Dr. Luis von Ahn: Hey guys.
TechFaster: How are you?
Dr. Luis von Ahn: Pretty good.
TechFaster: Can you tell us a little bit about Duolingo and what you guys are doing?
Dr. Luis von Ahn: Sure. So Duolingo is a language learning website. It is absolutely free. So it doesn’t have any ads, it doesn’t have any premium content or anything. It’s just totally free and as people are learning a language they get all kinds of different exercises. Some of them are you get to speak to a computer and it tells you whether you said it right or not. Some of them you have to translate little pieces of text, et cetera. And the reason Duolingo is totally free is because some of the exercises we give you are actually real world translations that nobody has ever translated before and we actually use that to help translate parts of the web. So while you’re also helping us to translate parts of the web so you know you could be translating Wikipedia for example while you’re learning Spanish.
TechFaster: that’s really awesome. Can you talk to us a little bit about your work on Games With a Purpose and Human Computation and maybe a little bit about how Duolingo employs the concept?
Dr. Luis von Ahn: Sure. So I guess it is now mostly called crowdsourcing. It used to be called human computation. It’s something that I started doing maybe around 2001. The idea was to try to – there are a lot of problems that computers cannot solve very well. For example telling you what’s inside an image, something that computers cannot do very well. So determining whether an image has a cat or a dog in it, a computer just can’t do it very well. But humans can.
So the idea was and still is, it still works, the idea is to try to get a lot of people on the internet to help us solve a lot of the problems that computers cannot yet solve. Nowadays a lot of this is called crowdsourcing where you think of a problem, a large scale problem and then you try to get people online to help solve it for you. One of the first things that I did with that was a game that was called the ESP game and later it got acquired by Google. It could be called the Google Image Labeler where the idea was it was a game that was really fun.
It was played by several millions of people and as they were playing it they were helping to improve image search on Google. So they were actually taking random images from the web and writing descriptions for them that could be used to improving the search on the web.
TechFaster: Interesting. Beyond Duolingo which is very cool, what do you see some of the future implications of crowdsourcing being?
Dr. Luis von Ahn: Well, there’s a lot of different types of applications. I think the idea is just we’ll see what we can do when we can put millions of people on the internet working for a common task. Over the years we’ve seen a lot of really good examples of this. Wikipedia is a good one. Another one is my previous work with reCAPTCHA where the idea was while people were using reCAPTCHA they were helping us to digitize books and it’s actually been quite successful. It’s digitizing about 2 million books a year, just people over the internet helping to digitize about 2 million books a year. So I don’t know exactly what’s going to happen in the future, but I think there’s still a lot more application of it.
TechFaster: That’s great. So how many users are you guys up to on Duolingo?
Dr. Luis von Ahn: Let’s see. That’s an interesting question because there’s multiple answers. So in total we have over a million. I don’t know exactly how much over a million, maybe 1.3 million or so, around and daily actives we have about a hundred and something thousand. So that’s the total number of people that come every day, about a hundred and something thousand.
TechFaster: That’s awesome. So you guys are transcribing the entire internet, translating it.
Dr. Luis von Ahn: Yeah.
TechFaster: That is neat. So in your article Games With a Purpose you talk about AI kind of learning from us. Do you foresee a point where artificial intelligence will be able to replicate human intelligence?
Dr. Luis von Ahn: Yeah, sure. I don’t know when that will be, but I don’t see why not. I think it will happen at some point. It may take a while, but there’s no fundamental limitations.
TechFaster: Interesting. What’s the next step for Duolingo? What is it, Portuguese, German, Spanish, French and am I missing one?
Dr. Luis von Ahn: Italian.
TechFaster: What’s next?
Dr. Luis von Ahn: Well, probably the big next thing that we’re going to do – well, we’re working on a bunch of different things. One of the bigger ones is we’re going to do an Android app. A lot of people ask us about it. So almost every time that we say anything online, so if we post on Facebook, we post on Twitter, whatever we do we get like 50 questions all of them the same saying where’s the Android app? So we’re working on it. Sorry it’s taken longer than the iPhone app. Partly it’s because it’s harder to develop for Android. So we’re working on that. We’ll have other languages, but I would say the big deal with other languages is we’re going to get other people to add more languages for us as opposed to us adding them. We’re very slow in adding other languages because it takes a lot of effort, but we’re going to hope – we’re hoping that pretty soon the community is going to start adding languages, helping us to add languages. So hopefully that will allow us to expand to a lot of languages.
TechFaster: I was wondering too when the different characters come into play? That might get a little difficult like Chinese.
Dr. Luis von Ahn: Yes. We don’t know exactly how that’s going to work. We’re working on it, but I cannot tell you exactly how it’s going to work because right now everything we do uses the same characters basically.
TechFaster: So of your user base currently, which language is the most popular.
Dr. Luis von Ahn: English, learning English. So I would say about 55% of the people who are on Duolingo are learning English. Then in order of popularity then comes Spanish then French then German, then Italian and then Portuguese. That’s what we’ve got.
TechFaster: What are you guys learning from this information that you’re collecting? Anything insightful?
Dr. Luis von Ahn: A lot. So one of the things that we’re doing that I’m probably most excited about here is that while we’re watching people learn, because see we have so many people learning and we watch everything they do, we watch everything that people do. So whenever they enter an answer, whenever they enter a wrong answer, what exact wrong answer they entered, what words they don’t know, what words they know. We watch everything and based on that we’re running a lot of diferent experiments about how to teach you better.
So if you use Duolingo right now, your version of Duolingo will be slightly different than everybody else’s because we’re running an individualized experiment on you and then we compare the results and then try to see who’s learning better and then using that we’re always improving the website. So to me that’s the most interesting thing is that something that we can really use to improve education that doesn’t – before education was online, before you could learn anything online, this was simply not possible. Trying to teach somebody. Somebody would write a book. The feedback loop is really slow for a book because you write a book then some people use it and then maybe five years later you write the second edition of the book.
That’s a really slow feedback loop whereas here if we have an idea that for example we could try – maybe it is better to teach you adverbs before adjectives in Spanish for example. We just have an idea, we can try it and literally within a couple of hours we can get the answer for that. So we can just definitively say it is actually better to do this concept before this other concept because people actually learn better. So using that I’m hoping that we’ll be able to really blow everything else out of the water in terms of how fast you can learn a language.
TechFaster: Well, this concept go into other subjects as well?
Dr. Luis von Ahn: Yeah. Right now we’re really only concentrating on language. Maybe other things later, but right now it’s just language for us. Usually everything I work on is just very specific stuff.
TechFaster: great. Well, that about wraps it up. Do you have any closing comments you want to leave us with? Any big announcements?
Dr. Luis von Ahn: No, nothing big. The only thing is we very recently just announced this study that we ran on how well people learn on Duolingo and it turns out – this is one of the things that I’m just really happy about. It turns out that people can learn – so we did a study about how fast people learn using Duolingo and it turns out in about 34 hours of using Duolingo, you get the equivalent as one semester of college.
To me what’s really impressive about that is that one semester of college takes a lot longer than 34 hours usually. It’s more than 34 hours of work. So this has that in a very real sense and according to the scientific study that we didn’t run it. It was an external set of scientists that ran it, shows that actually people learn faster online than offline in this case. It’s not super surprising.
My guess is that if you had a really, really good one on one tutor you could do better than our program, but usually college classes you’re sitting there with another 50 people and it’s not clear that you’re getting all that much attention whereas here the computer is giving you a lot of attention. You know it’s a computer, not a human, but it’s still giving you a lot of attention. So it’s not that surprising that you can do a little bit than offline which I’m very excited about.
TechFaster: That brings up the point. Do you see – maybe 50 years in the future, do you see like physical universities being a thing of the past?
Dr. Luis von Ahn: Yeah. I don’t know exactly if they’re going to be a thing of the past, but something has got to happen. The thing about universities, I’m also a Computer Science Professor so it’s not like I don’t know about this, but they’re very inefficient. There’s huge inefficiencies in terms of – the whole lecturing thing is quite inefficient. You’re talking to 100 other people that are sitting in your class. Half of them are asleep. The whole thing is quite inefficient. I think we can a lot better.
Right now it may just be the beginnings right now and there are many people saying well, online education doesn’t work, et cetera. There’s all kinds of things, but we’re just at the beginning and I see no reason why we can’t do significantly better mainly because we can measure what everybody is doing really closely whereas as opposed to in the offline world you simply can’t measure.
TechFaster: that is fascinating. I think the biggest barrier is getting rid of the blackboard.
Dr. Luis von Ahn: Oh my god, blackboard, yeah.
TechFaster: Terrible experiences with that. I hated the Spanish workbook where I had to like fill it out and it was painstaking. That was horrible. I think if Duolingo had been around when I was in Spanish class I would have really enjoyed Spanish rather than not so much.
Dr. Luis von Ahn: Yeah. We hear that all the time from people saying I wish Duolingo had been around when I was in high school. We hear that all the time.
TechFaster: Yeah. The Spanish workbooks are terrible. Great, well thanks so much for joining us, Luis.
Dr. Luis von Ahn: Sure. Thanks for having me.
TechFaster: Yeah, no problem.
Dr. Luis von Ahn: All right, bye guys.