Video Of The Week: Luis von Ahn on Technology Based Tutoring

Luis is the founder and CEO of our portfolio company DuoLingo. In this video, he explains how he thinks technology can replicate the experience of a one on one tutor.

#hacking education

Comments (Archived):

  1. pointsnfigures

    agree with his points about one on one tutoring, and also agree with his points we aren’t there yet in a static video format. Professor John Cochrane tried to teach a PhD level Asset Pricing MOOC at Chicago http://johnhcochrane.blogsp… Teaching language is easier because it can also be learned aurally. Teaching other topics are harder. But, we have to try because it’s not only the right thing to do-it’s better for all of us the more educated everyone becomes.

  2. Matt Kruza

    Completely agree with the premise. Certainly static video is not the answer, but honestly its almost like a binary tree of interactive videos? Small bit of video… then test / question.. if get 5 right move on, if not move it brings up 3 new short videos on the areas you struggle. Combine this with the human element after say three times of video failing (could be in person if you form a stud group, but the idea that just dawned on me is then you could have a remote tutor that steps in after video has failed after x number of times.) Curious of thoughts as I have long thought about the binary / updatable video interaction sequence.. but the idea of after 3 or 4 videos failing to teach you the concept brining in an on demand tutor for like 1 -10 minutes to get past that issue just popped in my head as I was typing this.. think there is some applicability there.

    1. Matt Kruza

      Alternatively instead of a remote teacher.. could be another “student” or “learner”.. potentially layered with the remote teacher for the most “extreme” times of being stuck. Through using this iterative model one teacher may be able to help 100 – 100 learn effectively at a time: 1. 1000 students start with the basic video : 70% get it right and move on 2. The 300 students who missed some questions get iterative videos (another 20% of the original – 67% of remaining get this after 3 more cycles) 3. The 100 or so left pair up with other students / learners and say 90 of them are able to help them learn 4. The remaining 10 (1% of original) need help of the tutor Obviously numbers will vary.. but wanted to clarify the conceptual model

      1. Mike Zamansky

        This is something of a “one room schoolhouse” model. I’ve used it successfully but you have to pick your spots (topics) and it only works with the right players. Even the best teachers don’t reach every student when using this pass down model, if you don’t have the right “advanced” students kids can spend a lot of time spinning their wheels.

        1. pointsnfigures

 is a place where content is created by community, and interaction takes place. Math/Science focused

        2. Matt Kruza

          Interesting analogy to one room schoolhouse. Would this maybe be different since you could be doing this on the internet with peer participation from the internet? Lot more talent to draw from? This could apply to students (younger), but this is more for adult continuous learning, and maybe high school and college / more independent study. Would take care of the lack of “advanced” students? (these students could either get free access to the materias and/or get paid as well to incent them to help others… and its also fun to be the “expert” or teacher in a peer set.. did that continually from middle school to college even without being paid)

  3. William Mougayar

    I hope he wasn’t implying that 1:1 is a replacement to classes. I think it’s an augmentation of learning.But at the end of day, the biggest outcomes are 100% related to how much a student puts into it, whether tutored or not. You have to be committed to putting time into learning.I once asked a friend of mine who is an executive at Berlitz how long it would take me to learn Italian. He said, it all depends on you & how much time and effort you are willing to put into it.

    1. awaldstein

      is there anything we do that that is not true of william?

      1. William Mougayar

        true, but maybe even more so for languages learning? it’s hard if done later in life.

        1. awaldstein

          willingness to learn is simply everything a product of public education not private nor ivy league that is the largest divide.If you are motivated you learn. the system doesn’t provide the motivation nor the incentive.

          1. LE

            willingness to learn is simply everything william.Depending on the subject, to be sure, learning (things which requires memorization) when you are older is harder, regardless of motivation.That doesn’t mean you can’t learn but there is no question that it takes more effort as we age. You pickup many things as you age but you for sure loose the ability to memorize. Which is really important in learning certain things. (The good news for me is I never had a good memory so I’ve developed crutches to help with that part that are helpful)Otoh as you get older you have more of a base of information to go on that makes it easier for you to learn certain things and to recognize patterns and suck information in in chunks. [1] I can process many things that I see in business or in computers [2] or in human nature as chunks so consequently I may very well be able to learn “better” than someone younger without that extensive base of information. Regardless of how quickly they can memorize. I will “get it” much quicker. Full machine speed as I like to say. That said I’d rather be young![1] A chunk is like knowing 212 is an area code so it doesn’t require memorizing “212” as “2” “1” “2”. Or a pattern like you do with wines (that I can’t).[2] For example when you have been using for computers a long time (relevant to programming as well I’m sure Mike Z would agree) you know that languages or machines do certain things a certain way and knowing that you just have to figure out the way that the new machine you are trying to learn (or language) does the thing that you learned how to do in the previous language. And with web search it’s almost trivial to do that “how do I read a text file in in perl” and so on.

          2. Rick

            “That doesn’t mean you can’t learn but there is no question that it takes more effort as we age.”.I not sure about that. I think it just takes interest. I recently learned a new programming language. I didn’t really want to learn a new programming language. I just wanted something to do for a couple weeks while the snow flies. The interest came for wanting something to do because I had very little to no interest in learning a new programming language. In fact I don’t even want to program anymore. I haven’t used the new language since I removed it from my system..What we face today with the education system is that it’s not customized to the customer. I recently was going to go back to college. I signed up and told them what I wanted to learn and more importantly what I *didn’t* want to learn..Within two weeks of started the classes we were starting to focus on most of the things I didn’t want to learn! I had to leave the program. Because unlike some kid who doesn’t know any better I wasn’t going to waste my time learning what they wanted to sell.

          3. kev polonski

            Statistically, for K-12 age kids, a stable home with an involved Dad and Mom is the single biggest factor affecting outcome, educational or anything else.

          4. awaldstein

            Then I was as lucky as they come and turned out fine.Economics simply didn’t matter in having a strong home and me finding my own drives from it.

          5. Tyler

            +1As the husband of a first grade teacher at a title 1 school, it’s tough to listen to her talk about many of her kids’ home life, or lack thereof.

  4. Richard

    stat of the year: Productivity growth in education since 1980 is 1/100th of that of consumer electronics! (As measured by CPI)

    1. Matt Kruza

      baumol’s cost disease in action. Probably negative productivity growth in medical procedures? (I get there is new tech in medicine.. but forget cost productivity).… . The flip side of this is that most of decent paying job growth has been in education and medical works.. so..

      1. Richard

        Flip, Flip side is that Education and Medicine are two huge piles of debt for the private and public sectors

        1. pointsnfigures

          hugely regulated

          1. Matt Kruza

            agreed on regulation and on the huge piles of debt. Basically education and medical industry actually produce a decent product (very good medicine.. varying qualities on education.. but pretty decent). But the issue is cost. Biggest cost is people… 10 million in education and 15-20 million in health care. So obviously those 30 millon or so care about wages (as totally makes sense) but the other 270 million consumers of those services want / need lower prices. Its not technology.. almost all political and around distribution of assets now… that’s why it is so hard for startups to “compete”.. .but over the next decade I think a tipping point will be reached

  5. Mike Zamansky

    A worthy goal but we’re not there yet. Over the past few months, I’ve witnessed first hand groups of VERY bright motivated individuals working to teach themselves to code using a variety of resources including MOOC and popular online learning sites.The results have been discouraging. In each case the learners made progress but the key was that they ultimately had access to people to help them along the way.Right now, all of these technologies are great for autodidacts and as supplements when a teacher is available.My fear is that municipalities looking to cut costs will continue to see these as a mechanism for replacing teachers not supporting them which will leave the have nots with even less.

    1. LE

      I agree with you. One of the things a teacher (or someone knowledgeable) can answer is whether you are even studying the correct thing to get you to where you need to go in the first place.

  6. LE

    I like the “if you have money you can just find the best tutor in the world and they can teach you” (roughly). Exactly how does someone go about finding “the best tutor in the world” exactly? Not to mention that people that are “the best at anything” typically aren’t tutors. (A good reason to allow unpaid internships however.)

  7. LE

    he explains how he thinks technology can replicate the experience of a one on one tutor.I watched the video but I don’t see at all how he explained how technology can replicate the experience of a one on one tutor. All I heard is how duolingo saying they wanted to be “as good as a one on one tutor” but not how they were going to do that or why they think they do that now.

  8. Kirsten Lambertsen

    To reply to some of the comments here, I took the context of this video to be: if you need/want to buckle down and learn something very specific as quickly as possible, tutor beats classroom. It’s probably hard to argue with that. I would be surprised if he is actually arguing for the end of the classroom, altogether.

  9. Vasudev Ram

    I had blogged briefly earlier about Duolingo and Luis, here:…Checked out his profile then, interesting, and many achievements:…- including work on CAPTCHAs, crowsourcing, and being the inventor of reCAPTCHA and founder of DuoLingo.I am taking a DuoLingo language course on the web and found the experience good.It is not yet like a human tutor, though. That’s a tough goal.

  10. Nicolas Rodriguez M

    Thanks for reminding me of this app Fred, gotta rust of my german and already got my wife into english. She told: “The most interesting thing about this app is the absence of pressure from classmates (anxiety).”

  11. normal guy

    Chris Sacca said Google is second to none to solve science problems, it sucks solving social. I think foreign languages is a difficult subject to be completely technology driven because of the heavy social aspect involved. The components that are present in a foreign language are constantly changing making it unique when compared to other subjects. Unless one-on-one technology tutoring has been replicated for other exact science related subjects such as math & physics we are decades away from thinking about having a foreign language technology driven one-on-one tutoring. Think about a few colorful topics that are included in foreign language instruction:Slangs – Every country has its own and even native speakers can’t keep up with them.Culture, Traditions & Customs – These are difficult things to be explained by a machine. The machine can teach you when the Independence Day is, it cannot answer the thousand questions a student can come up about the Independence Day.Accent – Every country has a different accent for every language spoken out there. It is easier for a tutor from the US who is also fluent in Italian to teach English to an Italian. It is much harder for a Chinese tutor to teach English to an Italian even if the tutor speaks English fluently.4) Phrasal verbs. Google Translate project is in its eighth year, it can translate ‘1 word’ from a language to another but it massively fails to translate ‘2 words’ (hence all of the Google Translate parodies).I don’t want to sound pessimist as technology can change and definitely can improve but there are certain other innovations waiting to happen before technology replicates a one-on-one foreign language tutoring. I am just a guy who speaks 6 languages working on his own foreign language solution, not a tutor 🙂