With AI, Anyone Can Be a Coder Now | Thomas Dohmke | TED

翻译人员: Yip Yan Yeung 校对人员: Gentle Yang

You know, I'm one of these adults

我是个依然喜欢玩乐高的成年人。

that actually still loves playing with LEGO.

I loved them way back in the '80s in Berlin when I grew up,

80 年代，我在柏林 长大的时候就爱玩乐高，

and I still love them.

现在仍然喜欢玩。

And these days, I build LEGO with my kids on Saturday afternoons.

如今，我会在星期六下午 和孩子们一起搭乐高。

And the reason that my love for LEGO

我对乐高的热爱 之所以长青，很简单，

has remained evergreen is, quite simply,

that LEGO is a system for realizing creativity

是因为乐高是一个几乎没有门槛的 实现创造力的系统。

with almost no barrier to entry.

And I’m not only a LEGO dad,

我不仅是乐高迷爸爸，

I'm also the CEO of GitHub.

还是 GitHub 的首席执行官。

And if you don't know GitHub,

如果你不了解 GitHub，

you can think of it as the home of coding.

你可以把它看作是编程之家。

It's where all the software developers,

这是所有软件开发人员，

the chief nerds of our society,

即我们社会中的首席书呆子们，

collaborate together.

共同合作的地方。

And it's part of our mission to make it as easy as possible

我们的一部分使命 就是让每一位开发人员

for every developer to build small and big ideas with code.

都能尽可能轻松地用代码 实现大大小小的想法。

But in contrast to LEGO,

但是与乐高形成鲜明对比的是，

the process of building software feels daunting to most people.

开发软件的过程 对大多数人来说是望而却步的。

This all started to change

2022 年末，ChatGPT 的出现 改变了一切。

when ChatGPT came along in late 2022.

Now we live in a world where intelligent machines understand us

我们现在生活在一个 智能机器对我们的理解

as much as we understand them.

与我们对它们的理解相当的世界中。

All because of language.

这一切全都是因为语言。

And this will forever change the way we create software.

这将永远改变我们开发软件的方式。

Up until now, in order to create software,

截至目前，要开发软件，

you had to be a professional software developer.

你必须是一名专业的软件开发人员。

You had to understand, speak and interpret the highly complex,

你必须能理解、能表达、能解释

sometimes nonsensical language of a machine that we call code.

高度复杂又时常无意义的机器语言， 我们称之为“代码”。

Modern code still looks like hieroglyphics to most people.

对于大多数人来说， 如今的代码仍然像象形文字。

Here's an example.

举个例子：

This, from the early 1940s,

图上的代码来自 19 世纪 40 年代早期，

is the world's first computer programming language,

是世界上第一种计算机编程语言，

called Plankalkül.

叫作 “Plankalkül”。

It set the foundation for the modern code that we use today.

它为我们今天使用的 现代代码奠定了基础。

And as you can see, it's a few numbers,

如你所见，有一些数字、

some bubbles and some big-ass brackets.

一些气泡和一些巨型大括号。

Not much humanity here, right?

没什么“人性化”的元素，对吧？

Flash forward about 20 years

快进 20 年， 到名为 “COBOL” 的编程语言。

to the programming language called COBOL.

COBOL was invented during the Eisenhower years,

COBOL 是在艾森豪威尔时代发明的，

but it remains an important language

但它仍然是我们许多 头部金融机构使用的重要语言。

for many of our largest financial institutions.

Wall Street, your savings account, your credit cards,

华尔街、你的储蓄账户、 你的信用卡，

all run on this today.

现在都靠这种语言运行。

And we see some familiar words here.

我们看到了一些熟悉的词语。

But structurally, I think this doesn't make much sense to most of you.

但是从结构上讲， 我认为很多人都看不太懂。

Flash forward another 30 years to 1991,

再快进 30 年，到了 1991 年，

and we saw the birth of Python,

我们见证了 Python 的诞生，

one of the most popular programming languages in this era of AI.

它是如今的 AI 时代 最受欢迎的编程语言之一。

In 80 years, we went from bubbles to brackets,

80 年来，我们从气泡走到了括号，

to blips of English,

再到零星的英语，

and yet, we got nowhere near as close as the intuitiveness of human language.

但是远不及 人类语言的通俗易懂。

But then came June 2020,

但是随后到了 2020 年 6 月，

and we got early access to OpenAI's large language model,

我们抢先体验了 OpenAI 的大语言模型，

then called GPT-3.

即后来的 GPT-3。

It was COVID, we were all on lockdown,

当时是疫情时期，我们都被封控在家，

I remember we were on a video call together.

我记得我们当时一起开了个视频会议。

We fed random programming exercises into this raw model,

我们向这个原始模型 输入了随机的编程练习，

and like magic,

就像魔法一样，

it solved 93 percent of them during the first few takes.

它在前几次尝试中 解答了 93% 的练习。

We at GitHub recognized we had something remarkable in our hands,

在 GitHub 的我们意识到 我们手上有一些不得了的东西，

and we quickly turned around a novel developer tool

于是我们迅速开发了一款

名为 GitHub Copilot 的新型开发者工具：

called GitHub Copilot:

an AI assistant that predicts and completes code

一款为软件开发者 预测并完成代码的 AI 助手。

for software developers.

Copilot is now the most adopted AI developer tool on the planet.

Copilot 如今是地球上 使用最广泛的 AI 开发者工具。

The age of programming has been reborn.

编程时代已经重生。

But the possibilities of the breakthrough

但是突破的可能性带来的 远不止它们的商业成果。

went further than just these business results.

Because the large language models that power ChatGPT and Copilot

由于驱动 ChatGPT 和 Copilot 的大语言模型

are trained on a vast library of human information,

是由海量的人类信息库训练而来的，

they understand and interpret nearly every human language,

它们几乎可以理解和解释所有人类语言，

every major human language.

所有主要的人类语言。

They seem to get us.

它们似乎可以理解我们。

We have struck a new fusion

我们实现了人类语言 和机器语言之间的新融合。

between the language of a human and a machine.

With Copilot, any person can now build software in any human language

有了 Copilot，现在人人 都可以用任意人类语言

with a single written prompt.

编写提示词来开发软件。

Goodbye to the bubbles and the big-ass bracket.

再见了气泡， 再见了巨型大括号。

This is the most profound breakthrough to technology

这是自软件开发起源以来 技术领域最重大的突破。

since the genesis of software development itself.

Today, there are over 100 million developers on GitHub.

如今，GitHub 上 有超过 1 亿开发者。

That's about one percent of the world's population,

差不多是世界人口的 1% 左右。

you know, plus-minus.

I think that number is about to explode.

我认为这个数字即将爆炸式增长。

And I want to show you why, here on my MacBook.

我想告诉你为什么， 就用我的 Macbook。

We started it all with the original Copilot or how we say the OG Copilot,

一切从初始的 Copliot， 或者我们口中的 OG Copilot 开始，

and it literally just predicted and completed code in the editor.

它实际上就是在代码编辑器中 预测、完成代码。

You can think of the editor as, you know, the Google Docs for developers.

你可以把编辑器想象成 开发人员的谷歌文档。

And when you have a doc open, you know how it is, empty page,

你打开了一个文档， 你知道是什么样，就是空白页，

what do I actually want to do?

我想干什么呢？

And I mentioned LEGO.

我提到了乐高。

So let’s build a 3D LEGO brick on a web page.

那我们就在网页上 做一个 3D 的乐高积木吧。

So what developers do, you know, they start typing.

开发人员会开始打字。

And so I typed in the JavaScript file,

我在 JavaScript 文件中开始打字，

create a function to create a LEGO brick.

创建了一个构建乐高积木的函数。

And you can see here this gray text, we call this ghost text.

你可以看到这个灰色文本， 我们称之为“幽灵文本”。

This is coming from the large language model.

这是大语言模型生成的。

So now I can just press the tab key and press enter.

我可以按下 Tab 键， 再按下回车。

And I get another suggestion, you know, to create a LEGO tower.

我收到了另一条建议： 搭建一座乐高塔。

Maybe we do that later.

也许我们稍后再做。

Or I can just do: function draw LEGO brick.

或者我就这么输入： function drawLegoBrick。

And here again you see ghost text from Copilot right away available for me.

你可以再次看见 Copilot 提供的 幽灵文本立即出现在我眼前。

And if I like what I'm seeing here,

如果我喜欢看到的内容，

so I get into a mode of writing and understanding,

那么我就会进入编写和理解的模式，

I can just accept this.

我只要接受就行了。

Developers love that, right?

开发人员喜欢这样，对吧？

Because instead of writing ten lines of code themselves

因为比起自己编写十行代码，

or copy and pasting them from the internet,

或从网上复制粘贴，

they get them right in their editor.

他们可以直接在编辑器里得到代码。

They can stay in the flow.

他们可以保持思路。

Now what the OG Copilot didn’t offer me is a way to interact with this.

OG Copilot 没有给我提供的是 与它互动的方式。

I cannot ask questions,

我不能问问题，

I cannot, you know, instruct it to do different things.

我不能指挥它做各种事。

Last year we launched a new feature, Copilot chat,

去年我们推出了一个新功能， Copilot Chat，

and you can think about it as ChatGPT in your editor.

你可以把它想象成 编辑器中的 ChatGPT。

So I can open this up here in the sidebar.

我可以在侧边栏打开它。

And now I can tell it to create a whole web page

我可以让它为我创建一整个网页， 上面放一个 3D 的乐高积木。

with a 3D LEGO brick for me.

Now you know, similar to ChatGPT, it streams the response,

类似于 ChatGPT， 它会采用流式回答，

and it gives me not only some code

它不仅会给我一些代码，

but it actually gives me an explanation.

还会给我一个解释。

You know, it starts writing code,

它开始编写代码，

you can see the comments that explain what that code does.

你还能看到注释 解释这行代码的作用。

It uses an open-source library called Three.js.

它用了一个 叫做 Three.js 的开源库。

And so you can kind of see here the idea of this empowering developers

你可以从中看到增强开发者

and people that want to learn development.

及想要学习软件开发的人的能力的想法。

And it ends, you know, with another explanation.

最后它又给出了一个解释。

Now I can go here, inspect that code,

我可以找到这个位置， 检查这段代码，

and I can actually push that button to copy it into my file.

按下这个按钮， 复制进我的文件。

But I want to show you something else here.

但我想展示一些别的东西。

And you might have already seen this little mic icon.

你可能已经看到了 这个小小的麦克风图标。

I can use that to speak to Copilot.

我可以用它来和 Copilot 说话。

And I want to ask it, in German, what that code does

我想用德语问这段代码是干什么用的，

that is on the left side in the editor.

就是编辑器左边的这段。

(Speaking German) Can you explain to me what that code does?

（德语）你能解释一下 这段代码是干什么的吗？

And now Copilot responds again,

Copilot 又回复了，

but it responds in German to me, right?

但它是用德语回复我的，对吧？

So it says, if I loosely translate,

如果我粗略地翻译一下，它在说：

"Yes, of course, this JavaScript code defines a function

“当然可以，这段 JavaScript 代码定义了一个函数，

named ‘drawLEGOBrick.’”

叫做 ‘drawLegoBrick’。”

So you get the idea here.

你可以理解这个意思。

A six-year-old in Berlin, in Mumbai and Rio,

柏林、孟买和里约的六岁孩子

can now explore coding without their parents being around

现在可以在父母不在身边， 甚至没有技术背景的情况下探索编程。

or even having a technical background.

(Laughter)

（笑声）

I mean, you know.

我的意思是，你懂得。

(Applause)

（掌声）

Now what you also see is you still need to kind of figure out

你还可以看到， 你依然需要指出

how you put that all together, right?

如何把它们组合在一起，对吧？

There’s a lot of technical stuff here.

有很多技术方面的东西。

I have code. I have to iterate on my machine.

我有了代码， 我得在我的机器上进行迭代。

I have to figure out how to deploy this to the cloud

我必须弄清楚 如何将其部署到云端，

so I can share with my friends.

这样我才能与朋友共享。

But here is my LEGO brick now.

这就是我的乐高积木。

This is what it looks like

如果我作为开发者完成了这些步骤， 就会得到这样的结果，

if I've done all these steps as a developer,

you can see now it’s a nicely rotating brick.

可以看到这块旋转的漂亮积木。

I can actually use my mouse to turn it around.

我可以用鼠标转动它。

These are the anti-studs here, the studs,

这些是接合孔和凸起，

There's nice lighting effects.

光照效果很不错。

I can even zoom into this and zoom out of this.

我还可以放大缩小。

Now I don't want to do all this developer stuff anymore.

我再也不想做这些开发者的事了，

I just want to channel my creativity straight into reality.

我只想将创造力直接转化为现实。

And so for the first time ever on stage,

我将向你们展示 一个新产品的公开首秀，

I'm going to show you a new product that we call Copilot Workspace

它叫 Copilot Workspace，

that does exactly that.

做的就是这件事。

So here is my workspace.

这是我的工作区。

And you can already see there's not an editor anymore.

可以看到已经没有编辑器了。

I can just see a task, and I can enter a task.

我只看到一个任务， 然后输入一个任务。

And so now I have my LEGO brick,

我现在已经有了一块乐高积木，

I want to now expand the LEGO brick into a LEGO house.

想把乐高积木扩展成一座乐高屋。

Stack the bricks in the shape of a LEGO house.

“把乐高积木拼成 一个乐高屋的形状。”

And I can do that also in German and in other languages.

我也可以用德语 和其他语言做到这一点。

But for now, let's stick with English.

但是我们现在还是用英语吧。

I can save my task.

我可以保存我的任务。

And now what happens is that Copilot Workspace analyzes what I already have

现在 Copilot Workspace 会分析我已经拥有的东西，

and then describes what it proposes to me.

然后描述它向我提出的建议。

Basically, it reframes my ask into a plan or a specification.

它其实将我的请求 重新制定成了一份计划或规范。

And so you can see here, you know, it's all in natural language in our user.

你可以看到它都是 用自然语言输出给用户的。

Some file names, of course, but there is no code here.

当然，有些文件名， 但是这里没有代码。

It's all describing it in English.

都是用英语描述的。

I can actually go into this and edit it

我可以点进这一句，编辑，

and can make changes to this line,

修改这一句，

or I can go down here and add another item

我也可以点击下面，添加一条，

if I feel like the plan is not exactly what I want.

如果我觉得这个计划 不符合我想要的。

I can go a step further and generate a plan,

我可以再进一步， 生成一个计划，

and now an agent runs through all my files I already have

智能体会遍历我的所有文件，

and figures out how do I need to modify those files,

弄清楚我该如何修改这些文件，

or, you know, do I need to add files to my repository

或者我是否需要 将这些文件进入我的代码库，

so you know it wants to add a “createLEGOHouse” function

这样它就可以添加一个 “createLegoHouse” 的函数，

and call the “createLEGOHouse” afterwards.

之后再调用 “createLegoHouse” 函数。

Looks good to me, so let's implement this.

我觉得不错， 那我们来对这些文件进行操作。

And now Copilot uses my task, my specification,

现在 Copilot 使用 我的任务、 我的规范、

my plan to write code for me.

我的计划来为我编写代码。

You can see here two files are queued,

你可以看到这里有两个文件在排队，

the public/legoBrick.js file

public/legoBrick.js 文件，

and boom, there's already my code written for me, right?

然后一瞬间， 我的代码已经为我写好了，对吧？

I didn't have to touch code,

我不必接触代码，

I didn't have to even know what code is.

甚至不需要知道代码是什么。

Now I see here now it imports some new line into my file

我可以看到它向我的文件里 导入了几句新代码，

and has written, you know, lots of code here that does those changes.

还写了这里的很多代码来实现这些修改。

So you want to see what that looks like, did we get a LEGO house?

你们想看看结果如何吗？ 我们得到一座乐高屋了吗？

So here's a button that lets me open a live preview,

这里有一个按钮， 我可以打开一个实时预览，

so I can do this.

我可以这么做。

And now the bricks fall from the sky and I have a LEGO house.

积木从天而降， 我就得到了一座乐高屋。

And you know, this is not a picture, right --

这可不是一张图片——

(Applause)

（掌声）

Yes, thank you.

没错，谢谢。

This is all live, this is the power of code,

这都是实时的， 这就是代码的力量，

this is the power of streaming my creativity into reality

这是用自然语言 将我的创造力变成现实的力量。

with natural language.

Now one last thing.

现在还有最后一件事。

Thank you, Copilot,

谢谢你，Copilot，

you have always to be nice to the AI.

你得一直友好地对待 AI。

(Laughter)

（笑声）

(Applause)

（掌声）

Now, what you just saw were three leaps in three years.

你刚刚看到的是 三年内的三次飞跃。

Three leaps that are more progress

这三次飞跃比过去 100 年内 我们在计算机编程的无障碍方面

to the accessibility of computer programming

than we have made in the last 100.

取得的进步还要大。

Remember how I said that one percent of the world's population is a developer?

还记得我说过 全球 1% 的人口是开发者吗？

Now you can see how this will change.

你会看到这将发生怎样的变化。

Copilot Workspace may still be a developer tool right now,

Copilot Workspace 目前可能 仍然是开发者工具，

but soon enough these kind of developer tools will become mainstream.

但很快这类开发者工具将成为主流。

Because, going forward, every person, no matter what language they speak,

因为，展望未来，每个人， 无论他们说什么语言，

will also have the power to speak machine.

都会有能力与机器“说话”。

Any human language is now the only skill

开始计算机编程所需的唯一技能， 就是任意的人类语言。

that you need to start computer programming.

This will lead to a globalized groundswell of software developers,

这将带来软件开发人员的全球性激增，

and it will reshape the geography of our global economy.

并将重塑全球经济的地理格局。

And because of this,

因此，

I think by 2030, maybe even sooner,

我认为到 2030 年，也许更早，

we will have more than one billion software developers on GitHub.

我们将在 GitHub 上 迎来超过 10 亿的软件开发人员。

Think about that:

想像一下：

10 percent of the world’s population will not only control a computer

全球 10% 的人口 不仅能控制计算机，

but will also be able to create software

而且还能像骑自行车一样开发软件。

just [as] if they were riding a bicycle.

This will generate a new renaissance of human creativity with software.

这将通过软件 开创人类创造力的新复兴。

Now anyone here in this room could have a brilliant idea right now:

在座的任何人 现在都有了绝妙的主意：

a website, an application,

一个网站，一个应用程序，

a cool computer game, an amazing song,

一个很酷的电脑游戏，一首很棒的歌曲，

maybe even a cure for something.

甚至可能是治愈什么的方法。

For example, last year, over a couple of weeks,

比如去年，在几周的时间里，

I built an app that tracks all the flights I've ever taken in my life.

我开发了一个应用程序，可以追踪 我一生中乘坐过的所有航班。

Now I know what you're thinking.

我知道你在想什么。

What a freaking nerd, right?

真是个该死的书呆子，对吧？

And yeah, it's true, I love building stuff like this.

是的，没错， 我喜欢创造这样的东西。

And with the help of AI,

在 AI 的帮助下，

now I can do this in English or in German

现在我甚至可以在喝完一杯酒之前 用英语或德语完成这件事。

before I even finish a glass of wine.

And soon enough, this will be true for everyone here.

很快，人人都会有这样的体验。

The floodgates of nerditude have swung wide open.

书呆子的闸门已经敞开了。

(Laughter)

（笑声）

(Applause)

（掌声）

Now this doesn’t mean

这并不意味着

that everyone will become a professional software developer

每个人都会成为 专业的软件开发人员，

or even that they should.

或者每个人都应该成为专业开发者。

The profession of a professional software developer

专业软件开发人员的专业性不会丧失。

is not going anywhere.

There will always be demand for those that design and maintain

永远都有设计、维护 世界上最大的软件系统的需求。

the largest software systems in the world.

We are adding millions of lines of code every single day

我们每天都在向越来越复杂的系统中 添加成千上万行代码，

to ever more complex systems,

and we are barely keeping up with maintaining the existing ones.

但我们几乎无法 及时维护现有的代码。

Like any infrastructure in this world out there,

如同世界上存在的任何基础设施，

we need real experts to preserve and renew it.

我们需要真正的专家保护、更新它。

The point here is not a "will" or a "should."

关键不是“将要”还是“应该”，

It's that anyone can.

而是人人都“可以”。

All because the most powerful system that we have,

一切都是因为我们拥有的 最强大的系统，

any human language,

任何人类语言，

is now fused to the language of a machine.

现在都融入了机器的语言。

And very soon,

很快，

building software will be just as simple and joyful

开发软件就会像 搭乐高一样简单而愉快。

as stacking a LEGO.

(Speaking German) Thank you very much.

（德语）非常感谢。

(Applause)

（掌声）

Bilawal Sidhu: Gosh, I've got to say, one billion developers

拉瓦尔·西杜（Bilawal Sidhu）： 天哪，我不得不说，十亿开发者，

makes GitHub sound more like YouTube and TikTok than it is today.

让 GitHub 听起来比现在 更像 YouTube 和抖音了。

Just super exciting.

太令人激动了。

Got to ask you one question,

我想问你一个问题，

perhaps the elephant in the room.

可能是个大家心知肚明的问题。

Amazing talk.

精彩的演讲。

So you said the developer is still in charge.

你说主导权仍在开发者手中，

You also said, "We've had three leaps in three years."

你还说： “我们在三年内取得了三次飞跃。”

Fast forwarding a little bit,

稍微快进一点，

do you think humans will still need to be in the loop,

你认为人类依然需要参与进来，

or will these AI systems be able to autonomously build

还是这些 AI 系统 能够自主开发和维护软件？

and maintain software?

TD: You know, the way I always think and talk about it

TD：我一直在思考和谈论这个问题，

is that we called it Copilot for a reason.

我们称之为 Copilot （意为“副驾驶”）是有原因的。

We need a pilot.

我们需要驾驶员。

We need a pilot that is creative, that can decide what to do.

我们需要有创意的驾驶员， 决定要做些什么。

It’s kind of like a LEGO set.

它就像乐高套装。

You need to take this big problem and break it down into smaller problems,

你得把这个大问题 分解成小问题，

into small building blocks.

分解成小块。

And for that, you need a systems thinker.

为此，你需要系统的思考者。

You need a human that can figure out, am I building a point of sale system?

你需要一个人类来弄清楚： 我是要开发一个销售终端系统吗？

Am I building an iPhone app?

我是要开发一个 iPhone 应用吗？

Am I building a cool computer game?

我是要开发一款很酷的电脑游戏吗？

Am I building the next Facebook?

我是要打造下一个 Facebook 吗？

Those are very different systems.

这些都是截然不同的系统。

Now these building blocks, they will grow in size.

这些模块会越来越大。

Today it's, you know, a couple of lines of code,

现在的几行代码，

maybe a whole file,

也许是整个文件，

in the future, it might be a whole subsystem.

未来可能会是整个子系统。

So I get more work taken off my shoulders.

这样我要承担的工作就越来越少了。

But I'm still there, you know, covering the large system.

不过我依然会负责这个大型系统。

And as I mentioned, you know, we're still running COBOL systems from the '60s.

如我所说，我们仍在运行 60 年代的 COBOL 系统。

So we have lots of work to do.

所以我们还有很多工作要做。

BS: Absolutely.

BS：当然。

So we will be in charge orchestrating these systems

我们会负责在更高的抽象层次上 协调这些系统。

at a higher level of abstraction.

Thomas Dohmke, everybody, thank you.

谢谢托马斯·多姆克 （Thomas Dohmke），谢谢大家。

TD: Thank you so much.

TD：非常感谢。