Duolingo CEO Luis von Ahn thinks AI has a lot to teach us

Decoder · · Duration 1:19:55

Luis von Ahn says Duolingo’s Lily adapts to a learner’s level and remembers earlier conversation topics. He sees large language models as helpful for teaching, while keeping learners engaged remains difficult. Read 3 viewpoints with supporting evidence and source links.

Understand this piece

3 key points

Synthesis

  1. Lily adapts conversations to the learner’s level

    Von Ahn says Lily adapts to a learner’s level. Duolingo knows the learner’s level from their prior learning on the app.

    Supporting evidence 1

    Original excerpt

    First of all, she adapts to your level — we know your level because you’ve been learning on Duolingo, so we have a pretty good idea of what your level is.

    Luis von Ahn · Publisher transcript paragraph 8

    Context

    We have this cast of characters that our users love. One of them is an emo teen with purple hair who is very unimpressed by you. You can talk to her now and practice your conversation skills, and you can have really good conversations with her. There are a lot of things that are amazing about her.

    Read in source context →

    Continue exploring

    AI personalization →
  2. Lily remembers topics from earlier conversations

    Von Ahn says Lily remembers previous conversation topics. In his example, she remembered that he had told her he liked Nirvana.

    Supporting evidence 1

    Original excerpt

    The other thing is that she has memory, so she remembers the last time you talked about something. For example, I just had a conversation today where she remembered that, last time, we talked about the fact that I like Nirvana.

    Luis von Ahn · Publisher transcript paragraph 9

    Context

    She was telling me that her favorite song is “Smells Like Teen Spirit.”

    Read in source context →

    Continue exploring

    AI memory →
  3. Von Ahn says LLMs teach well but are not good at engagement

    Von Ahn says large language models have already helped Duolingo and will continue to do so, but do not solve all its problems. He says they are good at teaching and not good at engagement, which he calls the hardest part of education.

    Supporting evidence 1

    Original excerpt

    It’s somewhere in between. I do think that large language models are going to be very positive for Duolingo — they already are, and I think they’re going to continue being very positive. What is not true is that large language models solve all our problems. One of the biggest issues that people aren’t talking about, particularly with education, is that large language models are good at teaching you stuff. They’re not good at engagement. And that’s the hardest thing with education.

    Luis von Ahn · Publisher transcript paragraph 75

    Read in source context →

Key moments3

Short, attributed passages with the context to verify them. The full conversation stays with its publisher.

AI personalization

Lily adapts conversations to the learner’s level

Original excerpt

First of all, she adapts to your level — we know your level because you’ve been learning on Duolingo, so we have a pretty good idea of what your level is.
Context

We have this cast of characters that our users love. One of them is an emo teen with purple hair who is very unimpressed by you. You can talk to her now and practice your conversation skills, and you can have really good conversations with her. There are a lot of things that are amazing about her.

AI memory

Lily remembers topics from earlier conversations

Original excerpt

The other thing is that she has memory, so she remembers the last time you talked about something. For example, I just had a conversation today where she remembered that, last time, we talked about the fact that I like Nirvana.
Context

She was telling me that her favorite song is “Smells Like Teen Spirit.”

AI limitations in education

Von Ahn says LLMs teach well but are not good at engagement

Original excerpt

It’s somewhere in between. I do think that large language models are going to be very positive for Duolingo — they already are, and I think they’re going to continue being very positive. What is not true is that large language models solve all our problems. One of the biggest issues that people aren’t talking about, particularly with education, is that large language models are good at teaching you stuff. They’re not good at engagement. And that’s the hardest thing with education.

Source & methodology

These viewpoints are linked to their original sources. Paraphrases are labeled and are not verbatim quotes.

Open transcript or source material (opens in a new tab)Report an issue

Explore these viewpoints by person