire and pushes it to similar users, making the phenomenon more of a thing.

演算法會識別這個渴望，並將其推送給相似的使用者，讓這種現象更加明顯。

But again, just like how ChatGPT misrepresented the word delve, the algorithm is probably misrepresenting reality.

但話說回來，就像 ChatGPT 誤解了「delve」這個詞一樣，演算法很可能也誤解了現實。

Now more businesses are making Labubu content because they think that's the desire.

現在越來越多企業製作 Labubu 相關內容，因為他們認為這就是大家想要的。

More influencers are also making labubu trends because we have to tap into trends to go viral.

越來越多的網紅也在製作 Labubu 的潮流內容，因為我們必須跟上潮流才能爆紅。

And yet the algorithm is only showing you the visually provocative items that work in the video format.

然而，演算法只會顯示那些在影片格式中效果好的、視覺上引人注目的內容。

TikTok has a limited idea of who you are as a user, and

TikTok 對你的使用者身份的理解是有限的，

there's no way that matches up with your complex desires as a human being.

而且這根本無法符合你作為一個人複雜的渴望。

所以我們得到的是一個有偏見的輸入。

And that's assuming that social media is trying to faithfully represent reality, which it isn't.

而且這還是在假設社群媒體試圖忠實呈現現實，但事實並非如此。

Instead, it's only trying to do what's going to make money for them.

相反地，它只會試圖做那些能為他們賺錢的事情。

It's in Spotify's interest to have you listening to Hyper Pop, and

it's in TikTok's interest to have you looking at the Boo Boos, because that's commodifiable.

TikTok 希望你關注那些可愛的影片，因為這些東西才能變現。

So once again, we have this difference between reality and the representation of reality, where they're actually constantly influencing one another.

所以我們又看到了現實和現實的呈現之間存在差異，而且兩者其實一直在互相影響。

But it's incredibly dangerous to ignore that distinction, because this goes beyond our language and our consumptive behaviors.

但忽略這種區別是極其危險的，因為這影響的不僅是我們的語言和消費行為。

This affects the world we see as possible.

這會影響我們所認為可能的世界。

Evidence suggests that ChatGPT is more conservative when speaking the Farsi language, likely because the limited training texts in Iran reflect the more conservative political climate in the region.

有證據顯示 ChatGPT 在說波斯語時更為保守，這很可能是因為伊朗的訓練文本有限，反映了該地區更保守的政治氣候。

Does that mean that an Iranian ChatGPT user will think more conservative thoughts?

這是否意味著伊朗的 ChatGPT 使用者也會產生更保守的想法？

We know that Elon Musk regularly makes changes to his chatbot Grok when he doesn't like how it's responding, and

我們知道 Elon Musk 會在他不喜歡他的哈拉機器人 Grok 的回應方式時，經常對它進行修改，

that he uses his platform X to artificially amplify his tweets.

而且他會利用他的平臺 X 來人為地放大他的推文。

Does that mean that the millions of Grok and X users are subconsciously being trained to align with Musk's ideology?

這是否意味著數百萬的 Grok 和 X 使用者在潛意識層面被訓練成與 Musk 的意識形態一致？

We need to constantly remember that these aren't neutral tools.

我們要不斷記住，這些並不是中立的工具。

Everything that ends up in your social media feed or in your chatbot responses is actually filtered through many layers of of what's good for the platform, what makes money, and

所有出現在你社交媒體動態消息裡或哈拉機器人回應中的內容，實際上都經過了對平臺有利、能賺錢，以及

what conforms to the platform's incorrect idea about who you are.

符合平臺對你個人錯誤認知的層層過濾。

When we ignore this, we view reality through a constant survivorship bias, which affects our understanding of the world.

當我們忽略這一點，我們就會透過持續的倖存者偏差來看待現實，這會影響我們對世界的理解。

After all, if you're talking more like ChatGPT, you're probably thinking more like ChatGPT as well.

畢竟，如果你說話越來越像 ChatGPT，你可能也會越來越像 ChatGPT 一樣思考。

But you can fight this if you constantly ask yourself, why?

但如果你不斷問自己「為什麼？」，你就能對抗這種情況。

If you don't ask yourself these questions, their version of reality is going to become your version of reality, so stay real.

如果你不問自己這些問題，他們的現實就會變成你的現實，所以要保持真實。