Gaming as the Future of Education with Duolingo CEO Luis von Ahn

大家好，欢迎回到《无偏见》节目。

Hi, listeners, and welcome back to No Priors.

今天，我们邀请到了路易斯·冯·安。

Today, we're joined by Luis von Ahn.

路易斯在卡内基梅隆大学获得了计算机科学博士学位，之后创立了reCAPTCHA，2009年被谷歌收购。

Luis earned his PhD in computer science from Carnegie Mellon and went on to reCAPTCHA, which was acquired by Google in 2009.

现在他是Duolingo的联合创始人兼首席执行官，这是全球最受欢迎的教育应用。

He's now the cofounder and CEO of Duolingo, the world's most popular education app.

这款应用每月有超过一亿一千六百万用户，市值高达一千七百亿美元，还有一位假装过自己去世的猫头鹰吉祥物。

With over a 116,000,000 monthly users, a market cap of 17,000,000,000, and an owl mascot that faked its own death.

今天我们聊一聊人工智能在教育中的应用，为什么学习中最难的问题是动力，如何为公司品牌冒险，为什么轻松的漫画风格很重要，还有那个让我们走到今天的16000次A/B测试。

We're gonna talk about AI for education, why motivation is the hardest problem in learning, taking risks with your company brand, why vibe cartooning is important, and the 16,000 AB test that got us here.

路易斯，非常感谢你参加这次对话。

Luis, thank you so much for doing this.

谢谢你们邀请我。

Thank you for having me.

很多人知道Duolingo是做什么的，但我特别想听听你聊聊，除了现在这个语言学习App之外，你希望它未来变成什么样。

Lots of people know what Duolingo is, but I would love to hear you describe it and in terms of, you know, beyond the language learning app it is today, what you want it to become.

它确实是个语言学习App。

Well, it's it's a language learning app.

它是全球最受欢迎的语言学习方式。

It's the most popular way to learn languages in the world.

过去几年，我们还增加了数学和音乐课程。

As of the last couple of years, we also teach math and music.

很快，我们也会推出国际象棋课程。

And as of very soon, we will also teach chess.

我们的想法是，打造一个平台，让大家能在这里学习那些很多人想学、但又需要长时间投入的技能。

The idea is, you know, we're trying to be an app where you can go there and learn the things that a lot of people want to learn, but that also take a long time to learn.

你2011年创办Duolingo的时候，还是一名教授。

You were a professor when you started Duolingo in 2011.

希望这么说不会冒犯你——其实很多教授都会创业。

I hope it is not offensive to say that, like, lots of professors start companies.

很少有教授会创办这种游戏化的消费类公司。

Few of them start, like, gamified consumer companies.

这到底是怎么发生的？

How did this happen?

我根本没想到会搞一个游戏化公司。

It's not like I expected to start a gamified company.

我们一开始的情况是，我那时候还是教授。

The way we got started is I was a professor.

我有个博士生叫塞弗林，现在是首席技术官，也是联合创始人。

I had a PhD student named Severin, who is now the CTO and was a co founder.

当时我们正在给他找一个博士论文选题。

But we were looking for a PhD thesis topic for him.

我们最后决定做跟教育有关的东西，让计算机来教人一些东西。

And what we agreed on is we were gonna work on something related to education where computers would teach you something.

过了一段时间，我们觉得语言是个不错的教学方向。

After a while, we agreed that good topic to teach was languages.

特别是因为学英语。

In particular, because of learning English.

在世界上大多数国家，懂英语能显著提高你的收入潜力。

In in most countries in the world, knowledge of English increases your income potential.

对。

Mhmm.

全球有大约二十亿人在学习英语。

And there's like 2,000,000,000 people in the world learning English.

所以我们想，好吧。

So we thought, okay.

那就用电脑来教语言吧。

Well, let's teach languages and let's teach them with a computer.

后来我们开始做，结果遇到了一个问题。

And then we started working on it and we ran into this problem.

于是我先做了西班牙语课程，因为我本身就是母语为西班牙语的人。

So I made the first Spanish course because I'm a native Spanish speaker.

塞维林是母语为德语的人，他做了第一个德语课程。

And Severin is a native German speaker and he made the first German course.

我们约定好要互相学对方的语言。

And we agreed that we were gonna learn each other's language.

但遇到的问题是，我们根本提不起劲去学，实在太枯燥了。

The problem that we ran into is that we couldn't get ourselves to do it because it was so boring.

天哪。

Oh, no.

我心想，我的天啊。

I'm like, oh my god.

我真的不想学德语。

I did not wanna learn German.

他也不想学西班牙语。

He did not wanna learn Spanish.

我们特别担心，心想：连我们自己都坚持不下去，那怎么可能指望别人去做呢？

We were very worried because we're like, okay, well, if we ourselves can't get it, you know, we can't get ourselves to do it, then we can't expect anybody else to do it.

我们的解决办法就是尽可能把它变成一个游戏。

The solution was to turn it into a game as much as possible.

所以到我们上线的时候，已经挺有意思的了。

And so by the time we launched, it was pretty fun.

但主要是因为我们自己也想坚持下去，才这么做的。

But it's it's mainly because we wanted we were trying to get ourselves to do it.

而之所以会这样，是因为我们俩都不喜欢学语言。

And and part of the reason that that was the case is neither of us likes learning languages.

我们其实根本不是那种爱学语言的人。

We actually are not language lovers.

我们真的不喜欢学语言。

We don't like learning languages.

我觉得正因如此，我们做出来的产品更适合普通人，而不是那种痴迷于学语言的人。

And I think because of that, we made a product that works for the average person as opposed to for people who are obsessed with learning languages.

这里面有太多内容我想慢慢聊，但我想先回过头说说最开始的困惑。

So much there I would like to unpack, but I wanna go back to just initial confusion.

这为什么适合当一个博士课题呢？

What makes this like a good PhD topic?

我的意思是，你们当时对哪个计算机科学问题感兴趣？

Like, what what was the computer science problem you were interested in?

计算机科学上的问题其实就是怎么把东西教给别人。

The computer science problem was basically trying to teach things to people.

这算是计算机科学的问题吗？

Is that a computer science problem?

你的意思是，怎么让计算机来做到这一点？

Well, you know, how to get computers to do it?

这里面有很多门道，比如怎么做到个性化适应，怎么用数据来根据学生的情况调整。

There's a lot there, you know, how to do adaptive, you know, how you use the data to adapt to the to the student.

还有动机问题，以及人机交互方面的挑战。

There's also motivation problems and and, you know, like human computer interaction.

所以当时能做的方向其实特别多。

So there's there was a lot there that we could have used.

顺便说一下，那时候人工智能才刚起步，我们开始能想到用模型来训练，不过这里说的可不是大语言模型。

And by the way, also, at the time, this was, you know, AI at the time was just starting to get to the point where we could have thought, you know, thought about training models to and this is not large language models.

就是训练一些分类器，让它们能更好地教点东西之类的。

This is just training classifiers to kinda teach better or something like that.

这里面其实有很多东西，完全够写一篇博士论文了。

So there's a lot there that could have been a PhD thesis topic.

你决定要让那些缺乏持久动力的普通人，也能更容易地通过电脑学语言。

You've decided that you need to make it easier for normal humans who do not have infinite motivation to learn languages from computers.

你最早做的一件事是什么？就是能帮人克服学语言太枯燥这个心理障碍的？

What was the first thing that you did that worked that, like, helped people get over the motivation hurdle of it being so boring?

我们最早做对的一件事，就是把课程从三十分钟缩短到两分钟。

The first thing that we did that worked was making lessons not thirty minutes long, but two minutes long.

这带来的改变真的很大。

It makes a big difference.

但这不是因为大家花的时间变少了。

And it's not because people were spending less time.

两分钟太棒了，因为任何时候你都可以想，就两分钟而已。

It's that two minutes is is amazing because at any point in time, just think to yourself, oh, it's just two minutes.

但要是三十分钟呢？你现在要是有人问你，你愿意花三十分钟做点什么吗？

But the time investment for thirty minutes, you know, if you're right now and you're like, if I ask you, do you wanna do something that takes thirty minutes?

你肯定会说，哎呀，这可有点难。

You're like, oh, man.

我还不确定要不要做。

I don't know about that.

可当你觉得只花两分钟的时候，说不定你一做就做了三十分钟。

Even though when the you think the time investment is only gonna be two minutes, you may spend the thirty minutes.

但关键是，你觉得自己随时都能开始，也随时都能停下。

But it's just it's just like, yeah, I can start right now because I can end anytime.

所以，这是第一个真正起效的办法。

So that's the first thing that worked.

我们后来还做了很多其他有效的事，但这个才是第一个带来巨大改变的关键。

We did many other things that worked big, but that was the first thing that that really was a big big game changer.

从行为角度来看，有没有什么让你特别意外的有效做法？

Is there anything that really surprised you from a behavior perspective that worked?

有几项。

A few things.

我完全没想到连续打卡会有这么大的威力。

I did not expect streaks to be this powerful.

说实话，连续打卡的影响力真的很强。

I mean, a streak is extremely powerful.

你能解释一下这是什么意思吗？

Can you explain what that is?

就是一个统计你连续多少天做了某件事的计数器。

It's just a counter that measures the number of days you've done something in a row.

比如在多邻国，你的连续打卡天数就是你连续多少天用它来学习语言。

In for Duolingo, the Duolingo streak is the number of days that you've used Duolingo in a row to learn a language.

它的效果简直惊人。

It's incredibly powerful.

人们真的会经常谈起自己的连续打卡记录。

People really talk about their streaks.

我们有整整一千万活跃用户，他们的连续打卡天数超过了365天。

We have 10,000,000 active users that have a streak longer than 365.

也就是说，过去一年甚至更久，他们每天都坚持使用了。

So that means they haven't missed a day in the last year or longer.

所以我真没想到连续打卡会有这么大的威力。

So I did not expect streaks to be this powerful.

还有一件事我也没想到，就是我们那些提醒用户回来使用的通知。

The other one that I did not expect this we have these notifications to get you to come back to the product.

我完全没料到那种带点‘冷嘲热讽’风格的通知效果会这么好。

I did not expect the passive aggressive ones to work so well.

具体是这样的：

This is how it went.

我们以前发的提醒就只是简单地说：‘回来学西班牙语吧。’

We we were sending you notifications that just said something like, hey, come back to learn Spanish.

如果你第一天没回来，我们会给你发一条通知。

If you did not come back the first day, we would send you a notification.

如果你第二天还没回来，我们还是会发通知提醒你。

If you did not come back the second day, we would send you a notification.

但要是连续五天没动静，我们就不再发了。

But after five days of inactivity, we stopped.

后来我突然想到，既然决定不发通知了，至少该告诉你一声。

At some point, it occurred to me that if we're stopping to send you a notification, we should probably let know.

于是第五天我们改发了一条通知，上面只写了：好。

So we started sending this notification the fifth day that said Mhmm.

这些提醒好像没什么效果了。

These reminders don't seem to be working.

我们暂时先停止发送了。

We're gonna stop sending them for now.

结果发现，这种方式特别管用，用户会觉得我们放弃他们了，反而更想回来。

It turns out this is very powerful at getting people to come back because they feel like we've given up on them.

我根本没想到会这样，说实话，我们当初根本没想着靠这个让人回来，但结果却特别有效。

I did not expect to I mean, we didn't even do that with with the goal of getting people to come back, but it turns out that's actually very powerful.

我没学过教育学。

I have not studied education.

所以这个想法并不是基于什么专业知识。

And so this doesn't come from a place of knowledge.

但我觉得，让事情变简单、降低入门门槛，这背后其实有不少争议。

But this idea of, like, making it easy and, like, making the, you know, barrier to entry lower, I I feel like there's some controversy around it.

它引发了一些问题，比如：

It raises questions like, okay.

学习本来就是难的，这没法回避。

Learning is hard and and, like, there is no way around it.

我有个聪明的朋友说，学习就像给大脑去健身，道理很简单。

Like, a smart friend of mine compared learning to, like, going to the gym, for your mind, like, it's very simple.

你只需要坚持去做，别无他法，就得愿意花力气，认真动脑筋。

You just have to do it, and there's no other way besides, like, being willing to put the effort and think really hard about it.

你对这种观点怎么看？

How do you react to that point of view?

如果所有做教育类应用的人都这么想，我会非常开心，因为那样我们就没竞争对手了。

I would be very, very happy if everybody who is working on an education app thought that way Because then we will have no competitors.

其实，学习最难的地方在于动力。

It just turns out that the hardest thing about learning is motivation.

顺便说一句，我觉得这在健身上也是一样的。

By the way, I believe that to be true of the gym too.

很多人整天争论椭圆机和跑步机哪个更好，但归根结底，最重要的是你得动起来。

You know, it's like a lot of people talk about like, you can do you can spend all your time debating whether an elliptical machine is better than a treadmill.

真正关键的是你真的去做了。

But in the end, what matters is that you do it.

这一点占了百分之九十的重要性。

That that's like 90% of what matters.

学习任何东西也是同样的道理。

It's the same with learning something.

你可以花大量时间纠结到底该读哪本书，是用App学好，还是找家教更好。

You can spend all your time debating whether you should read that book or that or this book or whether you should do it through an app or you should do it to a tutor.

当然，这些方法的效果各有不同。

There there's, of course, differences in how effective each of these are.

但最重要的是，你得真正行动起来。

But what matters the most is that you actually do it.

如果你想让人真正开始学，就得让入门变得尽可能简单，降低门槛，激发动力。

And if you wanna get people to actually do it, you have to make it as easy as possible to get started, to get in there, and to be motivated.

我认为这就是我们能快速增长的原因。

I believe that is the reason why we have grown so much.

说起来还挺有意思的。

I mean, and and in fact, it's funny.

我们经常看到一些语言学习App冒出来，说我们跟Duolingo一样，只是没有游戏化设计。

I mean, we see a lot of language learning apps that pop up that say, oh, we're like Duolingo, but without the gamification.

每次看到这种说法，我都觉得挺好。

And every time I see that, I'm like, great.

你就这么做吧。

You do that.

继续吧。

Carry on.

世界上99%的人对任何事情都提不起多大干劲。

99% of the world's population just is not that motivated for any activity.

当然，总有那么一小部分人特别有动力，这很好。

And, you know, there's the few people that are extremely motivated, good for them.

但绝大多数人就是没那么有动力。

The vast majority of people are just not.

如果你是工程师，或者从事其他很多类型的知识工作，那种心流状态几乎是神圣的。

If you are an engineer or, you know, in many other types of knowledge work, the idea of flow is almost holy.

对吧？

Right?

你得沉浸在环境里，熟悉代码库，认真思考一段时间，否则根本没法推进，更别提学到新东西了。

Like, you need to be, like, in the context and understand the code base and, like, be really thinking about it for a while or you won't make any progress, much less like learn something new.

你有想过这个问题吗？

Do you think about that at all?

因为很多人说，我在地铁上，

When because people are like, I'm on the subway.

我就用Duolingo学两分钟。

I'm gonna Duolingo for two minutes.

这种碎片化学习，对整体体验有什么影响？

Like, how does that factor into the experience?

我们其实没太考虑过这个。

We don't really think about that too much.

我明白这个道理，也觉得有道理。

I mean, I understand the concept and it makes sense.

但对我们来说，最重要的是让用户多花时间在App里。

But I mean, for us, the single most important thing is to get people time on the app.

这个我们心里清楚。

You know, we know.

结果发现，对于英语母语者来说，达到不错的西班牙语水平大概需要五百个小时。

Turns out for an English speaker, getting to a pretty good spot in Spanish takes about five hundred hours.

哦，真的吗。

Oh, wow.

明白了。

Okay.

我们只需要累计满这五百个小时就行。

We just gotta clock those five hundred hours.

只要把这时间攒够了就行。

We just gotta clock those.

就这么简单。

Like, that's it.

如果你状态没那么好，会不会变成六百个小时？

And, you know, is it six hundred hours because you're not as, you know, in flow?

还是说，如果你状态特别好，可能四百个小时就够了？

Or is it four hundred hours because you're, like, really in flow?

当然。

Sure.

那里头肯定有些浮动。

There's probably a variance there.

但归根结底，你得花上大约五百个小时，才能把西班牙语学到一个不错的水平。

But in the end, we have to clock something like five hundred hours for you to get to a good spot in Spanish.

顺便说一句，学中文的话，同样的门槛是两千个小时。

And by the way, that same number for Chinese is two thousand hours.

对。

Yes.

作为一个拥有中国文学硕士学位的人，我觉得中文根本不是那种适合拿来学的语言。

As somebody with a, you know, master's degree in Chinese lit, it's not a language meant to be learned.

确实不像。”

It really doesn't seem to be.

我对你关于教育中人类行为的看法，感觉有点现实得近乎冷峻，不够理想主义，但作为学习者，我却觉得特别有启发，因为我正在努力学习。

I find your perspective on the human behavior around education both, like, a little bit dark, like, not particularly idealistic, and yet incredibly inspiring to me as a sort of end user because I am working.

我一直在认真做这件事。

I'm engaged in the work.

我有三个孩子。

I have three kids.

理论上我也做些别的事情。

I theoretically do some other things.

但最让我想在空闲时间去做的，还是提升技能。

And yet the thing I would like most be interested in doing with my free time is skill acquisition.

对啊。

Mhmm.

没错吧？

Right?

比如，把棋下得更好。

Like, be a better chess player.

或者提升中文水平——虽然看起来挺没指望的，又或者多学点数学。

Improve my Chinese as, like, you know, fruitless as that seems to be or learn more math.

好多不同的事情。

Many different things.

但如果你问我，一天忙了十六个小时，还打算在最后挤出一个半小时，用这种特别累的方式继续干吗？

And yet, if you ask me, are you going to carve out, you know, an hour and a half at the end of a day where you've been attempting to be productive for sixteen hours and do more of that in a really painful way?

答案很明显：不会。

The answer is clearly no.

对吧？

Right?

没错。

Right.

但如果你问我，能不能想象在未来几年里，每次挤两分钟，总共攒够六百个小时去做这件事？

And yet if you ask me, could you imagine putting six hundred hours into it two minutes at a time over the next few years?

说实话，这样倒是可以做到。

Like, actually, that seems doable.

在消费类产品领域干久了，你会明白一些事。

After working enough in in consumer products, you realize a number of things.

说实话，现在没人读书了。

I mean, people don't read.

真的没人读。

They just don't.

人都懒。

People are lazy.

如果让他们选，他们嘴上可能说一堆，但实际一有空，还是更愿意刷Instagram或者TikTok。

And if given the choice, they may tell you all kinds of things, but the reality is that if given the choice, they'd rather scroll on Instagram or TikTok.

这其实就是现实。

Like, that's just that's just reality.

但这里面也有积极的一面，就是你可以把这些行为利用起来，做点真正有用的事。

The thing that is positive in all of this is, like, you can take some of that behavior and do do something useful with it anyway, perhaps.

我们做Duolingo，就是想这么做。

That's what we're trying to do with Duolingo.

我们想做的，就是把那些让玩家沉迷手游的套路用过来，但用来让人真正学到东西。

I mean, that's that's what we're trying to do is to use a lot of the the same tricks that are used to keep people engaged with with mobile games or whatever it is, but to get them to learn something.

说到底，这正是多邻国的核心所在。

I mean, that's that's really at the crux of what we do at Duolingo.

多邻国有一个挺独特的优势，那就是当机器学习能力开始快速进步时，你们已经具备了一定的规模。

Duolingo has a a pretty unique position in that you were a company with some scale when the capabilities in machine learning started sort of accelerating in their progress expanding.

你们当时是怎么应对的？作为产品团队，内部是怎么调整的？

How did you handle that and sort of how did you react internally as a product organization?

对我们来说，这完全是好事。

I mean, for us, it's been really positive.

毕竟，这些是语言模型，而我们教的正是语言。

The fact that, you know, they are language models, and, you know, and we teach languages.

这简直是天作之合。

It's a really perfect application.

它们带来的好处主要体现在两个方面。

They've been really positive in two ways.

要教一门语言，首先得准备大量的教学内容。

To teach something, one of the things you have to do is create a lot of content.

这就是你想要教给用户的内容。

This is the content that you're trying to teach.

以前我们制作这些内容，一半靠人工，一半靠自动化。

We used to make that content kinda half by hand, half automatically.

说是人工做的，但背后有一大堆自动化工具，帮内容创作者提速。

Like, it was kinda by hand, but it there's all kinds of automation around it to to make people to make the the people that were creating the content go faster.

但即便如此，进度还是有点慢。

But still, this was going relatively slow.

过去几年，我们彻底改造了整个内容生产流程，现在完全依赖大语言模型。

Over the last couple of years, what we did is we've retooled this whole content creation pipeline to be entirely based on large language models.

我的意思是，现在全都是AI在做。

I mean, it's all based on AI.

人工基本不再参与了，或者最多只是极少数情况偶尔介入。

So humans are not involved any longer or or very little, if if if at all.

这样一来，我们就能量产以前根本不可能实现的海量内容。

And what that has allowed us to do is just create massive amounts of content that were just not possible before.

所以我们现在制作的课程数量大大增加了。

So we're just we're creating many more courses.

比如，这正是我们目前正在推进的一项工作，刚刚上线。

So for example, that's one of the things that we're that we're doing right now, and we're just launching it right now.

以前我们只教40种语言。

We used to have we teach 40 languages.

但我们教语言的方式是，比如教西班牙语给英语母语者，这和教西班牙语给德语母语者不一样，也和教西班牙语给汉语母语者不同，等等。

But the way we teach languages is we teach for example, we teach Spanish for English speakers, which is a different course than Spanish for German speakers, which is a different course than Spanish for Chinese speakers, etcetera.

所以以前我们虽然教40种语言，但都只针对英语母语者。

So we used to teach 40 languages but for English speakers.

如果你是德语母语者，那就只能学其中的四种。

So if you were a native German speaker, you could only learn like four of them.

现在有了AI，因为我们能快得多，

What we're able to do with AI now is because we can go so much faster.

所以我们现在基本上能用40种语言，面向每一种母语背景的人授课。

At this point, we basically teach the 40 languages to every base language.

这大大增加了课程的数量。

And we're we're that that's that's a major increase in the number of courses.

这就是我们现在能做的一件事。

So that's that's an example of something that we can do.

另一个我们以前做不到、现在能实现的是口语练习。

The other example of something that we can do now that we couldn't do before is practicing conversation.

过去我们能教词汇、教阅读，但真正的对话，我们只知道通过真人来练习。

You know, historically, could teach you vocabulary, how to read, but this actual conversation, the only way we knew how to practice it was with another human.

我们试过这种方式，但发现大多数人不愿意用自己还不熟练的语言跟别人对话。

And we experiment with doing that, but it turns out most people don't want to talk to another human in a language that they are not very comfortable with.

因为那样会让人很难受。

Because it feels bad.

确实会让人难堪。

It feels bad.

这会让人感到羞愧。

It's it gives you shame.

这会让你觉得挺难受的。

It gives you, you know, you're it's not good.

但现在有了大语言模型，你可以跟AI练习对话，而且完全不用担心被评判。

But now with large language models, you can actually practice conversation with an AI, and then you you do so without feeling judged.

所以这方面的需求增长得很快。

And so we're seeing a lot of uptake on that.

这对我们的帮助真的很大。

So it's it's it's been really good for us.

你还对哪些AI改变产品功能的方面感到兴奋？

What else are you excited about in terms of, like, AI changing the product itself?

比如角色扮演、解释我的答案这些功能。

Like role play and explain my answer and these things.

你觉得哪些功能目前最有效、最重要？

Like, what do you think matters or works so far?

我特别感兴趣的是，用AI练习真实的语言交流，比如日常对话。

You know, I'm excited about generally practicing real world language like conversation.

我对这一点特别感兴趣。

I'm I'm excited about that.

除了语言之外，我对教数学也非常兴奋。

Outside of language, I'm really excited about teaching math.

利用大语言模型，我们能把数学教学做得好得多。

We can do a much better job at teaching math with large language models.

我们确实有一门数学课程，但现在正彻底重新设计，让它更像一对一辅导。

And we have a math course, but it is is you know, we're completely retooling it to be a lot more like a tutor.

再说一遍，我还是想回到辅导这个话题。

Now, you know, again, I'll go back to tutors.

辅导老师在提升学习效果方面确实很出色。

Tutors can be really good at kinda learning outcomes.

但他们有一个很大的问题。

They have this one big problem.

太枯燥了。

They're really boring.

所以我们正在想办法，让这个系统看起来像个家教，同时还要有游戏化的设计。

So we're trying to find ways to make it a quote unquote tutor, but that is also gamified.

结果发现，大多数人宁愿玩《糖果传奇》，也不愿意坐在那儿听家教讲课。

Just turns out most people would rather play Candy Crush than sit there in front of a tutor.

我们想打造一个东西，效果能接近家教，有趣程度又能媲美《糖果传奇》。

So we're trying to come up with something that is as effective as a tutor, but as fun as Candy Crush.

说实话，我们可能最终做出的东西，效果能达到家教的九成，趣味性也能达到《糖果传奇》的九成，但这两者结合在一起，整体效果肯定会好很多。

The reality is we'll probably come up with something that is 90% as effective as a tutor and 90% as fun as Candy Crush, But at least the combo of this will be a lot more effective.

所以是数学、音乐、国际象棋。

So math, music, chess

现在是国际象棋。

Chess now.

国际象棋现在。

Chess now.

对。

Yes.

你们是怎么决定让多邻国教语言之外的其他内容的？

How do you decide what Duolingo can teach beyond languages?

我们考虑开设的新课程，首先得面向庞大的用户群体。

What we're looking at for subjects to teach are, one, we're looking at very large audience.

我们需要找的是那种几亿人都想学的东西。

So we need something that hundreds of millions of people wanna learn.

这背后是有原因的，毕竟我们终究是个手机应用。

There's a reason for that, and it's because we're an app at the end of the day.

我们不可能向每个用户收三万美元。

And we cannot charge $30,000 to each user.

我们只收你十美元。

We charge you $10.

要想靠一门新课程赚到可观的收入，必须得有大量人来学。

In order to make a significant amount of money from a new subject, it has to be learned by a lot of people.

如果一门课太小众，就算收十美元，但只有几百人学，那根本不值得我们花时间。

Otherwise, these very niche subjects just you know, if we charge you $10 and there's only a 100 people learning it, this is just not worth our time.

所以它们必须拥有庞大的潜在用户群体。

So they have to be they have to be a large potential audience.

另外，我们会找那些需要很长时间才能掌握的内容。

The other thing is we look for things that take a long time to learn.

如果你两小时就能学会，那还不如直接去看个YouTube视频。

You know, if you can learn something in two hours, you probably should just go watch a YouTube video.

两小时的内容，视频完全能搞定。

And that does a perfectly good job if it's just like two hours.

我们关注的是那些真正需要几百小时才能学好的东西。

We look for things that really take hundreds of hours to learn.

还有，我们觉得这些内容对世界有好处，而且手机应用能做得很好。

And then things that we think are are good for the world and we can do a good job with a mobile app.

还有一点，内部必须有人对这个主题特别热情。

Oh, there's one extra thing which is internally we need to have somebody that's really excited about it.

之前所有这些课程都是这样。

So this has been true for all of these other subjects.

你知道，有人对音乐特别热情，也有人对国际象棋特别着迷，正是这种热情推动了这一切。

You know, there's somebody that's very excited about music, somebody that's very excited about chess, and that that's what has done it.

明白了。

Okay.

除了从技术角度考虑哪些适合多邻国的商业模式，你认为现在有哪些东西是计算机很难教的吗？

Even beyond, like, what makes sense for Duolingo as a business from a technology perspective, do you think there are things that are, like, hard to teach with computers for now?

我不确定有没有这样的东西。

I don't know if there are.

确实有一些领域和我们做多邻国的方式不一样。

There's things that are different, that's certainly different from what we do at Duolingo.

比如历史。

So for example, history.

教历史可不是靠这种反复练习的。

The way you teach history is not with these drills.

多邻国在练习这方面确实很拿手。

You know, the Duolingo is really good at drills.

教历史可能得靠制作精良的视频。

The way you teach history is probably with really well produced videos.

这大概是最好的方式了。

It's probably the best way to do it.

也许人工智能在这方面还能做得更好。

And maybe AI can get better at that.

但我觉得，确实有些东西和Duolingo的做法不一样。

But so I think there are just things that are different than Duolingo.

不过说到底，我不确定有什么是电脑真正教不了的。

But ultimately, I'm not sure that there's anything that computers can't really teach you.

我觉得，总体来看，电脑将来一定能很好地教所有东西。

I think, overall, we'll we'll be able to teach everything really well with computers.

你觉得有没有一种可能，人工智能会构成威胁？

Do you think there's any version of the world where AI is a threat?

比如，大型语言模型会对Duolingo构成威胁吗？

Like, large language models are a threat to Duolingo?

嗯，没错。

I mean, sure.

它们确实构成威胁，但这也正是我们这个时代用AI和大型语言模型时最让人担忧的地方之一。

They're a threat to I mean, but, know, this is one of the things that is scary about the world that we live in with with AI and large language models.

我们正在经历一场平台上的重大转变，某种意义上的。

We're undergoing a platform shift, I mean, of of some sort.

我不确定另一边会变成什么样。

I don't know what's gonna happen on the other side.

我倒不是特别担心，但谁也说不准。

I'm not super worried, but you just never know.

这不只是对Duolingo而言。

And it's not just for Duolingo.

可能影响的范围会更广。

So it could be all kinds of things.

对吧？

Right?

也可能对奈飞构成威胁。

I mean, could be a threat for Netflix.

比如，一个大型语言模型只要点一下按钮，就能给你生成一部完美的电影。

Like, it could be that just a large language model just press a button and it makes you the perfect movie.

所以，我不确定这种事会不会真的发生。

And then so, know, I don't know if that will happen or not.

对吧？

Right?

我真的说不准。

I I just don't know.

所以这跟前面说的其实挺像的。

So it's it's kind of a similar thing.

谁知道将来会怎样呢？

Like, who knows what will happen?

目前来看，还不像是个重大威胁。

At the moment, it doesn't look like it's it's a major threat.

不过，你的猜测和我一样，都只是猜而已。

But your guess is as good as mine, I think.

你们拥有庞大的用户覆盖、海量数据，还有很强的工程文化去推动这件事。

I mean, you guys have a huge amount of distribution and a huge amount of data and the interest in engineering culture to go after it.

所以。

So

在我们看来，我们拥有广泛的用户基础。

The way we see it is we have large distribution.

我们掌握着独一无二的语言学习行为数据。

We have data on how people are learning languages that is unique.

我的意思是，我们有大量用户在使用。

I mean, we have that, you know, a a lot of people.

而且品牌在语言学习领域确实很重要，我们在这一块的品牌口碑不错。

It's also the case that brand ends up mattering quite a bit, and we do have a good brand for language learning.

所以，综合这些因素，我们希望这对我们会很有利。

So the combination of all of this, we're hoping, will be good for us.

咱们能聊聊品牌吗？

Can we talk about brand for a minute?

因为Duolingo的品牌确实很特别。

Because it it is Duolingo has a very unique one.

在品牌调性上，你们可能比绝大多数上市公司都更敢冒险、更鲜明。

There is perhaps more risk and more distinctiveness in brand voice than the vast majority of public companies might take.

你们当然是家消费公司，但很多公司根本不敢这么玩。

You are, of course, a consumer company, but there are plenty of companies who don't don't take those risks.

你们这种风格是从哪儿来的？

Like, where does that come from for you guys?

为什么那只猫头鹰‘死了’？

Why did the owl die?

他没死。

He didn't die.

他只是假装死了。

He just faked his death.

好吧。

Okay.

这更奇怪了。

That's even weirder.

我们可不是第一天就决定要搞一个疯疯癫癫的品牌，让猫头鹰干些离谱的事。

This it's not like we, on day one, decided, you know, we're gonna have a brand that is unhinged that where the owl does weird stuff.

这一切都是慢慢发展起来的。

It just kind of evolved over time.

我们的品牌语气也是逐渐形成的。

Our brand voice evolved over time.

其实一开始，产品里的猫头鹰只是催你学习催得有点紧。

What happened was in the product, the owl was a little pushy to get you to do your lessons.

后来网上就开始有人拿猫头鹰编段子了。

And then the Internet just started coming up with memes about the owl doing crazy stuff.

比如有人说，猫头鹰为了让你上课，连你家人都敢绑架。

Like, you know, they started coming up with, like, you know, with the owls willing to kidnap your family for you to do a your lesson.

这些全是网友编出来的，跟我们没关系。

That all was invented by the Internet, not by us.

但看到这种情况后，我们就顺势而为了。

But as we saw that that was happening, we started leaning into it.

为什么不呢？

We thought, why not?

我们越这么干，效果就越好。

And the more we lean into it, the better it worked.

最后我们找到了一个特别能引起共鸣的方向。

And so we just found something that resonated.

就是这个点，让我们找到了感觉。

You know, that got us going.

几年前，我们招了一个刚入职的初级市场员工，他跟我们说：嘿。

Then a few years ago, we had a very junior marketing employee that we had just hired who said, hey.

我们这儿有个以前公司活动用过的旧猫头鹰服装，他说：嘿。

We have this old owl suit here that we had used for, like, events inside the company and just said, hey.

我能拿这个拍点抖音视频吗？

Can I make some TikToks out of it?

其实我一开始是反对的。

And I actually was against it.

我觉得没人会对这个感兴趣。

I'm like, I don't I don't know if anybody's gonna be interested in this.

结果我们一发到抖音上，那只猫头鹰就干些特别傻的事，比如扭屁股、摔跤之类的。

And it turned out that when we put it on TikTok and the owl was it really was just the owl doing really dumb stuff, like, you know, twerking and falling down or whatever.

很多视频突然就火了，有意思的是，这些视频里根本没提学语言、下载多邻国之类的话。

A lot of these videos started going viral, and it was interesting because none of the videos said, you know, learn a language on Duolingo or subscribe to Duolingo, anything.

纯粹就是猫头鹰在搞怪。

It really was just the owl doing weird stuff.

它自己就慢慢火起来了。

It just, you know, it took a life of its own.

后来我们把这套玩法系统化了，也清楚哪些风险值得冒。

We formalized a lot of it, and we have a pretty good idea of the types of risks we're willing to take.