The Big Interview: Luis von Ahn on Duolingo, AI, and Why the World Still Needs Language Learning
Luis von Ahn could have retired to a beach paradise years ago. Best known as the CEO of leading learning app Duolingo, von Ahn first made his mark in the 2000s as the inventor of CAPTCHA—the notoriously frustrating little online checks that prove you’re not a robot. After selling reCAPTCHA to Google in 2009, he didn’t waste any time launching his next venture: a company rooted in his childhood growing up in Guatemala, that is now one of the most prominent education platforms in the world.
Von Ahn’s mother, a doctor, poured every extra dollar she earned into sending him to private school, opening opportunities that most of his peers never got. As he tells me in this week’s Big Interview, that experience is why he founded Duolingo more than a decade ago, with a core goal of making high-quality education free and widely accessible to everyone. Today, the platform serves more than 130 million users worldwide, from immigrants learning new languages to A-list celebrities like George Clooney.
While inequality inspired von Ahn’s mission, his company now sits at the center of a far more urgent modern conversation: artificial intelligence. As AI rapidly reshapes how people learn, how companies operate, and how workers assess their own value, I wanted to know how it is shaping Duolingo’s internal processes, expansion plans, and long-term sustainability. If AI can translate almost anything in any format, easily simulate conversation, generate lesson plans, and personalize instruction… does the world still need Duolingo?
Von Ahn’s position is unequivocal: not only is Duolingo already benefiting from generative AI, but he says people will continue to seek out Duolingo’s gamified, motivation-driven approach to learning. In our conversation, he opens up about building a mission-driven public company within Wall Street’s constraints, why he doesn’t lose sleep over dips in Duolingo’s share price, and how Duolingo keeps learners engaged in ways AI alone cannot.
This interview has been edited for length and clarity.
KATIE DRUMMOND: Luis von Ahn, welcome to The Big Interview.
LUIS VON AHN: Thank you for having me.
We always open these conversations with a few quick warmup questions for your brain. Are you ready?
Sure.
What’s a language you’d desperately love to master, but haven’t gotten there yet?
Swedish. I’m learning it now, but I still need to get a lot better. My wife is Swedish.
That’s a pretty great motivation. You’d better get on it!
I’m on it.
What job do you think AI should never do?
A lot of jobs. I think any job where humans need inspiration, like teaching. Humans need to be inspired by other people, and it’s really hard to get that from AI.
I agree. AI definitely has an inspiration problem. You were 28 when you got the MacArthur “genius” grant. What did you do with the money?
I put it in the bank. I was really happy and proud to receive it, but that’s all I did with it at first. Eventually, most of it ended up going toward getting Duolingo off the ground, I think.
Which language has the most ridiculous grammar rules?
Finnish and Hungarian are really hard, with really strange rules. But generally, I don’t think it’s about the rules being actually ridiculous. Any language that’s very far from your native language just feels weird and ridiculous by default.
My sister is learning Mandarin right now, and I know she’d 100% agree with that.
It is a hard language to learn.
She’s having a rough time. She uses Duolingo, actually, one day at a time. Now, you invented CAPTCHA. Do you want to apologize to me and our audience now, or save it for later?
Look, I am sorry. By the way, it wasn’t just me.
Sure.
It was a team effort with my PhD advisor. But yeah, I apologize. I’m sorry.
Thank you, we appreciate that. So does the entire internet. Alright, let’s dive in. Let’s set the stage: you started Duolingo roughly 15 years ago, right?
Yeah, around 2011.
Back then, the original idea was a crowdsourced translation tool. Fast forward to 2026, I can translate English to any language instantly right on my AirPods. With generative AI and all the new tools that have come from that technology, there’s a lingering question: what does this mean for learning? Not just language learning, but how we think about learning as an activity overall. How would you describe Duolingo today, big picture? What has changed?
Generally, Duolingo is a platform for learning anything. We primarily teach languages, but that’s not all we do anymore. We also teach math, music, and chess now.
I think the desire to learn hasn’t gone down at all. Humans still need to learn new things; it just makes life fuller and better if you know more.
For what it’s worth, computer translation has been almost perfect for major languages for a decade already. Google Translate was essentially perfect between English and Spanish back in 2015, for example. But we haven’t seen demand for language learning go down at all. It’s actually gone up.
There are two big reasons for that. The biggest one, at least for our users, is that most people learning a non-native language do it as a hobby. Whether a computer can translate something for you doesn’t matter when it’s a hobby you enjoy.
A great example is chess. Computers have been better at chess than humans since the first one beat a world champion, but people still love learning to play chess. It just doesn’t matter. That’s the big reason.
The other reason is that half of our users are learning English, and for most of them that’s not a hobby. Knowing English directly improves your life in all kinds of ways, usually it means you can make more money, that’s pretty direct. For example, if you’re a waiter living in a non-English speaking country and you learn English, you can get a higher-paying job at a tourist hotel, right? So we just haven’t seen demand for language learning drop at all.
That’s really interesting. I want to go back to your background, which makes Duolingo’s origin story make so much sense. Can you talk about your home life growing up? Is there a memory that stands out as transformative?
I grew up in Guatemala. It’s funny, people can paint it in a really bleak way— I was born and grew up during Guatemala’s civil war, which sounds terrible, but it wasn’t that bad for me. I lived in Guatemala City, which was pretty sheltered from the worst of the conflict, and I grew up in a middle-class neighborhood, so it was pretty safe. It was just me, my mom, and my grandma living together, the three of us.
There are a lot of memories, but one that totally changed my life was when my mom came home one day with a computer. I’d never used one before, I was 7 years old. What I really wanted was a Nintendo. She brought me a Commodore 64. I was actually pretty upset! It’s so funny, my mom has never used a computer in her entire life, even to this day. She doesn’t have a smartphone, just a flip phone, and she’s never been on the internet.
I respect that, honestly. Good for her.
But she got me that computer, and that changed everything. This was the mid-to-late 1980s, and computers were not easy to use back then. I had to teach myself how to use it, it was super frustrating for a whole year until I could do anything useful with it. But I stuck with it, and that changed my whole life.
You went to the American School of Guatemala. You’ve talked before about how it gave you a window into the world and shaped your worldview. Can you expand on that?
I was very fortunate that my mother was a physician. For context: doctors aren’t nearly as wealthy in Guatemala as they are here in the US. My mom worked for the Guatemalan government, so we were straight-up middle class, but she spent all of her extra income, after paying for food and basic needs, on my education.
That school was where all the kids of rich people, ambassadors’ kids, everyone like that went. Even though I wasn’t rich, I got the same education rich kids got, and that made me see the world differently. Everyone else in my neighborhood went to public school or a lower-tier middle-class school. I could see the difference firsthand in what I was learning, and the opportunities that education opened for me. Because of that education, I was able to come to the US for college, and everything that followed.
That completely changed how I see education. A lot of people say education is a great equalizer that brings social classes together. I always saw the opposite: generally, wealthy people can get way better education than anyone else.
Absolutely.
We’re sitting here in New York. People with money in New York can get an incredible education, they pay $60,000 a year for it, and it’s probably worth it. But people without much money, especially in poor countries, sometimes barely even learn to read and write. That’s all they get. That’s why I wanted to start Duolingo: to build something that gives great education to everyone in the world. That’s what we’ve done, and I’m very proud that part of the vision has come true.
Today, Duolingo has more than 100 million active users, and it serves the full spectrum: from Syrian refugees across Europe learning the language of their new country, to wealthy famous people like George Clooney, who uses it. I saw that headline when I was researching for this, and I immediately clicked it.
It’s not just him, lots of wealthy famous people use it. So wealthy people already have access to way better opportunities than Syrian refugees, but on Duolingo, everyone uses the exact same system. More money can’t buy you a better Duolingo. That’s exactly what we wanted to do.
I’m curious: you run a public company, which comes with a lot of financial expectations. How do you keep that equal access model, where George Clooney gets the same experience as a Syrian refugee who needs to learn English?
Ultimately, we’re a very mission-driven company. We really care about this mission of building the best education in the world and making it universally available. But the thing is, over the long term, I’m pretty sure this mission can also build a very big, successful business.
Right now we have around 135 million active users, give or take. If we can get to 500 million, then a billion active users, that’s a huge business. Not everyone pays: about 10 percent of our active users pay, and the other 90 percent use it for free. So the George Clooneys of the world pay, and other people don’t. That’s right.
That still adds up to a large business. I’m perfectly fine with that. By the way, our free users tell their friends about Duolingo, and some of those friends end up paying. We love our free users.
I don’t want to dig too much into CAPTCHA, but looking back at your career: you created reCAPTCHA and sold it to Google. My understanding is that sale gave you enough money to retire at a very young age, if you wanted. You could have built a dream house anywhere, bought sports cars, done whatever people do when they make that kind of money young. But you didn’t. Why?
In 2009 I sold reCAPTCHA to Google, and I did think about retiring.
When you thought about it, what did your ideal retirement look like?
A house on a beach, but you know what’s funny? I hate sand.
So that killed the dream right there, and you got back to work.
That wasn’t even the main reason. The reason was I knew it would be really boring. I just always need to be working on something. So I pretty quickly started working on other projects.
And that was 2009, so it didn’t take you long to get from that to founding Duolingo?
No, only about a year and a half.
I heard Bill Gates once tried to convince you to join Microsoft, and you turned him down because you wanted to “do your own thing.” That’s a pretty bold thing to say to Bill Gates. Do you think you’d be a terrible employee?
I don’t know if I’d be terrible. I just had a bunch of things I really wanted to do on my own. Microsoft offered me a role at Microsoft Research, which is an amazing place, don’t get me wrong. But I just really wanted to do my own thing.
I don’t think I would have been that bad of an employee, actually. I’m a rule-follower by nature, so I would have followed the rules. But I’m way happier doing what I want to do. A big part of that is the education mission: I really want to give education to a lot of people. As I get older, I start thinking: what do I want to be remembered for?
You’re already thinking about that?
I am. Not that I’m dying tomorrow, but if I am remembered for something, what will it be? I want it to be something related to education.
Can you talk about building and scaling Duolingo from the ground up? As a first-time CEO, going from zero to nearly 1,000 employees now, what were the biggest challenges along the way?
The job of CEO changes a lot as the company grows. When you only have two or three employees, the job is totally different than when you have 100, or when you have close to 1,000 like we do now. The job changes quite a bit, so one of the hardest parts is growing alongside the company.
Not everyone may agree with this, but I think when a company is going from zero to 20 people, the best thing a CEO can do is micromanage.
Just be in the weeds with all 19 other employees?
Yes. If it still worked at 1,000 people, I’d still do it. It just stops working after a certain point.
Sounds exhausting.
At some point you have to figure out what’s not working anymore, and evolve your approach. The other big challenge for me personally is that I’m naturally conflict-avoidant. That’s not always compatible with being a good CEO.
Yeah, I can imagine that wouldn’t work.
The first person I had to fire at Duolingo, I ended up having to do it three times. The first two times, I sugarcoated it so much that they didn’t even realize they were being fired, and they came back to work the next day. No joke, that actually happened.
Oh wow. What happened the second time?
I can barely even explain it. I sugarcoated it the second time too, I thought I was being firmer, but I still wasn’t. The third time I had to be pretty blunt, I was like, “You are being fired. Do you understand what that means?”
I’ve learned a lot from that. The biggest lesson is: if you have something hard to say to someone, don’t beat around the bush. Just say it, then clean up the aftermath.
There’s a kindness to direct communication.
That took me many years to learn. Guatemalan culture is very roundabout. I joke that in Guatemala, the way to say yes is just “yes.” The way to say no is you say, “yeah …” That’s how people say no. I’ve had to learn to be a lot more direct.
Duolingo has always been headquartered in Pittsburgh. A lot of people don’t realize there’s a thriving tech scene there. What has it meant for the company to be based in Pittsburgh, good or bad, instead of the more obvious choice of Silicon Valley?
I think Pittsburgh was great for us early on, and it still is. There are a number of reasons, but one big one is that education, what we focus on, has never been a passing tech fad. In Silicon Valley, everyone’s always chasing the latest fad. There’s nothing wrong with that, right now the fad is AI, which is great. But if you want to build something long-term in education, it’s been really nice to be in a place where we’ve had time and space to develop our product slowly. It took us years to build the system we have today, and it’s been a blessing that we’ve been left alone to do that work.
Yep.
The other big benefit of Pittsburgh is that it allowed us to build an executive team that I think is better than we would have gotten anywhere else. We had a lot of top executive talent that ended up in Pittsburgh for personal reasons: they had aging parents, they had other reasons they needed to be here. If they’d been in San Francisco, they would have taken jobs at much bigger companies, since many of them had worked at big tech firms before. But because they were in Pittsburgh, we were the best opportunity in town, so they joined us.
You’re definitely the best tech game in town in Pittsburgh.
To be fair, we also have large offices in other places now, including one in New York where I spend more than half my time now. So having multiple locations helps, but overall Pittsburgh has really helped us.
Well, don’t give up on Pittsburgh. I have a real soft spot for the city.
Don’t worry.
You mentioned AI earlier in the context of Silicon Valley fads. Back in 2025 you sent an internal memo saying Duolingo would shift to an “AI-first” approach, where AI wouldn’t just boost productivity, it would help you advance your core mission. That memo got posted to your LinkedIn, and the public comments were pretty scathing. A lot of people even joked that AI should replace CEOs, as I’m sure you saw. It definitely wasn’t the best press Duolingo has ever gotten. Can you
