How to make learning as addictive as social media

I’m from Guatemala, and here’s a public service announcement: that’s where Guatemala is. And just to clarify, that’s not where prisoners are kept—that’s called Guantanamo. They’re not the same place. Guatemala is located just below Mexico, and for the Americans in the audience, let me put this in perspective for you—it really applies in many ways: you can think of Guatemala as Mexico’s Mexico. Just like the U.S. doesn’t want illegal immigration from Mexico, Mexico doesn’t want illegal immigration from Guatemala. It’s a smaller, poorer country. But, I’ll tell you this—it has much better Mexican food.

Guatemala is indeed a very poor country, and many people talk about education as the key to bridging the gap between different social classes. However, I’ve always seen it as something that deepens inequality. What happens in reality is that people with a lot of money can buy a good education, which helps them maintain their wealth. On the other hand, people with less money barely learn to read and write, which prevents them from earning much. This is especially true in poorer countries.

Now, I was lucky to receive an education meant for the wealthy, even though I didn’t grow up rich. That’s because I’m an only child, and my mother, a single parent, dedicated all her resources to my education. This allowed me to go to college in the U.S. and eventually earn a Ph.D. in computer science. About ten years ago, I decided I wanted to do something that would provide equal access to education for everyone. And that’s what I want to talk to you about today—making education accessible to all.

At that time, I was a computer science professor at Carnegie Mellon University, and I decided to work on this goal with my Ph.D. student, Severin. The problem of education was so vast that we decided to focus on teaching just one thing. The question was: What should we teach? Should we teach math? We both love math, but we didn’t go down that route. We then thought about teaching computer science. But in the end, we decided that, perhaps surprisingly to people in North America, the best subject to start with was foreign languages. Let me explain why.

First, there’s a huge demand for language learning. Approximately two billion people worldwide are learning a foreign language, both in and outside of schools. Interestingly, about 80% of them are learning English. The reason is that English can dramatically improve one’s life. In many countries, knowing English can significantly increase your earning potential. That’s why we decided to focus on foreign languages—because learning another language, particularly English, can directly boost your income.

This is quite different from many other subjects. For example, knowing math alone won’t necessarily increase your income, because you usually need to learn math first to understand physics, and then study engineering to make more money. However, if you’re a waiter and you learn English, you can work at a hotel and earn more money.

We also realized that in order to reach a broad audience, we had to use mobile phones, particularly smartphones. Building schools worldwide would be far too expensive, but the majority of people already have access to a smartphone, and that number continues to grow. So, we decided to create a way for people to learn languages on their mobile phones, making it accessible to everyone. That’s how Duolingo was born. Thank you.

To ensure that Duolingo was accessible to both the rich and the poor, we adopted a freemium model. This means you can use it for free, but if you don’t want to see ads at the end of a lesson, you can pay a subscription fee. Most of Duolingo’s revenue comes from people in wealthier countries, like the U.S. and Canada, who subscribe to remove the ads. The people who don’t pay typically come from poorer countries like Brazil, Vietnam, or Guatemala. I love this model because it’s a small form of wealth redistribution—rich people are essentially paying for the education of everyone.

However, there’s a significant challenge when it comes to delivering education via smartphones. Smartphones come with some of the most addictive distractions humanity has ever created—TikTok, Instagram, mobile games. Delivering education on a smartphone is like offering people broccoli and placing the most delicious dessert right next to it. If we want to truly educate everyone, it’s not enough to make learning accessible; we also need to make it engaging.

With Duolingo, we’ve managed to make education enjoyable. We’ve done this by using the same psychological techniques that apps like Instagram, TikTok, and mobile games use to keep people engaged—but we’ve applied them to learning. One of the most effective techniques is the concept of a streak, which tracks the number of consecutive days someone has used the app. If you break the streak, it resets to zero, and people don’t want to lose it. This simple idea works.

For example, we have over three million daily active users who have maintained their streaks for over a year. Now, fun fact: Which country has the longest streaks on Duolingo? It’s Japan. And the shortest? Latin America, of course. But hey, we’re fun, right?

Notifications are another important tool to keep people engaged. While notifications can be annoying in some cases, when used thoughtfully, they can remind people to learn. Duolingo has a sophisticated AI system that determines the best time to send notifications and the message content, maximizing the chances people will return. Interestingly, the best time to send a notification is 24 hours after someone last used the app—because if you were free at 3 PM yesterday, you’re probably free at the same time today.

And if people don’t use Duolingo for seven days, we stop sending them notifications. But we don’t just stop there. We also send a message saying, “Hey, these reminders don’t seem to be working. We’ll stop sending them for now.” You’d be surprised—this passive-aggressive message often gets people to return to the app. It’s funny, but it works.

Now, Duolingo has become a part of popular culture. There are thousands of memes, and even SNL skits about our green owl mascot, who is both passive-aggressive and persistent in reminding people to learn.

Duolingo’s success shows that it’s possible to get people to engage with learning in the same way that social media and games engage them. However, I don’t believe we can make an educational app as addictive as TikTok or Instagram, because at the end of the day, we’re trying to teach people something meaningful. But that’s okay. When you learn something, you derive meaning from it, while scrolling through Instagram often feels like a waste of time.

So, it’s fine if an educational app is only 80-90% as engaging as TikTok because people’s intrinsic motivation will make up for the rest. And that’s the key point: We can use the same techniques that keep people engaged with social media to get them to do something meaningful. Duolingo has proven that we can reach hundreds of millions of people, and my hope is that we can apply the same idea to other subjects like math, physics, or whatever else, using mobile phones to make education accessible to all.

In the future, I hope screen time isn’t viewed negatively, but instead seen as a tool for delivering quality education to everyone, rich or poor, through mobile phones.

But the most important takeaway from today: please, I beg you, do your language lessons today. Thank you!

Helen Walters: Luis, thank you so much. Can you elaborate a little more on the last point? How do you think we can apply this model to other subjects?

Luis von Ahn: I think this approach can work especially well for subjects that require repetition, such as reading or elementary math. Anything learned through repetition can be gamified in the way Duolingo does. However, subjects that require more explanation, like advanced physics, might need high-quality videos. Sal Khan is doing an excellent job with that. But for anything that involves repetition, we can definitely apply the same methods.