When Luis von Ahn was 8 years old, growing up in Guatemala City, his mother bought him a computer. He was upset because he actually wanted a Nintendo, he recalls. “But you know, it changed my life because I learned how to use a computer.” Developing that skill eventually led to his creation of the educational app Duolingo, which is used by millions around the world to learn languages, and now other subjects, such as math and music. The company went public in 2021 and is currently valued at $5.7 billion.
“Education has always been a big thing for me,” says von Ahn, 47. His mother, a doctor and single parent, didn’t earn much, but spent any spare money she had on ensuring von Ahn had access to the best education in Guatemala, he says. “The other kids in my neighborhood probably went to public school. It was very clear that I was years ahead of them in terms of what I was learning, and that always stuck with me,” he says. “And that's probably the biggest reason why we started Duolingo. The idea is to try to give education to everyone, regardless of whether you don't have a lot of money or you have a lot of money, you can use the same system.” Von Ahn believes the app has achieved that goal, and says its users range from across the socioeconomic spectrum.
TIME spoke with von Ahn in May about Duolingo’s story, user engagement, and how he’s approaching AI integration.
This conversation has been condensed and edited for clarity.
Tell me a bit about the evolution of Duolingo.
We started very small in 2011. We wanted to do something that would give equal access to education to everyone. At first, we didn't want to teach languages. At first, we thought we wanted to teach math, but we decided to teach languages, and the reason for that was because of English. In almost every country in the world, if you know English, you can make more money. There are a little under 2 billion people in the world learning English. Math has a billion people learning it. So English is basically the most learned subject that I know of in the whole world. So we thought, OK, let's do a thing that teaches languages for free. We launched it. It was very successful from the minute we launched it. It became the most downloaded app in the education category, and it has been since then. And we reached all kinds of really amazing numbers relatively early on. For example, at some point, there were more people learning languages on Duolingo in the U.S. than there are people learning languages across all U.S. high schools combined, and now that's true in most countries in the world. Not in every country—particularly in China, it's not yet true—but in most countries we teach more people languages than the public education system. So we're very proud of that.
For the first four or five years, we made no money. We didn't have any ads. There was no subscription. Nothing. It was just free. Then we started monetizing around 2017, but one of the crazy things that happened is that we had hired all these people, and they had come to work at Duolingo to give free education. And when I said to everyone, we need to make money to sustain ourselves, there was a revolt. People did not want to—most of the employees were like, “Why? Why do we need to make money?” Which is crazy if you think about it, but I had to explain to them that you actually have to pay salaries, and that's how we have to make money, etc. But eventually, we found a good way to make money, which is our freemium model: everybody can use Duolingo entirely for free. But if you don't pay, you have to see some ads, and if you don't like ads, you can pay to subscribe, and it gets rid of the ads.
That worked, and then after that, we started at least being self-sustaining. At some point, we became profitable. We're now a publicly traded company, so we're very happy with that. The other thing that has happened over time is we've added more subjects. We started with just languages, but now we have math, music, and chess, and we're very happy with those. I'm particularly excited about math. I think over time, hopefully, it'll become almost as large as languages. Languages are much larger, but I'm hoping that math at some point will become as large as it.
Is English the most popular language?
Yeah, for sure. Half of our users are learning English.
And where is it most used?
We have users in every single country in the world. The largest market is the U.S. in terms of number of users. Second is China, and then third [place] oscillates between Brazil, the U.K., and Germany.
Are there complications to operating in China?
Yes. We're fine now, but China has a lot of regulation, and we have to go through that, but we're happy. China's not the number one fastest growing country, but it is relatively fast growing, and our expectation is that within one or two years China will be our largest market. It's just growing really fast. There's just a lot of people in China, also a lot of people who want to learn English. So you keep growing.
How many languages have you learned, or are you learning?
I got pretty good at French to the point where I can watch TV shows in French. My pronunciation is not particularly good, but I can watch TV shows pretty well. I am currently learning Swedish because my wife is Swedish, and now Swedish is an excellent language. It just turns out not a particularly useful language to learn because all Swedes seem to know English. But I am learning it because my wife is Swedish. I'm not as good as I am at French.
I have to confess I don't use Duolingo, but I did speak to some friends who use it and got some feedback. A couple of things that I heard a few times were that they don't like the guilt-tripping constant reminders, and also the need to pay to progress…
You don't need to pay to progress. You need to pay to turn off ads. And if you don't pay, you sort of cannot do too much per hour. That's basically it.
But there's also the AI function…
Oh yeah, as in, you have to pay if you want to do the practicing conversation with AI. You have to pay because that also costs us money. Yeah.
I'm sure you hear similar feedback. How do you take that on board and how do you integrate that into improving the app?
Well, a couple of things. In terms of the guilt-tripping, we try a lot of things. The thing about learning something by yourself—and this is really what has made us the most popular education app in the world—the hardest thing about learning something by yourself is staying motivated. It's just really hard to stay motivated. Most people will tell you that they want to learn something, but their actions don't say the same thing. I mean, their actions are [that] they're going to go to Instagram or TikTok or whatever. One of the main things I've learned here at Duolingo over the last 15 years is what people say, what people do, and what people think—completely different things. So people will tell you they really want to learn something, etc. But it's very hard to stay motivated to learn something like a language. To learn a language to a relatively good point takes hundreds, if not thousands, of hours. That is just the case. There's no faster way to do that. Your brain needs that many hours to learn it. So we need to find ways to get people to continue doing it, and we just try everything we can. We try to make it fun. Try to make it as gamified as possible. We send you reminders. Some of the reminders are guilt-tripping, and it just turns out this works. And that's the idea. The idea is to try to get you to be motivated to continue working. I mean, learning.
It sounds like your workforce is very passionate…
Mission driven. Yes.
There was a story about assessment of employees based on AI usage, which was rolled back. What happened and what did you learn from that experience?
Last year I sent a memo internally that, honestly internally it was not controversial. I mean, it was like, “Hey, we’ve got to use AI in as [many] places as we can.” There was one part that, at the time was not controversial, but we have since flipped, which is we're going to assess people's performance based on their use of AI. That's not going to be the only thing, but one of the things I look at is are you using AI or not? Because we really do believe that AI can make you a lot more productive. I sent that memo. Externally, people read that memo and thought that I was going to fire all our employees. This is actually not what happened. We didn't fire anybody. We in fact hired more employees last year, and part of the reason is it's just now an employee is way more valuable than before because one employee with AI can just do way more than a single employee before AI. So I want to keep hiring, but internally, what we learned over the last year was that there were people who did their job, and they had a lot of impact. But they're like, “AI doesn't really apply to my job. Do you want me to just use AI for no reason?” And they had very good arguments. So, we've gone back to just, your performance is based on the impact you have on the company, regardless of what you do. So yeah, we shouldn't have done that, but we did.
You talked about offering something to users that isn't Instagram or TikTok. There are people who are also thinking about how they use AI and even not wanting to use it. How do you balance improving the app using AI integration against keeping those loyal users who like using Duolingo because it's not social media or ChatGPT?
Ultimately, our goal is to teach you things that take a long time to learn. That's our goal. It turns out AI can really help us teach you those things much better. A really good example is conversational practice. Before AI, there was no way for you to practice conversation on Duolingo, and the only way we could have done it before AI was to pair you up with another human. But this is one of those cases where people say they want to talk to another human in a language that they are not comfortable with, but people do not actually talk to another human in a language that they're not comfortable with. We tried many times. I believe about 5% of humanity's population is willing to do that, and the other 90% to 95% doesn't actually want to. Even though they say they do, they don't actually want to do that because it just requires too much. People are ashamed. Before AI, we simply could not get you to practice conversation. It just was a hole in your learning. Now with AI, you can actually fully practice conversation, and that you learn the language better. So, for our usage of AI, we have this golden rule internally in the company, which is we are only going to use AI for the benefit of our learners, and so that's it. In fact, the overwhelming majority of our usage of AI is either in speeding up our development process or in teaching better. And so, I understand that some people may not love using ChatGPT, etc. But you know, ultimately, we're just trying to teach you better.
