Name: 為什麼 Python 是 AI 的語言？ Python 之父 Guido van Rossum 深度解析！ (Why Python is the language of AI: insights from Guido van Rossum (creator of Python))
Uploaded: 2025-12-22T12:56:30.000Z
Duration: 8 min 7 s

過去完成式

In every stage of Python's development, I've always been surprised by how far it had come.

在 Python 的每個開發階段，我總是對它走到今天這一步感到驚訝。

情態副詞

So it's exponential growth, slowly but surely.

所以這是指數級的成長，緩慢但穩定。

轉承詞

I created Python because I needed a language that was easier to use than C.

我創造 Python 是因為我需要一種比 C 更容易使用的語言。

And in the context the only alternative was shell scripts, which was totally inadequate for what I wanted to do.

在當時的環境下，唯一的選擇是 shell scripts，這對於我想做的事情來說完全不夠用。

I wanted something that was much safer than C, that took care of memory allocation and of all sort of out of bounds indexing, but

我想要一種比 C 安全得多，能處理記憶體配置和各種越界索引，但

was still a sort of an actual programming language.

又仍然是一種真正的程式語言。

The first thing that happened was I showed my Python prototype when it was about three months old.

第一件事就是我展示了我的 Python 原型，那時它大約三個月大。

It spread throughout the institute where I worked.

它在我的研究機構裡傳播開來。

CWI 在阿姆斯特丹，它是一個研究實驗室。

Everybody has sort of the need for random little scripts to process some one off bit of data.

每個人都有處理一些單次數據片段的隨機小腳本的需求。

They gave me feedback, they started fixing bugs on their own.

他們給了我回饋，他們開始自己修復錯誤。

Within a year we realized we had something with potential value for many coders out there.

一年內我們意識到我們擁有的東西對許多編碼者來說具有潛在價值。

And we worked hard on the first open source release that was an immediate success.

我們努力進行了第一次開源發布，那是一個即時的成功。

所以儘管如此，這並不是我的主要工作。

I had a lot of fun talking to early users, implementing feature requests that people made, applying patches that people made.

我很享受與早期使用者交流、實施大家提出的功能請求、套用大家提交的補丁。

One of the first patches I think was affixed to the make file so it would run on some esoteric UNIX version.

我認為最早的補丁之一是附加到 make 檔案上，這樣它就可以在某個晦澀的 UNIX 版本上運行。

Soon enough there, there was like a whole community out there that was talking about Python.

很快地，那裡就形成了一個整個社群在談論 Python。

And because they were all so excited about Python, they thought it was such a cool tool, they were evangelizing.

因為他們都對 Python 如此興奮，認為它是一個很棒的工具，他們在傳教。

Not something I like to do, but also not something I discouraged anyone from doing.

這不是我喜歡做的事，但也不是我阻止任何人做的事。

When people said, oh, Python is so cool, in a little bit of fan mail I said thank you for and now tell all your friends.

當人們說，喔，Python 很酷，我會在一些粉絲信裡說謝謝，然後叫他們告訴所有朋友。

I was a big fan of Monty Python and I thought that the existing sort of patterns for naming languages like acronyms or famous people or even mythology were boring.

我是 Monty Python 的忠實粉絲，我認為現有的命名語言模式，像是縮寫、名人或神話，都很無聊。

I wanted to counteract that as a tiny act of rebellion with something from pop culture.

我想要用流行文化的東西來反擊，作為一個小小的反叛行為。

而且，Python 是個六個字母的單字。

It's very easy to remember because it has all these other connotations of snakes and comedy.

它很容易記住，因為它還有蛇和喜劇的聯想。

It's a place where people can get help with projects.

這是一個大家可以獲得專案協助的地方。

Or again, I mostly encourage the community to, to do more of that.

或者再次，我主要鼓勵社群多做這件事。

There's a general story that once a particular community, whether that's bioscience or databases or AI or machine learning, once a particular language has some use there and it sort of looks like it's a good solution.

有個普遍的說法是，一旦某個特定社群，無論是生物科學、資料庫、AI 或機器學習，一旦某個特定語言在那裡有了一些用途，而且看起來是個不錯的解決方案。

It reaches a point where people sort of, when they develop new pieces of software, whether that's libraries or applications, they will say, let's do that also in Python, because then we can leverage the stuff in Python that's already there.

它會達到一個點，當人們開發新的軟體時，無論是函式庫還是應用程式，他們都會說，我們也用 Python 來做吧，因為這樣我們就可以利用 Python 中已有的東西。

And the same, of course, works just as well in the Java world and the JavaScript or TypeScript world.

當然，這在 Java 世界、JavaScript 或 TypeScript 世界裡也同樣適用。

And through Python's good support for large arrays of floating point numbers of various sizes, that was sort of not built in, but

透過 Python 對各種大小的大型浮點數陣列的良好支援，這有點像是沒有內建，但

became a powerful and sort of well supported third party extension.

卻成為了一個強大且支援良好的第三方擴充功能。

So the people now writing things for AI are also familiar with Python because they started out in machine learning.

所以現在寫 AI 相關程式的人也熟悉 Python，因為他們一開始是從機器學習入門的。

Probably maintaining the stability and backwards compatibility are of utmost importance to the current core development community, including especially the Steering Council.

對於目前的核心開發社群，尤其是指導委員會來說，維護穩定性和向後相容性大概是最重要的。

For every new feature we have to very carefully consider is this sort of breaking existing code, which is something that makes it hard to add new keywords to the language.

對於每一個新功能，我們都必須非常仔細地考慮它是否會破壞現有的程式碼，這使得為語言添加新關鍵字變得困難。

And because that was so hard, we have invented a feature called soft keywords, which are keywords that are context sensitive.

因為這很困難，所以我們發明瞭一個叫做「軟關鍵字」的功能，它們是根據上下文敏感的關鍵字。

We have at least one complete new complex statement that uses two new context sensitive soft keywords.

我們至少有一個完整的新複雜語句，使用了兩個新的上下文敏感軟關鍵字。

There are other features under consideration that also use soft keywords.

還有其他一些正在考慮中的功能也使用了軟關鍵字。

And where it came from actually was that at some point I suddenly got this bug in my mind.

而它實際上是怎麼來的呢？有一次我腦子裡突然出現了一個錯誤。

I wanted to redesign the, the parser completely using very different principles, a different sort of way of building parsers that was much more flexible for the syntax of the language.

我想徹底重新設計解析器，使用非常不同的原則，一種更靈活地構建解析器的方式，來處理語言的語法。

And soft keywords more or less just rolled out of that.

而軟關鍵字差不多就是從那裡發展出來的。

Python has traditionally been server side.

And JavaScript of course, is almost purely client side.

當然，JavaScript 幾乎是純客戶端的。

TypeScript piggybacks on JavaScript by adding a type system that's very similar to Python's type system.

TypeScript 透過加入一個與 Python 的類型系統非常相似的類型系統，來搭上 JavaScript 的便車。

Those languages, I expect will continue to be popular because they are popular with the human users.

我預期這些程式語言會持續受歡迎，因為牠們深受人類使用者的喜愛。

I expect that AI will eventually be just as helpful when you're writing C or C or Java.

我預期 AI 在你撰寫 C 或 C++ 或 Java 時，最終也會同樣有幫助。

I had not anticipated Python to become so important for AI.

我沒預料到 Python 在 AI 領域會變得如此重要。

The role of Python in a world where AI is even more ubiquitous as it is today, you still need to write code and Python is still a fantastic tool to write code.

在 AI 比現在更加無所不在的世界裡，Python 的角色依然是，你還是需要寫程式碼，而 Python 仍是撰寫程式碼的絕佳工具。

And because Python is very human friendly, that almost automatically makes it also AI friendly, because the same brain cells that help us interpret human language and generate human language also help us understand programming languages.

而且因為 Python 對人類非常友善，這幾乎也讓它對 AI 友善，因為幫助我們理解人類語言和產生人類語言的腦細胞，也同樣幫助我們理解程式語言。

Because the languages have been designed to sort of put all kinds of things that are similar to human languages.

因為這些程式語言的設計，就是要把各種與人類語言相似的東西都放進去。

I mean, if and for are keywords that weren't chosen randomly.

我的意思是，if 和 for 這兩個關鍵字都不是隨便選的。

And so I expect that AIs will just learn to work with whatever language works well for humans.

所以我預期 AI 也會學著去處理對人類來說最有效的語言。