digest
I'm from Guatemala. This is a public service announcement: that is where Guatemala is, not where they keep the prisoners. That is called Guantanamo, which is not the same place. Guatemala is right below Mexico. For the Americans in the audience, let this sink in, because it really applies in most ways: you can think of it as Mexico. Just like the US doesn't want illegal immigration from Mexico, Mexico doesn't want illegal immigration from Guatemala. It's a smaller country and a poorer country, and, well, what can I tell you, it has much better Mexican food.
Guatemala is a very poor country. A lot of people talk about education as something that brings equality to different social classes, but I always saw it as the opposite, something that brings inequality. What happens in practice is that people who have a lot of money can buy themselves a really good education and therefore continue having a lot of money, whereas people who don't have very much money barely learn how to read and write and therefore never make a lot of money. This is especially true in poor countries.
I was fortunate that I received a rich person's education, even though I didn't grow up rich. It's because I'm an only child and my mother, who was a single mother, spent all of her resources on my education, and this allowed me to come to college in the US and eventually get a PhD in computer science. Because of all of this, about 10 years ago I decided I wanted to do something that would give equal access to education to everyone. Oh, by the way, this is what I want to talk to you about today: giving equal access to education to everyone.
At the time, I was a professor of computer science at Carnegie Mellon University, and I decided to work on this with my PhD students. The way my brain works, all of education is just too general of a problem, so I decided to start by teaching one thing. Then I started thinking, okay, what should we teach first? Should we teach math? We both loved math, me and my co-founder. We didn't decide to do math. Then we started thinking maybe we should teach computer science, but eventually, and this is maybe surprising to people in North America, we decided that the best subject to start with was teaching foreign languages. Let me tell you why. There are a number of reasons. One of them is that there's a huge audience for it. There are about two billion people in the world learning a foreign language, both in school and outside of schools. Most of these people, by the way, are learning English. About 80% of them are learning English. In this map right here, all the countries in green are countries in which people are predominantly learning English, and the reason for that is because English can truly transform your life. In most countries in the world, knowledge of English can significantly increase your income potential.
So this is another reason why we decided to start with foreign languages: you can directly make more money if you learn another language, in particular English. This is pretty different from many other subjects. For example, take math. In the case of math, just knowledge of math does not increase your income potential, because usually you have to learn math to then learn physics, to then become a civil engineer. That's how you make more money. Whereas with languages, if you were a waiter and you learn English, now you can be a waiter at a hotel and make more money. So we decided to get started by teaching languages.
We also convinced ourselves that the only way to really reach a lot of people was by using a mobile phone or a smartphone. In particular, building schools all over the world is simply too expensive. On the other hand, most of the world's population already has access to a smartphone, and the trend is that that fraction is only going to increase. So we decided that we would make a way to learn foreign languages on a mobile phone that was accessible to everyone, and then we called it Duolingo. Thank you.
In order to truly be accessible to everyone, rich and poor, Duolingo uses a premium model to support itself. What that means is that you can learn as much as you want without ever having to pay, but if you don't pay, you may have to see an ad at the end of a lesson. If you don't like ads, you can also pay to subscribe to turn off the ads. It turns out that the vast majority of the revenue for Duolingo comes from people who pay to subscribe to turn off the ads. Who are these people? They're usually well-off people in rich countries like the US and Canada. The people who don't pay to subscribe usually come from poorer countries like Brazil, Vietnam, or Guatemala. What I like about this model is that it is a small form of wealth redistribution, because we're basically getting the rich people to pay for the education of everyone. So I like that.
With smartphones we can reach a lot of people, and we can even get the rich people to pay for the whole thing. This is great. However, if you're trying to deliver education with a smartphone, you run into a humongous problem, and it is that smartphones come equipped with some of the most addictive drugs that humanity has ever engineered: TikTok, Instagram, mobile games. Delivering education over a smartphone is like hoping that people will eat their broccoli, but right next to it you put the most delicious dessert ever made. If you really want to deliver education to everyone, not only do you have to make it accessible, but also you have to make it so that people want to actually learn. With Duolingo we've been able to do this, and at the highest level. The way we've done this is by making the broccoli taste like dessert. I'll say it another way: we've used the same psychological techniques that apps like Instagram, TikTok, or mobile games use to keep people engaged, but in this case we use them to keep people engaged with education.
Let me give you some examples of these techniques. One of the most powerful ones is the notion of a streak. A streak is just a counter that measures the number of days that you've used the product consecutively. You take that number, you put it very prominently in your product, and then people come back every day. The reason people come back every day is because if they don't come back, that number resets to zero, and people don't want to lose their streak. It works. On one side, streaks have been criticized, for example for getting teens addicted to Snapchat, but in the case of an educational app, streaks get people to come back to study every day. To give you an idea of the power of streaks, in the case of Duolingo we have over 3 million daily active users who have a streak longer than 365 days. That means they haven't missed a day in the last year or longer. Now, fun fact about streaks: what country do you think has the longest average streaks for an educational app? It's Japan, of course. Shortest ever streaks: Latin America, baby. But we're fun, hey, we're fun.
Another important mechanism to get people to come back to your product is notifications. On one side, notifications can be really spammy and annoying, but in the case of an educational product, people actually want to be reminded to learn. In the case of Duolingo, we have a very sophisticated AI system that basically chooses when to send the notification and also what to say in each notification to maximize the probability that people come back. Interestingly, even after all this sophistication, it turns out that the algorithm for choosing what time to send you a notification is pretty simple. You know what the best time to send people a notification is? I'll tell you, and there's an easy explanation: if you were free yesterday at 3 p.m., you're probably free today at 3 p.m. as well. So this is what a very sophisticated, millions-of-dollars AI found. It's funny.
With notifications, you shouldn't be spamming, and we're not spamming with Duolingo. We actually stop sending notifications after seven days of inactivity, so if you don't use Duolingo for seven days, we stop sending you notifications. At some point it occurred to us that if we're going to stop sending people notifications, we should let them know. So we started sending this notification to people saying, "Hey, these reminders don't seem to be working, we'll stop sending them for now." You know what people do when they get this notification? They come back. Passive aggressive works, for my mother and for Duolingo. These passive-aggressive notifications are really good at getting people to come back, because they feel like our green owl mascot has given up on them, so they come back. Speaking of our green owl mascot, by the way, because all our notifications come from our green owl mascot, and he's passive aggressive and also pretty pushy, this has given rise to a lot of memes on the internet that make fun of the great lengths he will go to get you back to learn a language. Here's one of my favorite ones. It basically looks like you forgot your Spanish lessons, and then there's an intruder. Presumably the owl broke into your house to get you to learn a language. Duolingo has entered the zeitgeist. There are thousands of memes, there are SNL skits about it, and it's because we've managed to get people to want to learn a language by using the same techniques that mobile games and social media use to get people engaged.
This is a really important point. I don't actually believe there's a way to make an educational app as engaging as something like TikTok or Instagram or mobile games. The reason I don't believe that is because ultimately you have to teach people something, and it's hard to compete with cats and celebrities. But the good news is that I don't think you have to. Here's the thing: when you're learning something, you get meaning out of it, whereas when you're scrolling for two hours on Instagram, a lot of times afterwards you feel like you just wasted your time. So I think it's actually okay if your educational product is only 80 or 90% as engaging as something like TikTok, because the other 10 or 20% will be provided by people's internal motivation. Though of course, not much more than that. This is really a key point: if you want to get people to do something meaningful, you can use the same techniques that apps like social media use to get people to do it. Even if you're not as engaging as those apps are, you can still get hundreds of millions of people to use your product. In the case of Duolingo, for example, there are more people learning languages on Duolingo in the United States than there are people learning languages across all US high schools combined. This is true in most countries in the world.
My hope is that, I know we can do this, but my hope is that as humanity we can do what Duolingo has done for learning languages for all other subjects, where we can get people to learn math with mobile phones, millions of people to learn math with mobile phones, or physics, or whatever. I hope for a future in which screen time is not a bad thing, in which we can deliver high-quality education to everyone, rich or poor, using a mobile phone. But the single most important thing I can end this talk with is a reminder: please, pretty please, I beg you, do your language lessons today. Thank you very much.
Thank you, Luis. Thank you so much.
I wonder if you can say a little bit more about that last point. How do you think you can apply this type of thinking to other subjects? You mentioned math and things like that. How do you do that?
Yeah. In particular, subjects that are learned through repetition. It turns out most things that are really meaningful are learned through thousands and thousands of repetitions. You learn to read through repetition, you learn elementary school math through repetition. Most things that you can learn through repetition, you can actually gamify and turn into something like Duolingo, where people just do it a lot and do it in a fun way. It's a little harder for things like explanations. That's probably going to require some really good videos. Salman is doing a really good job with that. But for things that require a lot of repetition, I think we can use the same methods.
Amazing. Thank you so much for sharing.
Thank you.