Luis Von Ahn

Whether you do it to an open AI or whether you do it through Microsoft or whatever, the price is going down for a number of reasons. Everything's getting more efficient, and also chips are getting cheaper over time, et cetera. So the price is going down. At the moment, there's a good amount of cushion, but we expect that over time there will be an even larger amount of cushion.

Whether you do it to an open AI or whether you do it through Microsoft or whatever, the price is going down for a number of reasons. Everything's getting more efficient, and also chips are getting cheaper over time, et cetera. So the price is going down. At the moment, there's a good amount of cushion, but we expect that over time there will be an even larger amount of cushion.

So at least for our application, I'm not particularly concerned in terms of margins. I think for our application, the margins work out.

So at least for our application, I'm not particularly concerned in terms of margins. I think for our application, the margins work out.

So at least for our application, I'm not particularly concerned in terms of margins. I think for our application, the margins work out.

I think the answer is yes, but I think the operative term is somewhat. They are somewhat interchangeable. So we've tried to build our technology stacks so that you can interchange them. But the reality is that you interchange them and you start getting wonky stuff because you probably spent a lot of time really testing your way into the right queries. You may have done some fine-tuning, etc.

I think the answer is yes, but I think the operative term is somewhat. They are somewhat interchangeable. So we've tried to build our technology stacks so that you can interchange them. But the reality is that you interchange them and you start getting wonky stuff because you probably spent a lot of time really testing your way into the right queries. You may have done some fine-tuning, etc.

I think the answer is yes, but I think the operative term is somewhat. They are somewhat interchangeable. So we've tried to build our technology stacks so that you can interchange them. But the reality is that you interchange them and you start getting wonky stuff because you probably spent a lot of time really testing your way into the right queries. You may have done some fine-tuning, etc.

So you can interchange them, but if you do, you probably need to spend a few months making sure that the wonkiness goes away.

So you can interchange them, but if you do, you probably need to spend a few months making sure that the wonkiness goes away.

So you can interchange them, but if you do, you probably need to spend a few months making sure that the wonkiness goes away.

It's somewhere in between. Let's see, I do think that large language models are going to be very positive for Duolingo, already are, and I think they're going to continue being very positive. What is not true is that, oh my God, large language models just solve all our problems.

It's somewhere in between. Let's see, I do think that large language models are going to be very positive for Duolingo, already are, and I think they're going to continue being very positive. What is not true is that, oh my God, large language models just solve all our problems.

It's somewhere in between. Let's see, I do think that large language models are going to be very positive for Duolingo, already are, and I think they're going to continue being very positive. What is not true is that, oh my God, large language models just solve all our problems.

I mean, one of the biggest issues that people don't seem to, or at least people are not talking about that, particularly with education when it comes to large language models, is large language models are good at teaching you stuff. They're not good at engagement. And that's the hardest thing with education.

I mean, one of the biggest issues that people don't seem to, or at least people are not talking about that, particularly with education when it comes to large language models, is large language models are good at teaching you stuff. They're not good at engagement. And that's the hardest thing with education.

I mean, one of the biggest issues that people don't seem to, or at least people are not talking about that, particularly with education when it comes to large language models, is large language models are good at teaching you stuff. They're not good at engagement. And that's the hardest thing with education.

I mean, the hardest thing about me trying to teach you something is just keeping you engaged. Somehow people forget, you know, some people saying like, oh my God, you can learn quantum physics with ChatGPT. And yeah, sure. But that's just not that impressive. You can learn quantum physics with a book. The technology to learn this has been around for a long time. It's called a book and it works.

I mean, the hardest thing about me trying to teach you something is just keeping you engaged. Somehow people forget, you know, some people saying like, oh my God, you can learn quantum physics with ChatGPT. And yeah, sure. But that's just not that impressive. You can learn quantum physics with a book. The technology to learn this has been around for a long time. It's called a book and it works.

I mean, the hardest thing about me trying to teach you something is just keeping you engaged. Somehow people forget, you know, some people saying like, oh my God, you can learn quantum physics with ChatGPT. And yeah, sure. But that's just not that impressive. You can learn quantum physics with a book. The technology to learn this has been around for a long time. It's called a book and it works.