WEBVTT - Duolingo CEO Luis Von Ahn Talks AI Powering Doulingo to All Time Highs

0:00:00.000 --> 0:00:01.880
<v Speaker 1>All right, we want to switch gears here and talk

0:00:02.000 --> 0:00:05.640
<v Speaker 1>about Dual Lingo. AI innovations powering Dual Lingo to all

0:00:05.760 --> 0:00:09.280
<v Speaker 1>time highs. The company revealing this week two new AI

0:00:09.440 --> 0:00:12.760
<v Speaker 1>powered innovations to the platform, which of course teaches over

0:00:12.880 --> 0:00:16.360
<v Speaker 1>forty languages as well as math and music. Now, since

0:00:16.400 --> 0:00:19.040
<v Speaker 1>the announcement, the Saka Sine price target hikes from both

0:00:19.079 --> 0:00:21.800
<v Speaker 1>evercore and need them and for more. I am thrilled

0:00:21.840 --> 0:00:23.720
<v Speaker 1>to say we are joined now by Luis van On.

0:00:23.880 --> 0:00:28.160
<v Speaker 1>He is the co founder, CEO and chairman of Dual Lingo.

0:00:28.320 --> 0:00:31.360
<v Speaker 1>So of those two innovations, let's start with video call

0:00:31.720 --> 0:00:35.240
<v Speaker 1>with Lily. My understanding is basically it's a video call

0:00:35.440 --> 0:00:37.919
<v Speaker 1>and I will be speaking with one of your characters.

0:00:37.920 --> 0:00:40.320
<v Speaker 1>She has purple hair, her name is Lily, and the

0:00:40.400 --> 0:00:43.800
<v Speaker 1>goal really is conversation. And with that in mind, are

0:00:43.800 --> 0:00:47.920
<v Speaker 1>you trying to target more advanced language learners who maybe

0:00:47.960 --> 0:00:50.600
<v Speaker 1>already have a handle on conjugations and how to ask

0:00:50.600 --> 0:00:51.120
<v Speaker 1>for the check.

0:00:53.000 --> 0:00:54.480
<v Speaker 2>Yeah, we're trying to target everybody.

0:00:54.480 --> 0:00:57.720
<v Speaker 3>The idea is, historically we haven't been able to get

0:00:57.760 --> 0:01:01.720
<v Speaker 3>people to practice conversation very well with couters, but now

0:01:01.800 --> 0:01:04.679
<v Speaker 3>with AI, you actually can have a conversation with a

0:01:04.920 --> 0:01:10.400
<v Speaker 3>fully automated character, and the character adjusts to your level. Now,

0:01:10.560 --> 0:01:13.039
<v Speaker 3>what we're really excited about is that it turns out

0:01:13.040 --> 0:01:16.640
<v Speaker 3>most people, actually, when they're just language beginners or intermediate,

0:01:16.760 --> 0:01:19.280
<v Speaker 3>they don't want to have a conversation with another human because.

0:01:19.040 --> 0:01:20.200
<v Speaker 2>They're too shy about it.

0:01:20.240 --> 0:01:23.160
<v Speaker 3>But having a conversation with an animated character actually allows

0:01:23.160 --> 0:01:25.880
<v Speaker 3>them to practice their conversation. So we're hoping this will

0:01:25.920 --> 0:01:28.200
<v Speaker 3>really bring to a lingo to the next level in

0:01:28.280 --> 0:01:29.320
<v Speaker 3>terms of teaching languages.

0:01:29.560 --> 0:01:32.520
<v Speaker 1>It's fascinating, of course, the human psychology element and the

0:01:32.600 --> 0:01:36.840
<v Speaker 1>decision to keep an animated character there as the voice

0:01:36.840 --> 0:01:40.160
<v Speaker 1>and face of Lily. So my understanding is that Video

0:01:40.240 --> 0:01:43.760
<v Speaker 1>Call with Lily basically it falls under your due Lingo

0:01:43.959 --> 0:01:47.560
<v Speaker 1>Max subscription thirty dollars a month. You just started rolling

0:01:47.560 --> 0:01:49.920
<v Speaker 1>that out in April on With this in mind and

0:01:49.920 --> 0:01:52.240
<v Speaker 1>with the launch of Video Call with Lily, should we

0:01:52.280 --> 0:01:55.640
<v Speaker 1>be expecting any upgrades to your forecasts here?

0:01:57.880 --> 0:01:59.920
<v Speaker 3>Well, I'm not really going to talk about forecasts or anything,

0:01:59.920 --> 0:02:02.800
<v Speaker 3>but we're really excited about, you know, about the uptake.

0:02:02.840 --> 0:02:05.400
<v Speaker 2>I think people are really loving being able to.

0:02:05.400 --> 0:02:08.680
<v Speaker 3>Practice their conversation, and this is particularly true of English learners.

0:02:08.760 --> 0:02:11.239
<v Speaker 3>So people who are learning English, they really care a

0:02:11.280 --> 0:02:12.200
<v Speaker 3>lot about conversation.

0:02:12.360 --> 0:02:13.520
<v Speaker 2>So we're excited about that.

0:02:15.320 --> 0:02:17.880
<v Speaker 1>And I mean to the point on English when we

0:02:17.919 --> 0:02:21.280
<v Speaker 1>spoke a couple months ago, I believe less than fifty

0:02:21.280 --> 0:02:24.840
<v Speaker 1>percent of your daily actor users we're English learners. I

0:02:24.880 --> 0:02:27.800
<v Speaker 1>know that is a big initiative for you. So with

0:02:27.880 --> 0:02:29.960
<v Speaker 1>that in mind, where are you when it comes to

0:02:30.000 --> 0:02:31.960
<v Speaker 1>reaching that fifty percent threshold.

0:02:33.080 --> 0:02:34.880
<v Speaker 3>Yeah, Actually, the threshold we want to reach is about

0:02:34.880 --> 0:02:36.680
<v Speaker 3>eighty percent. It turns out that the number of people

0:02:36.680 --> 0:02:38.280
<v Speaker 3>in the world, the fraction of people in the world

0:02:38.320 --> 0:02:40.079
<v Speaker 3>that are learning a language, that are learning English is

0:02:40.080 --> 0:02:43.000
<v Speaker 3>about eighty percent, whereas of our daily active users this

0:02:43.120 --> 0:02:45.880
<v Speaker 3>is under fifty percent. So we're hoping that, you know,

0:02:45.919 --> 0:02:47.880
<v Speaker 3>it'll get to eighty percent. This will take us a while,

0:02:48.480 --> 0:02:51.000
<v Speaker 3>but you know, we're doing everything we can. And one

0:02:51.000 --> 0:02:53.680
<v Speaker 3>of the important things is are these AI innovations that

0:02:53.800 --> 0:02:56.840
<v Speaker 3>allow us to practice more natural conversation.

0:02:57.800 --> 0:03:00.640
<v Speaker 4>I think what's interesting also about this AI revolution, and

0:03:00.680 --> 0:03:03.480
<v Speaker 4>as it pertains to language and companies that work on

0:03:03.600 --> 0:03:06.919
<v Speaker 4>language like you do, is what it takes to get

0:03:06.960 --> 0:03:10.120
<v Speaker 4>there and advance and grow. What's fascinating about dual LINGO

0:03:10.160 --> 0:03:13.240
<v Speaker 4>to me is that you guys have a nearly thirteen

0:03:13.280 --> 0:03:15.239
<v Speaker 4>billion dollar company that at the end of the last

0:03:15.320 --> 0:03:18.400
<v Speaker 4>year only had about or less than rather eight hundred

0:03:18.440 --> 0:03:21.560
<v Speaker 4>people working at it. What does it mean for the

0:03:21.600 --> 0:03:25.000
<v Speaker 4>workforce for you to be expanding this way in AI?

0:03:26.720 --> 0:03:27.600
<v Speaker 2>Yeah, we're quite lean.

0:03:27.680 --> 0:03:29.920
<v Speaker 3>We still have about it's a little over eight hundred

0:03:29.919 --> 0:03:32.679
<v Speaker 3>people that we have working on due LINGO, and most

0:03:32.720 --> 0:03:35.640
<v Speaker 3>of them are working on the actual product, so mostly

0:03:35.680 --> 0:03:39.400
<v Speaker 3>are their engineers. It just turns out that, you know,

0:03:39.440 --> 0:03:42.240
<v Speaker 3>engineers are quite scalable, and it's getting even better with

0:03:42.280 --> 0:03:45.680
<v Speaker 3>a lot of THEAI improvements because a lot of the

0:03:45.680 --> 0:03:48.520
<v Speaker 3>coding that used to be done entirely manually now can

0:03:48.560 --> 0:03:50.800
<v Speaker 3>be done in part with the help of a computer.

0:03:51.840 --> 0:03:54.480
<v Speaker 3>So we're going to continue growing the number of employees,

0:03:55.240 --> 0:03:57.840
<v Speaker 3>but I don't think you should expect us to quadruple

0:03:57.920 --> 0:04:00.280
<v Speaker 3>the employees anytime soon or anything like that. Is going

0:04:00.320 --> 0:04:03.559
<v Speaker 3>to slowly engine because we get a lot of efficiencies

0:04:03.600 --> 0:04:03.960
<v Speaker 3>with AI.

0:04:04.560 --> 0:04:07.600
<v Speaker 1>Well, continuing along these lines, it's interesting. Late last year,

0:04:07.640 --> 0:04:10.240
<v Speaker 1>of course, you just decided not to renew the contracts

0:04:10.280 --> 0:04:14.520
<v Speaker 1>about ten percent of your contracted workforce, opting instead to

0:04:14.600 --> 0:04:17.400
<v Speaker 1>use AI in some cases and with that in mind,

0:04:17.400 --> 0:04:20.200
<v Speaker 1>of course, AI not just enhancing the product that you offer,

0:04:20.279 --> 0:04:24.040
<v Speaker 1>but maybe enhancing the productivity of your workforce you expect

0:04:24.040 --> 0:04:26.679
<v Speaker 1>to grow the workforce. Where are you hiring right now?

0:04:28.240 --> 0:04:32.359
<v Speaker 3>I mean the main places we're hiring are highly skilled engineers,

0:04:32.440 --> 0:04:35.080
<v Speaker 3>product managers, and designers. I mean, the way we see

0:04:35.120 --> 0:04:38.640
<v Speaker 3>it is, you know, there's some road work that can

0:04:38.680 --> 0:04:41.840
<v Speaker 3>be entirely substituted by AI, But for the majority of

0:04:42.440 --> 0:04:45.520
<v Speaker 3>our employees or the types of works we do, what's

0:04:45.560 --> 0:04:47.120
<v Speaker 3>going to happen is we're just going to be able

0:04:47.160 --> 0:04:50.240
<v Speaker 3>to see each one of them get more efficient. So

0:04:50.279 --> 0:04:52.760
<v Speaker 3>we're not going to be doing anything like layoffs or anything.

0:04:52.760 --> 0:04:54.760
<v Speaker 3>We're just going to get a lot more efficiency because

0:04:54.800 --> 0:04:57.240
<v Speaker 3>what happens is somebody who used to be an animator

0:04:57.480 --> 0:05:00.240
<v Speaker 3>now can still be an animator.

0:05:00.240 --> 0:05:01.200
<v Speaker 2>They will be working on.

0:05:01.160 --> 0:05:03.200
<v Speaker 3>Kind of the more conceptual stuff as opposed to having

0:05:03.200 --> 0:05:05.239
<v Speaker 3>to spend a lot of time on you know, tiny

0:05:05.279 --> 0:05:08.680
<v Speaker 3>little movements of the characters, which now can be done

0:05:09.400 --> 0:05:10.039
<v Speaker 3>mostly by AI.

0:05:10.160 --> 0:05:11.840
<v Speaker 4>I'd love to go back to Katie's point a little

0:05:11.880 --> 0:05:14.680
<v Speaker 4>bit as well about kind of the higher price tiers

0:05:14.680 --> 0:05:17.400
<v Speaker 4>that you're offering people. What do you find people coming

0:05:17.440 --> 0:05:20.120
<v Speaker 4>to do a lingo for I'm a user myself. I

0:05:20.360 --> 0:05:22.960
<v Speaker 4>want to learn Spanish. I use it every so often.

0:05:23.000 --> 0:05:25.039
<v Speaker 4>I'm not using it in my job every day. Do

0:05:25.120 --> 0:05:28.920
<v Speaker 4>you find people learning at these higher price tiers to

0:05:29.320 --> 0:05:32.480
<v Speaker 4>get different jobs across the world? And are is that

0:05:32.640 --> 0:05:35.159
<v Speaker 4>a rationale for why people are willing to pay for it?

0:05:35.160 --> 0:05:36.480
<v Speaker 4>Will employers pay for it?

0:05:38.160 --> 0:05:38.400
<v Speaker 2>Yeah?

0:05:38.440 --> 0:05:41.160
<v Speaker 3>So there's there's many reasons for learning a language. The

0:05:41.279 --> 0:05:43.880
<v Speaker 3>two big ones that we have among our users, and

0:05:43.920 --> 0:05:47.840
<v Speaker 3>they're about half half. One half is hobby. These are

0:05:47.839 --> 0:05:50.400
<v Speaker 3>people who don't actually need to learn a language, but

0:05:50.440 --> 0:05:52.880
<v Speaker 3>they just think to themselves, well, uh, you know, I

0:05:52.960 --> 0:05:54.920
<v Speaker 3>used to, I don't know, play a lot of candy

0:05:54.960 --> 0:05:58.160
<v Speaker 3>crush or do a lot of Instagram, and now I'm

0:05:58.240 --> 0:06:00.159
<v Speaker 3>using Duolingo and at least I'm learning.

0:06:00.080 --> 0:06:02.320
<v Speaker 2>In some Spanish. That's one group.

0:06:02.400 --> 0:06:04.400
<v Speaker 3>And then the other big group are people who are

0:06:04.440 --> 0:06:07.680
<v Speaker 3>doing this to actually learn a language and doing so

0:06:07.839 --> 0:06:10.640
<v Speaker 3>for either school or work. And usually those are people

0:06:10.640 --> 0:06:12.800
<v Speaker 3>who are learning English. And so those are the two

0:06:12.960 --> 0:06:15.800
<v Speaker 3>big buckets. It's easy to see that the people who

0:06:15.800 --> 0:06:18.039
<v Speaker 3>are learning English are willing to pay because they really,

0:06:18.120 --> 0:06:20.200
<v Speaker 3>you know, they actually want to get a result. But

0:06:20.240 --> 0:06:22.960
<v Speaker 3>it turns out that the people who are hobbyists are

0:06:23.000 --> 0:06:25.520
<v Speaker 3>also interested in paying, because, as with all hobbies, you

0:06:25.600 --> 0:06:28.560
<v Speaker 3>usually get started with something you know, you get started small,

0:06:28.640 --> 0:06:30.200
<v Speaker 3>but as you get more and more into it, you

0:06:30.640 --> 0:06:33.440
<v Speaker 3>want to have more features and then you end up paying.

0:06:33.480 --> 0:06:36.360
<v Speaker 3>So we're seeing appetite to pay in both big groups.

0:06:36.600 --> 0:06:38.560
<v Speaker 1>I'm glad you brought up candy Crush. I'm not much

0:06:38.600 --> 0:06:40.680
<v Speaker 1>of a candy Crush player. I really like two dots,

0:06:40.680 --> 0:06:43.599
<v Speaker 1>but that's a different conversation. But as part of of

0:06:43.640 --> 0:06:45.920
<v Speaker 1>course the innovations that we're announced this week, we mentioned

0:06:46.000 --> 0:06:49.560
<v Speaker 1>video Call with Lily. There's also Adventures, which are a

0:06:49.640 --> 0:06:52.239
<v Speaker 1>mini game Louis. So it's interesting that you're also trying

0:06:52.240 --> 0:06:55.120
<v Speaker 1>to tap into that gaming mindset. Maybe people on the

0:06:55.160 --> 0:06:57.640
<v Speaker 1>subway just want to have some fun here.

0:06:59.200 --> 0:07:02.040
<v Speaker 3>Yes, it's always been gamified, and I think that's one

0:07:02.040 --> 0:07:03.760
<v Speaker 3>of the big reasons why we've grown so much. It's

0:07:03.800 --> 0:07:05.440
<v Speaker 3>because it turns out that learning a language and do

0:07:05.520 --> 0:07:08.120
<v Speaker 3>ling is easy and fun. But we're really leaning into it,

0:07:08.160 --> 0:07:10.080
<v Speaker 3>and it turns out there is one aspect of learning

0:07:10.080 --> 0:07:13.480
<v Speaker 3>a language, which is these situational things where you're you know,

0:07:13.520 --> 0:07:16.760
<v Speaker 3>you may have to buy a croissant, or you know,

0:07:16.800 --> 0:07:18.520
<v Speaker 3>whatever it is get yourself out of trouble, that you

0:07:18.520 --> 0:07:21.440
<v Speaker 3>can really turn into a little bit of an adventure game.

0:07:22.240 --> 0:07:24.960
<v Speaker 3>So some of the lessons are now going to be

0:07:25.120 --> 0:07:27.280
<v Speaker 3>as opposed to just having to do certain you know,

0:07:27.320 --> 0:07:29.680
<v Speaker 3>translations or listening to things, You're going to be put

0:07:29.720 --> 0:07:33.080
<v Speaker 3>into a full screen where you're in a situation where

0:07:33.120 --> 0:07:36.160
<v Speaker 3>you may have to buy something or you know, clean

0:07:36.240 --> 0:07:39.720
<v Speaker 3>up something, or ask for directions or something, and you know,

0:07:39.800 --> 0:07:41.280
<v Speaker 3>it's a whole game that you can that you can fail,

0:07:41.320 --> 0:07:42.160
<v Speaker 3>and it's pretty fun.

0:07:42.400 --> 0:07:43.000
<v Speaker 2>And we don't have.

0:07:43.000 --> 0:07:44.560
<v Speaker 1>Much time left, but I do want to talk a

0:07:44.600 --> 0:07:46.840
<v Speaker 1>little bit about the stock movement. Of course, do a

0:07:46.880 --> 0:07:49.960
<v Speaker 1>lingo shares hitting an all time high this week, up

0:07:50.080 --> 0:07:52.000
<v Speaker 1>mightily for the year. You take a look at what

0:07:52.000 --> 0:07:54.280
<v Speaker 1>that means for market cap. Ashanal I mentioned you're close

0:07:54.320 --> 0:07:57.320
<v Speaker 1>to thirteen billion dollars in market cap right now. You

0:07:57.360 --> 0:08:00.160
<v Speaker 1>were under seven billion a year ago. I know that

0:08:00.200 --> 0:08:02.880
<v Speaker 1>you have plans to hire more, but is there any

0:08:02.920 --> 0:08:04.440
<v Speaker 1>potential M and A in your future?

0:08:06.800 --> 0:08:10.320
<v Speaker 3>I mean, look, we're always considering, you know, how to

0:08:10.360 --> 0:08:12.920
<v Speaker 3>expand to different things, as you mentioned, we are you know,

0:08:12.960 --> 0:08:16.960
<v Speaker 3>we historically have mainly taught languages. We of late now,

0:08:17.000 --> 0:08:19.840
<v Speaker 3>also teach math and music. We're always seeing what other

0:08:19.880 --> 0:08:22.000
<v Speaker 3>subjects we can teach, and there may be some M

0:08:22.000 --> 0:08:24.760
<v Speaker 3>and A opportunities there. However, you know, there's there's nothing

0:08:24.880 --> 0:08:26.960
<v Speaker 3>very specific that we're looking at. I mean generally, we're

0:08:27.000 --> 0:08:30.160
<v Speaker 3>just doing our thing and you know, getting better at

0:08:30.280 --> 0:08:32.640
<v Speaker 3>our main market, which is teaching languages.

0:08:32.880 --> 0:08:34.920
<v Speaker 1>All right, Louis, that's a good place to leave it.

0:08:35.000 --> 0:08:38.079
<v Speaker 1>Really appreciate your time today. That, of course is Louis

0:08:38.160 --> 0:08:40.599
<v Speaker 1>von On. He is the CEO of Doueling Go