← ASU+GSV Summit
ASU+GSV Summit
How AI Can Transform Math Learning with Goblins Founder Sawyer Altman | GSVtv Podcast
00:18:07
Loading summary
Transcript61 lines
- [00:00]
A
Foreign.
- [00:07]
B
We are here at the ASU GSV Summit 2026 with Sawyer Altman, CEO of goblins. It's a GSV Cup 50 cup company this year and doing incredibly interesting work with math education using AI. Sawyer has a really interesting background. He founded and sold a charitable energy drink company. He founded an education nonprofit. He was a founding designer at X1. He's taught storytelling. He's taught computer science. He's taught design. He's taught restorative justice. He's taught constructivist teaching to over a thousand students of all ages in 12 countries, most recently in New York City and now with Goblins. This is a tool that lets kids speak, draw and get help from live avatars. Welcome to the podcast. Thanks for having me. Yeah. So first off, let's start. I just named some of this very wide background. You've done a lot of different things. What did being in the classroom teach you about ed tech? And what sort of was the spark that led to found God bless.
- [01:07]
A
Yeah, yeah, Good question. And with all this, that whole path I took there, you know, it makes a lot more sense in retrospect. A lot of our lives in the innocence, but here we are. Yeah. You know, I think the one thing I came to appreciate from my experience teaching was, and I recommend really everyone have that experience in their lifetime is extremely informative whether you go into edtech or not. I just came to appreciate that a single classroom is such a cross section of so much of society. You inherit your students, socioeconomic status, what the previous teacher in your subject left them with, whether it's foundational gaps or certain attitudes that they've inherited and so on. What kids had for lunch that day, you know, if they had tater tots, they're going to be a little sleepier. You're going to get that and so on. There's just so much complexity that you're dealing with. And certainly that manifested in the problem that led to Goblins more concretely in that I had 30 kids in my class and there was one of me and their heads were just all over the place.
- [02:10]
B
Yeah.
- [02:11]
A
And so I really wished I had amazing experiences with my students, working with them. One on one. I can identify specific misconception. I could start filling in those gaps. But, you know, there was just not enough of me to go around. Yeah. I was looking for something that could really amplify me and ultimately I didn't see anyone building it.
- [02:30]
B
Yeah. So not enough of you to go around. And you sometimes refer to Goblins as a math teacher, cloning Device. Right. Almost like the transmogrifier. Right. The Calvin and Hobbes. You can make multiple pieces of yourself to go around. So what does that mean? You have this teacher avatar in the system and you use AI to really duplicate and amplify the effect of a great teacher. What does that look like?
- [02:52]
A
It's a great question. And Calvin and Hobbes I think is an apt comparison. It's meant to be a little tongue in cheek, Right?
- [02:57]
B
Sure.
- [02:57]
A
You know, one of the things we want to say to teachers is like, hey, this is going to make your life easier. Often we'll say to teachers like, imagine you could be in 30 places at once. And that's kind of like a holy Grail. It's like, no, it's literally impossible. We all know it is like technically impossible. Yeah. But like this notion that we could get you closer to that is appealing. Okay, well how would you do that? Is often what saying that opens them up. So.
- [03:18]
B
Exactly.
- [03:19]
A
And so, yeah, you mentioned the cloning feature. The key thing with Goblins is we're able to get close to what a great teacher does, which is they can look at a student's whole work just authentically as the student has shown up their whole process. And they can make those inferences that I was talking about working one on one with students and then they could build their conceptual foundation with them. And essentially we're doing that same thing in real time over authentically what the student has shown us on their whiteboard or through this open ended conversation. And that is, you know, that takes an enormous burden off the teacher in that we're able to do a lot of that differentiation in real time. We're also, because we can see the whole process, we can gain insight into where their class is at with much higher fidelity and way less effort actually than if they were to manually look at all of the final answers the students submitted and were like, grade it one by one. And so there's a lot of ways that we can provide that application effect. And then to the point of you mentioned this cloning concept. We like teachers take a selfie, record their voice for 30 seconds, and then they've got this little Pixar, like avatar.
- [04:24]
B
Right?
- [04:25]
A
It's deliberately Pixar. Yeah, yeah. It's certainly meant to be. Again, a little tongue in cheek. But part of that is helping drive home this idea of like, hey, this is an extension of you. And that's a little gimmicky admittedly, but kind of helps bring the whole thing together. And then the second bit Is, you know, we really do believe strongly in the connection that the teacher and the student. Yeah. I mean, really, with just about everything. With Goblins, the first thing we've tried to solve for is the motivation, the student engagement, and this notion that the student is in a context in which they feel like they belong, in which they feel like, hey, someone cares about me and my success. Yeah, we think that's huge. We serve. We want to amplify that. Certainly quite the contrary of that possibly going away in the age of AI, we would like. We would not like that to happen. We want to reinforce that connection. And so, yeah, in a way, this concept helps maintain that connection, helps teachers sort of step into this new future more so on their own terms.
- [05:26]
B
I think it's very intuitive when you see it in action. Right. You just did the pitch at the GSV cup today, and there's sort of a junior Mr. Altman in the corner that looks like you as your glasses and haircut. And. And if.
- [05:36]
A
If.
- [05:37]
B
If Mr. Altman, is your math teacher in school having a sort of cartoon version that gives you advice that can respond to your process. When you show your work and you're writing or talking to. To Goblins, it can respond to you. I think I. I imagine for students, that seems very natural.
- [05:51]
A
Although I. I didn't like to be called Mr. Altman, I was always Sawyer to them.
- [05:54]
B
Fair enough.
- [05:55]
A
It always felt a little, you know, the whole world. There's somebody.
- [05:58]
B
A junior Sawyer, the math teacher. There you go.
- [06:01]
A
But, yeah, the key is, the hope is it feels authentic to the teacher also.
- [06:06]
B
Exactly. And it builds on the relationship that's already being built in the classroom over time. The built relationship of trust and mentoring and all the things you want in a classroom relationship. So for those who haven't seen Goblins in action, the avatar is one part of it, but a big part of it is the process, which you named. It's turning. Instead of just a final answer on a worksheet, it's a student is recording their voice. They're doing whiteboard. They're explaining their thinking. You tell our listeners what it actually looks like.
- [06:32]
A
Yeah. Well, the student could be working on content that the teacher is assigned. They can search by standards, align content, curriculum line content. Just pull unit four, lesson three, right off the shelf. Here's the problem set. And so easy for them to kind of push the content into the student's hands. Or we have a mode, we call it super adaptive mode. We kind of learn the student over time. We've seen where they Might have struggled before. And then we can just start filling in the gaps.
- [06:57]
B
Great.
- [06:57]
A
Anywhere we notice there's an issue. And then in practice, you know, it could look a bunch of different ways. Depends on the student. And that's kind of the beautiful thing about it. It's very flexible and it's very individualized. You know, we, a student could start a problem by saying, I don't know, I don't know repeatedly. We have a student who could say, I've never done this before, et cetera. And so it can go in a bunch of different directions. Or you could have a student who's shown most of the work but tripped up on their third operation line number three. And so we're going to circle that. We'll say like, hey, you were doing great up until here. You got the right idea. But just like check that arithmetic. And so we can figure out, is it a misconception? Is it just a simple operational error? And then we can flow with them in real time. Hopefully. The feel is that it is like having your teacher there. And a lot of students say it's an incredible thing. If I'm at home, I may not have, especially as you get into the higher level math, I may not have parents who know how to do that or that or remember what it was like. The whole thing about new math and math has changed. They feel like, finally, here's someone that supports me. And some students might be in class, they feel shy. They feel like maybe they don't have that awesome connection with the teacher. Also, they have someone who doesn't judge them and is patient. And it seems like, wow, they, this thing really gets me. Yeah, yeah.
- [08:21]
B
You mentioned, you know, flowing in real time or real time differentiation. I feel like that's a real through line and a sort of a real key feature of what Goblins is doing. It's, it's responsive, it's giving feedback in adaptive feedback in super real time. You don't have to wait like those of us who went to school before AI, you'd give your math homework and you'd get it back three days later with a red pen on it. That's. We don't live in that world anymore. And I think that's very good for. From a learning science perspective. Tell us what the real time means to you.
- [08:47]
A
Yeah, well, you know, this previous. We've been looking for individualization and personalization and education for a long time. You guys talking about this since computers, you were first conceived?
- [08:57]
B
Practically, yes.
- [08:58]
A
And especially starting in the 2010s or so it became possible using machine learning algorithms to get closer to figuring out what the students should be working on. But it always was just predicting the next problem the students should be working on.
- [09:16]
B
Right.
- [09:16]
A
And it wasn't really, you know, genuine instruction. Yeah. And so a lot of students, you know, you could get something wrong. You know, it's basically their algorithm is, you know, did you get it right? Did you get it wrong? Get it right, get it wrong. Right. And it would kind of push you back or slide you up. And a lot of these students that say you're an algebra one and you're really struggling with all these foundational skills, you could end up in this long demoralizing slide all the way back to fourth grade.
- [09:36]
B
That's right.
- [09:36]
A
And first of all, you know, now you feel pretty lousy. And then it doesn't really help you build your way back up. You're just showing you the next problem. It takes an enormous amount of motivation for that student to start climbing again.
- [09:48]
B
That's true.
- [09:48]
A
On their own. And so an incredible thing about large language models which this new technology which is key to how Galvans works is that we can have this open ended conversation and when we notice a student is significantly struggling, we can create mini lessons that we can actually start filling in the blanks that we can. The students feel like, wow, I actually. This thing can support me. Yes. It's not just going to turn me down.
- [10:12]
B
Yeah. And get back to grade level as quickly as possible. Rather than that. I like that phrase, the demoralizing slide. I think we underestimate how demoralizing it really is and how it can cause like a lifelong disinterest in any subject. But especially unfortunately, that's where it often happens.
- [10:28]
A
Yeah. You know, technology is a lot about timing. You know, the notion of personalization, individualization, it's an extremely attractive idea and I think it's just really becoming possible.
- [10:42]
B
I agree. We've talked about personalization or differentiate, individualization, differentiation. There's like 10 synonyms for a decade or more in edtech and I feel like we are finally seeing it in real time. It's amazing. Well, one thing that differentiates you as a founder is you also have worked with the Gates Foundation. That's not something as a small startup. That's usually not the type of, type of funder and the type of partner you always look at. Tell us how that came about and what other entrepreneurs who are starting out might see from a funder like that as opposed to a, to an early stage vc.
- [11:13]
A
Yeah, yeah. It's a great question. Yeah, we're extremely fortunate. We're very grateful to the Gates Foundation. I mean when it was actually here's a plug for Asugsville here at the conference, I had showed up at the end of an event, I had a phone demo literally on my phone. We actually started developing for the phone originally. And I said, hey guys, you got two minutes. And they checked it out and they said, hey, we've got a, I think it was an RFI coming out. And they said, send it in. That was interesting. And they took a shot on us when we were, you know, two people, literally, me and my co founder.
- [11:44]
B
Yeah, it's amazing.
- [11:45]
A
It was wonderful. And that's, I mean, you know, the edtech is a tough market. Yeah. And the appetite from your venture is, especially from the generalists, is partly there from the EdTech folks. You know, the, a lot of them want to see your traction and a great product demo isn't enough. And from the Gates foundation point of view, they saw it and they said, hey, I think this could be great pedagogy. And we want to see this turn from this is a great demo, but I want to, is it low latency, is it high quality? How does it work at scale, all these kind of things. They wanted to see that happen.
- [12:22]
B
It's amazing. You have said you want to inspire a billion kids to love to learn. It's a massive ambition. You're still a pretty small team. What's your theory of change about how this type of education, this real time differentiation with maybe teacher avatars with responsive adaptive feedback, how can this change how math education works at that capacity?
- [12:44]
A
Yeah, it's a great question. You know, I realize a billion is, you know. Right. But it changes the way you think about the problem about a billion. Like inherently to reach a billion you have to do something ridiculously scalable. And so it kind of purifies. It's not the same as we're going to do this for 10,000, we're going to do this for a million, which is still an ambitious number.
- [13:05]
B
It is.
- [13:06]
A
And so I think what it centers us on as a company is we're pretty obsessed with this idea of creating like mind blowing pedagogy. The only way you're really going to inspire students is you got to blow their minds a couple of times. And so that's really what we're aspiring towards. And the, I think there's, I mean already a lot of things that like just became available technologically. You know, as soon as it's Available. We're playing, playing with it and trying to figure out how can we integrate it and make it awesome. And I think there's a lot of stuff also coming down the pike that is about to be impossible. One thing might be simulations via world models. This idea of whatever the student's interest is. Let's say, let's take a trigonometry problem. Step one of the trigonometry problem is you have a couple of sides and you are trying to figure out what's the angle.
- [13:56]
B
Yep.
- [13:56]
A
Okay, cool. Next level of this is you, you know, call it a. You know, there's a quarterback at the, at the point of the angle you're trying to solve for.
- [14:04]
B
Right.
- [14:04]
A
And then you know this line. You know that line. So what angle do you have to throw at? Yeah, that's a little more interesting. Kids can connect to that, you know, depending on the student. Vastly more interesting would be you just created a real time simulation and it feels like, you know, you are the quarterback. It's in totally 3D and it's immersive. And you watch the quarterback, excuse me, the linebacker, start running and you can see sort of the trail behind them. You can see the distance in real time. And the clock sort of slows down and you have to pivot to throw. That could be. I mean, it just. Intrinsically, it sort of problematizes. I can see myself in this scenario. I can see how this might be useful. Now, obviously the quarterback's not doing that calculation, but it just creates the sense of intuition. Oh, that's what that angle is. That's what trigonometry used for.
- [14:57]
B
That's right. And so it's a conceptual understanding and it's a scenario that could raise interest and get people inspired. Exactly.
- [15:04]
A
And no matter what the interest, I'm sure there's a trigonometry version for someone who cares about dance or someone for sure, crazy swish D or whatever it is.
- [15:10]
B
For sure.
- [15:11]
A
And yeah, I think. And then there's a lot of ways that we can take these technologies and I think there's a couple of these besides that that are going to become possible.
- [15:22]
B
This resonates a lot with me. I was a tutor in my younger years and I remember teaching one student who was a dancer about. About degrees by talking about ballet, spinning.
- [15:30]
A
There it is.
- [15:31]
B
And it's exactly there. But you're still drawing on paper. It would be totally amazing to have it in a simulation. I think we are close to that. It's pretty exciting. Yeah. Yeah, it's amazing. You have Had a really interesting strategy for growth. You landed 50,000 students across 24 states partially by cold calling state districts yourself. I mean, you mentioned going up to the gates folks with a demo. I think that kind of, you know, entrepreneurial bravery is powerful. Tell us about how you do that and what you would recommend for others who want to sort of take on that type of mantle.
- [16:03]
A
Yeah, well, you know, I think the keyword is partially, you know, we, you know, cold calling is a great way to start kind of breaking your way into a territory and then from there you hope that the product does well and you start getting referrals and you start pulling on the threads that way. But, but it's true. I mean it, it. Cold calling landed us some really great partners.
- [16:21]
B
Yeah.
- [16:21]
A
And you know, it turns out, I think, I think the insight there is, you know, a lot of folks just aren't willing to schlep. Yeah, yeah. There's. There's a Paul Graham essay about this is at least 10 years old now. Yeah. And. And frankly, I mean, what a lot of people aren't willing to do is where the opportunity is. Yep. There's. If for whatever reason there's a, there's sort of this conventional wisdom that cold calling just doesn't work in education. I don't believe it. I mean, well, we've seen that very well. If you're persistent, you've got a shot. You just have to say the right thing. And frankly, there is still this sort of relationship development aspects of this. You still got to get to know the person. Still. One of the folks on our team does a lot of cold calling still. They can tell you for a lot of resistant principals across New York City, their kids names and when those birthdays are and so on, you know, when done right, you know, it's a strategy that can work.
- [17:13]
B
Yeah, there's, there's a, there's a famous sort of apropos of nothing, but it makes me think of famous story about Estee Lauder or the cosmetics giant who. One of her strategies for getting people to notice her, she'd go into department stores and literally drop a bottle of her perfume on the floor in department stores. Which of course filled the entire department store with the smell of. Exactly. And I, I think there's, you know, there's some corollary to that. In education it helps to have a product that's really beautiful, that's video based where you can. But yeah, bravery matters. And that common wisdom with everybody believes cold calling doesn't work. Nobody does it. And you're the only one cold calling. Right? Exactly. It can happen. Sawyer Altman is the CEO and co founder of goblins. It's a GSV Cup 50 company this year and doing really, really interesting work in math education with AI. Thanks so much for being here.
- [18:01]
A
Thank you, Alex. Appreciate it.